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{]} --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 -Wdisallowed-function-list=@var{sym},@var{sym},@dots{} @gol
237 -Wno-div-by-zero -Wempty-body -Wenum-compare -Wno-endif-labels @gol
238 -Werror -Werror=* @gol
239 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
240 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
241 -Wformat-security -Wformat-y2k @gol
242 -Wframe-larger-than=@var{len} -Wignored-qualifiers @gol
243 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
244 -Winit-self -Winline @gol
245 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
246 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
247 -Wlogical-op -Wlong-long @gol
248 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
249 -Wmissing-format-attribute -Wmissing-include-dirs @gol
250 -Wmissing-noreturn -Wno-mudflap @gol
251 -Wno-multichar -Wnonnull -Wno-overflow @gol
252 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
253 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
254 -Wpointer-arith -Wno-pointer-to-int-cast @gol
255 -Wredundant-decls @gol
256 -Wreturn-type -Wsequence-point -Wshadow @gol
257 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
258 -Wstrict-aliasing -Wstrict-aliasing=n @gol
259 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
260 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
261 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
262 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
263 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
264 -Wunused-value -Wunused-variable @gol
265 -Wvariadic-macros -Wvla @gol
266 -Wvolatile-register-var -Wwrite-strings}
268 @item C and Objective-C-only Warning Options
269 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
270 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
271 -Wold-style-declaration -Wold-style-definition @gol
272 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
273 -Wdeclaration-after-statement -Wpointer-sign}
275 @item Debugging Options
276 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
277 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
278 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
279 -fdump-noaddr -fdump-unnumbered @gol
280 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
281 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
282 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
283 -fdump-statistics @gol
285 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
289 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-nrv -fdump-tree-vect @gol
298 -fdump-tree-sink @gol
299 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
300 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
301 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
302 -ftree-vectorizer-verbose=@var{n} @gol
303 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
304 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
305 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
306 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
307 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
308 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
309 -ftest-coverage -ftime-report -fvar-tracking @gol
310 -g -g@var{level} -gcoff -gdwarf-2 @gol
311 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
312 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
313 -fdebug-prefix-map=@var{old}=@var{new} @gol
314 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
315 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
316 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
317 -print-multi-directory -print-multi-lib @gol
318 -print-prog-name=@var{program} -print-search-dirs -Q @gol
319 -print-sysroot -print-sysroot-headers-suffix @gol
322 @item Optimization Options
323 @xref{Optimize Options,,Options that Control Optimization}.
325 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
326 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
327 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
328 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
329 -fcheck-data-deps -fconserve-stack -fcprop-registers -fcrossjumping @gol
330 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
331 -fdata-sections -fdce -fdce @gol
332 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
333 -fearly-inlining -fexpensive-optimizations -ffast-math @gol
334 -ffinite-math-only -ffloat-store -fforward-propagate @gol
335 -ffunction-sections -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
336 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
337 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
338 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg -fipa-pta @gol
339 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
340 -fipa-type-escape -fira-algorithm=@var{algorithm} @gol
341 -fira-region=@var{region} -fira-coalesce -fno-ira-share-save-slots @gol
342 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
343 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
344 -floop-block -floop-interchange -floop-strip-mine @gol
345 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
346 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
347 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
348 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
349 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
350 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
351 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
352 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
353 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
354 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
355 -fprofile-generate=@var{path} @gol
356 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
357 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
358 -freorder-blocks-and-partition -freorder-functions @gol
359 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
360 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
361 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
362 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
363 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
364 -fselective-scheduling -fselective-scheduling2 @gol
365 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
366 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
367 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
368 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
369 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
370 -ftree-copyrename -ftree-dce @gol
371 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
372 -ftree-loop-distribution @gol
373 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
374 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc @gol
375 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
376 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
377 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
378 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
379 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
381 --param @var{name}=@var{value}
382 -O -O0 -O1 -O2 -O3 -Os}
384 @item Preprocessor Options
385 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
386 @gccoptlist{-A@var{question}=@var{answer} @gol
387 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
388 -C -dD -dI -dM -dN @gol
389 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
390 -idirafter @var{dir} @gol
391 -include @var{file} -imacros @var{file} @gol
392 -iprefix @var{file} -iwithprefix @var{dir} @gol
393 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
394 -imultilib @var{dir} -isysroot @var{dir} @gol
395 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
396 -P -fworking-directory -remap @gol
397 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
398 -Xpreprocessor @var{option}}
400 @item Assembler Option
401 @xref{Assembler Options,,Passing Options to the Assembler}.
402 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
405 @xref{Link Options,,Options for Linking}.
406 @gccoptlist{@var{object-file-name} -l@var{library} @gol
407 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
408 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
409 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
412 @item Directory Options
413 @xref{Directory Options,,Options for Directory Search}.
414 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
415 -specs=@var{file} -I- --sysroot=@var{dir}}
418 @c I wrote this xref this way to avoid overfull hbox. -- rms
419 @xref{Target Options}.
420 @gccoptlist{-V @var{version} -b @var{machine}}
422 @item Machine Dependent Options
423 @xref{Submodel Options,,Hardware Models and Configurations}.
424 @c This list is ordered alphanumerically by subsection name.
425 @c Try and put the significant identifier (CPU or system) first,
426 @c so users have a clue at guessing where the ones they want will be.
429 @gccoptlist{-EB -EL @gol
430 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
431 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
434 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
435 -mabi=@var{name} @gol
436 -mapcs-stack-check -mno-apcs-stack-check @gol
437 -mapcs-float -mno-apcs-float @gol
438 -mapcs-reentrant -mno-apcs-reentrant @gol
439 -msched-prolog -mno-sched-prolog @gol
440 -mlittle-endian -mbig-endian -mwords-little-endian @gol
441 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
442 -mthumb-interwork -mno-thumb-interwork @gol
443 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
444 -mstructure-size-boundary=@var{n} @gol
445 -mabort-on-noreturn @gol
446 -mlong-calls -mno-long-calls @gol
447 -msingle-pic-base -mno-single-pic-base @gol
448 -mpic-register=@var{reg} @gol
449 -mnop-fun-dllimport @gol
450 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
451 -mpoke-function-name @gol
453 -mtpcs-frame -mtpcs-leaf-frame @gol
454 -mcaller-super-interworking -mcallee-super-interworking @gol
456 -mword-relocations @gol
457 -mfix-cortex-m3-ldrd}
460 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
461 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
463 @emph{Blackfin Options}
464 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
465 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
466 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
467 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
468 -mno-id-shared-library -mshared-library-id=@var{n} @gol
469 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
470 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
471 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
475 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
476 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
477 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
478 -mstack-align -mdata-align -mconst-align @gol
479 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
480 -melf -maout -melinux -mlinux -sim -sim2 @gol
481 -mmul-bug-workaround -mno-mul-bug-workaround}
484 @gccoptlist{-mmac -mpush-args}
486 @emph{Darwin Options}
487 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
488 -arch_only -bind_at_load -bundle -bundle_loader @gol
489 -client_name -compatibility_version -current_version @gol
491 -dependency-file -dylib_file -dylinker_install_name @gol
492 -dynamic -dynamiclib -exported_symbols_list @gol
493 -filelist -flat_namespace -force_cpusubtype_ALL @gol
494 -force_flat_namespace -headerpad_max_install_names @gol
496 -image_base -init -install_name -keep_private_externs @gol
497 -multi_module -multiply_defined -multiply_defined_unused @gol
498 -noall_load -no_dead_strip_inits_and_terms @gol
499 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
500 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
501 -private_bundle -read_only_relocs -sectalign @gol
502 -sectobjectsymbols -whyload -seg1addr @gol
503 -sectcreate -sectobjectsymbols -sectorder @gol
504 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
505 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
506 -segprot -segs_read_only_addr -segs_read_write_addr @gol
507 -single_module -static -sub_library -sub_umbrella @gol
508 -twolevel_namespace -umbrella -undefined @gol
509 -unexported_symbols_list -weak_reference_mismatches @gol
510 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
511 -mkernel -mone-byte-bool}
513 @emph{DEC Alpha Options}
514 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
515 -mieee -mieee-with-inexact -mieee-conformant @gol
516 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
517 -mtrap-precision=@var{mode} -mbuild-constants @gol
518 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
519 -mbwx -mmax -mfix -mcix @gol
520 -mfloat-vax -mfloat-ieee @gol
521 -mexplicit-relocs -msmall-data -mlarge-data @gol
522 -msmall-text -mlarge-text @gol
523 -mmemory-latency=@var{time}}
525 @emph{DEC Alpha/VMS Options}
526 @gccoptlist{-mvms-return-codes}
529 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
530 -mhard-float -msoft-float @gol
531 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
532 -mdouble -mno-double @gol
533 -mmedia -mno-media -mmuladd -mno-muladd @gol
534 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
535 -mlinked-fp -mlong-calls -malign-labels @gol
536 -mlibrary-pic -macc-4 -macc-8 @gol
537 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
538 -moptimize-membar -mno-optimize-membar @gol
539 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
540 -mvliw-branch -mno-vliw-branch @gol
541 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
542 -mno-nested-cond-exec -mtomcat-stats @gol
546 @emph{GNU/Linux Options}
547 @gccoptlist{-muclibc}
549 @emph{H8/300 Options}
550 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
553 @gccoptlist{-march=@var{architecture-type} @gol
554 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
555 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
556 -mfixed-range=@var{register-range} @gol
557 -mjump-in-delay -mlinker-opt -mlong-calls @gol
558 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
559 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
560 -mno-jump-in-delay -mno-long-load-store @gol
561 -mno-portable-runtime -mno-soft-float @gol
562 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
563 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
564 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
565 -munix=@var{unix-std} -nolibdld -static -threads}
567 @emph{i386 and x86-64 Options}
568 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
569 -mfpmath=@var{unit} @gol
570 -masm=@var{dialect} -mno-fancy-math-387 @gol
571 -mno-fp-ret-in-387 -msoft-float @gol
572 -mno-wide-multiply -mrtd -malign-double @gol
573 -mpreferred-stack-boundary=@var{num}
574 -mincoming-stack-boundary=@var{num}
575 -mcld -mcx16 -msahf -mrecip @gol
576 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
578 -msse4a -m3dnow -mpopcnt -mabm -msse5 @gol
579 -mthreads -mno-align-stringops -minline-all-stringops @gol
580 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
581 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
582 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
583 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
584 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
585 -mcmodel=@var{code-model} @gol
586 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
587 -mfused-madd -mno-fused-madd -msse2avx}
590 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
591 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
592 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
593 -minline-float-divide-max-throughput @gol
594 -minline-int-divide-min-latency @gol
595 -minline-int-divide-max-throughput @gol
596 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
597 -mno-dwarf2-asm -mearly-stop-bits @gol
598 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
599 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
600 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
601 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
602 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
603 -mno-sched-prefer-non-data-spec-insns @gol
604 -mno-sched-prefer-non-control-spec-insns @gol
605 -mno-sched-count-spec-in-critical-path}
607 @emph{M32R/D Options}
608 @gccoptlist{-m32r2 -m32rx -m32r @gol
610 -malign-loops -mno-align-loops @gol
611 -missue-rate=@var{number} @gol
612 -mbranch-cost=@var{number} @gol
613 -mmodel=@var{code-size-model-type} @gol
614 -msdata=@var{sdata-type} @gol
615 -mno-flush-func -mflush-func=@var{name} @gol
616 -mno-flush-trap -mflush-trap=@var{number} @gol
620 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
622 @emph{M680x0 Options}
623 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
624 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
625 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
626 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
627 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
628 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
629 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
630 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
633 @emph{M68hc1x Options}
634 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
635 -mauto-incdec -minmax -mlong-calls -mshort @gol
636 -msoft-reg-count=@var{count}}
639 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
640 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
641 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
642 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
643 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
646 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
647 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
648 -mips64 -mips64r2 @gol
649 -mips16 -mno-mips16 -mflip-mips16 @gol
650 -minterlink-mips16 -mno-interlink-mips16 @gol
651 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
652 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
653 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
654 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
655 -mfpu=@var{fpu-type} @gol
656 -msmartmips -mno-smartmips @gol
657 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
658 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
659 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
660 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
661 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
662 -membedded-data -mno-embedded-data @gol
663 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
664 -mcode-readable=@var{setting} @gol
665 -msplit-addresses -mno-split-addresses @gol
666 -mexplicit-relocs -mno-explicit-relocs @gol
667 -mcheck-zero-division -mno-check-zero-division @gol
668 -mdivide-traps -mdivide-breaks @gol
669 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
670 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
671 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
672 -mfix-r10000 -mno-fix-r10000 -mfix-vr4120 -mno-fix-vr4120 @gol
673 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
674 -mflush-func=@var{func} -mno-flush-func @gol
675 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
676 -mfp-exceptions -mno-fp-exceptions @gol
677 -mvr4130-align -mno-vr4130-align}
680 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
681 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
682 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
683 -mno-base-addresses -msingle-exit -mno-single-exit}
685 @emph{MN10300 Options}
686 @gccoptlist{-mmult-bug -mno-mult-bug @gol
687 -mam33 -mno-am33 @gol
688 -mam33-2 -mno-am33-2 @gol
689 -mreturn-pointer-on-d0 @gol
692 @emph{PDP-11 Options}
693 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
694 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
695 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
696 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
697 -mbranch-expensive -mbranch-cheap @gol
698 -msplit -mno-split -munix-asm -mdec-asm}
700 @emph{picoChip Options}
701 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N}
702 -msymbol-as-address -mno-inefficient-warnings}
704 @emph{PowerPC Options}
705 See RS/6000 and PowerPC Options.
707 @emph{RS/6000 and PowerPC Options}
708 @gccoptlist{-mcpu=@var{cpu-type} @gol
709 -mtune=@var{cpu-type} @gol
710 -mpower -mno-power -mpower2 -mno-power2 @gol
711 -mpowerpc -mpowerpc64 -mno-powerpc @gol
712 -maltivec -mno-altivec @gol
713 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
714 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
715 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
716 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
717 -mnew-mnemonics -mold-mnemonics @gol
718 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
719 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
720 -malign-power -malign-natural @gol
721 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
722 -msingle-float -mdouble-float -msimple-fpu @gol
723 -mstring -mno-string -mupdate -mno-update @gol
724 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
725 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
726 -mstrict-align -mno-strict-align -mrelocatable @gol
727 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
728 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
729 -mdynamic-no-pic -maltivec -mswdiv @gol
730 -mprioritize-restricted-insns=@var{priority} @gol
731 -msched-costly-dep=@var{dependence_type} @gol
732 -minsert-sched-nops=@var{scheme} @gol
733 -mcall-sysv -mcall-netbsd @gol
734 -maix-struct-return -msvr4-struct-return @gol
735 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
736 -misel -mno-isel @gol
737 -misel=yes -misel=no @gol
739 -mspe=yes -mspe=no @gol
741 -mgen-cell-microcode -mwarn-cell-microcode @gol
742 -mvrsave -mno-vrsave @gol
743 -mmulhw -mno-mulhw @gol
744 -mdlmzb -mno-dlmzb @gol
745 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
746 -mprototype -mno-prototype @gol
747 -msim -mmvme -mads -myellowknife -memb -msdata @gol
748 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
750 @emph{S/390 and zSeries Options}
751 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
752 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
753 -mlong-double-64 -mlong-double-128 @gol
754 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
755 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
756 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
757 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
758 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
761 @gccoptlist{-meb -mel @gol
765 -mscore5 -mscore5u -mscore7 -mscore7d}
768 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
769 -m4-nofpu -m4-single-only -m4-single -m4 @gol
770 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
771 -m5-64media -m5-64media-nofpu @gol
772 -m5-32media -m5-32media-nofpu @gol
773 -m5-compact -m5-compact-nofpu @gol
774 -mb -ml -mdalign -mrelax @gol
775 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
776 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
777 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
778 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
779 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
783 @gccoptlist{-mcpu=@var{cpu-type} @gol
784 -mtune=@var{cpu-type} @gol
785 -mcmodel=@var{code-model} @gol
786 -m32 -m64 -mapp-regs -mno-app-regs @gol
787 -mfaster-structs -mno-faster-structs @gol
788 -mfpu -mno-fpu -mhard-float -msoft-float @gol
789 -mhard-quad-float -msoft-quad-float @gol
790 -mimpure-text -mno-impure-text -mlittle-endian @gol
791 -mstack-bias -mno-stack-bias @gol
792 -munaligned-doubles -mno-unaligned-doubles @gol
793 -mv8plus -mno-v8plus -mvis -mno-vis
794 -threads -pthreads -pthread}
797 @gccoptlist{-mwarn-reloc -merror-reloc @gol
798 -msafe-dma -munsafe-dma @gol
800 -msmall-mem -mlarge-mem -mstdmain @gol
801 -mfixed-range=@var{register-range}}
803 @emph{System V Options}
804 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
807 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
808 -mprolog-function -mno-prolog-function -mspace @gol
809 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
810 -mapp-regs -mno-app-regs @gol
811 -mdisable-callt -mno-disable-callt @gol
817 @gccoptlist{-mg -mgnu -munix}
819 @emph{VxWorks Options}
820 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
821 -Xbind-lazy -Xbind-now}
823 @emph{x86-64 Options}
824 See i386 and x86-64 Options.
826 @emph{Xstormy16 Options}
829 @emph{Xtensa Options}
830 @gccoptlist{-mconst16 -mno-const16 @gol
831 -mfused-madd -mno-fused-madd @gol
832 -mserialize-volatile -mno-serialize-volatile @gol
833 -mtext-section-literals -mno-text-section-literals @gol
834 -mtarget-align -mno-target-align @gol
835 -mlongcalls -mno-longcalls}
837 @emph{zSeries Options}
838 See S/390 and zSeries Options.
840 @item Code Generation Options
841 @xref{Code Gen Options,,Options for Code Generation Conventions}.
842 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
843 -ffixed-@var{reg} -fexceptions @gol
844 -fnon-call-exceptions -funwind-tables @gol
845 -fasynchronous-unwind-tables @gol
846 -finhibit-size-directive -finstrument-functions @gol
847 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
848 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
849 -fno-common -fno-ident @gol
850 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
851 -fno-jump-tables @gol
852 -frecord-gcc-switches @gol
853 -freg-struct-return -fshort-enums @gol
854 -fshort-double -fshort-wchar @gol
855 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
856 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
857 -fno-stack-limit -fargument-alias -fargument-noalias @gol
858 -fargument-noalias-global -fargument-noalias-anything @gol
859 -fleading-underscore -ftls-model=@var{model} @gol
860 -ftrapv -fwrapv -fbounds-check @gol
865 * Overall Options:: Controlling the kind of output:
866 an executable, object files, assembler files,
867 or preprocessed source.
868 * C Dialect Options:: Controlling the variant of C language compiled.
869 * C++ Dialect Options:: Variations on C++.
870 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
872 * Language Independent Options:: Controlling how diagnostics should be
874 * Warning Options:: How picky should the compiler be?
875 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
876 * Optimize Options:: How much optimization?
877 * Preprocessor Options:: Controlling header files and macro definitions.
878 Also, getting dependency information for Make.
879 * Assembler Options:: Passing options to the assembler.
880 * Link Options:: Specifying libraries and so on.
881 * Directory Options:: Where to find header files and libraries.
882 Where to find the compiler executable files.
883 * Spec Files:: How to pass switches to sub-processes.
884 * Target Options:: Running a cross-compiler, or an old version of GCC.
887 @node Overall Options
888 @section Options Controlling the Kind of Output
890 Compilation can involve up to four stages: preprocessing, compilation
891 proper, assembly and linking, always in that order. GCC is capable of
892 preprocessing and compiling several files either into several
893 assembler input files, or into one assembler input file; then each
894 assembler input file produces an object file, and linking combines all
895 the object files (those newly compiled, and those specified as input)
896 into an executable file.
898 @cindex file name suffix
899 For any given input file, the file name suffix determines what kind of
904 C source code which must be preprocessed.
907 C source code which should not be preprocessed.
910 C++ source code which should not be preprocessed.
913 Objective-C source code. Note that you must link with the @file{libobjc}
914 library to make an Objective-C program work.
917 Objective-C source code which should not be preprocessed.
921 Objective-C++ source code. Note that you must link with the @file{libobjc}
922 library to make an Objective-C++ program work. Note that @samp{.M} refers
923 to a literal capital M@.
926 Objective-C++ source code which should not be preprocessed.
929 C, C++, Objective-C or Objective-C++ header file to be turned into a
934 @itemx @var{file}.cxx
935 @itemx @var{file}.cpp
936 @itemx @var{file}.CPP
937 @itemx @var{file}.c++
939 C++ source code which must be preprocessed. Note that in @samp{.cxx},
940 the last two letters must both be literally @samp{x}. Likewise,
941 @samp{.C} refers to a literal capital C@.
945 Objective-C++ source code which must be preprocessed.
948 Objective-C++ source code which should not be preprocessed.
953 @itemx @var{file}.hxx
954 @itemx @var{file}.hpp
955 @itemx @var{file}.HPP
956 @itemx @var{file}.h++
957 @itemx @var{file}.tcc
958 C++ header file to be turned into a precompiled header.
961 @itemx @var{file}.for
962 @itemx @var{file}.ftn
963 Fixed form Fortran source code which should not be preprocessed.
966 @itemx @var{file}.FOR
967 @itemx @var{file}.fpp
968 @itemx @var{file}.FPP
969 @itemx @var{file}.FTN
970 Fixed form Fortran source code which must be preprocessed (with the traditional
974 @itemx @var{file}.f95
975 @itemx @var{file}.f03
976 @itemx @var{file}.f08
977 Free form Fortran source code which should not be preprocessed.
980 @itemx @var{file}.F95
981 @itemx @var{file}.F03
982 @itemx @var{file}.F08
983 Free form Fortran source code which must be preprocessed (with the
984 traditional preprocessor).
986 @c FIXME: Descriptions of Java file types.
993 Ada source code file which contains a library unit declaration (a
994 declaration of a package, subprogram, or generic, or a generic
995 instantiation), or a library unit renaming declaration (a package,
996 generic, or subprogram renaming declaration). Such files are also
1000 Ada source code file containing a library unit body (a subprogram or
1001 package body). Such files are also called @dfn{bodies}.
1003 @c GCC also knows about some suffixes for languages not yet included:
1014 @itemx @var{file}.sx
1015 Assembler code which must be preprocessed.
1018 An object file to be fed straight into linking.
1019 Any file name with no recognized suffix is treated this way.
1023 You can specify the input language explicitly with the @option{-x} option:
1026 @item -x @var{language}
1027 Specify explicitly the @var{language} for the following input files
1028 (rather than letting the compiler choose a default based on the file
1029 name suffix). This option applies to all following input files until
1030 the next @option{-x} option. Possible values for @var{language} are:
1032 c c-header c-cpp-output
1033 c++ c++-header c++-cpp-output
1034 objective-c objective-c-header objective-c-cpp-output
1035 objective-c++ objective-c++-header objective-c++-cpp-output
1036 assembler assembler-with-cpp
1038 f77 f77-cpp-input f95 f95-cpp-input
1043 Turn off any specification of a language, so that subsequent files are
1044 handled according to their file name suffixes (as they are if @option{-x}
1045 has not been used at all).
1047 @item -pass-exit-codes
1048 @opindex pass-exit-codes
1049 Normally the @command{gcc} program will exit with the code of 1 if any
1050 phase of the compiler returns a non-success return code. If you specify
1051 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1052 numerically highest error produced by any phase that returned an error
1053 indication. The C, C++, and Fortran frontends return 4, if an internal
1054 compiler error is encountered.
1057 If you only want some of the stages of compilation, you can use
1058 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1059 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1060 @command{gcc} is to stop. Note that some combinations (for example,
1061 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1066 Compile or assemble the source files, but do not link. The linking
1067 stage simply is not done. The ultimate output is in the form of an
1068 object file for each source file.
1070 By default, the object file name for a source file is made by replacing
1071 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1073 Unrecognized input files, not requiring compilation or assembly, are
1078 Stop after the stage of compilation proper; do not assemble. The output
1079 is in the form of an assembler code file for each non-assembler input
1082 By default, the assembler file name for a source file is made by
1083 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1085 Input files that don't require compilation are ignored.
1089 Stop after the preprocessing stage; do not run the compiler proper. The
1090 output is in the form of preprocessed source code, which is sent to the
1093 Input files which don't require preprocessing are ignored.
1095 @cindex output file option
1098 Place output in file @var{file}. This applies regardless to whatever
1099 sort of output is being produced, whether it be an executable file,
1100 an object file, an assembler file or preprocessed C code.
1102 If @option{-o} is not specified, the default is to put an executable
1103 file in @file{a.out}, the object file for
1104 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1105 assembler file in @file{@var{source}.s}, a precompiled header file in
1106 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1111 Print (on standard error output) the commands executed to run the stages
1112 of compilation. Also print the version number of the compiler driver
1113 program and of the preprocessor and the compiler proper.
1117 Like @option{-v} except the commands are not executed and all command
1118 arguments are quoted. This is useful for shell scripts to capture the
1119 driver-generated command lines.
1123 Use pipes rather than temporary files for communication between the
1124 various stages of compilation. This fails to work on some systems where
1125 the assembler is unable to read from a pipe; but the GNU assembler has
1130 If you are compiling multiple source files, this option tells the driver
1131 to pass all the source files to the compiler at once (for those
1132 languages for which the compiler can handle this). This will allow
1133 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1134 language for which this is supported is C@. If you pass source files for
1135 multiple languages to the driver, using this option, the driver will invoke
1136 the compiler(s) that support IMA once each, passing each compiler all the
1137 source files appropriate for it. For those languages that do not support
1138 IMA this option will be ignored, and the compiler will be invoked once for
1139 each source file in that language. If you use this option in conjunction
1140 with @option{-save-temps}, the compiler will generate multiple
1142 (one for each source file), but only one (combined) @file{.o} or
1147 Print (on the standard output) a description of the command line options
1148 understood by @command{gcc}. If the @option{-v} option is also specified
1149 then @option{--help} will also be passed on to the various processes
1150 invoked by @command{gcc}, so that they can display the command line options
1151 they accept. If the @option{-Wextra} option has also been specified
1152 (prior to the @option{--help} option), then command line options which
1153 have no documentation associated with them will also be displayed.
1156 @opindex target-help
1157 Print (on the standard output) a description of target-specific command
1158 line options for each tool. For some targets extra target-specific
1159 information may also be printed.
1161 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1162 Print (on the standard output) a description of the command line
1163 options understood by the compiler that fit into a specific class.
1164 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1165 @samp{params}, or @var{language}:
1168 @item @samp{optimizers}
1169 This will display all of the optimization options supported by the
1172 @item @samp{warnings}
1173 This will display all of the options controlling warning messages
1174 produced by the compiler.
1177 This will display target-specific options. Unlike the
1178 @option{--target-help} option however, target-specific options of the
1179 linker and assembler will not be displayed. This is because those
1180 tools do not currently support the extended @option{--help=} syntax.
1183 This will display the values recognized by the @option{--param}
1186 @item @var{language}
1187 This will display the options supported for @var{language}, where
1188 @var{language} is the name of one of the languages supported in this
1192 This will display the options that are common to all languages.
1195 It is possible to further refine the output of the @option{--help=}
1196 option by adding a comma separated list of qualifiers after the
1197 class. These can be any from the following list:
1200 @item @samp{undocumented}
1201 Display only those options which are undocumented.
1204 Display options which take an argument that appears after an equal
1205 sign in the same continuous piece of text, such as:
1206 @samp{--help=target}.
1208 @item @samp{separate}
1209 Display options which take an argument that appears as a separate word
1210 following the original option, such as: @samp{-o output-file}.
1213 Thus for example to display all the undocumented target-specific
1214 switches supported by the compiler the following can be used:
1217 --help=target,undocumented
1220 The sense of a qualifier can be inverted by prefixing it with the
1221 @var{^} character, so for example to display all binary warning
1222 options (i.e., ones that are either on or off and that do not take an
1223 argument), which have a description the following can be used:
1226 --help=warnings,^joined,^undocumented
1229 A class can also be used as a qualifier, although this usually
1230 restricts the output by so much that there is nothing to display. One
1231 case where it does work however is when one of the classes is
1232 @var{target}. So for example to display all the target-specific
1233 optimization options the following can be used:
1236 --help=target,optimizers
1239 The @option{--help=} option can be repeated on the command line. Each
1240 successive use will display its requested class of options, skipping
1241 those that have already been displayed.
1243 If the @option{-Q} option appears on the command line before the
1244 @option{--help=} option, then the descriptive text displayed by
1245 @option{--help=} is changed. Instead of describing the displayed
1246 options, an indication is given as to whether the option is enabled,
1247 disabled or set to a specific value (assuming that the compiler
1248 knows this at the point where the @option{--help=} option is used).
1250 Here is a truncated example from the ARM port of @command{gcc}:
1253 % gcc -Q -mabi=2 --help=target -c
1254 The following options are target specific:
1256 -mabort-on-noreturn [disabled]
1260 The output is sensitive to the effects of previous command line
1261 options, so for example it is possible to find out which optimizations
1262 are enabled at @option{-O2} by using:
1265 -O2 --help=optimizers
1268 Alternatively you can discover which binary optimizations are enabled
1269 by @option{-O3} by using:
1272 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1273 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1274 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1279 Display the version number and copyrights of the invoked GCC@.
1283 Invoke all subcommands under a wrapper program. It takes a single
1284 comma separated list as an argument, which will be used to invoke
1288 gcc -c t.c -wrapper gdb,--args
1291 This will invoke all subprograms of gcc under "gdb --args",
1292 thus cc1 invocation will be "gdb --args cc1 ...".
1294 @include @value{srcdir}/../libiberty/at-file.texi
1298 @section Compiling C++ Programs
1300 @cindex suffixes for C++ source
1301 @cindex C++ source file suffixes
1302 C++ source files conventionally use one of the suffixes @samp{.C},
1303 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1304 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1305 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1306 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1307 files with these names and compiles them as C++ programs even if you
1308 call the compiler the same way as for compiling C programs (usually
1309 with the name @command{gcc}).
1313 However, the use of @command{gcc} does not add the C++ library.
1314 @command{g++} is a program that calls GCC and treats @samp{.c},
1315 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1316 files unless @option{-x} is used, and automatically specifies linking
1317 against the C++ library. This program is also useful when
1318 precompiling a C header file with a @samp{.h} extension for use in C++
1319 compilations. On many systems, @command{g++} is also installed with
1320 the name @command{c++}.
1322 @cindex invoking @command{g++}
1323 When you compile C++ programs, you may specify many of the same
1324 command-line options that you use for compiling programs in any
1325 language; or command-line options meaningful for C and related
1326 languages; or options that are meaningful only for C++ programs.
1327 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1328 explanations of options for languages related to C@.
1329 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1330 explanations of options that are meaningful only for C++ programs.
1332 @node C Dialect Options
1333 @section Options Controlling C Dialect
1334 @cindex dialect options
1335 @cindex language dialect options
1336 @cindex options, dialect
1338 The following options control the dialect of C (or languages derived
1339 from C, such as C++, Objective-C and Objective-C++) that the compiler
1343 @cindex ANSI support
1347 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1348 equivalent to @samp{-std=c++98}.
1350 This turns off certain features of GCC that are incompatible with ISO
1351 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1352 such as the @code{asm} and @code{typeof} keywords, and
1353 predefined macros such as @code{unix} and @code{vax} that identify the
1354 type of system you are using. It also enables the undesirable and
1355 rarely used ISO trigraph feature. For the C compiler,
1356 it disables recognition of C++ style @samp{//} comments as well as
1357 the @code{inline} keyword.
1359 The alternate keywords @code{__asm__}, @code{__extension__},
1360 @code{__inline__} and @code{__typeof__} continue to work despite
1361 @option{-ansi}. You would not want to use them in an ISO C program, of
1362 course, but it is useful to put them in header files that might be included
1363 in compilations done with @option{-ansi}. Alternate predefined macros
1364 such as @code{__unix__} and @code{__vax__} are also available, with or
1365 without @option{-ansi}.
1367 The @option{-ansi} option does not cause non-ISO programs to be
1368 rejected gratuitously. For that, @option{-pedantic} is required in
1369 addition to @option{-ansi}. @xref{Warning Options}.
1371 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1372 option is used. Some header files may notice this macro and refrain
1373 from declaring certain functions or defining certain macros that the
1374 ISO standard doesn't call for; this is to avoid interfering with any
1375 programs that might use these names for other things.
1377 Functions that would normally be built in but do not have semantics
1378 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1379 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1380 built-in functions provided by GCC}, for details of the functions
1385 Determine the language standard. @xref{Standards,,Language Standards
1386 Supported by GCC}, for details of these standard versions. This option
1387 is currently only supported when compiling C or C++.
1389 The compiler can accept several base standards, such as @samp{c89} or
1390 @samp{c++98}, and GNU dialects of those standards, such as
1391 @samp{gnu89} or @samp{gnu++98}. By specifying a base standard, the
1392 compiler will accept all programs following that standard and those
1393 using GNU extensions that do not contradict it. For example,
1394 @samp{-std=c89} turns off certain features of GCC that are
1395 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1396 keywords, but not other GNU extensions that do not have a meaning in
1397 ISO C90, such as omitting the middle term of a @code{?:}
1398 expression. On the other hand, by specifying a GNU dialect of a
1399 standard, all features the compiler support are enabled, even when
1400 those features change the meaning of the base standard and some
1401 strict-conforming programs may be rejected. The particular standard
1402 is used by @option{-pedantic} to identify which features are GNU
1403 extensions given that version of the standard. For example
1404 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1405 comments, while @samp{-std=gnu99 -pedantic} would not.
1407 A value for this option must be provided; possible values are
1412 Support all ISO C90 programs (certain GNU extensions that conflict
1413 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1415 @item iso9899:199409
1416 ISO C90 as modified in amendment 1.
1422 ISO C99. Note that this standard is not yet fully supported; see
1423 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1424 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1427 GNU dialect of ISO C90 (including some C99 features). This
1428 is the default for C code.
1432 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1433 this will become the default. The name @samp{gnu9x} is deprecated.
1436 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1440 GNU dialect of @option{-std=c++98}. This is the default for
1444 The working draft of the upcoming ISO C++0x standard. This option
1445 enables experimental features that are likely to be included in
1446 C++0x. The working draft is constantly changing, and any feature that is
1447 enabled by this flag may be removed from future versions of GCC if it is
1448 not part of the C++0x standard.
1451 GNU dialect of @option{-std=c++0x}. This option enables
1452 experimental features that may be removed in future versions of GCC.
1455 @item -fgnu89-inline
1456 @opindex fgnu89-inline
1457 The option @option{-fgnu89-inline} tells GCC to use the traditional
1458 GNU semantics for @code{inline} functions when in C99 mode.
1459 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1460 is accepted and ignored by GCC versions 4.1.3 up to but not including
1461 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1462 C99 mode. Using this option is roughly equivalent to adding the
1463 @code{gnu_inline} function attribute to all inline functions
1464 (@pxref{Function Attributes}).
1466 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1467 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1468 specifies the default behavior). This option was first supported in
1469 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1471 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1472 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1473 in effect for @code{inline} functions. @xref{Common Predefined
1474 Macros,,,cpp,The C Preprocessor}.
1476 @item -aux-info @var{filename}
1478 Output to the given filename prototyped declarations for all functions
1479 declared and/or defined in a translation unit, including those in header
1480 files. This option is silently ignored in any language other than C@.
1482 Besides declarations, the file indicates, in comments, the origin of
1483 each declaration (source file and line), whether the declaration was
1484 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1485 @samp{O} for old, respectively, in the first character after the line
1486 number and the colon), and whether it came from a declaration or a
1487 definition (@samp{C} or @samp{F}, respectively, in the following
1488 character). In the case of function definitions, a K&R-style list of
1489 arguments followed by their declarations is also provided, inside
1490 comments, after the declaration.
1494 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1495 keyword, so that code can use these words as identifiers. You can use
1496 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1497 instead. @option{-ansi} implies @option{-fno-asm}.
1499 In C++, this switch only affects the @code{typeof} keyword, since
1500 @code{asm} and @code{inline} are standard keywords. You may want to
1501 use the @option{-fno-gnu-keywords} flag instead, which has the same
1502 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1503 switch only affects the @code{asm} and @code{typeof} keywords, since
1504 @code{inline} is a standard keyword in ISO C99.
1507 @itemx -fno-builtin-@var{function}
1508 @opindex fno-builtin
1509 @cindex built-in functions
1510 Don't recognize built-in functions that do not begin with
1511 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1512 functions provided by GCC}, for details of the functions affected,
1513 including those which are not built-in functions when @option{-ansi} or
1514 @option{-std} options for strict ISO C conformance are used because they
1515 do not have an ISO standard meaning.
1517 GCC normally generates special code to handle certain built-in functions
1518 more efficiently; for instance, calls to @code{alloca} may become single
1519 instructions that adjust the stack directly, and calls to @code{memcpy}
1520 may become inline copy loops. The resulting code is often both smaller
1521 and faster, but since the function calls no longer appear as such, you
1522 cannot set a breakpoint on those calls, nor can you change the behavior
1523 of the functions by linking with a different library. In addition,
1524 when a function is recognized as a built-in function, GCC may use
1525 information about that function to warn about problems with calls to
1526 that function, or to generate more efficient code, even if the
1527 resulting code still contains calls to that function. For example,
1528 warnings are given with @option{-Wformat} for bad calls to
1529 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1530 known not to modify global memory.
1532 With the @option{-fno-builtin-@var{function}} option
1533 only the built-in function @var{function} is
1534 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1535 function is named that is not built-in in this version of GCC, this
1536 option is ignored. There is no corresponding
1537 @option{-fbuiltin-@var{function}} option; if you wish to enable
1538 built-in functions selectively when using @option{-fno-builtin} or
1539 @option{-ffreestanding}, you may define macros such as:
1542 #define abs(n) __builtin_abs ((n))
1543 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1548 @cindex hosted environment
1550 Assert that compilation takes place in a hosted environment. This implies
1551 @option{-fbuiltin}. A hosted environment is one in which the
1552 entire standard library is available, and in which @code{main} has a return
1553 type of @code{int}. Examples are nearly everything except a kernel.
1554 This is equivalent to @option{-fno-freestanding}.
1556 @item -ffreestanding
1557 @opindex ffreestanding
1558 @cindex hosted environment
1560 Assert that compilation takes place in a freestanding environment. This
1561 implies @option{-fno-builtin}. A freestanding environment
1562 is one in which the standard library may not exist, and program startup may
1563 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1564 This is equivalent to @option{-fno-hosted}.
1566 @xref{Standards,,Language Standards Supported by GCC}, for details of
1567 freestanding and hosted environments.
1571 @cindex openmp parallel
1572 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1573 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1574 compiler generates parallel code according to the OpenMP Application
1575 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}. This option
1576 implies @option{-pthread}, and thus is only supported on targets that
1577 have support for @option{-pthread}.
1579 @item -fms-extensions
1580 @opindex fms-extensions
1581 Accept some non-standard constructs used in Microsoft header files.
1583 Some cases of unnamed fields in structures and unions are only
1584 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1585 fields within structs/unions}, for details.
1589 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1590 options for strict ISO C conformance) implies @option{-trigraphs}.
1592 @item -no-integrated-cpp
1593 @opindex no-integrated-cpp
1594 Performs a compilation in two passes: preprocessing and compiling. This
1595 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1596 @option{-B} option. The user supplied compilation step can then add in
1597 an additional preprocessing step after normal preprocessing but before
1598 compiling. The default is to use the integrated cpp (internal cpp)
1600 The semantics of this option will change if "cc1", "cc1plus", and
1601 "cc1obj" are merged.
1603 @cindex traditional C language
1604 @cindex C language, traditional
1606 @itemx -traditional-cpp
1607 @opindex traditional-cpp
1608 @opindex traditional
1609 Formerly, these options caused GCC to attempt to emulate a pre-standard
1610 C compiler. They are now only supported with the @option{-E} switch.
1611 The preprocessor continues to support a pre-standard mode. See the GNU
1612 CPP manual for details.
1614 @item -fcond-mismatch
1615 @opindex fcond-mismatch
1616 Allow conditional expressions with mismatched types in the second and
1617 third arguments. The value of such an expression is void. This option
1618 is not supported for C++.
1620 @item -flax-vector-conversions
1621 @opindex flax-vector-conversions
1622 Allow implicit conversions between vectors with differing numbers of
1623 elements and/or incompatible element types. This option should not be
1626 @item -funsigned-char
1627 @opindex funsigned-char
1628 Let the type @code{char} be unsigned, like @code{unsigned char}.
1630 Each kind of machine has a default for what @code{char} should
1631 be. It is either like @code{unsigned char} by default or like
1632 @code{signed char} by default.
1634 Ideally, a portable program should always use @code{signed char} or
1635 @code{unsigned char} when it depends on the signedness of an object.
1636 But many programs have been written to use plain @code{char} and
1637 expect it to be signed, or expect it to be unsigned, depending on the
1638 machines they were written for. This option, and its inverse, let you
1639 make such a program work with the opposite default.
1641 The type @code{char} is always a distinct type from each of
1642 @code{signed char} or @code{unsigned char}, even though its behavior
1643 is always just like one of those two.
1646 @opindex fsigned-char
1647 Let the type @code{char} be signed, like @code{signed char}.
1649 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1650 the negative form of @option{-funsigned-char}. Likewise, the option
1651 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1653 @item -fsigned-bitfields
1654 @itemx -funsigned-bitfields
1655 @itemx -fno-signed-bitfields
1656 @itemx -fno-unsigned-bitfields
1657 @opindex fsigned-bitfields
1658 @opindex funsigned-bitfields
1659 @opindex fno-signed-bitfields
1660 @opindex fno-unsigned-bitfields
1661 These options control whether a bit-field is signed or unsigned, when the
1662 declaration does not use either @code{signed} or @code{unsigned}. By
1663 default, such a bit-field is signed, because this is consistent: the
1664 basic integer types such as @code{int} are signed types.
1667 @node C++ Dialect Options
1668 @section Options Controlling C++ Dialect
1670 @cindex compiler options, C++
1671 @cindex C++ options, command line
1672 @cindex options, C++
1673 This section describes the command-line options that are only meaningful
1674 for C++ programs; but you can also use most of the GNU compiler options
1675 regardless of what language your program is in. For example, you
1676 might compile a file @code{firstClass.C} like this:
1679 g++ -g -frepo -O -c firstClass.C
1683 In this example, only @option{-frepo} is an option meant
1684 only for C++ programs; you can use the other options with any
1685 language supported by GCC@.
1687 Here is a list of options that are @emph{only} for compiling C++ programs:
1691 @item -fabi-version=@var{n}
1692 @opindex fabi-version
1693 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1694 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1695 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1696 the version that conforms most closely to the C++ ABI specification.
1697 Therefore, the ABI obtained using version 0 will change as ABI bugs
1700 The default is version 2.
1702 @item -fno-access-control
1703 @opindex fno-access-control
1704 Turn off all access checking. This switch is mainly useful for working
1705 around bugs in the access control code.
1709 Check that the pointer returned by @code{operator new} is non-null
1710 before attempting to modify the storage allocated. This check is
1711 normally unnecessary because the C++ standard specifies that
1712 @code{operator new} will only return @code{0} if it is declared
1713 @samp{throw()}, in which case the compiler will always check the
1714 return value even without this option. In all other cases, when
1715 @code{operator new} has a non-empty exception specification, memory
1716 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1717 @samp{new (nothrow)}.
1719 @item -fconserve-space
1720 @opindex fconserve-space
1721 Put uninitialized or runtime-initialized global variables into the
1722 common segment, as C does. This saves space in the executable at the
1723 cost of not diagnosing duplicate definitions. If you compile with this
1724 flag and your program mysteriously crashes after @code{main()} has
1725 completed, you may have an object that is being destroyed twice because
1726 two definitions were merged.
1728 This option is no longer useful on most targets, now that support has
1729 been added for putting variables into BSS without making them common.
1731 @item -ffriend-injection
1732 @opindex ffriend-injection
1733 Inject friend functions into the enclosing namespace, so that they are
1734 visible outside the scope of the class in which they are declared.
1735 Friend functions were documented to work this way in the old Annotated
1736 C++ Reference Manual, and versions of G++ before 4.1 always worked
1737 that way. However, in ISO C++ a friend function which is not declared
1738 in an enclosing scope can only be found using argument dependent
1739 lookup. This option causes friends to be injected as they were in
1742 This option is for compatibility, and may be removed in a future
1745 @item -fno-elide-constructors
1746 @opindex fno-elide-constructors
1747 The C++ standard allows an implementation to omit creating a temporary
1748 which is only used to initialize another object of the same type.
1749 Specifying this option disables that optimization, and forces G++ to
1750 call the copy constructor in all cases.
1752 @item -fno-enforce-eh-specs
1753 @opindex fno-enforce-eh-specs
1754 Don't generate code to check for violation of exception specifications
1755 at runtime. This option violates the C++ standard, but may be useful
1756 for reducing code size in production builds, much like defining
1757 @samp{NDEBUG}. This does not give user code permission to throw
1758 exceptions in violation of the exception specifications; the compiler
1759 will still optimize based on the specifications, so throwing an
1760 unexpected exception will result in undefined behavior.
1763 @itemx -fno-for-scope
1765 @opindex fno-for-scope
1766 If @option{-ffor-scope} is specified, the scope of variables declared in
1767 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1768 as specified by the C++ standard.
1769 If @option{-fno-for-scope} is specified, the scope of variables declared in
1770 a @i{for-init-statement} extends to the end of the enclosing scope,
1771 as was the case in old versions of G++, and other (traditional)
1772 implementations of C++.
1774 The default if neither flag is given to follow the standard,
1775 but to allow and give a warning for old-style code that would
1776 otherwise be invalid, or have different behavior.
1778 @item -fno-gnu-keywords
1779 @opindex fno-gnu-keywords
1780 Do not recognize @code{typeof} as a keyword, so that code can use this
1781 word as an identifier. You can use the keyword @code{__typeof__} instead.
1782 @option{-ansi} implies @option{-fno-gnu-keywords}.
1784 @item -fno-implicit-templates
1785 @opindex fno-implicit-templates
1786 Never emit code for non-inline templates which are instantiated
1787 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1788 @xref{Template Instantiation}, for more information.
1790 @item -fno-implicit-inline-templates
1791 @opindex fno-implicit-inline-templates
1792 Don't emit code for implicit instantiations of inline templates, either.
1793 The default is to handle inlines differently so that compiles with and
1794 without optimization will need the same set of explicit instantiations.
1796 @item -fno-implement-inlines
1797 @opindex fno-implement-inlines
1798 To save space, do not emit out-of-line copies of inline functions
1799 controlled by @samp{#pragma implementation}. This will cause linker
1800 errors if these functions are not inlined everywhere they are called.
1802 @item -fms-extensions
1803 @opindex fms-extensions
1804 Disable pedantic warnings about constructs used in MFC, such as implicit
1805 int and getting a pointer to member function via non-standard syntax.
1807 @item -fno-nonansi-builtins
1808 @opindex fno-nonansi-builtins
1809 Disable built-in declarations of functions that are not mandated by
1810 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1811 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1813 @item -fno-operator-names
1814 @opindex fno-operator-names
1815 Do not treat the operator name keywords @code{and}, @code{bitand},
1816 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1817 synonyms as keywords.
1819 @item -fno-optional-diags
1820 @opindex fno-optional-diags
1821 Disable diagnostics that the standard says a compiler does not need to
1822 issue. Currently, the only such diagnostic issued by G++ is the one for
1823 a name having multiple meanings within a class.
1826 @opindex fpermissive
1827 Downgrade some diagnostics about nonconformant code from errors to
1828 warnings. Thus, using @option{-fpermissive} will allow some
1829 nonconforming code to compile.
1833 Enable automatic template instantiation at link time. This option also
1834 implies @option{-fno-implicit-templates}. @xref{Template
1835 Instantiation}, for more information.
1839 Disable generation of information about every class with virtual
1840 functions for use by the C++ runtime type identification features
1841 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1842 of the language, you can save some space by using this flag. Note that
1843 exception handling uses the same information, but it will generate it as
1844 needed. The @samp{dynamic_cast} operator can still be used for casts that
1845 do not require runtime type information, i.e.@: casts to @code{void *} or to
1846 unambiguous base classes.
1850 Emit statistics about front-end processing at the end of the compilation.
1851 This information is generally only useful to the G++ development team.
1853 @item -ftemplate-depth-@var{n}
1854 @opindex ftemplate-depth
1855 Set the maximum instantiation depth for template classes to @var{n}.
1856 A limit on the template instantiation depth is needed to detect
1857 endless recursions during template class instantiation. ANSI/ISO C++
1858 conforming programs must not rely on a maximum depth greater than 17.
1860 @item -fno-threadsafe-statics
1861 @opindex fno-threadsafe-statics
1862 Do not emit the extra code to use the routines specified in the C++
1863 ABI for thread-safe initialization of local statics. You can use this
1864 option to reduce code size slightly in code that doesn't need to be
1867 @item -fuse-cxa-atexit
1868 @opindex fuse-cxa-atexit
1869 Register destructors for objects with static storage duration with the
1870 @code{__cxa_atexit} function rather than the @code{atexit} function.
1871 This option is required for fully standards-compliant handling of static
1872 destructors, but will only work if your C library supports
1873 @code{__cxa_atexit}.
1875 @item -fno-use-cxa-get-exception-ptr
1876 @opindex fno-use-cxa-get-exception-ptr
1877 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1878 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1879 if the runtime routine is not available.
1881 @item -fvisibility-inlines-hidden
1882 @opindex fvisibility-inlines-hidden
1883 This switch declares that the user does not attempt to compare
1884 pointers to inline methods where the addresses of the two functions
1885 were taken in different shared objects.
1887 The effect of this is that GCC may, effectively, mark inline methods with
1888 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1889 appear in the export table of a DSO and do not require a PLT indirection
1890 when used within the DSO@. Enabling this option can have a dramatic effect
1891 on load and link times of a DSO as it massively reduces the size of the
1892 dynamic export table when the library makes heavy use of templates.
1894 The behavior of this switch is not quite the same as marking the
1895 methods as hidden directly, because it does not affect static variables
1896 local to the function or cause the compiler to deduce that
1897 the function is defined in only one shared object.
1899 You may mark a method as having a visibility explicitly to negate the
1900 effect of the switch for that method. For example, if you do want to
1901 compare pointers to a particular inline method, you might mark it as
1902 having default visibility. Marking the enclosing class with explicit
1903 visibility will have no effect.
1905 Explicitly instantiated inline methods are unaffected by this option
1906 as their linkage might otherwise cross a shared library boundary.
1907 @xref{Template Instantiation}.
1909 @item -fvisibility-ms-compat
1910 @opindex fvisibility-ms-compat
1911 This flag attempts to use visibility settings to make GCC's C++
1912 linkage model compatible with that of Microsoft Visual Studio.
1914 The flag makes these changes to GCC's linkage model:
1918 It sets the default visibility to @code{hidden}, like
1919 @option{-fvisibility=hidden}.
1922 Types, but not their members, are not hidden by default.
1925 The One Definition Rule is relaxed for types without explicit
1926 visibility specifications which are defined in more than one different
1927 shared object: those declarations are permitted if they would have
1928 been permitted when this option was not used.
1931 In new code it is better to use @option{-fvisibility=hidden} and
1932 export those classes which are intended to be externally visible.
1933 Unfortunately it is possible for code to rely, perhaps accidentally,
1934 on the Visual Studio behavior.
1936 Among the consequences of these changes are that static data members
1937 of the same type with the same name but defined in different shared
1938 objects will be different, so changing one will not change the other;
1939 and that pointers to function members defined in different shared
1940 objects may not compare equal. When this flag is given, it is a
1941 violation of the ODR to define types with the same name differently.
1945 Do not use weak symbol support, even if it is provided by the linker.
1946 By default, G++ will use weak symbols if they are available. This
1947 option exists only for testing, and should not be used by end-users;
1948 it will result in inferior code and has no benefits. This option may
1949 be removed in a future release of G++.
1953 Do not search for header files in the standard directories specific to
1954 C++, but do still search the other standard directories. (This option
1955 is used when building the C++ library.)
1958 In addition, these optimization, warning, and code generation options
1959 have meanings only for C++ programs:
1962 @item -fno-default-inline
1963 @opindex fno-default-inline
1964 Do not assume @samp{inline} for functions defined inside a class scope.
1965 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1966 functions will have linkage like inline functions; they just won't be
1969 @item -Wabi @r{(C++ and Objective-C++ only)}
1972 Warn when G++ generates code that is probably not compatible with the
1973 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1974 all such cases, there are probably some cases that are not warned about,
1975 even though G++ is generating incompatible code. There may also be
1976 cases where warnings are emitted even though the code that is generated
1979 You should rewrite your code to avoid these warnings if you are
1980 concerned about the fact that code generated by G++ may not be binary
1981 compatible with code generated by other compilers.
1983 The known incompatibilities at this point include:
1988 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1989 pack data into the same byte as a base class. For example:
1992 struct A @{ virtual void f(); int f1 : 1; @};
1993 struct B : public A @{ int f2 : 1; @};
1997 In this case, G++ will place @code{B::f2} into the same byte
1998 as@code{A::f1}; other compilers will not. You can avoid this problem
1999 by explicitly padding @code{A} so that its size is a multiple of the
2000 byte size on your platform; that will cause G++ and other compilers to
2001 layout @code{B} identically.
2004 Incorrect handling of tail-padding for virtual bases. G++ does not use
2005 tail padding when laying out virtual bases. For example:
2008 struct A @{ virtual void f(); char c1; @};
2009 struct B @{ B(); char c2; @};
2010 struct C : public A, public virtual B @{@};
2014 In this case, G++ will not place @code{B} into the tail-padding for
2015 @code{A}; other compilers will. You can avoid this problem by
2016 explicitly padding @code{A} so that its size is a multiple of its
2017 alignment (ignoring virtual base classes); that will cause G++ and other
2018 compilers to layout @code{C} identically.
2021 Incorrect handling of bit-fields with declared widths greater than that
2022 of their underlying types, when the bit-fields appear in a union. For
2026 union U @{ int i : 4096; @};
2030 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2031 union too small by the number of bits in an @code{int}.
2034 Empty classes can be placed at incorrect offsets. For example:
2044 struct C : public B, public A @{@};
2048 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2049 it should be placed at offset zero. G++ mistakenly believes that the
2050 @code{A} data member of @code{B} is already at offset zero.
2053 Names of template functions whose types involve @code{typename} or
2054 template template parameters can be mangled incorrectly.
2057 template <typename Q>
2058 void f(typename Q::X) @{@}
2060 template <template <typename> class Q>
2061 void f(typename Q<int>::X) @{@}
2065 Instantiations of these templates may be mangled incorrectly.
2069 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2070 @opindex Wctor-dtor-privacy
2071 @opindex Wno-ctor-dtor-privacy
2072 Warn when a class seems unusable because all the constructors or
2073 destructors in that class are private, and it has neither friends nor
2074 public static member functions.
2076 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2077 @opindex Wnon-virtual-dtor
2078 @opindex Wno-non-virtual-dtor
2079 Warn when a class has virtual functions and accessible non-virtual
2080 destructor, in which case it would be possible but unsafe to delete
2081 an instance of a derived class through a pointer to the base class.
2082 This warning is also enabled if -Weffc++ is specified.
2084 @item -Wreorder @r{(C++ and Objective-C++ only)}
2086 @opindex Wno-reorder
2087 @cindex reordering, warning
2088 @cindex warning for reordering of member initializers
2089 Warn when the order of member initializers given in the code does not
2090 match the order in which they must be executed. For instance:
2096 A(): j (0), i (1) @{ @}
2100 The compiler will rearrange the member initializers for @samp{i}
2101 and @samp{j} to match the declaration order of the members, emitting
2102 a warning to that effect. This warning is enabled by @option{-Wall}.
2105 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2108 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2111 Warn about violations of the following style guidelines from Scott Meyers'
2112 @cite{Effective C++} book:
2116 Item 11: Define a copy constructor and an assignment operator for classes
2117 with dynamically allocated memory.
2120 Item 12: Prefer initialization to assignment in constructors.
2123 Item 14: Make destructors virtual in base classes.
2126 Item 15: Have @code{operator=} return a reference to @code{*this}.
2129 Item 23: Don't try to return a reference when you must return an object.
2133 Also warn about violations of the following style guidelines from
2134 Scott Meyers' @cite{More Effective C++} book:
2138 Item 6: Distinguish between prefix and postfix forms of increment and
2139 decrement operators.
2142 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2146 When selecting this option, be aware that the standard library
2147 headers do not obey all of these guidelines; use @samp{grep -v}
2148 to filter out those warnings.
2150 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2151 @opindex Wstrict-null-sentinel
2152 @opindex Wno-strict-null-sentinel
2153 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2154 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2155 to @code{__null}. Although it is a null pointer constant not a null pointer,
2156 it is guaranteed to be of the same size as a pointer. But this use is
2157 not portable across different compilers.
2159 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2160 @opindex Wno-non-template-friend
2161 @opindex Wnon-template-friend
2162 Disable warnings when non-templatized friend functions are declared
2163 within a template. Since the advent of explicit template specification
2164 support in G++, if the name of the friend is an unqualified-id (i.e.,
2165 @samp{friend foo(int)}), the C++ language specification demands that the
2166 friend declare or define an ordinary, nontemplate function. (Section
2167 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2168 could be interpreted as a particular specialization of a templatized
2169 function. Because this non-conforming behavior is no longer the default
2170 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2171 check existing code for potential trouble spots and is on by default.
2172 This new compiler behavior can be turned off with
2173 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2174 but disables the helpful warning.
2176 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2177 @opindex Wold-style-cast
2178 @opindex Wno-old-style-cast
2179 Warn if an old-style (C-style) cast to a non-void type is used within
2180 a C++ program. The new-style casts (@samp{dynamic_cast},
2181 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2182 less vulnerable to unintended effects and much easier to search for.
2184 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2185 @opindex Woverloaded-virtual
2186 @opindex Wno-overloaded-virtual
2187 @cindex overloaded virtual fn, warning
2188 @cindex warning for overloaded virtual fn
2189 Warn when a function declaration hides virtual functions from a
2190 base class. For example, in:
2197 struct B: public A @{
2202 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2210 will fail to compile.
2212 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2213 @opindex Wno-pmf-conversions
2214 @opindex Wpmf-conversions
2215 Disable the diagnostic for converting a bound pointer to member function
2218 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2219 @opindex Wsign-promo
2220 @opindex Wno-sign-promo
2221 Warn when overload resolution chooses a promotion from unsigned or
2222 enumerated type to a signed type, over a conversion to an unsigned type of
2223 the same size. Previous versions of G++ would try to preserve
2224 unsignedness, but the standard mandates the current behavior.
2229 A& operator = (int);
2239 In this example, G++ will synthesize a default @samp{A& operator =
2240 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2243 @node Objective-C and Objective-C++ Dialect Options
2244 @section Options Controlling Objective-C and Objective-C++ Dialects
2246 @cindex compiler options, Objective-C and Objective-C++
2247 @cindex Objective-C and Objective-C++ options, command line
2248 @cindex options, Objective-C and Objective-C++
2249 (NOTE: This manual does not describe the Objective-C and Objective-C++
2250 languages themselves. See @xref{Standards,,Language Standards
2251 Supported by GCC}, for references.)
2253 This section describes the command-line options that are only meaningful
2254 for Objective-C and Objective-C++ programs, but you can also use most of
2255 the language-independent GNU compiler options.
2256 For example, you might compile a file @code{some_class.m} like this:
2259 gcc -g -fgnu-runtime -O -c some_class.m
2263 In this example, @option{-fgnu-runtime} is an option meant only for
2264 Objective-C and Objective-C++ programs; you can use the other options with
2265 any language supported by GCC@.
2267 Note that since Objective-C is an extension of the C language, Objective-C
2268 compilations may also use options specific to the C front-end (e.g.,
2269 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2270 C++-specific options (e.g., @option{-Wabi}).
2272 Here is a list of options that are @emph{only} for compiling Objective-C
2273 and Objective-C++ programs:
2276 @item -fconstant-string-class=@var{class-name}
2277 @opindex fconstant-string-class
2278 Use @var{class-name} as the name of the class to instantiate for each
2279 literal string specified with the syntax @code{@@"@dots{}"}. The default
2280 class name is @code{NXConstantString} if the GNU runtime is being used, and
2281 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2282 @option{-fconstant-cfstrings} option, if also present, will override the
2283 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2284 to be laid out as constant CoreFoundation strings.
2287 @opindex fgnu-runtime
2288 Generate object code compatible with the standard GNU Objective-C
2289 runtime. This is the default for most types of systems.
2291 @item -fnext-runtime
2292 @opindex fnext-runtime
2293 Generate output compatible with the NeXT runtime. This is the default
2294 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2295 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2298 @item -fno-nil-receivers
2299 @opindex fno-nil-receivers
2300 Assume that all Objective-C message dispatches (e.g.,
2301 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2302 is not @code{nil}. This allows for more efficient entry points in the runtime
2303 to be used. Currently, this option is only available in conjunction with
2304 the NeXT runtime on Mac OS X 10.3 and later.
2306 @item -fobjc-call-cxx-cdtors
2307 @opindex fobjc-call-cxx-cdtors
2308 For each Objective-C class, check if any of its instance variables is a
2309 C++ object with a non-trivial default constructor. If so, synthesize a
2310 special @code{- (id) .cxx_construct} instance method that will run
2311 non-trivial default constructors on any such instance variables, in order,
2312 and then return @code{self}. Similarly, check if any instance variable
2313 is a C++ object with a non-trivial destructor, and if so, synthesize a
2314 special @code{- (void) .cxx_destruct} method that will run
2315 all such default destructors, in reverse order.
2317 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2318 thusly generated will only operate on instance variables declared in the
2319 current Objective-C class, and not those inherited from superclasses. It
2320 is the responsibility of the Objective-C runtime to invoke all such methods
2321 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2322 will be invoked by the runtime immediately after a new object
2323 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2324 be invoked immediately before the runtime deallocates an object instance.
2326 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2327 support for invoking the @code{- (id) .cxx_construct} and
2328 @code{- (void) .cxx_destruct} methods.
2330 @item -fobjc-direct-dispatch
2331 @opindex fobjc-direct-dispatch
2332 Allow fast jumps to the message dispatcher. On Darwin this is
2333 accomplished via the comm page.
2335 @item -fobjc-exceptions
2336 @opindex fobjc-exceptions
2337 Enable syntactic support for structured exception handling in Objective-C,
2338 similar to what is offered by C++ and Java. This option is
2339 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2348 @@catch (AnObjCClass *exc) @{
2355 @@catch (AnotherClass *exc) @{
2358 @@catch (id allOthers) @{
2368 The @code{@@throw} statement may appear anywhere in an Objective-C or
2369 Objective-C++ program; when used inside of a @code{@@catch} block, the
2370 @code{@@throw} may appear without an argument (as shown above), in which case
2371 the object caught by the @code{@@catch} will be rethrown.
2373 Note that only (pointers to) Objective-C objects may be thrown and
2374 caught using this scheme. When an object is thrown, it will be caught
2375 by the nearest @code{@@catch} clause capable of handling objects of that type,
2376 analogously to how @code{catch} blocks work in C++ and Java. A
2377 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2378 any and all Objective-C exceptions not caught by previous @code{@@catch}
2381 The @code{@@finally} clause, if present, will be executed upon exit from the
2382 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2383 regardless of whether any exceptions are thrown, caught or rethrown
2384 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2385 of the @code{finally} clause in Java.
2387 There are several caveats to using the new exception mechanism:
2391 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2392 idioms provided by the @code{NSException} class, the new
2393 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2394 systems, due to additional functionality needed in the (NeXT) Objective-C
2398 As mentioned above, the new exceptions do not support handling
2399 types other than Objective-C objects. Furthermore, when used from
2400 Objective-C++, the Objective-C exception model does not interoperate with C++
2401 exceptions at this time. This means you cannot @code{@@throw} an exception
2402 from Objective-C and @code{catch} it in C++, or vice versa
2403 (i.e., @code{throw @dots{} @@catch}).
2406 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2407 blocks for thread-safe execution:
2410 @@synchronized (ObjCClass *guard) @{
2415 Upon entering the @code{@@synchronized} block, a thread of execution shall
2416 first check whether a lock has been placed on the corresponding @code{guard}
2417 object by another thread. If it has, the current thread shall wait until
2418 the other thread relinquishes its lock. Once @code{guard} becomes available,
2419 the current thread will place its own lock on it, execute the code contained in
2420 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2421 making @code{guard} available to other threads).
2423 Unlike Java, Objective-C does not allow for entire methods to be marked
2424 @code{@@synchronized}. Note that throwing exceptions out of
2425 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2426 to be unlocked properly.
2430 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2432 @item -freplace-objc-classes
2433 @opindex freplace-objc-classes
2434 Emit a special marker instructing @command{ld(1)} not to statically link in
2435 the resulting object file, and allow @command{dyld(1)} to load it in at
2436 run time instead. This is used in conjunction with the Fix-and-Continue
2437 debugging mode, where the object file in question may be recompiled and
2438 dynamically reloaded in the course of program execution, without the need
2439 to restart the program itself. Currently, Fix-and-Continue functionality
2440 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2445 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2446 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2447 compile time) with static class references that get initialized at load time,
2448 which improves run-time performance. Specifying the @option{-fzero-link} flag
2449 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2450 to be retained. This is useful in Zero-Link debugging mode, since it allows
2451 for individual class implementations to be modified during program execution.
2455 Dump interface declarations for all classes seen in the source file to a
2456 file named @file{@var{sourcename}.decl}.
2458 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2459 @opindex Wassign-intercept
2460 @opindex Wno-assign-intercept
2461 Warn whenever an Objective-C assignment is being intercepted by the
2464 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2465 @opindex Wno-protocol
2467 If a class is declared to implement a protocol, a warning is issued for
2468 every method in the protocol that is not implemented by the class. The
2469 default behavior is to issue a warning for every method not explicitly
2470 implemented in the class, even if a method implementation is inherited
2471 from the superclass. If you use the @option{-Wno-protocol} option, then
2472 methods inherited from the superclass are considered to be implemented,
2473 and no warning is issued for them.
2475 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2477 @opindex Wno-selector
2478 Warn if multiple methods of different types for the same selector are
2479 found during compilation. The check is performed on the list of methods
2480 in the final stage of compilation. Additionally, a check is performed
2481 for each selector appearing in a @code{@@selector(@dots{})}
2482 expression, and a corresponding method for that selector has been found
2483 during compilation. Because these checks scan the method table only at
2484 the end of compilation, these warnings are not produced if the final
2485 stage of compilation is not reached, for example because an error is
2486 found during compilation, or because the @option{-fsyntax-only} option is
2489 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2490 @opindex Wstrict-selector-match
2491 @opindex Wno-strict-selector-match
2492 Warn if multiple methods with differing argument and/or return types are
2493 found for a given selector when attempting to send a message using this
2494 selector to a receiver of type @code{id} or @code{Class}. When this flag
2495 is off (which is the default behavior), the compiler will omit such warnings
2496 if any differences found are confined to types which share the same size
2499 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2500 @opindex Wundeclared-selector
2501 @opindex Wno-undeclared-selector
2502 Warn if a @code{@@selector(@dots{})} expression referring to an
2503 undeclared selector is found. A selector is considered undeclared if no
2504 method with that name has been declared before the
2505 @code{@@selector(@dots{})} expression, either explicitly in an
2506 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2507 an @code{@@implementation} section. This option always performs its
2508 checks as soon as a @code{@@selector(@dots{})} expression is found,
2509 while @option{-Wselector} only performs its checks in the final stage of
2510 compilation. This also enforces the coding style convention
2511 that methods and selectors must be declared before being used.
2513 @item -print-objc-runtime-info
2514 @opindex print-objc-runtime-info
2515 Generate C header describing the largest structure that is passed by
2520 @node Language Independent Options
2521 @section Options to Control Diagnostic Messages Formatting
2522 @cindex options to control diagnostics formatting
2523 @cindex diagnostic messages
2524 @cindex message formatting
2526 Traditionally, diagnostic messages have been formatted irrespective of
2527 the output device's aspect (e.g.@: its width, @dots{}). The options described
2528 below can be used to control the diagnostic messages formatting
2529 algorithm, e.g.@: how many characters per line, how often source location
2530 information should be reported. Right now, only the C++ front end can
2531 honor these options. However it is expected, in the near future, that
2532 the remaining front ends would be able to digest them correctly.
2535 @item -fmessage-length=@var{n}
2536 @opindex fmessage-length
2537 Try to format error messages so that they fit on lines of about @var{n}
2538 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2539 the front ends supported by GCC@. If @var{n} is zero, then no
2540 line-wrapping will be done; each error message will appear on a single
2543 @opindex fdiagnostics-show-location
2544 @item -fdiagnostics-show-location=once
2545 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2546 reporter to emit @emph{once} source location information; that is, in
2547 case the message is too long to fit on a single physical line and has to
2548 be wrapped, the source location won't be emitted (as prefix) again,
2549 over and over, in subsequent continuation lines. This is the default
2552 @item -fdiagnostics-show-location=every-line
2553 Only meaningful in line-wrapping mode. Instructs the diagnostic
2554 messages reporter to emit the same source location information (as
2555 prefix) for physical lines that result from the process of breaking
2556 a message which is too long to fit on a single line.
2558 @item -fdiagnostics-show-option
2559 @opindex fdiagnostics-show-option
2560 This option instructs the diagnostic machinery to add text to each
2561 diagnostic emitted, which indicates which command line option directly
2562 controls that diagnostic, when such an option is known to the
2563 diagnostic machinery.
2565 @item -Wcoverage-mismatch
2566 @opindex Wcoverage-mismatch
2567 Warn if feedback profiles do not match when using the
2568 @option{-fprofile-use} option.
2569 If a source file was changed between @option{-fprofile-gen} and
2570 @option{-fprofile-use}, the files with the profile feedback can fail
2571 to match the source file and GCC can not use the profile feedback
2572 information. By default, GCC emits an error message in this case.
2573 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2574 error. GCC does not use appropriate feedback profiles, so using this
2575 option can result in poorly optimized code. This option is useful
2576 only in the case of very minor changes such as bug fixes to an
2581 @node Warning Options
2582 @section Options to Request or Suppress Warnings
2583 @cindex options to control warnings
2584 @cindex warning messages
2585 @cindex messages, warning
2586 @cindex suppressing warnings
2588 Warnings are diagnostic messages that report constructions which
2589 are not inherently erroneous but which are risky or suggest there
2590 may have been an error.
2592 The following language-independent options do not enable specific
2593 warnings but control the kinds of diagnostics produced by GCC.
2596 @cindex syntax checking
2598 @opindex fsyntax-only
2599 Check the code for syntax errors, but don't do anything beyond that.
2603 Inhibit all warning messages.
2608 Make all warnings into errors.
2613 Make the specified warning into an error. The specifier for a warning
2614 is appended, for example @option{-Werror=switch} turns the warnings
2615 controlled by @option{-Wswitch} into errors. This switch takes a
2616 negative form, to be used to negate @option{-Werror} for specific
2617 warnings, for example @option{-Wno-error=switch} makes
2618 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2619 is in effect. You can use the @option{-fdiagnostics-show-option}
2620 option to have each controllable warning amended with the option which
2621 controls it, to determine what to use with this option.
2623 Note that specifying @option{-Werror=}@var{foo} automatically implies
2624 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2627 @item -Wfatal-errors
2628 @opindex Wfatal-errors
2629 @opindex Wno-fatal-errors
2630 This option causes the compiler to abort compilation on the first error
2631 occurred rather than trying to keep going and printing further error
2636 You can request many specific warnings with options beginning
2637 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2638 implicit declarations. Each of these specific warning options also
2639 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2640 example, @option{-Wno-implicit}. This manual lists only one of the
2641 two forms, whichever is not the default. For further,
2642 language-specific options also refer to @ref{C++ Dialect Options} and
2643 @ref{Objective-C and Objective-C++ Dialect Options}.
2648 Issue all the warnings demanded by strict ISO C and ISO C++;
2649 reject all programs that use forbidden extensions, and some other
2650 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2651 version of the ISO C standard specified by any @option{-std} option used.
2653 Valid ISO C and ISO C++ programs should compile properly with or without
2654 this option (though a rare few will require @option{-ansi} or a
2655 @option{-std} option specifying the required version of ISO C)@. However,
2656 without this option, certain GNU extensions and traditional C and C++
2657 features are supported as well. With this option, they are rejected.
2659 @option{-pedantic} does not cause warning messages for use of the
2660 alternate keywords whose names begin and end with @samp{__}. Pedantic
2661 warnings are also disabled in the expression that follows
2662 @code{__extension__}. However, only system header files should use
2663 these escape routes; application programs should avoid them.
2664 @xref{Alternate Keywords}.
2666 Some users try to use @option{-pedantic} to check programs for strict ISO
2667 C conformance. They soon find that it does not do quite what they want:
2668 it finds some non-ISO practices, but not all---only those for which
2669 ISO C @emph{requires} a diagnostic, and some others for which
2670 diagnostics have been added.
2672 A feature to report any failure to conform to ISO C might be useful in
2673 some instances, but would require considerable additional work and would
2674 be quite different from @option{-pedantic}. We don't have plans to
2675 support such a feature in the near future.
2677 Where the standard specified with @option{-std} represents a GNU
2678 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2679 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2680 extended dialect is based. Warnings from @option{-pedantic} are given
2681 where they are required by the base standard. (It would not make sense
2682 for such warnings to be given only for features not in the specified GNU
2683 C dialect, since by definition the GNU dialects of C include all
2684 features the compiler supports with the given option, and there would be
2685 nothing to warn about.)
2687 @item -pedantic-errors
2688 @opindex pedantic-errors
2689 Like @option{-pedantic}, except that errors are produced rather than
2695 This enables all the warnings about constructions that some users
2696 consider questionable, and that are easy to avoid (or modify to
2697 prevent the warning), even in conjunction with macros. This also
2698 enables some language-specific warnings described in @ref{C++ Dialect
2699 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2701 @option{-Wall} turns on the following warning flags:
2703 @gccoptlist{-Waddress @gol
2704 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2706 -Wchar-subscripts @gol
2708 -Wimplicit-function-declaration @gol
2711 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2712 -Wmissing-braces @gol
2718 -Wsequence-point @gol
2719 -Wsign-compare @r{(only in C++)} @gol
2720 -Wstrict-aliasing @gol
2721 -Wstrict-overflow=1 @gol
2724 -Wuninitialized @gol
2725 -Wunknown-pragmas @gol
2726 -Wunused-function @gol
2729 -Wunused-variable @gol
2730 -Wvolatile-register-var @gol
2733 Note that some warning flags are not implied by @option{-Wall}. Some of
2734 them warn about constructions that users generally do not consider
2735 questionable, but which occasionally you might wish to check for;
2736 others warn about constructions that are necessary or hard to avoid in
2737 some cases, and there is no simple way to modify the code to suppress
2738 the warning. Some of them are enabled by @option{-Wextra} but many of
2739 them must be enabled individually.
2745 This enables some extra warning flags that are not enabled by
2746 @option{-Wall}. (This option used to be called @option{-W}. The older
2747 name is still supported, but the newer name is more descriptive.)
2749 @gccoptlist{-Wclobbered @gol
2751 -Wignored-qualifiers @gol
2752 -Wmissing-field-initializers @gol
2753 -Wmissing-parameter-type @r{(C only)} @gol
2754 -Wold-style-declaration @r{(C only)} @gol
2755 -Woverride-init @gol
2758 -Wuninitialized @gol
2759 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2762 The option @option{-Wextra} also prints warning messages for the
2768 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2769 @samp{>}, or @samp{>=}.
2772 (C++ only) An enumerator and a non-enumerator both appear in a
2773 conditional expression.
2776 (C++ only) Ambiguous virtual bases.
2779 (C++ only) Subscripting an array which has been declared @samp{register}.
2782 (C++ only) Taking the address of a variable which has been declared
2786 (C++ only) A base class is not initialized in a derived class' copy
2791 @item -Wchar-subscripts
2792 @opindex Wchar-subscripts
2793 @opindex Wno-char-subscripts
2794 Warn if an array subscript has type @code{char}. This is a common cause
2795 of error, as programmers often forget that this type is signed on some
2797 This warning is enabled by @option{-Wall}.
2801 @opindex Wno-comment
2802 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2803 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2804 This warning is enabled by @option{-Wall}.
2809 @opindex ffreestanding
2810 @opindex fno-builtin
2811 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2812 the arguments supplied have types appropriate to the format string
2813 specified, and that the conversions specified in the format string make
2814 sense. This includes standard functions, and others specified by format
2815 attributes (@pxref{Function Attributes}), in the @code{printf},
2816 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2817 not in the C standard) families (or other target-specific families).
2818 Which functions are checked without format attributes having been
2819 specified depends on the standard version selected, and such checks of
2820 functions without the attribute specified are disabled by
2821 @option{-ffreestanding} or @option{-fno-builtin}.
2823 The formats are checked against the format features supported by GNU
2824 libc version 2.2. These include all ISO C90 and C99 features, as well
2825 as features from the Single Unix Specification and some BSD and GNU
2826 extensions. Other library implementations may not support all these
2827 features; GCC does not support warning about features that go beyond a
2828 particular library's limitations. However, if @option{-pedantic} is used
2829 with @option{-Wformat}, warnings will be given about format features not
2830 in the selected standard version (but not for @code{strfmon} formats,
2831 since those are not in any version of the C standard). @xref{C Dialect
2832 Options,,Options Controlling C Dialect}.
2834 Since @option{-Wformat} also checks for null format arguments for
2835 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2837 @option{-Wformat} is included in @option{-Wall}. For more control over some
2838 aspects of format checking, the options @option{-Wformat-y2k},
2839 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2840 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2841 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2844 @opindex Wformat-y2k
2845 @opindex Wno-format-y2k
2846 If @option{-Wformat} is specified, also warn about @code{strftime}
2847 formats which may yield only a two-digit year.
2849 @item -Wno-format-contains-nul
2850 @opindex Wno-format-contains-nul
2851 @opindex Wformat-contains-nul
2852 If @option{-Wformat} is specified, do not warn about format strings that
2855 @item -Wno-format-extra-args
2856 @opindex Wno-format-extra-args
2857 @opindex Wformat-extra-args
2858 If @option{-Wformat} is specified, do not warn about excess arguments to a
2859 @code{printf} or @code{scanf} format function. The C standard specifies
2860 that such arguments are ignored.
2862 Where the unused arguments lie between used arguments that are
2863 specified with @samp{$} operand number specifications, normally
2864 warnings are still given, since the implementation could not know what
2865 type to pass to @code{va_arg} to skip the unused arguments. However,
2866 in the case of @code{scanf} formats, this option will suppress the
2867 warning if the unused arguments are all pointers, since the Single
2868 Unix Specification says that such unused arguments are allowed.
2870 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2871 @opindex Wno-format-zero-length
2872 @opindex Wformat-zero-length
2873 If @option{-Wformat} is specified, do not warn about zero-length formats.
2874 The C standard specifies that zero-length formats are allowed.
2876 @item -Wformat-nonliteral
2877 @opindex Wformat-nonliteral
2878 @opindex Wno-format-nonliteral
2879 If @option{-Wformat} is specified, also warn if the format string is not a
2880 string literal and so cannot be checked, unless the format function
2881 takes its format arguments as a @code{va_list}.
2883 @item -Wformat-security
2884 @opindex Wformat-security
2885 @opindex Wno-format-security
2886 If @option{-Wformat} is specified, also warn about uses of format
2887 functions that represent possible security problems. At present, this
2888 warns about calls to @code{printf} and @code{scanf} functions where the
2889 format string is not a string literal and there are no format arguments,
2890 as in @code{printf (foo);}. This may be a security hole if the format
2891 string came from untrusted input and contains @samp{%n}. (This is
2892 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2893 in future warnings may be added to @option{-Wformat-security} that are not
2894 included in @option{-Wformat-nonliteral}.)
2898 @opindex Wno-format=2
2899 Enable @option{-Wformat} plus format checks not included in
2900 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2901 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2903 @item -Wnonnull @r{(C and Objective-C only)}
2905 @opindex Wno-nonnull
2906 Warn about passing a null pointer for arguments marked as
2907 requiring a non-null value by the @code{nonnull} function attribute.
2909 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2910 can be disabled with the @option{-Wno-nonnull} option.
2912 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2914 @opindex Wno-init-self
2915 Warn about uninitialized variables which are initialized with themselves.
2916 Note this option can only be used with the @option{-Wuninitialized} option.
2918 For example, GCC will warn about @code{i} being uninitialized in the
2919 following snippet only when @option{-Winit-self} has been specified:
2930 @item -Wimplicit-int @r{(C and Objective-C only)}
2931 @opindex Wimplicit-int
2932 @opindex Wno-implicit-int
2933 Warn when a declaration does not specify a type.
2934 This warning is enabled by @option{-Wall}.
2936 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
2937 @opindex Wimplicit-function-declaration
2938 @opindex Wno-implicit-function-declaration
2939 Give a warning whenever a function is used before being declared. In
2940 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2941 enabled by default and it is made into an error by
2942 @option{-pedantic-errors}. This warning is also enabled by
2947 @opindex Wno-implicit
2948 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2949 This warning is enabled by @option{-Wall}.
2951 @item -Wignored-qualifiers @r{(C and C++ only)}
2952 @opindex Wignored-qualifiers
2953 @opindex Wno-ignored-qualifiers
2954 Warn if the return type of a function has a type qualifier
2955 such as @code{const}. For ISO C such a type qualifier has no effect,
2956 since the value returned by a function is not an lvalue.
2957 For C++, the warning is only emitted for scalar types or @code{void}.
2958 ISO C prohibits qualified @code{void} return types on function
2959 definitions, so such return types always receive a warning
2960 even without this option.
2962 This warning is also enabled by @option{-Wextra}.
2967 Warn if the type of @samp{main} is suspicious. @samp{main} should be
2968 a function with external linkage, returning int, taking either zero
2969 arguments, two, or three arguments of appropriate types. This warning
2970 is enabled by default in C++ and is enabled by either @option{-Wall}
2971 or @option{-pedantic}.
2973 @item -Wmissing-braces
2974 @opindex Wmissing-braces
2975 @opindex Wno-missing-braces
2976 Warn if an aggregate or union initializer is not fully bracketed. In
2977 the following example, the initializer for @samp{a} is not fully
2978 bracketed, but that for @samp{b} is fully bracketed.
2981 int a[2][2] = @{ 0, 1, 2, 3 @};
2982 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2985 This warning is enabled by @option{-Wall}.
2987 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2988 @opindex Wmissing-include-dirs
2989 @opindex Wno-missing-include-dirs
2990 Warn if a user-supplied include directory does not exist.
2993 @opindex Wparentheses
2994 @opindex Wno-parentheses
2995 Warn if parentheses are omitted in certain contexts, such
2996 as when there is an assignment in a context where a truth value
2997 is expected, or when operators are nested whose precedence people
2998 often get confused about.
3000 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3001 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3002 interpretation from that of ordinary mathematical notation.
3004 Also warn about constructions where there may be confusion to which
3005 @code{if} statement an @code{else} branch belongs. Here is an example of
3020 In C/C++, every @code{else} branch belongs to the innermost possible
3021 @code{if} statement, which in this example is @code{if (b)}. This is
3022 often not what the programmer expected, as illustrated in the above
3023 example by indentation the programmer chose. When there is the
3024 potential for this confusion, GCC will issue a warning when this flag
3025 is specified. To eliminate the warning, add explicit braces around
3026 the innermost @code{if} statement so there is no way the @code{else}
3027 could belong to the enclosing @code{if}. The resulting code would
3044 This warning is enabled by @option{-Wall}.
3046 @item -Wsequence-point
3047 @opindex Wsequence-point
3048 @opindex Wno-sequence-point
3049 Warn about code that may have undefined semantics because of violations
3050 of sequence point rules in the C and C++ standards.
3052 The C and C++ standards defines the order in which expressions in a C/C++
3053 program are evaluated in terms of @dfn{sequence points}, which represent
3054 a partial ordering between the execution of parts of the program: those
3055 executed before the sequence point, and those executed after it. These
3056 occur after the evaluation of a full expression (one which is not part
3057 of a larger expression), after the evaluation of the first operand of a
3058 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3059 function is called (but after the evaluation of its arguments and the
3060 expression denoting the called function), and in certain other places.
3061 Other than as expressed by the sequence point rules, the order of
3062 evaluation of subexpressions of an expression is not specified. All
3063 these rules describe only a partial order rather than a total order,
3064 since, for example, if two functions are called within one expression
3065 with no sequence point between them, the order in which the functions
3066 are called is not specified. However, the standards committee have
3067 ruled that function calls do not overlap.
3069 It is not specified when between sequence points modifications to the
3070 values of objects take effect. Programs whose behavior depends on this
3071 have undefined behavior; the C and C++ standards specify that ``Between
3072 the previous and next sequence point an object shall have its stored
3073 value modified at most once by the evaluation of an expression.
3074 Furthermore, the prior value shall be read only to determine the value
3075 to be stored.''. If a program breaks these rules, the results on any
3076 particular implementation are entirely unpredictable.
3078 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3079 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3080 diagnosed by this option, and it may give an occasional false positive
3081 result, but in general it has been found fairly effective at detecting
3082 this sort of problem in programs.
3084 The standard is worded confusingly, therefore there is some debate
3085 over the precise meaning of the sequence point rules in subtle cases.
3086 Links to discussions of the problem, including proposed formal
3087 definitions, may be found on the GCC readings page, at
3088 @w{@uref{http://gcc.gnu.org/readings.html}}.
3090 This warning is enabled by @option{-Wall} for C and C++.
3093 @opindex Wreturn-type
3094 @opindex Wno-return-type
3095 Warn whenever a function is defined with a return-type that defaults
3096 to @code{int}. Also warn about any @code{return} statement with no
3097 return-value in a function whose return-type is not @code{void}
3098 (falling off the end of the function body is considered returning
3099 without a value), and about a @code{return} statement with a
3100 expression in a function whose return-type is @code{void}.
3102 For C++, a function without return type always produces a diagnostic
3103 message, even when @option{-Wno-return-type} is specified. The only
3104 exceptions are @samp{main} and functions defined in system headers.
3106 This warning is enabled by @option{-Wall}.
3111 Warn whenever a @code{switch} statement has an index of enumerated type
3112 and lacks a @code{case} for one or more of the named codes of that
3113 enumeration. (The presence of a @code{default} label prevents this
3114 warning.) @code{case} labels outside the enumeration range also
3115 provoke warnings when this option is used.
3116 This warning is enabled by @option{-Wall}.
3118 @item -Wswitch-default
3119 @opindex Wswitch-default
3120 @opindex Wno-switch-default
3121 Warn whenever a @code{switch} statement does not have a @code{default}
3125 @opindex Wswitch-enum
3126 @opindex Wno-switch-enum
3127 Warn whenever a @code{switch} statement has an index of enumerated type
3128 and lacks a @code{case} for one or more of the named codes of that
3129 enumeration. @code{case} labels outside the enumeration range also
3130 provoke warnings when this option is used.
3134 @opindex Wno-sync-nand
3135 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3136 built-in functions are used. These functions changed semantics in GCC 4.4.
3140 @opindex Wno-trigraphs
3141 Warn if any trigraphs are encountered that might change the meaning of
3142 the program (trigraphs within comments are not warned about).
3143 This warning is enabled by @option{-Wall}.
3145 @item -Wunused-function
3146 @opindex Wunused-function
3147 @opindex Wno-unused-function
3148 Warn whenever a static function is declared but not defined or a
3149 non-inline static function is unused.
3150 This warning is enabled by @option{-Wall}.
3152 @item -Wunused-label
3153 @opindex Wunused-label
3154 @opindex Wno-unused-label
3155 Warn whenever a label is declared but not used.
3156 This warning is enabled by @option{-Wall}.
3158 To suppress this warning use the @samp{unused} attribute
3159 (@pxref{Variable Attributes}).
3161 @item -Wunused-parameter
3162 @opindex Wunused-parameter
3163 @opindex Wno-unused-parameter
3164 Warn whenever a function parameter is unused aside from its declaration.
3166 To suppress this warning use the @samp{unused} attribute
3167 (@pxref{Variable Attributes}).
3169 @item -Wunused-variable
3170 @opindex Wunused-variable
3171 @opindex Wno-unused-variable
3172 Warn whenever a local variable or non-constant static variable is unused
3173 aside from its declaration.
3174 This warning is enabled by @option{-Wall}.
3176 To suppress this warning use the @samp{unused} attribute
3177 (@pxref{Variable Attributes}).
3179 @item -Wunused-value
3180 @opindex Wunused-value
3181 @opindex Wno-unused-value
3182 Warn whenever a statement computes a result that is explicitly not
3183 used. To suppress this warning cast the unused expression to
3184 @samp{void}. This includes an expression-statement or the left-hand
3185 side of a comma expression that contains no side effects. For example,
3186 an expression such as @samp{x[i,j]} will cause a warning, while
3187 @samp{x[(void)i,j]} will not.
3189 This warning is enabled by @option{-Wall}.
3194 All the above @option{-Wunused} options combined.
3196 In order to get a warning about an unused function parameter, you must
3197 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3198 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3200 @item -Wuninitialized
3201 @opindex Wuninitialized
3202 @opindex Wno-uninitialized
3203 Warn if an automatic variable is used without first being initialized
3204 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3205 warn if a non-static reference or non-static @samp{const} member
3206 appears in a class without constructors.
3208 If you want to warn about code which uses the uninitialized value of the
3209 variable in its own initializer, use the @option{-Winit-self} option.
3211 These warnings occur for individual uninitialized or clobbered
3212 elements of structure, union or array variables as well as for
3213 variables which are uninitialized or clobbered as a whole. They do
3214 not occur for variables or elements declared @code{volatile}. Because
3215 these warnings depend on optimization, the exact variables or elements
3216 for which there are warnings will depend on the precise optimization
3217 options and version of GCC used.
3219 Note that there may be no warning about a variable that is used only
3220 to compute a value that itself is never used, because such
3221 computations may be deleted by data flow analysis before the warnings
3224 These warnings are made optional because GCC is not smart
3225 enough to see all the reasons why the code might be correct
3226 despite appearing to have an error. Here is one example of how
3247 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3248 always initialized, but GCC doesn't know this. Here is
3249 another common case:
3254 if (change_y) save_y = y, y = new_y;
3256 if (change_y) y = save_y;
3261 This has no bug because @code{save_y} is used only if it is set.
3263 @cindex @code{longjmp} warnings
3264 This option also warns when a non-volatile automatic variable might be
3265 changed by a call to @code{longjmp}. These warnings as well are possible
3266 only in optimizing compilation.
3268 The compiler sees only the calls to @code{setjmp}. It cannot know
3269 where @code{longjmp} will be called; in fact, a signal handler could
3270 call it at any point in the code. As a result, you may get a warning
3271 even when there is in fact no problem because @code{longjmp} cannot
3272 in fact be called at the place which would cause a problem.
3274 Some spurious warnings can be avoided if you declare all the functions
3275 you use that never return as @code{noreturn}. @xref{Function
3278 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3280 @item -Wunknown-pragmas
3281 @opindex Wunknown-pragmas
3282 @opindex Wno-unknown-pragmas
3283 @cindex warning for unknown pragmas
3284 @cindex unknown pragmas, warning
3285 @cindex pragmas, warning of unknown
3286 Warn when a #pragma directive is encountered which is not understood by
3287 GCC@. If this command line option is used, warnings will even be issued
3288 for unknown pragmas in system header files. This is not the case if
3289 the warnings were only enabled by the @option{-Wall} command line option.
3292 @opindex Wno-pragmas
3294 Do not warn about misuses of pragmas, such as incorrect parameters,
3295 invalid syntax, or conflicts between pragmas. See also
3296 @samp{-Wunknown-pragmas}.
3298 @item -Wstrict-aliasing
3299 @opindex Wstrict-aliasing
3300 @opindex Wno-strict-aliasing
3301 This option is only active when @option{-fstrict-aliasing} is active.
3302 It warns about code which might break the strict aliasing rules that the
3303 compiler is using for optimization. The warning does not catch all
3304 cases, but does attempt to catch the more common pitfalls. It is
3305 included in @option{-Wall}.
3306 It is equivalent to @option{-Wstrict-aliasing=3}
3308 @item -Wstrict-aliasing=n
3309 @opindex Wstrict-aliasing=n
3310 @opindex Wno-strict-aliasing=n
3311 This option is only active when @option{-fstrict-aliasing} is active.
3312 It warns about code which might break the strict aliasing rules that the
3313 compiler is using for optimization.
3314 Higher levels correspond to higher accuracy (fewer false positives).
3315 Higher levels also correspond to more effort, similar to the way -O works.
3316 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3319 Level 1: Most aggressive, quick, least accurate.
3320 Possibly useful when higher levels
3321 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3322 false negatives. However, it has many false positives.
3323 Warns for all pointer conversions between possibly incompatible types,
3324 even if never dereferenced. Runs in the frontend only.
3326 Level 2: Aggressive, quick, not too precise.
3327 May still have many false positives (not as many as level 1 though),
3328 and few false negatives (but possibly more than level 1).
3329 Unlike level 1, it only warns when an address is taken. Warns about
3330 incomplete types. Runs in the frontend only.
3332 Level 3 (default for @option{-Wstrict-aliasing}):
3333 Should have very few false positives and few false
3334 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3335 Takes care of the common punn+dereference pattern in the frontend:
3336 @code{*(int*)&some_float}.
3337 If optimization is enabled, it also runs in the backend, where it deals
3338 with multiple statement cases using flow-sensitive points-to information.
3339 Only warns when the converted pointer is dereferenced.
3340 Does not warn about incomplete types.
3342 @item -Wstrict-overflow
3343 @itemx -Wstrict-overflow=@var{n}
3344 @opindex Wstrict-overflow
3345 @opindex Wno-strict-overflow
3346 This option is only active when @option{-fstrict-overflow} is active.
3347 It warns about cases where the compiler optimizes based on the
3348 assumption that signed overflow does not occur. Note that it does not
3349 warn about all cases where the code might overflow: it only warns
3350 about cases where the compiler implements some optimization. Thus
3351 this warning depends on the optimization level.
3353 An optimization which assumes that signed overflow does not occur is
3354 perfectly safe if the values of the variables involved are such that
3355 overflow never does, in fact, occur. Therefore this warning can
3356 easily give a false positive: a warning about code which is not
3357 actually a problem. To help focus on important issues, several
3358 warning levels are defined. No warnings are issued for the use of
3359 undefined signed overflow when estimating how many iterations a loop
3360 will require, in particular when determining whether a loop will be
3364 @item -Wstrict-overflow=1
3365 Warn about cases which are both questionable and easy to avoid. For
3366 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3367 compiler will simplify this to @code{1}. This level of
3368 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3369 are not, and must be explicitly requested.
3371 @item -Wstrict-overflow=2
3372 Also warn about other cases where a comparison is simplified to a
3373 constant. For example: @code{abs (x) >= 0}. This can only be
3374 simplified when @option{-fstrict-overflow} is in effect, because
3375 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3376 zero. @option{-Wstrict-overflow} (with no level) is the same as
3377 @option{-Wstrict-overflow=2}.
3379 @item -Wstrict-overflow=3
3380 Also warn about other cases where a comparison is simplified. For
3381 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3383 @item -Wstrict-overflow=4
3384 Also warn about other simplifications not covered by the above cases.
3385 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3387 @item -Wstrict-overflow=5
3388 Also warn about cases where the compiler reduces the magnitude of a
3389 constant involved in a comparison. For example: @code{x + 2 > y} will
3390 be simplified to @code{x + 1 >= y}. This is reported only at the
3391 highest warning level because this simplification applies to many
3392 comparisons, so this warning level will give a very large number of
3396 @item -Warray-bounds
3397 @opindex Wno-array-bounds
3398 @opindex Warray-bounds
3399 This option is only active when @option{-ftree-vrp} is active
3400 (default for -O2 and above). It warns about subscripts to arrays
3401 that are always out of bounds. This warning is enabled by @option{-Wall}.
3403 @item -Wno-div-by-zero
3404 @opindex Wno-div-by-zero
3405 @opindex Wdiv-by-zero
3406 Do not warn about compile-time integer division by zero. Floating point
3407 division by zero is not warned about, as it can be a legitimate way of
3408 obtaining infinities and NaNs.
3410 @item -Wsystem-headers
3411 @opindex Wsystem-headers
3412 @opindex Wno-system-headers
3413 @cindex warnings from system headers
3414 @cindex system headers, warnings from
3415 Print warning messages for constructs found in system header files.
3416 Warnings from system headers are normally suppressed, on the assumption
3417 that they usually do not indicate real problems and would only make the
3418 compiler output harder to read. Using this command line option tells
3419 GCC to emit warnings from system headers as if they occurred in user
3420 code. However, note that using @option{-Wall} in conjunction with this
3421 option will @emph{not} warn about unknown pragmas in system
3422 headers---for that, @option{-Wunknown-pragmas} must also be used.
3425 @opindex Wfloat-equal
3426 @opindex Wno-float-equal
3427 Warn if floating point values are used in equality comparisons.
3429 The idea behind this is that sometimes it is convenient (for the
3430 programmer) to consider floating-point values as approximations to
3431 infinitely precise real numbers. If you are doing this, then you need
3432 to compute (by analyzing the code, or in some other way) the maximum or
3433 likely maximum error that the computation introduces, and allow for it
3434 when performing comparisons (and when producing output, but that's a
3435 different problem). In particular, instead of testing for equality, you
3436 would check to see whether the two values have ranges that overlap; and
3437 this is done with the relational operators, so equality comparisons are
3440 @item -Wtraditional @r{(C and Objective-C only)}
3441 @opindex Wtraditional
3442 @opindex Wno-traditional
3443 Warn about certain constructs that behave differently in traditional and
3444 ISO C@. Also warn about ISO C constructs that have no traditional C
3445 equivalent, and/or problematic constructs which should be avoided.
3449 Macro parameters that appear within string literals in the macro body.
3450 In traditional C macro replacement takes place within string literals,
3451 but does not in ISO C@.
3454 In traditional C, some preprocessor directives did not exist.
3455 Traditional preprocessors would only consider a line to be a directive
3456 if the @samp{#} appeared in column 1 on the line. Therefore
3457 @option{-Wtraditional} warns about directives that traditional C
3458 understands but would ignore because the @samp{#} does not appear as the
3459 first character on the line. It also suggests you hide directives like
3460 @samp{#pragma} not understood by traditional C by indenting them. Some
3461 traditional implementations would not recognize @samp{#elif}, so it
3462 suggests avoiding it altogether.
3465 A function-like macro that appears without arguments.
3468 The unary plus operator.
3471 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3472 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3473 constants.) Note, these suffixes appear in macros defined in the system
3474 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3475 Use of these macros in user code might normally lead to spurious
3476 warnings, however GCC's integrated preprocessor has enough context to
3477 avoid warning in these cases.
3480 A function declared external in one block and then used after the end of
3484 A @code{switch} statement has an operand of type @code{long}.
3487 A non-@code{static} function declaration follows a @code{static} one.
3488 This construct is not accepted by some traditional C compilers.
3491 The ISO type of an integer constant has a different width or
3492 signedness from its traditional type. This warning is only issued if
3493 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3494 typically represent bit patterns, are not warned about.
3497 Usage of ISO string concatenation is detected.
3500 Initialization of automatic aggregates.
3503 Identifier conflicts with labels. Traditional C lacks a separate
3504 namespace for labels.
3507 Initialization of unions. If the initializer is zero, the warning is
3508 omitted. This is done under the assumption that the zero initializer in
3509 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3510 initializer warnings and relies on default initialization to zero in the
3514 Conversions by prototypes between fixed/floating point values and vice
3515 versa. The absence of these prototypes when compiling with traditional
3516 C would cause serious problems. This is a subset of the possible
3517 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3520 Use of ISO C style function definitions. This warning intentionally is
3521 @emph{not} issued for prototype declarations or variadic functions
3522 because these ISO C features will appear in your code when using
3523 libiberty's traditional C compatibility macros, @code{PARAMS} and
3524 @code{VPARAMS}. This warning is also bypassed for nested functions
3525 because that feature is already a GCC extension and thus not relevant to
3526 traditional C compatibility.
3529 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3530 @opindex Wtraditional-conversion
3531 @opindex Wno-traditional-conversion
3532 Warn if a prototype causes a type conversion that is different from what
3533 would happen to the same argument in the absence of a prototype. This
3534 includes conversions of fixed point to floating and vice versa, and
3535 conversions changing the width or signedness of a fixed point argument
3536 except when the same as the default promotion.
3538 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3539 @opindex Wdeclaration-after-statement
3540 @opindex Wno-declaration-after-statement
3541 Warn when a declaration is found after a statement in a block. This
3542 construct, known from C++, was introduced with ISO C99 and is by default
3543 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3544 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3549 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3551 @item -Wno-endif-labels
3552 @opindex Wno-endif-labels
3553 @opindex Wendif-labels
3554 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3559 Warn whenever a local variable shadows another local variable, parameter or
3560 global variable or whenever a built-in function is shadowed.
3562 @item -Wlarger-than=@var{len}
3563 @opindex Wlarger-than=@var{len}
3564 @opindex Wlarger-than-@var{len}
3565 Warn whenever an object of larger than @var{len} bytes is defined.
3567 @item -Wframe-larger-than=@var{len}
3568 @opindex Wframe-larger-than
3569 Warn if the size of a function frame is larger than @var{len} bytes.
3570 The computation done to determine the stack frame size is approximate
3571 and not conservative.
3572 The actual requirements may be somewhat greater than @var{len}
3573 even if you do not get a warning. In addition, any space allocated
3574 via @code{alloca}, variable-length arrays, or related constructs
3575 is not included by the compiler when determining
3576 whether or not to issue a warning.
3578 @item -Wunsafe-loop-optimizations
3579 @opindex Wunsafe-loop-optimizations
3580 @opindex Wno-unsafe-loop-optimizations
3581 Warn if the loop cannot be optimized because the compiler could not
3582 assume anything on the bounds of the loop indices. With
3583 @option{-funsafe-loop-optimizations} warn if the compiler made
3586 @item -Wno-pedantic-ms-format
3587 @opindex Wno-pedantic-ms-format
3588 @opindex Wpedantic-ms-format
3589 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3590 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3591 depending on the MS runtime, when you are using the options @option{-Wformat}
3592 and @option{-pedantic} without gnu-extensions.
3594 @item -Wpointer-arith
3595 @opindex Wpointer-arith
3596 @opindex Wno-pointer-arith
3597 Warn about anything that depends on the ``size of'' a function type or
3598 of @code{void}. GNU C assigns these types a size of 1, for
3599 convenience in calculations with @code{void *} pointers and pointers
3600 to functions. In C++, warn also when an arithmetic operation involves
3601 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3604 @opindex Wtype-limits
3605 @opindex Wno-type-limits
3606 Warn if a comparison is always true or always false due to the limited
3607 range of the data type, but do not warn for constant expressions. For
3608 example, warn if an unsigned variable is compared against zero with
3609 @samp{<} or @samp{>=}. This warning is also enabled by
3612 @item -Wbad-function-cast @r{(C and Objective-C only)}
3613 @opindex Wbad-function-cast
3614 @opindex Wno-bad-function-cast
3615 Warn whenever a function call is cast to a non-matching type.
3616 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3618 @item -Wc++-compat @r{(C and Objective-C only)}
3619 Warn about ISO C constructs that are outside of the common subset of
3620 ISO C and ISO C++, e.g.@: request for implicit conversion from
3621 @code{void *} to a pointer to non-@code{void} type.
3623 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3624 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3625 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3626 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3630 @opindex Wno-cast-qual
3631 Warn whenever a pointer is cast so as to remove a type qualifier from
3632 the target type. For example, warn if a @code{const char *} is cast
3633 to an ordinary @code{char *}.
3636 @opindex Wcast-align
3637 @opindex Wno-cast-align
3638 Warn whenever a pointer is cast such that the required alignment of the
3639 target is increased. For example, warn if a @code{char *} is cast to
3640 an @code{int *} on machines where integers can only be accessed at
3641 two- or four-byte boundaries.
3643 @item -Wwrite-strings
3644 @opindex Wwrite-strings
3645 @opindex Wno-write-strings
3646 When compiling C, give string constants the type @code{const
3647 char[@var{length}]} so that copying the address of one into a
3648 non-@code{const} @code{char *} pointer will get a warning. These
3649 warnings will help you find at compile time code that can try to write
3650 into a string constant, but only if you have been very careful about
3651 using @code{const} in declarations and prototypes. Otherwise, it will
3652 just be a nuisance. This is why we did not make @option{-Wall} request
3655 When compiling C++, warn about the deprecated conversion from string
3656 literals to @code{char *}. This warning is enabled by default for C++
3661 @opindex Wno-clobbered
3662 Warn for variables that might be changed by @samp{longjmp} or
3663 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3666 @opindex Wconversion
3667 @opindex Wno-conversion
3668 Warn for implicit conversions that may alter a value. This includes
3669 conversions between real and integer, like @code{abs (x)} when
3670 @code{x} is @code{double}; conversions between signed and unsigned,
3671 like @code{unsigned ui = -1}; and conversions to smaller types, like
3672 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3673 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3674 changed by the conversion like in @code{abs (2.0)}. Warnings about
3675 conversions between signed and unsigned integers can be disabled by
3676 using @option{-Wno-sign-conversion}.
3678 For C++, also warn for conversions between @code{NULL} and non-pointer
3679 types; confusing overload resolution for user-defined conversions; and
3680 conversions that will never use a type conversion operator:
3681 conversions to @code{void}, the same type, a base class or a reference
3682 to them. Warnings about conversions between signed and unsigned
3683 integers are disabled by default in C++ unless
3684 @option{-Wsign-conversion} is explicitly enabled.
3687 @opindex Wempty-body
3688 @opindex Wno-empty-body
3689 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3690 while} statement. This warning is also enabled by @option{-Wextra}.
3692 @item -Wenum-compare @r{(C++ and Objective-C++ only)}
3693 @opindex Wenum-compare
3694 @opindex Wno-enum-compare
3695 Warn about a comparison between values of different enum types. This
3696 warning is enabled by default.
3698 @item -Wsign-compare
3699 @opindex Wsign-compare
3700 @opindex Wno-sign-compare
3701 @cindex warning for comparison of signed and unsigned values
3702 @cindex comparison of signed and unsigned values, warning
3703 @cindex signed and unsigned values, comparison warning
3704 Warn when a comparison between signed and unsigned values could produce
3705 an incorrect result when the signed value is converted to unsigned.
3706 This warning is also enabled by @option{-Wextra}; to get the other warnings
3707 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3709 @item -Wsign-conversion
3710 @opindex Wsign-conversion
3711 @opindex Wno-sign-conversion
3712 Warn for implicit conversions that may change the sign of an integer
3713 value, like assigning a signed integer expression to an unsigned
3714 integer variable. An explicit cast silences the warning. In C, this
3715 option is enabled also by @option{-Wconversion}.
3719 @opindex Wno-address
3720 Warn about suspicious uses of memory addresses. These include using
3721 the address of a function in a conditional expression, such as
3722 @code{void func(void); if (func)}, and comparisons against the memory
3723 address of a string literal, such as @code{if (x == "abc")}. Such
3724 uses typically indicate a programmer error: the address of a function
3725 always evaluates to true, so their use in a conditional usually
3726 indicate that the programmer forgot the parentheses in a function
3727 call; and comparisons against string literals result in unspecified
3728 behavior and are not portable in C, so they usually indicate that the
3729 programmer intended to use @code{strcmp}. This warning is enabled by
3733 @opindex Wlogical-op
3734 @opindex Wno-logical-op
3735 Warn about suspicious uses of logical operators in expressions.
3736 This includes using logical operators in contexts where a
3737 bit-wise operator is likely to be expected.
3739 @item -Waggregate-return
3740 @opindex Waggregate-return
3741 @opindex Wno-aggregate-return
3742 Warn if any functions that return structures or unions are defined or
3743 called. (In languages where you can return an array, this also elicits
3746 @item -Wno-attributes
3747 @opindex Wno-attributes
3748 @opindex Wattributes
3749 Do not warn if an unexpected @code{__attribute__} is used, such as
3750 unrecognized attributes, function attributes applied to variables,
3751 etc. This will not stop errors for incorrect use of supported
3754 @item -Wno-builtin-macro-redefined
3755 @opindex Wno-builtin-macro-redefined
3756 @opindex Wbuiltin-macro-redefined
3757 Do not warn if certain built-in macros are redefined. This suppresses
3758 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
3759 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
3761 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3762 @opindex Wstrict-prototypes
3763 @opindex Wno-strict-prototypes
3764 Warn if a function is declared or defined without specifying the
3765 argument types. (An old-style function definition is permitted without
3766 a warning if preceded by a declaration which specifies the argument
3769 @item -Wold-style-declaration @r{(C and Objective-C only)}
3770 @opindex Wold-style-declaration
3771 @opindex Wno-old-style-declaration
3772 Warn for obsolescent usages, according to the C Standard, in a
3773 declaration. For example, warn if storage-class specifiers like
3774 @code{static} are not the first things in a declaration. This warning
3775 is also enabled by @option{-Wextra}.
3777 @item -Wold-style-definition @r{(C and Objective-C only)}
3778 @opindex Wold-style-definition
3779 @opindex Wno-old-style-definition
3780 Warn if an old-style function definition is used. A warning is given
3781 even if there is a previous prototype.
3783 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3784 @opindex Wmissing-parameter-type
3785 @opindex Wno-missing-parameter-type
3786 A function parameter is declared without a type specifier in K&R-style
3793 This warning is also enabled by @option{-Wextra}.
3795 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3796 @opindex Wmissing-prototypes
3797 @opindex Wno-missing-prototypes
3798 Warn if a global function is defined without a previous prototype
3799 declaration. This warning is issued even if the definition itself
3800 provides a prototype. The aim is to detect global functions that fail
3801 to be declared in header files.
3803 @item -Wmissing-declarations
3804 @opindex Wmissing-declarations
3805 @opindex Wno-missing-declarations
3806 Warn if a global function is defined without a previous declaration.
3807 Do so even if the definition itself provides a prototype.
3808 Use this option to detect global functions that are not declared in
3809 header files. In C++, no warnings are issued for function templates,
3810 or for inline functions, or for functions in anonymous namespaces.
3812 @item -Wmissing-field-initializers
3813 @opindex Wmissing-field-initializers
3814 @opindex Wno-missing-field-initializers
3818 Warn if a structure's initializer has some fields missing. For
3819 example, the following code would cause such a warning, because
3820 @code{x.h} is implicitly zero:
3823 struct s @{ int f, g, h; @};
3824 struct s x = @{ 3, 4 @};
3827 This option does not warn about designated initializers, so the following
3828 modification would not trigger a warning:
3831 struct s @{ int f, g, h; @};
3832 struct s x = @{ .f = 3, .g = 4 @};
3835 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3836 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3838 @item -Wmissing-noreturn
3839 @opindex Wmissing-noreturn
3840 @opindex Wno-missing-noreturn
3841 Warn about functions which might be candidates for attribute @code{noreturn}.
3842 Note these are only possible candidates, not absolute ones. Care should
3843 be taken to manually verify functions actually do not ever return before
3844 adding the @code{noreturn} attribute, otherwise subtle code generation
3845 bugs could be introduced. You will not get a warning for @code{main} in
3846 hosted C environments.
3848 @item -Wmissing-format-attribute
3849 @opindex Wmissing-format-attribute
3850 @opindex Wno-missing-format-attribute
3853 Warn about function pointers which might be candidates for @code{format}
3854 attributes. Note these are only possible candidates, not absolute ones.
3855 GCC will guess that function pointers with @code{format} attributes that
3856 are used in assignment, initialization, parameter passing or return
3857 statements should have a corresponding @code{format} attribute in the
3858 resulting type. I.e.@: the left-hand side of the assignment or
3859 initialization, the type of the parameter variable, or the return type
3860 of the containing function respectively should also have a @code{format}
3861 attribute to avoid the warning.
3863 GCC will also warn about function definitions which might be
3864 candidates for @code{format} attributes. Again, these are only
3865 possible candidates. GCC will guess that @code{format} attributes
3866 might be appropriate for any function that calls a function like
3867 @code{vprintf} or @code{vscanf}, but this might not always be the
3868 case, and some functions for which @code{format} attributes are
3869 appropriate may not be detected.
3871 @item -Wno-multichar
3872 @opindex Wno-multichar
3874 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3875 Usually they indicate a typo in the user's code, as they have
3876 implementation-defined values, and should not be used in portable code.
3878 @item -Wnormalized=<none|id|nfc|nfkc>
3879 @opindex Wnormalized=
3882 @cindex character set, input normalization
3883 In ISO C and ISO C++, two identifiers are different if they are
3884 different sequences of characters. However, sometimes when characters
3885 outside the basic ASCII character set are used, you can have two
3886 different character sequences that look the same. To avoid confusion,
3887 the ISO 10646 standard sets out some @dfn{normalization rules} which
3888 when applied ensure that two sequences that look the same are turned into
3889 the same sequence. GCC can warn you if you are using identifiers which
3890 have not been normalized; this option controls that warning.
3892 There are four levels of warning that GCC supports. The default is
3893 @option{-Wnormalized=nfc}, which warns about any identifier which is
3894 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3895 recommended form for most uses.
3897 Unfortunately, there are some characters which ISO C and ISO C++ allow
3898 in identifiers that when turned into NFC aren't allowable as
3899 identifiers. That is, there's no way to use these symbols in portable
3900 ISO C or C++ and have all your identifiers in NFC@.
3901 @option{-Wnormalized=id} suppresses the warning for these characters.
3902 It is hoped that future versions of the standards involved will correct
3903 this, which is why this option is not the default.
3905 You can switch the warning off for all characters by writing
3906 @option{-Wnormalized=none}. You would only want to do this if you
3907 were using some other normalization scheme (like ``D''), because
3908 otherwise you can easily create bugs that are literally impossible to see.
3910 Some characters in ISO 10646 have distinct meanings but look identical
3911 in some fonts or display methodologies, especially once formatting has
3912 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3913 LETTER N'', will display just like a regular @code{n} which has been
3914 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3915 normalization scheme to convert all these into a standard form as
3916 well, and GCC will warn if your code is not in NFKC if you use
3917 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3918 about every identifier that contains the letter O because it might be
3919 confused with the digit 0, and so is not the default, but may be
3920 useful as a local coding convention if the programming environment is
3921 unable to be fixed to display these characters distinctly.
3923 @item -Wno-deprecated
3924 @opindex Wno-deprecated
3925 @opindex Wdeprecated
3926 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
3928 @item -Wno-deprecated-declarations
3929 @opindex Wno-deprecated-declarations
3930 @opindex Wdeprecated-declarations
3931 Do not warn about uses of functions (@pxref{Function Attributes}),
3932 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3933 Attributes}) marked as deprecated by using the @code{deprecated}
3937 @opindex Wno-overflow
3939 Do not warn about compile-time overflow in constant expressions.
3941 @item -Woverride-init @r{(C and Objective-C only)}
3942 @opindex Woverride-init
3943 @opindex Wno-override-init
3947 Warn if an initialized field without side effects is overridden when
3948 using designated initializers (@pxref{Designated Inits, , Designated
3951 This warning is included in @option{-Wextra}. To get other
3952 @option{-Wextra} warnings without this one, use @samp{-Wextra
3953 -Wno-override-init}.
3958 Warn if a structure is given the packed attribute, but the packed
3959 attribute has no effect on the layout or size of the structure.
3960 Such structures may be mis-aligned for little benefit. For
3961 instance, in this code, the variable @code{f.x} in @code{struct bar}
3962 will be misaligned even though @code{struct bar} does not itself
3963 have the packed attribute:
3970 @} __attribute__((packed));
3978 @item -Wpacked-bitfield-compat
3979 @opindex Wpacked-bitfield-compat
3980 @opindex Wno-packed-bitfield-compat
3981 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
3982 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
3983 the change can lead to differences in the structure layout. GCC
3984 informs you when the offset of such a field has changed in GCC 4.4.
3985 For example there is no longer a 4-bit padding between field @code{a}
3986 and @code{b} in this structure:
3993 @} __attribute__ ((packed));
3996 This warning is enabled by default. Use
3997 @option{-Wno-packed-bitfield-compat} to disable this warning.
4002 Warn if padding is included in a structure, either to align an element
4003 of the structure or to align the whole structure. Sometimes when this
4004 happens it is possible to rearrange the fields of the structure to
4005 reduce the padding and so make the structure smaller.
4007 @item -Wredundant-decls
4008 @opindex Wredundant-decls
4009 @opindex Wno-redundant-decls
4010 Warn if anything is declared more than once in the same scope, even in
4011 cases where multiple declaration is valid and changes nothing.
4013 @item -Wnested-externs @r{(C and Objective-C only)}
4014 @opindex Wnested-externs
4015 @opindex Wno-nested-externs
4016 Warn if an @code{extern} declaration is encountered within a function.
4018 @item -Wunreachable-code
4019 @opindex Wunreachable-code
4020 @opindex Wno-unreachable-code
4021 Warn if the compiler detects that code will never be executed.
4023 This option is intended to warn when the compiler detects that at
4024 least a whole line of source code will never be executed, because
4025 some condition is never satisfied or because it is after a
4026 procedure that never returns.
4028 It is possible for this option to produce a warning even though there
4029 are circumstances under which part of the affected line can be executed,
4030 so care should be taken when removing apparently-unreachable code.
4032 For instance, when a function is inlined, a warning may mean that the
4033 line is unreachable in only one inlined copy of the function.
4035 This option is not made part of @option{-Wall} because in a debugging
4036 version of a program there is often substantial code which checks
4037 correct functioning of the program and is, hopefully, unreachable
4038 because the program does work. Another common use of unreachable
4039 code is to provide behavior which is selectable at compile-time.
4044 Warn if a function can not be inlined and it was declared as inline.
4045 Even with this option, the compiler will not warn about failures to
4046 inline functions declared in system headers.
4048 The compiler uses a variety of heuristics to determine whether or not
4049 to inline a function. For example, the compiler takes into account
4050 the size of the function being inlined and the amount of inlining
4051 that has already been done in the current function. Therefore,
4052 seemingly insignificant changes in the source program can cause the
4053 warnings produced by @option{-Winline} to appear or disappear.
4055 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4056 @opindex Wno-invalid-offsetof
4057 @opindex Winvalid-offsetof
4058 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4059 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4060 to a non-POD type is undefined. In existing C++ implementations,
4061 however, @samp{offsetof} typically gives meaningful results even when
4062 applied to certain kinds of non-POD types. (Such as a simple
4063 @samp{struct} that fails to be a POD type only by virtue of having a
4064 constructor.) This flag is for users who are aware that they are
4065 writing nonportable code and who have deliberately chosen to ignore the
4068 The restrictions on @samp{offsetof} may be relaxed in a future version
4069 of the C++ standard.
4071 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4072 @opindex Wno-int-to-pointer-cast
4073 @opindex Wint-to-pointer-cast
4074 Suppress warnings from casts to pointer type of an integer of a
4077 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4078 @opindex Wno-pointer-to-int-cast
4079 @opindex Wpointer-to-int-cast
4080 Suppress warnings from casts from a pointer to an integer type of a
4084 @opindex Winvalid-pch
4085 @opindex Wno-invalid-pch
4086 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4087 the search path but can't be used.
4091 @opindex Wno-long-long
4092 Warn if @samp{long long} type is used. This is default. To inhibit
4093 the warning messages, use @option{-Wno-long-long}. Flags
4094 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
4095 only when @option{-pedantic} flag is used.
4097 @item -Wvariadic-macros
4098 @opindex Wvariadic-macros
4099 @opindex Wno-variadic-macros
4100 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4101 alternate syntax when in pedantic ISO C99 mode. This is default.
4102 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4107 Warn if variable length array is used in the code.
4108 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4109 the variable length array.
4111 @item -Wvolatile-register-var
4112 @opindex Wvolatile-register-var
4113 @opindex Wno-volatile-register-var
4114 Warn if a register variable is declared volatile. The volatile
4115 modifier does not inhibit all optimizations that may eliminate reads
4116 and/or writes to register variables. This warning is enabled by
4119 @item -Wdisabled-optimization
4120 @opindex Wdisabled-optimization
4121 @opindex Wno-disabled-optimization
4122 Warn if a requested optimization pass is disabled. This warning does
4123 not generally indicate that there is anything wrong with your code; it
4124 merely indicates that GCC's optimizers were unable to handle the code
4125 effectively. Often, the problem is that your code is too big or too
4126 complex; GCC will refuse to optimize programs when the optimization
4127 itself is likely to take inordinate amounts of time.
4129 @item -Wpointer-sign @r{(C and Objective-C only)}
4130 @opindex Wpointer-sign
4131 @opindex Wno-pointer-sign
4132 Warn for pointer argument passing or assignment with different signedness.
4133 This option is only supported for C and Objective-C@. It is implied by
4134 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4135 @option{-Wno-pointer-sign}.
4137 @item -Wstack-protector
4138 @opindex Wstack-protector
4139 @opindex Wno-stack-protector
4140 This option is only active when @option{-fstack-protector} is active. It
4141 warns about functions that will not be protected against stack smashing.
4144 @opindex Wno-mudflap
4145 Suppress warnings about constructs that cannot be instrumented by
4148 @item -Woverlength-strings
4149 @opindex Woverlength-strings
4150 @opindex Wno-overlength-strings
4151 Warn about string constants which are longer than the ``minimum
4152 maximum'' length specified in the C standard. Modern compilers
4153 generally allow string constants which are much longer than the
4154 standard's minimum limit, but very portable programs should avoid
4155 using longer strings.
4157 The limit applies @emph{after} string constant concatenation, and does
4158 not count the trailing NUL@. In C89, the limit was 509 characters; in
4159 C99, it was raised to 4095. C++98 does not specify a normative
4160 minimum maximum, so we do not diagnose overlength strings in C++@.
4162 This option is implied by @option{-pedantic}, and can be disabled with
4163 @option{-Wno-overlength-strings}.
4165 @item -Wdisallowed-function-list=@var{sym},@var{sym},@dots{}
4166 @opindex Wdisallowed-function-list
4168 If any of @var{sym} is called, GCC will issue a warning. This can be useful
4169 in enforcing coding conventions that ban calls to certain functions, for
4170 example, @code{alloca}, @code{malloc}, etc.
4173 @node Debugging Options
4174 @section Options for Debugging Your Program or GCC
4175 @cindex options, debugging
4176 @cindex debugging information options
4178 GCC has various special options that are used for debugging
4179 either your program or GCC:
4184 Produce debugging information in the operating system's native format
4185 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4188 On most systems that use stabs format, @option{-g} enables use of extra
4189 debugging information that only GDB can use; this extra information
4190 makes debugging work better in GDB but will probably make other debuggers
4192 refuse to read the program. If you want to control for certain whether
4193 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4194 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4196 GCC allows you to use @option{-g} with
4197 @option{-O}. The shortcuts taken by optimized code may occasionally
4198 produce surprising results: some variables you declared may not exist
4199 at all; flow of control may briefly move where you did not expect it;
4200 some statements may not be executed because they compute constant
4201 results or their values were already at hand; some statements may
4202 execute in different places because they were moved out of loops.
4204 Nevertheless it proves possible to debug optimized output. This makes
4205 it reasonable to use the optimizer for programs that might have bugs.
4207 The following options are useful when GCC is generated with the
4208 capability for more than one debugging format.
4212 Produce debugging information for use by GDB@. This means to use the
4213 most expressive format available (DWARF 2, stabs, or the native format
4214 if neither of those are supported), including GDB extensions if at all
4219 Produce debugging information in stabs format (if that is supported),
4220 without GDB extensions. This is the format used by DBX on most BSD
4221 systems. On MIPS, Alpha and System V Release 4 systems this option
4222 produces stabs debugging output which is not understood by DBX or SDB@.
4223 On System V Release 4 systems this option requires the GNU assembler.
4225 @item -feliminate-unused-debug-symbols
4226 @opindex feliminate-unused-debug-symbols
4227 Produce debugging information in stabs format (if that is supported),
4228 for only symbols that are actually used.
4230 @item -femit-class-debug-always
4231 Instead of emitting debugging information for a C++ class in only one
4232 object file, emit it in all object files using the class. This option
4233 should be used only with debuggers that are unable to handle the way GCC
4234 normally emits debugging information for classes because using this
4235 option will increase the size of debugging information by as much as a
4240 Produce debugging information in stabs format (if that is supported),
4241 using GNU extensions understood only by the GNU debugger (GDB)@. The
4242 use of these extensions is likely to make other debuggers crash or
4243 refuse to read the program.
4247 Produce debugging information in COFF format (if that is supported).
4248 This is the format used by SDB on most System V systems prior to
4253 Produce debugging information in XCOFF format (if that is supported).
4254 This is the format used by the DBX debugger on IBM RS/6000 systems.
4258 Produce debugging information in XCOFF format (if that is supported),
4259 using GNU extensions understood only by the GNU debugger (GDB)@. The
4260 use of these extensions is likely to make other debuggers crash or
4261 refuse to read the program, and may cause assemblers other than the GNU
4262 assembler (GAS) to fail with an error.
4266 Produce debugging information in DWARF version 2 format (if that is
4267 supported). This is the format used by DBX on IRIX 6. With this
4268 option, GCC uses features of DWARF version 3 when they are useful;
4269 version 3 is upward compatible with version 2, but may still cause
4270 problems for older debuggers.
4274 Produce debugging information in VMS debug format (if that is
4275 supported). This is the format used by DEBUG on VMS systems.
4278 @itemx -ggdb@var{level}
4279 @itemx -gstabs@var{level}
4280 @itemx -gcoff@var{level}
4281 @itemx -gxcoff@var{level}
4282 @itemx -gvms@var{level}
4283 Request debugging information and also use @var{level} to specify how
4284 much information. The default level is 2.
4286 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4289 Level 1 produces minimal information, enough for making backtraces in
4290 parts of the program that you don't plan to debug. This includes
4291 descriptions of functions and external variables, but no information
4292 about local variables and no line numbers.
4294 Level 3 includes extra information, such as all the macro definitions
4295 present in the program. Some debuggers support macro expansion when
4296 you use @option{-g3}.
4298 @option{-gdwarf-2} does not accept a concatenated debug level, because
4299 GCC used to support an option @option{-gdwarf} that meant to generate
4300 debug information in version 1 of the DWARF format (which is very
4301 different from version 2), and it would have been too confusing. That
4302 debug format is long obsolete, but the option cannot be changed now.
4303 Instead use an additional @option{-g@var{level}} option to change the
4304 debug level for DWARF2.
4306 @item -feliminate-dwarf2-dups
4307 @opindex feliminate-dwarf2-dups
4308 Compress DWARF2 debugging information by eliminating duplicated
4309 information about each symbol. This option only makes sense when
4310 generating DWARF2 debugging information with @option{-gdwarf-2}.
4312 @item -femit-struct-debug-baseonly
4313 Emit debug information for struct-like types
4314 only when the base name of the compilation source file
4315 matches the base name of file in which the struct was defined.
4317 This option substantially reduces the size of debugging information,
4318 but at significant potential loss in type information to the debugger.
4319 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4320 See @option{-femit-struct-debug-detailed} for more detailed control.
4322 This option works only with DWARF 2.
4324 @item -femit-struct-debug-reduced
4325 Emit debug information for struct-like types
4326 only when the base name of the compilation source file
4327 matches the base name of file in which the type was defined,
4328 unless the struct is a template or defined in a system header.
4330 This option significantly reduces the size of debugging information,
4331 with some potential loss in type information to the debugger.
4332 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4333 See @option{-femit-struct-debug-detailed} for more detailed control.
4335 This option works only with DWARF 2.
4337 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4338 Specify the struct-like types
4339 for which the compiler will generate debug information.
4340 The intent is to reduce duplicate struct debug information
4341 between different object files within the same program.
4343 This option is a detailed version of
4344 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4345 which will serve for most needs.
4347 A specification has the syntax
4348 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4350 The optional first word limits the specification to
4351 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4352 A struct type is used directly when it is the type of a variable, member.
4353 Indirect uses arise through pointers to structs.
4354 That is, when use of an incomplete struct would be legal, the use is indirect.
4356 @samp{struct one direct; struct two * indirect;}.
4358 The optional second word limits the specification to
4359 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4360 Generic structs are a bit complicated to explain.
4361 For C++, these are non-explicit specializations of template classes,
4362 or non-template classes within the above.
4363 Other programming languages have generics,
4364 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4366 The third word specifies the source files for those
4367 structs for which the compiler will emit debug information.
4368 The values @samp{none} and @samp{any} have the normal meaning.
4369 The value @samp{base} means that
4370 the base of name of the file in which the type declaration appears
4371 must match the base of the name of the main compilation file.
4372 In practice, this means that
4373 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4374 but types declared in other header will not.
4375 The value @samp{sys} means those types satisfying @samp{base}
4376 or declared in system or compiler headers.
4378 You may need to experiment to determine the best settings for your application.
4380 The default is @samp{-femit-struct-debug-detailed=all}.
4382 This option works only with DWARF 2.
4384 @item -fno-merge-debug-strings
4385 @opindex fmerge-debug-strings
4386 @opindex fno-merge-debug-strings
4387 Direct the linker to merge together strings which are identical in
4388 different object files. This is not supported by all assemblers or
4389 linker. This decreases the size of the debug information in the
4390 output file at the cost of increasing link processing time. This is
4393 @item -fdebug-prefix-map=@var{old}=@var{new}
4394 @opindex fdebug-prefix-map
4395 When compiling files in directory @file{@var{old}}, record debugging
4396 information describing them as in @file{@var{new}} instead.
4398 @item -fno-dwarf2-cfi-asm
4399 @opindex fdwarf2-cfi-asm
4400 @opindex fno-dwarf2-cfi-asm
4401 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4402 instead of using GAS @code{.cfi_*} directives.
4404 @cindex @command{prof}
4407 Generate extra code to write profile information suitable for the
4408 analysis program @command{prof}. You must use this option when compiling
4409 the source files you want data about, and you must also use it when
4412 @cindex @command{gprof}
4415 Generate extra code to write profile information suitable for the
4416 analysis program @command{gprof}. You must use this option when compiling
4417 the source files you want data about, and you must also use it when
4422 Makes the compiler print out each function name as it is compiled, and
4423 print some statistics about each pass when it finishes.
4426 @opindex ftime-report
4427 Makes the compiler print some statistics about the time consumed by each
4428 pass when it finishes.
4431 @opindex fmem-report
4432 Makes the compiler print some statistics about permanent memory
4433 allocation when it finishes.
4435 @item -fpre-ipa-mem-report
4436 @opindex fpre-ipa-mem-report
4437 @item -fpost-ipa-mem-report
4438 @opindex fpost-ipa-mem-report
4439 Makes the compiler print some statistics about permanent memory
4440 allocation before or after interprocedural optimization.
4442 @item -fprofile-arcs
4443 @opindex fprofile-arcs
4444 Add code so that program flow @dfn{arcs} are instrumented. During
4445 execution the program records how many times each branch and call is
4446 executed and how many times it is taken or returns. When the compiled
4447 program exits it saves this data to a file called
4448 @file{@var{auxname}.gcda} for each source file. The data may be used for
4449 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4450 test coverage analysis (@option{-ftest-coverage}). Each object file's
4451 @var{auxname} is generated from the name of the output file, if
4452 explicitly specified and it is not the final executable, otherwise it is
4453 the basename of the source file. In both cases any suffix is removed
4454 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4455 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4456 @xref{Cross-profiling}.
4458 @cindex @command{gcov}
4462 This option is used to compile and link code instrumented for coverage
4463 analysis. The option is a synonym for @option{-fprofile-arcs}
4464 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4465 linking). See the documentation for those options for more details.
4470 Compile the source files with @option{-fprofile-arcs} plus optimization
4471 and code generation options. For test coverage analysis, use the
4472 additional @option{-ftest-coverage} option. You do not need to profile
4473 every source file in a program.
4476 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4477 (the latter implies the former).
4480 Run the program on a representative workload to generate the arc profile
4481 information. This may be repeated any number of times. You can run
4482 concurrent instances of your program, and provided that the file system
4483 supports locking, the data files will be correctly updated. Also
4484 @code{fork} calls are detected and correctly handled (double counting
4488 For profile-directed optimizations, compile the source files again with
4489 the same optimization and code generation options plus
4490 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4491 Control Optimization}).
4494 For test coverage analysis, use @command{gcov} to produce human readable
4495 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4496 @command{gcov} documentation for further information.
4500 With @option{-fprofile-arcs}, for each function of your program GCC
4501 creates a program flow graph, then finds a spanning tree for the graph.
4502 Only arcs that are not on the spanning tree have to be instrumented: the
4503 compiler adds code to count the number of times that these arcs are
4504 executed. When an arc is the only exit or only entrance to a block, the
4505 instrumentation code can be added to the block; otherwise, a new basic
4506 block must be created to hold the instrumentation code.
4509 @item -ftest-coverage
4510 @opindex ftest-coverage
4511 Produce a notes file that the @command{gcov} code-coverage utility
4512 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4513 show program coverage. Each source file's note file is called
4514 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4515 above for a description of @var{auxname} and instructions on how to
4516 generate test coverage data. Coverage data will match the source files
4517 more closely, if you do not optimize.
4519 @item -fdbg-cnt-list
4520 @opindex fdbg-cnt-list
4521 Print the name and the counter upperbound for all debug counters.
4523 @item -fdbg-cnt=@var{counter-value-list}
4525 Set the internal debug counter upperbound. @var{counter-value-list}
4526 is a comma-separated list of @var{name}:@var{value} pairs
4527 which sets the upperbound of each debug counter @var{name} to @var{value}.
4528 All debug counters have the initial upperbound of @var{UINT_MAX},
4529 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4530 e.g. With -fdbg-cnt=dce:10,tail_call:0
4531 dbg_cnt(dce) will return true only for first 10 invocations
4532 and dbg_cnt(tail_call) will return false always.
4534 @item -d@var{letters}
4535 @itemx -fdump-rtl-@var{pass}
4537 Says to make debugging dumps during compilation at times specified by
4538 @var{letters}. This is used for debugging the RTL-based passes of the
4539 compiler. The file names for most of the dumps are made by appending a
4540 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4541 from the name of the output file, if explicitly specified and it is not
4542 an executable, otherwise it is the basename of the source file. These
4543 switches may have different effects when @option{-E} is used for
4546 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4547 @option{-d} option @var{letters}. Here are the possible
4548 letters for use in @var{pass} and @var{letters}, and their meanings:
4552 @item -fdump-rtl-alignments
4553 @opindex fdump-rtl-alignments
4554 Dump after branch alignments have been computed.
4556 @item -fdump-rtl-asmcons
4557 @opindex fdump-rtl-asmcons
4558 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4560 @item -fdump-rtl-auto_inc_dec
4561 @opindex fdump-rtl-auto_inc_dec
4562 Dump after auto-inc-dec discovery. This pass is only run on
4563 architectures that have auto inc or auto dec instructions.
4565 @item -fdump-rtl-barriers
4566 @opindex fdump-rtl-barriers
4567 Dump after cleaning up the barrier instructions.
4569 @item -fdump-rtl-bbpart
4570 @opindex fdump-rtl-bbpart
4571 Dump after partitioning hot and cold basic blocks.
4573 @item -fdump-rtl-bbro
4574 @opindex fdump-rtl-bbro
4575 Dump after block reordering.
4577 @item -fdump-rtl-btl1
4578 @itemx -fdump-rtl-btl2
4579 @opindex fdump-rtl-btl2
4580 @opindex fdump-rtl-btl2
4581 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4582 after the two branch
4583 target load optimization passes.
4585 @item -fdump-rtl-bypass
4586 @opindex fdump-rtl-bypass
4587 Dump after jump bypassing and control flow optimizations.
4589 @item -fdump-rtl-combine
4590 @opindex fdump-rtl-combine
4591 Dump after the RTL instruction combination pass.
4593 @item -fdump-rtl-compgotos
4594 @opindex fdump-rtl-compgotos
4595 Dump after dumplicating the computed gotos.
4597 @item -fdump-rtl-ce1
4598 @itemx -fdump-rtl-ce2
4599 @itemx -fdump-rtl-ce3
4600 @opindex fdump-rtl-ce1
4601 @opindex fdump-rtl-ce2
4602 @opindex fdump-rtl-ce3
4603 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4604 @option{-fdump-rtl-ce3} enable dumping after the three
4605 if conversion passes.
4607 @itemx -fdump-rtl-cprop_hardreg
4608 @opindex fdump-rtl-cprop_hardreg
4609 Dump after hard register copy propagation.
4611 @itemx -fdump-rtl-csa
4612 @opindex fdump-rtl-csa
4613 Dump after combining stack adjustments.
4615 @item -fdump-rtl-cse1
4616 @itemx -fdump-rtl-cse2
4617 @opindex fdump-rtl-cse1
4618 @opindex fdump-rtl-cse2
4619 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4620 the two common sub-expression elimination passes.
4622 @itemx -fdump-rtl-dce
4623 @opindex fdump-rtl-dce
4624 Dump after the standalone dead code elimination passes.
4626 @itemx -fdump-rtl-dbr
4627 @opindex fdump-rtl-dbr
4628 Dump after delayed branch scheduling.
4630 @item -fdump-rtl-dce1
4631 @itemx -fdump-rtl-dce2
4632 @opindex fdump-rtl-dce1
4633 @opindex fdump-rtl-dce2
4634 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
4635 the two dead store elimination passes.
4638 @opindex fdump-rtl-eh
4639 Dump after finalization of EH handling code.
4641 @item -fdump-rtl-eh_ranges
4642 @opindex fdump-rtl-eh_ranges
4643 Dump after conversion of EH handling range regions.
4645 @item -fdump-rtl-expand
4646 @opindex fdump-rtl-expand
4647 Dump after RTL generation.
4649 @item -fdump-rtl-fwprop1
4650 @itemx -fdump-rtl-fwprop2
4651 @opindex fdump-rtl-fwprop1
4652 @opindex fdump-rtl-fwprop2
4653 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
4654 dumping after the two forward propagation passes.
4656 @item -fdump-rtl-gcse1
4657 @itemx -fdump-rtl-gcse2
4658 @opindex fdump-rtl-gcse1
4659 @opindex fdump-rtl-gcse2
4660 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
4661 after global common subexpression elimination.
4663 @item -fdump-rtl-init-regs
4664 @opindex fdump-rtl-init-regs
4665 Dump after the initialization of the registers.
4667 @item -fdump-rtl-initvals
4668 @opindex fdump-rtl-initvals
4669 Dump after the computation of the initial value sets.
4671 @itemx -fdump-rtl-into_cfglayout
4672 @opindex fdump-rtl-into_cfglayout
4673 Dump after converting to cfglayout mode.
4675 @item -fdump-rtl-ira
4676 @opindex fdump-rtl-ira
4677 Dump after iterated register allocation.
4679 @item -fdump-rtl-jump
4680 @opindex fdump-rtl-jump
4681 Dump after the second jump optimization.
4683 @item -fdump-rtl-loop2
4684 @opindex fdump-rtl-loop2
4685 @option{-fdump-rtl-loop2} enables dumping after the rtl
4686 loop optimization passes.
4688 @item -fdump-rtl-mach
4689 @opindex fdump-rtl-mach
4690 Dump after performing the machine dependent reorganization pass, if that
4693 @item -fdump-rtl-mode_sw
4694 @opindex fdump-rtl-mode_sw
4695 Dump after removing redundant mode switches.
4697 @item -fdump-rtl-rnreg
4698 @opindex fdump-rtl-rnreg
4699 Dump after register renumbering.
4701 @itemx -fdump-rtl-outof_cfglayout
4702 @opindex fdump-rtl-outof_cfglayout
4703 Dump after converting from cfglayout mode.
4705 @item -fdump-rtl-peephole2
4706 @opindex fdump-rtl-peephole2
4707 Dump after the peephole pass.
4709 @item -fdump-rtl-postreload
4710 @opindex fdump-rtl-postreload
4711 Dump after post-reload optimizations.
4713 @itemx -fdump-rtl-pro_and_epilogue
4714 @opindex fdump-rtl-pro_and_epilogue
4715 Dump after generating the function pro and epilogues.
4717 @item -fdump-rtl-regmove
4718 @opindex fdump-rtl-regmove
4719 Dump after the register move pass.
4721 @item -fdump-rtl-sched1
4722 @itemx -fdump-rtl-sched2
4723 @opindex fdump-rtl-sched1
4724 @opindex fdump-rtl-sched2
4725 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
4726 after the basic block scheduling passes.
4728 @item -fdump-rtl-see
4729 @opindex fdump-rtl-see
4730 Dump after sign extension elimination.
4732 @item -fdump-rtl-seqabstr
4733 @opindex fdump-rtl-seqabstr
4734 Dump after common sequence discovery.
4736 @item -fdump-rtl-shorten
4737 @opindex fdump-rtl-shorten
4738 Dump after shortening branches.
4740 @item -fdump-rtl-sibling
4741 @opindex fdump-rtl-sibling
4742 Dump after sibling call optimizations.
4744 @item -fdump-rtl-split1
4745 @itemx -fdump-rtl-split2
4746 @itemx -fdump-rtl-split3
4747 @itemx -fdump-rtl-split4
4748 @itemx -fdump-rtl-split5
4749 @opindex fdump-rtl-split1
4750 @opindex fdump-rtl-split2
4751 @opindex fdump-rtl-split3
4752 @opindex fdump-rtl-split4
4753 @opindex fdump-rtl-split5
4754 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
4755 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
4756 @option{-fdump-rtl-split5} enable dumping after five rounds of
4757 instruction splitting.
4759 @item -fdump-rtl-sms
4760 @opindex fdump-rtl-sms
4761 Dump after modulo scheduling. This pass is only run on some
4764 @item -fdump-rtl-stack
4765 @opindex fdump-rtl-stack
4766 Dump after conversion from GCC's "flat register file" registers to the
4767 x87's stack-like registers. This pass is only run on x86 variants.
4769 @item -fdump-rtl-subreg1
4770 @itemx -fdump-rtl-subreg2
4771 @opindex fdump-rtl-subreg1
4772 @opindex fdump-rtl-subreg2
4773 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
4774 the two subreg expansion passes.
4776 @item -fdump-rtl-unshare
4777 @opindex fdump-rtl-unshare
4778 Dump after all rtl has been unshared.
4780 @item -fdump-rtl-vartrack
4781 @opindex fdump-rtl-vartrack
4782 Dump after variable tracking.
4784 @item -fdump-rtl-vregs
4785 @opindex fdump-rtl-vregs
4786 Dump after converting virtual registers to hard registers.
4788 @item -fdump-rtl-web
4789 @opindex fdump-rtl-web
4790 Dump after live range splitting.
4792 @item -fdump-rtl-regclass
4793 @itemx -fdump-rtl-subregs_of_mode_init
4794 @itemx -fdump-rtl-subregs_of_mode_finish
4795 @itemx -fdump-rtl-dfinit
4796 @itemx -fdump-rtl-dfinish
4797 @opindex fdump-rtl-regclass
4798 @opindex fdump-rtl-subregs_of_mode_init
4799 @opindex fdump-rtl-subregs_of_mode_finish
4800 @opindex fdump-rtl-dfinit
4801 @opindex fdump-rtl-dfinish
4802 These dumps are defined but always produce empty files.
4804 @item -fdump-rtl-all
4805 @opindex fdump-rtl-all
4806 Produce all the dumps listed above.
4810 Annotate the assembler output with miscellaneous debugging information.
4814 Dump all macro definitions, at the end of preprocessing, in addition to
4819 Produce a core dump whenever an error occurs.
4823 Print statistics on memory usage, at the end of the run, to
4828 Annotate the assembler output with a comment indicating which
4829 pattern and alternative was used. The length of each instruction is
4834 Dump the RTL in the assembler output as a comment before each instruction.
4835 Also turns on @option{-dp} annotation.
4839 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
4840 dump a representation of the control flow graph suitable for viewing with VCG
4841 to @file{@var{file}.@var{pass}.vcg}.
4845 Just generate RTL for a function instead of compiling it. Usually used
4846 with @option{-fdump-rtl-expand}.
4850 Dump debugging information during parsing, to standard error.
4854 @opindex fdump-noaddr
4855 When doing debugging dumps, suppress address output. This makes it more
4856 feasible to use diff on debugging dumps for compiler invocations with
4857 different compiler binaries and/or different
4858 text / bss / data / heap / stack / dso start locations.
4860 @item -fdump-unnumbered
4861 @opindex fdump-unnumbered
4862 When doing debugging dumps, suppress instruction numbers and address output.
4863 This makes it more feasible to use diff on debugging dumps for compiler
4864 invocations with different options, in particular with and without
4867 @item -fdump-translation-unit @r{(C++ only)}
4868 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4869 @opindex fdump-translation-unit
4870 Dump a representation of the tree structure for the entire translation
4871 unit to a file. The file name is made by appending @file{.tu} to the
4872 source file name. If the @samp{-@var{options}} form is used, @var{options}
4873 controls the details of the dump as described for the
4874 @option{-fdump-tree} options.
4876 @item -fdump-class-hierarchy @r{(C++ only)}
4877 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4878 @opindex fdump-class-hierarchy
4879 Dump a representation of each class's hierarchy and virtual function
4880 table layout to a file. The file name is made by appending @file{.class}
4881 to the source file name. If the @samp{-@var{options}} form is used,
4882 @var{options} controls the details of the dump as described for the
4883 @option{-fdump-tree} options.
4885 @item -fdump-ipa-@var{switch}
4887 Control the dumping at various stages of inter-procedural analysis
4888 language tree to a file. The file name is generated by appending a switch
4889 specific suffix to the source file name. The following dumps are possible:
4893 Enables all inter-procedural analysis dumps.
4896 Dumps information about call-graph optimization, unused function removal,
4897 and inlining decisions.
4900 Dump after function inlining.
4904 @item -fdump-statistics-@var{option}
4905 @opindex -fdump-statistics
4906 Enable and control dumping of pass statistics in a separate file. The
4907 file name is generated by appending a suffix ending in @samp{.statistics}
4908 to the source file name. If the @samp{-@var{option}} form is used,
4909 @samp{-stats} will cause counters to be summed over the whole compilation unit
4910 while @samp{-details} will dump every event as the passes generate them.
4911 The default with no option is to sum counters for each function compiled.
4913 @item -fdump-tree-@var{switch}
4914 @itemx -fdump-tree-@var{switch}-@var{options}
4916 Control the dumping at various stages of processing the intermediate
4917 language tree to a file. The file name is generated by appending a switch
4918 specific suffix to the source file name. If the @samp{-@var{options}}
4919 form is used, @var{options} is a list of @samp{-} separated options that
4920 control the details of the dump. Not all options are applicable to all
4921 dumps, those which are not meaningful will be ignored. The following
4922 options are available
4926 Print the address of each node. Usually this is not meaningful as it
4927 changes according to the environment and source file. Its primary use
4928 is for tying up a dump file with a debug environment.
4930 Inhibit dumping of members of a scope or body of a function merely
4931 because that scope has been reached. Only dump such items when they
4932 are directly reachable by some other path. When dumping pretty-printed
4933 trees, this option inhibits dumping the bodies of control structures.
4935 Print a raw representation of the tree. By default, trees are
4936 pretty-printed into a C-like representation.
4938 Enable more detailed dumps (not honored by every dump option).
4940 Enable dumping various statistics about the pass (not honored by every dump
4943 Enable showing basic block boundaries (disabled in raw dumps).
4945 Enable showing virtual operands for every statement.
4947 Enable showing line numbers for statements.
4949 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4951 Enable showing the tree dump for each statement.
4953 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
4954 and @option{lineno}.
4957 The following tree dumps are possible:
4961 Dump before any tree based optimization, to @file{@var{file}.original}.
4964 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4967 @opindex fdump-tree-gimple
4968 Dump each function before and after the gimplification pass to a file. The
4969 file name is made by appending @file{.gimple} to the source file name.
4972 @opindex fdump-tree-cfg
4973 Dump the control flow graph of each function to a file. The file name is
4974 made by appending @file{.cfg} to the source file name.
4977 @opindex fdump-tree-vcg
4978 Dump the control flow graph of each function to a file in VCG format. The
4979 file name is made by appending @file{.vcg} to the source file name. Note
4980 that if the file contains more than one function, the generated file cannot
4981 be used directly by VCG@. You will need to cut and paste each function's
4982 graph into its own separate file first.
4985 @opindex fdump-tree-ch
4986 Dump each function after copying loop headers. The file name is made by
4987 appending @file{.ch} to the source file name.
4990 @opindex fdump-tree-ssa
4991 Dump SSA related information to a file. The file name is made by appending
4992 @file{.ssa} to the source file name.
4995 @opindex fdump-tree-alias
4996 Dump aliasing information for each function. The file name is made by
4997 appending @file{.alias} to the source file name.
5000 @opindex fdump-tree-ccp
5001 Dump each function after CCP@. The file name is made by appending
5002 @file{.ccp} to the source file name.
5005 @opindex fdump-tree-storeccp
5006 Dump each function after STORE-CCP@. The file name is made by appending
5007 @file{.storeccp} to the source file name.
5010 @opindex fdump-tree-pre
5011 Dump trees after partial redundancy elimination. The file name is made
5012 by appending @file{.pre} to the source file name.
5015 @opindex fdump-tree-fre
5016 Dump trees after full redundancy elimination. The file name is made
5017 by appending @file{.fre} to the source file name.
5020 @opindex fdump-tree-copyprop
5021 Dump trees after copy propagation. The file name is made
5022 by appending @file{.copyprop} to the source file name.
5024 @item store_copyprop
5025 @opindex fdump-tree-store_copyprop
5026 Dump trees after store copy-propagation. The file name is made
5027 by appending @file{.store_copyprop} to the source file name.
5030 @opindex fdump-tree-dce
5031 Dump each function after dead code elimination. The file name is made by
5032 appending @file{.dce} to the source file name.
5035 @opindex fdump-tree-mudflap
5036 Dump each function after adding mudflap instrumentation. The file name is
5037 made by appending @file{.mudflap} to the source file name.
5040 @opindex fdump-tree-sra
5041 Dump each function after performing scalar replacement of aggregates. The
5042 file name is made by appending @file{.sra} to the source file name.
5045 @opindex fdump-tree-sink
5046 Dump each function after performing code sinking. The file name is made
5047 by appending @file{.sink} to the source file name.
5050 @opindex fdump-tree-dom
5051 Dump each function after applying dominator tree optimizations. The file
5052 name is made by appending @file{.dom} to the source file name.
5055 @opindex fdump-tree-dse
5056 Dump each function after applying dead store elimination. The file
5057 name is made by appending @file{.dse} to the source file name.
5060 @opindex fdump-tree-phiopt
5061 Dump each function after optimizing PHI nodes into straightline code. The file
5062 name is made by appending @file{.phiopt} to the source file name.
5065 @opindex fdump-tree-forwprop
5066 Dump each function after forward propagating single use variables. The file
5067 name is made by appending @file{.forwprop} to the source file name.
5070 @opindex fdump-tree-copyrename
5071 Dump each function after applying the copy rename optimization. The file
5072 name is made by appending @file{.copyrename} to the source file name.
5075 @opindex fdump-tree-nrv
5076 Dump each function after applying the named return value optimization on
5077 generic trees. The file name is made by appending @file{.nrv} to the source
5081 @opindex fdump-tree-vect
5082 Dump each function after applying vectorization of loops. The file name is
5083 made by appending @file{.vect} to the source file name.
5086 @opindex fdump-tree-vrp
5087 Dump each function after Value Range Propagation (VRP). The file name
5088 is made by appending @file{.vrp} to the source file name.
5091 @opindex fdump-tree-all
5092 Enable all the available tree dumps with the flags provided in this option.
5095 @item -ftree-vectorizer-verbose=@var{n}
5096 @opindex ftree-vectorizer-verbose
5097 This option controls the amount of debugging output the vectorizer prints.
5098 This information is written to standard error, unless
5099 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5100 in which case it is output to the usual dump listing file, @file{.vect}.
5101 For @var{n}=0 no diagnostic information is reported.
5102 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5103 and the total number of loops that got vectorized.
5104 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5105 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5106 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5107 level that @option{-fdump-tree-vect-stats} uses.
5108 Higher verbosity levels mean either more information dumped for each
5109 reported loop, or same amount of information reported for more loops:
5110 If @var{n}=3, alignment related information is added to the reports.
5111 If @var{n}=4, data-references related information (e.g.@: memory dependences,
5112 memory access-patterns) is added to the reports.
5113 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
5114 that did not pass the first analysis phase (i.e., may not be countable, or
5115 may have complicated control-flow).
5116 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
5117 For @var{n}=7, all the information the vectorizer generates during its
5118 analysis and transformation is reported. This is the same verbosity level
5119 that @option{-fdump-tree-vect-details} uses.
5121 @item -frandom-seed=@var{string}
5122 @opindex frandom-string
5123 This option provides a seed that GCC uses when it would otherwise use
5124 random numbers. It is used to generate certain symbol names
5125 that have to be different in every compiled file. It is also used to
5126 place unique stamps in coverage data files and the object files that
5127 produce them. You can use the @option{-frandom-seed} option to produce
5128 reproducibly identical object files.
5130 The @var{string} should be different for every file you compile.
5132 @item -fsched-verbose=@var{n}
5133 @opindex fsched-verbose
5134 On targets that use instruction scheduling, this option controls the
5135 amount of debugging output the scheduler prints. This information is
5136 written to standard error, unless @option{-fdump-rtl-sched1} or
5137 @option{-fdump-rtl-sched2} is specified, in which case it is output
5138 to the usual dump listing file, @file{.sched} or @file{.sched2}
5139 respectively. However for @var{n} greater than nine, the output is
5140 always printed to standard error.
5142 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5143 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5144 For @var{n} greater than one, it also output basic block probabilities,
5145 detailed ready list information and unit/insn info. For @var{n} greater
5146 than two, it includes RTL at abort point, control-flow and regions info.
5147 And for @var{n} over four, @option{-fsched-verbose} also includes
5152 Store the usual ``temporary'' intermediate files permanently; place them
5153 in the current directory and name them based on the source file. Thus,
5154 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5155 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5156 preprocessed @file{foo.i} output file even though the compiler now
5157 normally uses an integrated preprocessor.
5159 When used in combination with the @option{-x} command line option,
5160 @option{-save-temps} is sensible enough to avoid over writing an
5161 input source file with the same extension as an intermediate file.
5162 The corresponding intermediate file may be obtained by renaming the
5163 source file before using @option{-save-temps}.
5167 Report the CPU time taken by each subprocess in the compilation
5168 sequence. For C source files, this is the compiler proper and assembler
5169 (plus the linker if linking is done). The output looks like this:
5176 The first number on each line is the ``user time'', that is time spent
5177 executing the program itself. The second number is ``system time'',
5178 time spent executing operating system routines on behalf of the program.
5179 Both numbers are in seconds.
5181 @item -fvar-tracking
5182 @opindex fvar-tracking
5183 Run variable tracking pass. It computes where variables are stored at each
5184 position in code. Better debugging information is then generated
5185 (if the debugging information format supports this information).
5187 It is enabled by default when compiling with optimization (@option{-Os},
5188 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5189 the debug info format supports it.
5191 @item -print-file-name=@var{library}
5192 @opindex print-file-name
5193 Print the full absolute name of the library file @var{library} that
5194 would be used when linking---and don't do anything else. With this
5195 option, GCC does not compile or link anything; it just prints the
5198 @item -print-multi-directory
5199 @opindex print-multi-directory
5200 Print the directory name corresponding to the multilib selected by any
5201 other switches present in the command line. This directory is supposed
5202 to exist in @env{GCC_EXEC_PREFIX}.
5204 @item -print-multi-lib
5205 @opindex print-multi-lib
5206 Print the mapping from multilib directory names to compiler switches
5207 that enable them. The directory name is separated from the switches by
5208 @samp{;}, and each switch starts with an @samp{@@} instead of the
5209 @samp{-}, without spaces between multiple switches. This is supposed to
5210 ease shell-processing.
5212 @item -print-prog-name=@var{program}
5213 @opindex print-prog-name
5214 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5216 @item -print-libgcc-file-name
5217 @opindex print-libgcc-file-name
5218 Same as @option{-print-file-name=libgcc.a}.
5220 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5221 but you do want to link with @file{libgcc.a}. You can do
5224 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5227 @item -print-search-dirs
5228 @opindex print-search-dirs
5229 Print the name of the configured installation directory and a list of
5230 program and library directories @command{gcc} will search---and don't do anything else.
5232 This is useful when @command{gcc} prints the error message
5233 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5234 To resolve this you either need to put @file{cpp0} and the other compiler
5235 components where @command{gcc} expects to find them, or you can set the environment
5236 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5237 Don't forget the trailing @samp{/}.
5238 @xref{Environment Variables}.
5240 @item -print-sysroot
5241 @opindex print-sysroot
5242 Print the target sysroot directory that will be used during
5243 compilation. This is the target sysroot specified either at configure
5244 time or using the @option{--sysroot} option, possibly with an extra
5245 suffix that depends on compilation options. If no target sysroot is
5246 specified, the option prints nothing.
5248 @item -print-sysroot-headers-suffix
5249 @opindex print-sysroot-headers-suffix
5250 Print the suffix added to the target sysroot when searching for
5251 headers, or give an error if the compiler is not configured with such
5252 a suffix---and don't do anything else.
5255 @opindex dumpmachine
5256 Print the compiler's target machine (for example,
5257 @samp{i686-pc-linux-gnu})---and don't do anything else.
5260 @opindex dumpversion
5261 Print the compiler version (for example, @samp{3.0})---and don't do
5266 Print the compiler's built-in specs---and don't do anything else. (This
5267 is used when GCC itself is being built.) @xref{Spec Files}.
5269 @item -feliminate-unused-debug-types
5270 @opindex feliminate-unused-debug-types
5271 Normally, when producing DWARF2 output, GCC will emit debugging
5272 information for all types declared in a compilation
5273 unit, regardless of whether or not they are actually used
5274 in that compilation unit. Sometimes this is useful, such as
5275 if, in the debugger, you want to cast a value to a type that is
5276 not actually used in your program (but is declared). More often,
5277 however, this results in a significant amount of wasted space.
5278 With this option, GCC will avoid producing debug symbol output
5279 for types that are nowhere used in the source file being compiled.
5282 @node Optimize Options
5283 @section Options That Control Optimization
5284 @cindex optimize options
5285 @cindex options, optimization
5287 These options control various sorts of optimizations.
5289 Without any optimization option, the compiler's goal is to reduce the
5290 cost of compilation and to make debugging produce the expected
5291 results. Statements are independent: if you stop the program with a
5292 breakpoint between statements, you can then assign a new value to any
5293 variable or change the program counter to any other statement in the
5294 function and get exactly the results you would expect from the source
5297 Turning on optimization flags makes the compiler attempt to improve
5298 the performance and/or code size at the expense of compilation time
5299 and possibly the ability to debug the program.
5301 The compiler performs optimization based on the knowledge it has of the
5302 program. Compiling multiple files at once to a single output file mode allows
5303 the compiler to use information gained from all of the files when compiling
5306 Not all optimizations are controlled directly by a flag. Only
5307 optimizations that have a flag are listed.
5314 Optimize. Optimizing compilation takes somewhat more time, and a lot
5315 more memory for a large function.
5317 With @option{-O}, the compiler tries to reduce code size and execution
5318 time, without performing any optimizations that take a great deal of
5321 @option{-O} turns on the following optimization flags:
5324 -fcprop-registers @gol
5327 -fdelayed-branch @gol
5329 -fguess-branch-probability @gol
5330 -fif-conversion2 @gol
5331 -fif-conversion @gol
5332 -finline-small-functions @gol
5333 -fipa-pure-const @gol
5334 -fipa-reference @gol
5336 -fsplit-wide-types @gol
5337 -ftree-builtin-call-dce @gol
5340 -ftree-copyrename @gol
5342 -ftree-dominator-opts @gol
5349 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5350 where doing so does not interfere with debugging.
5354 Optimize even more. GCC performs nearly all supported optimizations
5355 that do not involve a space-speed tradeoff.
5356 As compared to @option{-O}, this option increases both compilation time
5357 and the performance of the generated code.
5359 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5360 also turns on the following optimization flags:
5361 @gccoptlist{-fthread-jumps @gol
5362 -falign-functions -falign-jumps @gol
5363 -falign-loops -falign-labels @gol
5366 -fcse-follow-jumps -fcse-skip-blocks @gol
5367 -fdelete-null-pointer-checks @gol
5368 -fexpensive-optimizations @gol
5369 -fgcse -fgcse-lm @gol
5370 -findirect-inlining @gol
5371 -foptimize-sibling-calls @gol
5374 -freorder-blocks -freorder-functions @gol
5375 -frerun-cse-after-loop @gol
5376 -fsched-interblock -fsched-spec @gol
5377 -fschedule-insns -fschedule-insns2 @gol
5378 -fstrict-aliasing -fstrict-overflow @gol
5379 -ftree-switch-conversion @gol
5383 Please note the warning under @option{-fgcse} about
5384 invoking @option{-O2} on programs that use computed gotos.
5388 Optimize yet more. @option{-O3} turns on all optimizations specified
5389 by @option{-O2} and also turns on the @option{-finline-functions},
5390 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5391 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5395 Reduce compilation time and make debugging produce the expected
5396 results. This is the default.
5400 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5401 do not typically increase code size. It also performs further
5402 optimizations designed to reduce code size.
5404 @option{-Os} disables the following optimization flags:
5405 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5406 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5407 -fprefetch-loop-arrays -ftree-vect-loop-version}
5409 If you use multiple @option{-O} options, with or without level numbers,
5410 the last such option is the one that is effective.
5413 Options of the form @option{-f@var{flag}} specify machine-independent
5414 flags. Most flags have both positive and negative forms; the negative
5415 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5416 below, only one of the forms is listed---the one you typically will
5417 use. You can figure out the other form by either removing @samp{no-}
5420 The following options control specific optimizations. They are either
5421 activated by @option{-O} options or are related to ones that are. You
5422 can use the following flags in the rare cases when ``fine-tuning'' of
5423 optimizations to be performed is desired.
5426 @item -fno-default-inline
5427 @opindex fno-default-inline
5428 Do not make member functions inline by default merely because they are
5429 defined inside the class scope (C++ only). Otherwise, when you specify
5430 @w{@option{-O}}, member functions defined inside class scope are compiled
5431 inline by default; i.e., you don't need to add @samp{inline} in front of
5432 the member function name.
5434 @item -fno-defer-pop
5435 @opindex fno-defer-pop
5436 Always pop the arguments to each function call as soon as that function
5437 returns. For machines which must pop arguments after a function call,
5438 the compiler normally lets arguments accumulate on the stack for several
5439 function calls and pops them all at once.
5441 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5443 @item -fforward-propagate
5444 @opindex fforward-propagate
5445 Perform a forward propagation pass on RTL@. The pass tries to combine two
5446 instructions and checks if the result can be simplified. If loop unrolling
5447 is active, two passes are performed and the second is scheduled after
5450 This option is enabled by default at optimization levels @option{-O2},
5451 @option{-O3}, @option{-Os}.
5453 @item -fomit-frame-pointer
5454 @opindex fomit-frame-pointer
5455 Don't keep the frame pointer in a register for functions that
5456 don't need one. This avoids the instructions to save, set up and
5457 restore frame pointers; it also makes an extra register available
5458 in many functions. @strong{It also makes debugging impossible on
5461 On some machines, such as the VAX, this flag has no effect, because
5462 the standard calling sequence automatically handles the frame pointer
5463 and nothing is saved by pretending it doesn't exist. The
5464 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5465 whether a target machine supports this flag. @xref{Registers,,Register
5466 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5468 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5470 @item -foptimize-sibling-calls
5471 @opindex foptimize-sibling-calls
5472 Optimize sibling and tail recursive calls.
5474 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5478 Don't pay attention to the @code{inline} keyword. Normally this option
5479 is used to keep the compiler from expanding any functions inline.
5480 Note that if you are not optimizing, no functions can be expanded inline.
5482 @item -finline-small-functions
5483 @opindex finline-small-functions
5484 Integrate functions into their callers when their body is smaller than expected
5485 function call code (so overall size of program gets smaller). The compiler
5486 heuristically decides which functions are simple enough to be worth integrating
5489 Enabled at level @option{-O2}.
5491 @item -findirect-inlining
5492 @opindex findirect-inlining
5493 Inline also indirect calls that are discovered to be known at compile
5494 time thanks to previous inlining. This option has any effect only
5495 when inlining itself is turned on by the @option{-finline-functions}
5496 or @option{-finline-small-functions} options.
5498 Enabled at level @option{-O2}.
5500 @item -finline-functions
5501 @opindex finline-functions
5502 Integrate all simple functions into their callers. The compiler
5503 heuristically decides which functions are simple enough to be worth
5504 integrating in this way.
5506 If all calls to a given function are integrated, and the function is
5507 declared @code{static}, then the function is normally not output as
5508 assembler code in its own right.
5510 Enabled at level @option{-O3}.
5512 @item -finline-functions-called-once
5513 @opindex finline-functions-called-once
5514 Consider all @code{static} functions called once for inlining into their
5515 caller even if they are not marked @code{inline}. If a call to a given
5516 function is integrated, then the function is not output as assembler code
5519 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5521 @item -fearly-inlining
5522 @opindex fearly-inlining
5523 Inline functions marked by @code{always_inline} and functions whose body seems
5524 smaller than the function call overhead early before doing
5525 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5526 makes profiling significantly cheaper and usually inlining faster on programs
5527 having large chains of nested wrapper functions.
5531 @item -finline-limit=@var{n}
5532 @opindex finline-limit
5533 By default, GCC limits the size of functions that can be inlined. This flag
5534 allows coarse control of this limit. @var{n} is the size of functions that
5535 can be inlined in number of pseudo instructions.
5537 Inlining is actually controlled by a number of parameters, which may be
5538 specified individually by using @option{--param @var{name}=@var{value}}.
5539 The @option{-finline-limit=@var{n}} option sets some of these parameters
5543 @item max-inline-insns-single
5544 is set to @var{n}/2.
5545 @item max-inline-insns-auto
5546 is set to @var{n}/2.
5549 See below for a documentation of the individual
5550 parameters controlling inlining and for the defaults of these parameters.
5552 @emph{Note:} there may be no value to @option{-finline-limit} that results
5553 in default behavior.
5555 @emph{Note:} pseudo instruction represents, in this particular context, an
5556 abstract measurement of function's size. In no way does it represent a count
5557 of assembly instructions and as such its exact meaning might change from one
5558 release to an another.
5560 @item -fkeep-inline-functions
5561 @opindex fkeep-inline-functions
5562 In C, emit @code{static} functions that are declared @code{inline}
5563 into the object file, even if the function has been inlined into all
5564 of its callers. This switch does not affect functions using the
5565 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5566 inline functions into the object file.
5568 @item -fkeep-static-consts
5569 @opindex fkeep-static-consts
5570 Emit variables declared @code{static const} when optimization isn't turned
5571 on, even if the variables aren't referenced.
5573 GCC enables this option by default. If you want to force the compiler to
5574 check if the variable was referenced, regardless of whether or not
5575 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5577 @item -fmerge-constants
5578 @opindex fmerge-constants
5579 Attempt to merge identical constants (string constants and floating point
5580 constants) across compilation units.
5582 This option is the default for optimized compilation if the assembler and
5583 linker support it. Use @option{-fno-merge-constants} to inhibit this
5586 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5588 @item -fmerge-all-constants
5589 @opindex fmerge-all-constants
5590 Attempt to merge identical constants and identical variables.
5592 This option implies @option{-fmerge-constants}. In addition to
5593 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5594 arrays or initialized constant variables with integral or floating point
5595 types. Languages like C or C++ require each variable, including multiple
5596 instances of the same variable in recursive calls, to have distinct locations,
5597 so using this option will result in non-conforming
5600 @item -fmodulo-sched
5601 @opindex fmodulo-sched
5602 Perform swing modulo scheduling immediately before the first scheduling
5603 pass. This pass looks at innermost loops and reorders their
5604 instructions by overlapping different iterations.
5606 @item -fmodulo-sched-allow-regmoves
5607 @opindex fmodulo-sched-allow-regmoves
5608 Perform more aggressive SMS based modulo scheduling with register moves
5609 allowed. By setting this flag certain anti-dependences edges will be
5610 deleted which will trigger the generation of reg-moves based on the
5611 life-range analysis. This option is effective only with
5612 @option{-fmodulo-sched} enabled.
5614 @item -fno-branch-count-reg
5615 @opindex fno-branch-count-reg
5616 Do not use ``decrement and branch'' instructions on a count register,
5617 but instead generate a sequence of instructions that decrement a
5618 register, compare it against zero, then branch based upon the result.
5619 This option is only meaningful on architectures that support such
5620 instructions, which include x86, PowerPC, IA-64 and S/390.
5622 The default is @option{-fbranch-count-reg}.
5624 @item -fno-function-cse
5625 @opindex fno-function-cse
5626 Do not put function addresses in registers; make each instruction that
5627 calls a constant function contain the function's address explicitly.
5629 This option results in less efficient code, but some strange hacks
5630 that alter the assembler output may be confused by the optimizations
5631 performed when this option is not used.
5633 The default is @option{-ffunction-cse}
5635 @item -fno-zero-initialized-in-bss
5636 @opindex fno-zero-initialized-in-bss
5637 If the target supports a BSS section, GCC by default puts variables that
5638 are initialized to zero into BSS@. This can save space in the resulting
5641 This option turns off this behavior because some programs explicitly
5642 rely on variables going to the data section. E.g., so that the
5643 resulting executable can find the beginning of that section and/or make
5644 assumptions based on that.
5646 The default is @option{-fzero-initialized-in-bss}.
5648 @item -fmudflap -fmudflapth -fmudflapir
5652 @cindex bounds checking
5654 For front-ends that support it (C and C++), instrument all risky
5655 pointer/array dereferencing operations, some standard library
5656 string/heap functions, and some other associated constructs with
5657 range/validity tests. Modules so instrumented should be immune to
5658 buffer overflows, invalid heap use, and some other classes of C/C++
5659 programming errors. The instrumentation relies on a separate runtime
5660 library (@file{libmudflap}), which will be linked into a program if
5661 @option{-fmudflap} is given at link time. Run-time behavior of the
5662 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5663 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5666 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5667 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5668 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5669 instrumentation should ignore pointer reads. This produces less
5670 instrumentation (and therefore faster execution) and still provides
5671 some protection against outright memory corrupting writes, but allows
5672 erroneously read data to propagate within a program.
5674 @item -fthread-jumps
5675 @opindex fthread-jumps
5676 Perform optimizations where we check to see if a jump branches to a
5677 location where another comparison subsumed by the first is found. If
5678 so, the first branch is redirected to either the destination of the
5679 second branch or a point immediately following it, depending on whether
5680 the condition is known to be true or false.
5682 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5684 @item -fsplit-wide-types
5685 @opindex fsplit-wide-types
5686 When using a type that occupies multiple registers, such as @code{long
5687 long} on a 32-bit system, split the registers apart and allocate them
5688 independently. This normally generates better code for those types,
5689 but may make debugging more difficult.
5691 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5694 @item -fcse-follow-jumps
5695 @opindex fcse-follow-jumps
5696 In common subexpression elimination (CSE), scan through jump instructions
5697 when the target of the jump is not reached by any other path. For
5698 example, when CSE encounters an @code{if} statement with an
5699 @code{else} clause, CSE will follow the jump when the condition
5702 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5704 @item -fcse-skip-blocks
5705 @opindex fcse-skip-blocks
5706 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5707 follow jumps which conditionally skip over blocks. When CSE
5708 encounters a simple @code{if} statement with no else clause,
5709 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5710 body of the @code{if}.
5712 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5714 @item -frerun-cse-after-loop
5715 @opindex frerun-cse-after-loop
5716 Re-run common subexpression elimination after loop optimizations has been
5719 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5723 Perform a global common subexpression elimination pass.
5724 This pass also performs global constant and copy propagation.
5726 @emph{Note:} When compiling a program using computed gotos, a GCC
5727 extension, you may get better runtime performance if you disable
5728 the global common subexpression elimination pass by adding
5729 @option{-fno-gcse} to the command line.
5731 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5735 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5736 attempt to move loads which are only killed by stores into themselves. This
5737 allows a loop containing a load/store sequence to be changed to a load outside
5738 the loop, and a copy/store within the loop.
5740 Enabled by default when gcse is enabled.
5744 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5745 global common subexpression elimination. This pass will attempt to move
5746 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5747 loops containing a load/store sequence can be changed to a load before
5748 the loop and a store after the loop.
5750 Not enabled at any optimization level.
5754 When @option{-fgcse-las} is enabled, the global common subexpression
5755 elimination pass eliminates redundant loads that come after stores to the
5756 same memory location (both partial and full redundancies).
5758 Not enabled at any optimization level.
5760 @item -fgcse-after-reload
5761 @opindex fgcse-after-reload
5762 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5763 pass is performed after reload. The purpose of this pass is to cleanup
5766 @item -funsafe-loop-optimizations
5767 @opindex funsafe-loop-optimizations
5768 If given, the loop optimizer will assume that loop indices do not
5769 overflow, and that the loops with nontrivial exit condition are not
5770 infinite. This enables a wider range of loop optimizations even if
5771 the loop optimizer itself cannot prove that these assumptions are valid.
5772 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5773 if it finds this kind of loop.
5775 @item -fcrossjumping
5776 @opindex fcrossjumping
5777 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5778 resulting code may or may not perform better than without cross-jumping.
5780 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5782 @item -fauto-inc-dec
5783 @opindex fauto-inc-dec
5784 Combine increments or decrements of addresses with memory accesses.
5785 This pass is always skipped on architectures that do not have
5786 instructions to support this. Enabled by default at @option{-O} and
5787 higher on architectures that support this.
5791 Perform dead code elimination (DCE) on RTL@.
5792 Enabled by default at @option{-O} and higher.
5796 Perform dead store elimination (DSE) on RTL@.
5797 Enabled by default at @option{-O} and higher.
5799 @item -fif-conversion
5800 @opindex fif-conversion
5801 Attempt to transform conditional jumps into branch-less equivalents. This
5802 include use of conditional moves, min, max, set flags and abs instructions, and
5803 some tricks doable by standard arithmetics. The use of conditional execution
5804 on chips where it is available is controlled by @code{if-conversion2}.
5806 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5808 @item -fif-conversion2
5809 @opindex fif-conversion2
5810 Use conditional execution (where available) to transform conditional jumps into
5811 branch-less equivalents.
5813 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5815 @item -fdelete-null-pointer-checks
5816 @opindex fdelete-null-pointer-checks
5817 Use global dataflow analysis to identify and eliminate useless checks
5818 for null pointers. The compiler assumes that dereferencing a null
5819 pointer would have halted the program. If a pointer is checked after
5820 it has already been dereferenced, it cannot be null.
5822 In some environments, this assumption is not true, and programs can
5823 safely dereference null pointers. Use
5824 @option{-fno-delete-null-pointer-checks} to disable this optimization
5825 for programs which depend on that behavior.
5827 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5829 @item -fexpensive-optimizations
5830 @opindex fexpensive-optimizations
5831 Perform a number of minor optimizations that are relatively expensive.
5833 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5835 @item -foptimize-register-move
5837 @opindex foptimize-register-move
5839 Attempt to reassign register numbers in move instructions and as
5840 operands of other simple instructions in order to maximize the amount of
5841 register tying. This is especially helpful on machines with two-operand
5844 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5847 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5849 @item -fira-algorithm=@var{algorithm}
5850 Use specified coloring algorithm for the integrated register
5851 allocator. The @var{algorithm} argument should be @code{priority} or
5852 @code{CB}. The first algorithm specifies Chow's priority coloring,
5853 the second one specifies Chaitin-Briggs coloring. The second
5854 algorithm can be unimplemented for some architectures. If it is
5855 implemented, it is the default because Chaitin-Briggs coloring as a
5856 rule generates a better code.
5858 @item -fira-region=@var{region}
5859 Use specified regions for the integrated register allocator. The
5860 @var{region} argument should be one of @code{all}, @code{mixed}, or
5861 @code{one}. The first value means using all loops as register
5862 allocation regions, the second value which is the default means using
5863 all loops except for loops with small register pressure as the
5864 regions, and third one means using all function as a single region.
5865 The first value can give best result for machines with small size and
5866 irregular register set, the third one results in faster and generates
5867 decent code and the smallest size code, and the default value usually
5868 give the best results in most cases and for most architectures.
5870 @item -fira-coalesce
5871 @opindex fira-coalesce
5872 Do optimistic register coalescing. This option might be profitable for
5873 architectures with big regular register files.
5875 @item -fno-ira-share-save-slots
5876 @opindex fno-ira-share-save-slots
5877 Switch off sharing stack slots used for saving call used hard
5878 registers living through a call. Each hard register will get a
5879 separate stack slot and as a result function stack frame will be
5882 @item -fno-ira-share-spill-slots
5883 @opindex fno-ira-share-spill-slots
5884 Switch off sharing stack slots allocated for pseudo-registers. Each
5885 pseudo-register which did not get a hard register will get a separate
5886 stack slot and as a result function stack frame will be bigger.
5888 @item -fira-verbose=@var{n}
5889 @opindex fira-verbose
5890 Set up how verbose dump file for the integrated register allocator
5891 will be. Default value is 5. If the value is greater or equal to 10,
5892 the dump file will be stderr as if the value were @var{n} minus 10.
5894 @item -fdelayed-branch
5895 @opindex fdelayed-branch
5896 If supported for the target machine, attempt to reorder instructions
5897 to exploit instruction slots available after delayed branch
5900 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5902 @item -fschedule-insns
5903 @opindex fschedule-insns
5904 If supported for the target machine, attempt to reorder instructions to
5905 eliminate execution stalls due to required data being unavailable. This
5906 helps machines that have slow floating point or memory load instructions
5907 by allowing other instructions to be issued until the result of the load
5908 or floating point instruction is required.
5910 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5912 @item -fschedule-insns2
5913 @opindex fschedule-insns2
5914 Similar to @option{-fschedule-insns}, but requests an additional pass of
5915 instruction scheduling after register allocation has been done. This is
5916 especially useful on machines with a relatively small number of
5917 registers and where memory load instructions take more than one cycle.
5919 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5921 @item -fno-sched-interblock
5922 @opindex fno-sched-interblock
5923 Don't schedule instructions across basic blocks. This is normally
5924 enabled by default when scheduling before register allocation, i.e.@:
5925 with @option{-fschedule-insns} or at @option{-O2} or higher.
5927 @item -fno-sched-spec
5928 @opindex fno-sched-spec
5929 Don't allow speculative motion of non-load instructions. This is normally
5930 enabled by default when scheduling before register allocation, i.e.@:
5931 with @option{-fschedule-insns} or at @option{-O2} or higher.
5933 @item -fsched-spec-load
5934 @opindex fsched-spec-load
5935 Allow speculative motion of some load instructions. This only makes
5936 sense when scheduling before register allocation, i.e.@: with
5937 @option{-fschedule-insns} or at @option{-O2} or higher.
5939 @item -fsched-spec-load-dangerous
5940 @opindex fsched-spec-load-dangerous
5941 Allow speculative motion of more load instructions. This only makes
5942 sense when scheduling before register allocation, i.e.@: with
5943 @option{-fschedule-insns} or at @option{-O2} or higher.
5945 @item -fsched-stalled-insns
5946 @itemx -fsched-stalled-insns=@var{n}
5947 @opindex fsched-stalled-insns
5948 Define how many insns (if any) can be moved prematurely from the queue
5949 of stalled insns into the ready list, during the second scheduling pass.
5950 @option{-fno-sched-stalled-insns} means that no insns will be moved
5951 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5952 on how many queued insns can be moved prematurely.
5953 @option{-fsched-stalled-insns} without a value is equivalent to
5954 @option{-fsched-stalled-insns=1}.
5956 @item -fsched-stalled-insns-dep
5957 @itemx -fsched-stalled-insns-dep=@var{n}
5958 @opindex fsched-stalled-insns-dep
5959 Define how many insn groups (cycles) will be examined for a dependency
5960 on a stalled insn that is candidate for premature removal from the queue
5961 of stalled insns. This has an effect only during the second scheduling pass,
5962 and only if @option{-fsched-stalled-insns} is used.
5963 @option{-fno-sched-stalled-insns-dep} is equivalent to
5964 @option{-fsched-stalled-insns-dep=0}.
5965 @option{-fsched-stalled-insns-dep} without a value is equivalent to
5966 @option{-fsched-stalled-insns-dep=1}.
5968 @item -fsched2-use-superblocks
5969 @opindex fsched2-use-superblocks
5970 When scheduling after register allocation, do use superblock scheduling
5971 algorithm. Superblock scheduling allows motion across basic block boundaries
5972 resulting on faster schedules. This option is experimental, as not all machine
5973 descriptions used by GCC model the CPU closely enough to avoid unreliable
5974 results from the algorithm.
5976 This only makes sense when scheduling after register allocation, i.e.@: with
5977 @option{-fschedule-insns2} or at @option{-O2} or higher.
5979 @item -fsched2-use-traces
5980 @opindex fsched2-use-traces
5981 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5982 allocation and additionally perform code duplication in order to increase the
5983 size of superblocks using tracer pass. See @option{-ftracer} for details on
5986 This mode should produce faster but significantly longer programs. Also
5987 without @option{-fbranch-probabilities} the traces constructed may not
5988 match the reality and hurt the performance. This only makes
5989 sense when scheduling after register allocation, i.e.@: with
5990 @option{-fschedule-insns2} or at @option{-O2} or higher.
5994 Eliminate redundant sign extension instructions and move the non-redundant
5995 ones to optimal placement using lazy code motion (LCM).
5997 @item -freschedule-modulo-scheduled-loops
5998 @opindex freschedule-modulo-scheduled-loops
5999 The modulo scheduling comes before the traditional scheduling, if a loop
6000 was modulo scheduled we may want to prevent the later scheduling passes
6001 from changing its schedule, we use this option to control that.
6003 @item -fselective-scheduling
6004 @opindex fselective-scheduling
6005 Schedule instructions using selective scheduling algorithm. Selective
6006 scheduling runs instead of the first scheduler pass.
6008 @item -fselective-scheduling2
6009 @opindex fselective-scheduling2
6010 Schedule instructions using selective scheduling algorithm. Selective
6011 scheduling runs instead of the second scheduler pass.
6013 @item -fsel-sched-pipelining
6014 @opindex fsel-sched-pipelining
6015 Enable software pipelining of innermost loops during selective scheduling.
6016 This option has no effect until one of @option{-fselective-scheduling} or
6017 @option{-fselective-scheduling2} is turned on.
6019 @item -fsel-sched-pipelining-outer-loops
6020 @opindex fsel-sched-pipelining-outer-loops
6021 When pipelining loops during selective scheduling, also pipeline outer loops.
6022 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6024 @item -fcaller-saves
6025 @opindex fcaller-saves
6026 Enable values to be allocated in registers that will be clobbered by
6027 function calls, by emitting extra instructions to save and restore the
6028 registers around such calls. Such allocation is done only when it
6029 seems to result in better code than would otherwise be produced.
6031 This option is always enabled by default on certain machines, usually
6032 those which have no call-preserved registers to use instead.
6034 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6036 @item -fconserve-stack
6037 @opindex fconserve-stack
6038 Attempt to minimize stack usage. The compiler will attempt to use less
6039 stack space, even if that makes the program slower. This option
6040 implies setting the @option{large-stack-frame} parameter to 100
6041 and the @option{large-stack-frame-growth} parameter to 400.
6043 @item -ftree-reassoc
6044 @opindex ftree-reassoc
6045 Perform reassociation on trees. This flag is enabled by default
6046 at @option{-O} and higher.
6050 Perform partial redundancy elimination (PRE) on trees. This flag is
6051 enabled by default at @option{-O2} and @option{-O3}.
6055 Perform full redundancy elimination (FRE) on trees. The difference
6056 between FRE and PRE is that FRE only considers expressions
6057 that are computed on all paths leading to the redundant computation.
6058 This analysis is faster than PRE, though it exposes fewer redundancies.
6059 This flag is enabled by default at @option{-O} and higher.
6061 @item -ftree-copy-prop
6062 @opindex ftree-copy-prop
6063 Perform copy propagation on trees. This pass eliminates unnecessary
6064 copy operations. This flag is enabled by default at @option{-O} and
6067 @item -fipa-pure-const
6068 @opindex fipa-pure-const
6069 Discover which functions are pure or constant.
6070 Enabled by default at @option{-O} and higher.
6072 @item -fipa-reference
6073 @opindex fipa-reference
6074 Discover which static variables do not escape cannot escape the
6076 Enabled by default at @option{-O} and higher.
6078 @item -fipa-struct-reorg
6079 @opindex fipa-struct-reorg
6080 Perform structure reorganization optimization, that change C-like structures
6081 layout in order to better utilize spatial locality. This transformation is
6082 affective for programs containing arrays of structures. Available in two
6083 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6084 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
6085 to provide the safety of this transformation. It works only in whole program
6086 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6087 enabled. Structures considered @samp{cold} by this transformation are not
6088 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6090 With this flag, the program debug info reflects a new structure layout.
6094 Perform interprocedural pointer analysis. This option is experimental
6095 and does not affect generated code.
6099 Perform interprocedural constant propagation.
6100 This optimization analyzes the program to determine when values passed
6101 to functions are constants and then optimizes accordingly.
6102 This optimization can substantially increase performance
6103 if the application has constants passed to functions.
6104 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6106 @item -fipa-cp-clone
6107 @opindex fipa-cp-clone
6108 Perform function cloning to make interprocedural constant propagation stronger.
6109 When enabled, interprocedural constant propagation will perform function cloning
6110 when externally visible function can be called with constant arguments.
6111 Because this optimization can create multiple copies of functions,
6112 it may significantly increase code size
6113 (see @option{--param ipcp-unit-growth=@var{value}}).
6114 This flag is enabled by default at @option{-O3}.
6116 @item -fipa-matrix-reorg
6117 @opindex fipa-matrix-reorg
6118 Perform matrix flattening and transposing.
6119 Matrix flattening tries to replace a m-dimensional matrix
6120 with its equivalent n-dimensional matrix, where n < m.
6121 This reduces the level of indirection needed for accessing the elements
6122 of the matrix. The second optimization is matrix transposing that
6123 attemps to change the order of the matrix's dimensions in order to
6124 improve cache locality.
6125 Both optimizations need the @option{-fwhole-program} flag.
6126 Transposing is enabled only if profiling information is available.
6131 Perform forward store motion on trees. This flag is
6132 enabled by default at @option{-O} and higher.
6136 Perform sparse conditional constant propagation (CCP) on trees. This
6137 pass only operates on local scalar variables and is enabled by default
6138 at @option{-O} and higher.
6140 @item -ftree-switch-conversion
6141 Perform conversion of simple initializations in a switch to
6142 initializations from a scalar array. This flag is enabled by default
6143 at @option{-O2} and higher.
6147 Perform dead code elimination (DCE) on trees. This flag is enabled by
6148 default at @option{-O} and higher.
6150 @item -ftree-builtin-call-dce
6151 @opindex ftree-builtin-call-dce
6152 Perform conditional dead code elimination (DCE) for calls to builtin functions
6153 that may set @code{errno} but are otherwise side-effect free. This flag is
6154 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6157 @item -ftree-dominator-opts
6158 @opindex ftree-dominator-opts
6159 Perform a variety of simple scalar cleanups (constant/copy
6160 propagation, redundancy elimination, range propagation and expression
6161 simplification) based on a dominator tree traversal. This also
6162 performs jump threading (to reduce jumps to jumps). This flag is
6163 enabled by default at @option{-O} and higher.
6167 Perform dead store elimination (DSE) on trees. A dead store is a store into
6168 a memory location which will later be overwritten by another store without
6169 any intervening loads. In this case the earlier store can be deleted. This
6170 flag is enabled by default at @option{-O} and higher.
6174 Perform loop header copying on trees. This is beneficial since it increases
6175 effectiveness of code motion optimizations. It also saves one jump. This flag
6176 is enabled by default at @option{-O} and higher. It is not enabled
6177 for @option{-Os}, since it usually increases code size.
6179 @item -ftree-loop-optimize
6180 @opindex ftree-loop-optimize
6181 Perform loop optimizations on trees. This flag is enabled by default
6182 at @option{-O} and higher.
6184 @item -ftree-loop-linear
6185 @opindex ftree-loop-linear
6186 Perform linear loop transformations on tree. This flag can improve cache
6187 performance and allow further loop optimizations to take place.
6189 @item -floop-interchange
6190 Perform loop interchange transformations on loops. Interchanging two
6191 nested loops switches the inner and outer loops. For example, given a
6196 A(J, I) = A(J, I) * C
6200 loop interchange will transform the loop as if the user had written:
6204 A(J, I) = A(J, I) * C
6208 which can be beneficial when @code{N} is larger than the caches,
6209 because in Fortran, the elements of an array are stored in memory
6210 contiguously by column, and the original loop iterates over rows,
6211 potentially creating at each access a cache miss. This optimization
6212 applies to all the languages supported by GCC and is not limited to
6215 @item -floop-strip-mine
6216 Perform loop strip mining transformations on loops. Strip mining
6217 splits a loop into two nested loops. The outer loop has strides
6218 equal to the strip size and the inner loop has strides of the
6219 original loop within a strip. For example, given a loop like:
6225 loop strip mining will transform the loop as if the user had written:
6228 DO I = II, min (II + 3, N)
6233 This optimization applies to all the languages supported by GCC and is
6234 not limited to Fortran.
6237 Perform loop blocking transformations on loops. Blocking strip mines
6238 each loop in the loop nest such that the memory accesses of the
6239 element loops fit inside caches. For example, given a loop like:
6243 A(J, I) = B(I) + C(J)
6247 loop blocking will transform the loop as if the user had written:
6251 DO I = II, min (II + 63, N)
6252 DO J = JJ, min (JJ + 63, M)
6253 A(J, I) = B(I) + C(J)
6259 which can be beneficial when @code{M} is larger than the caches,
6260 because the innermost loop will iterate over a smaller amount of data
6261 that can be kept in the caches. This optimization applies to all the
6262 languages supported by GCC and is not limited to Fortran.
6264 @item -fcheck-data-deps
6265 @opindex fcheck-data-deps
6266 Compare the results of several data dependence analyzers. This option
6267 is used for debugging the data dependence analyzers.
6269 @item -ftree-loop-distribution
6270 Perform loop distribution. This flag can improve cache performance on
6271 big loop bodies and allow further loop optimizations, like
6272 parallelization or vectorization, to take place. For example, the loop
6289 @item -ftree-loop-im
6290 @opindex ftree-loop-im
6291 Perform loop invariant motion on trees. This pass moves only invariants that
6292 would be hard to handle at RTL level (function calls, operations that expand to
6293 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6294 operands of conditions that are invariant out of the loop, so that we can use
6295 just trivial invariantness analysis in loop unswitching. The pass also includes
6298 @item -ftree-loop-ivcanon
6299 @opindex ftree-loop-ivcanon
6300 Create a canonical counter for number of iterations in the loop for that
6301 determining number of iterations requires complicated analysis. Later
6302 optimizations then may determine the number easily. Useful especially
6303 in connection with unrolling.
6307 Perform induction variable optimizations (strength reduction, induction
6308 variable merging and induction variable elimination) on trees.
6310 @item -ftree-parallelize-loops=n
6311 @opindex ftree-parallelize-loops
6312 Parallelize loops, i.e., split their iteration space to run in n threads.
6313 This is only possible for loops whose iterations are independent
6314 and can be arbitrarily reordered. The optimization is only
6315 profitable on multiprocessor machines, for loops that are CPU-intensive,
6316 rather than constrained e.g.@: by memory bandwidth. This option
6317 implies @option{-pthread}, and thus is only supported on targets
6318 that have support for @option{-pthread}.
6322 Perform scalar replacement of aggregates. This pass replaces structure
6323 references with scalars to prevent committing structures to memory too
6324 early. This flag is enabled by default at @option{-O} and higher.
6326 @item -ftree-copyrename
6327 @opindex ftree-copyrename
6328 Perform copy renaming on trees. This pass attempts to rename compiler
6329 temporaries to other variables at copy locations, usually resulting in
6330 variable names which more closely resemble the original variables. This flag
6331 is enabled by default at @option{-O} and higher.
6335 Perform temporary expression replacement during the SSA->normal phase. Single
6336 use/single def temporaries are replaced at their use location with their
6337 defining expression. This results in non-GIMPLE code, but gives the expanders
6338 much more complex trees to work on resulting in better RTL generation. This is
6339 enabled by default at @option{-O} and higher.
6341 @item -ftree-vectorize
6342 @opindex ftree-vectorize
6343 Perform loop vectorization on trees. This flag is enabled by default at
6346 @item -ftree-vect-loop-version
6347 @opindex ftree-vect-loop-version
6348 Perform loop versioning when doing loop vectorization on trees. When a loop
6349 appears to be vectorizable except that data alignment or data dependence cannot
6350 be determined at compile time then vectorized and non-vectorized versions of
6351 the loop are generated along with runtime checks for alignment or dependence
6352 to control which version is executed. This option is enabled by default
6353 except at level @option{-Os} where it is disabled.
6355 @item -fvect-cost-model
6356 @opindex fvect-cost-model
6357 Enable cost model for vectorization.
6361 Perform Value Range Propagation on trees. This is similar to the
6362 constant propagation pass, but instead of values, ranges of values are
6363 propagated. This allows the optimizers to remove unnecessary range
6364 checks like array bound checks and null pointer checks. This is
6365 enabled by default at @option{-O2} and higher. Null pointer check
6366 elimination is only done if @option{-fdelete-null-pointer-checks} is
6371 Perform tail duplication to enlarge superblock size. This transformation
6372 simplifies the control flow of the function allowing other optimizations to do
6375 @item -funroll-loops
6376 @opindex funroll-loops
6377 Unroll loops whose number of iterations can be determined at compile
6378 time or upon entry to the loop. @option{-funroll-loops} implies
6379 @option{-frerun-cse-after-loop}. This option makes code larger,
6380 and may or may not make it run faster.
6382 @item -funroll-all-loops
6383 @opindex funroll-all-loops
6384 Unroll all loops, even if their number of iterations is uncertain when
6385 the loop is entered. This usually makes programs run more slowly.
6386 @option{-funroll-all-loops} implies the same options as
6387 @option{-funroll-loops},
6389 @item -fsplit-ivs-in-unroller
6390 @opindex fsplit-ivs-in-unroller
6391 Enables expressing of values of induction variables in later iterations
6392 of the unrolled loop using the value in the first iteration. This breaks
6393 long dependency chains, thus improving efficiency of the scheduling passes.
6395 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6396 same effect. However in cases the loop body is more complicated than
6397 a single basic block, this is not reliable. It also does not work at all
6398 on some of the architectures due to restrictions in the CSE pass.
6400 This optimization is enabled by default.
6402 @item -fvariable-expansion-in-unroller
6403 @opindex fvariable-expansion-in-unroller
6404 With this option, the compiler will create multiple copies of some
6405 local variables when unrolling a loop which can result in superior code.
6407 @item -fpredictive-commoning
6408 @opindex fpredictive-commoning
6409 Perform predictive commoning optimization, i.e., reusing computations
6410 (especially memory loads and stores) performed in previous
6411 iterations of loops.
6413 This option is enabled at level @option{-O3}.
6415 @item -fprefetch-loop-arrays
6416 @opindex fprefetch-loop-arrays
6417 If supported by the target machine, generate instructions to prefetch
6418 memory to improve the performance of loops that access large arrays.
6420 This option may generate better or worse code; results are highly
6421 dependent on the structure of loops within the source code.
6423 Disabled at level @option{-Os}.
6426 @itemx -fno-peephole2
6427 @opindex fno-peephole
6428 @opindex fno-peephole2
6429 Disable any machine-specific peephole optimizations. The difference
6430 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6431 are implemented in the compiler; some targets use one, some use the
6432 other, a few use both.
6434 @option{-fpeephole} is enabled by default.
6435 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6437 @item -fno-guess-branch-probability
6438 @opindex fno-guess-branch-probability
6439 Do not guess branch probabilities using heuristics.
6441 GCC will use heuristics to guess branch probabilities if they are
6442 not provided by profiling feedback (@option{-fprofile-arcs}). These
6443 heuristics are based on the control flow graph. If some branch probabilities
6444 are specified by @samp{__builtin_expect}, then the heuristics will be
6445 used to guess branch probabilities for the rest of the control flow graph,
6446 taking the @samp{__builtin_expect} info into account. The interactions
6447 between the heuristics and @samp{__builtin_expect} can be complex, and in
6448 some cases, it may be useful to disable the heuristics so that the effects
6449 of @samp{__builtin_expect} are easier to understand.
6451 The default is @option{-fguess-branch-probability} at levels
6452 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6454 @item -freorder-blocks
6455 @opindex freorder-blocks
6456 Reorder basic blocks in the compiled function in order to reduce number of
6457 taken branches and improve code locality.
6459 Enabled at levels @option{-O2}, @option{-O3}.
6461 @item -freorder-blocks-and-partition
6462 @opindex freorder-blocks-and-partition
6463 In addition to reordering basic blocks in the compiled function, in order
6464 to reduce number of taken branches, partitions hot and cold basic blocks
6465 into separate sections of the assembly and .o files, to improve
6466 paging and cache locality performance.
6468 This optimization is automatically turned off in the presence of
6469 exception handling, for linkonce sections, for functions with a user-defined
6470 section attribute and on any architecture that does not support named
6473 @item -freorder-functions
6474 @opindex freorder-functions
6475 Reorder functions in the object file in order to
6476 improve code locality. This is implemented by using special
6477 subsections @code{.text.hot} for most frequently executed functions and
6478 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6479 the linker so object file format must support named sections and linker must
6480 place them in a reasonable way.
6482 Also profile feedback must be available in to make this option effective. See
6483 @option{-fprofile-arcs} for details.
6485 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6487 @item -fstrict-aliasing
6488 @opindex fstrict-aliasing
6489 Allows the compiler to assume the strictest aliasing rules applicable to
6490 the language being compiled. For C (and C++), this activates
6491 optimizations based on the type of expressions. In particular, an
6492 object of one type is assumed never to reside at the same address as an
6493 object of a different type, unless the types are almost the same. For
6494 example, an @code{unsigned int} can alias an @code{int}, but not a
6495 @code{void*} or a @code{double}. A character type may alias any other
6498 @anchor{Type-punning}Pay special attention to code like this:
6511 The practice of reading from a different union member than the one most
6512 recently written to (called ``type-punning'') is common. Even with
6513 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6514 is accessed through the union type. So, the code above will work as
6515 expected. @xref{Structures unions enumerations and bit-fields
6516 implementation}. However, this code might not:
6527 Similarly, access by taking the address, casting the resulting pointer
6528 and dereferencing the result has undefined behavior, even if the cast
6529 uses a union type, e.g.:
6533 return ((union a_union *) &d)->i;
6537 The @option{-fstrict-aliasing} option is enabled at levels
6538 @option{-O2}, @option{-O3}, @option{-Os}.
6540 @item -fstrict-overflow
6541 @opindex fstrict-overflow
6542 Allow the compiler to assume strict signed overflow rules, depending
6543 on the language being compiled. For C (and C++) this means that
6544 overflow when doing arithmetic with signed numbers is undefined, which
6545 means that the compiler may assume that it will not happen. This
6546 permits various optimizations. For example, the compiler will assume
6547 that an expression like @code{i + 10 > i} will always be true for
6548 signed @code{i}. This assumption is only valid if signed overflow is
6549 undefined, as the expression is false if @code{i + 10} overflows when
6550 using twos complement arithmetic. When this option is in effect any
6551 attempt to determine whether an operation on signed numbers will
6552 overflow must be written carefully to not actually involve overflow.
6554 This option also allows the compiler to assume strict pointer
6555 semantics: given a pointer to an object, if adding an offset to that
6556 pointer does not produce a pointer to the same object, the addition is
6557 undefined. This permits the compiler to conclude that @code{p + u >
6558 p} is always true for a pointer @code{p} and unsigned integer
6559 @code{u}. This assumption is only valid because pointer wraparound is
6560 undefined, as the expression is false if @code{p + u} overflows using
6561 twos complement arithmetic.
6563 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6564 that integer signed overflow is fully defined: it wraps. When
6565 @option{-fwrapv} is used, there is no difference between
6566 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
6567 integers. With @option{-fwrapv} certain types of overflow are
6568 permitted. For example, if the compiler gets an overflow when doing
6569 arithmetic on constants, the overflowed value can still be used with
6570 @option{-fwrapv}, but not otherwise.
6572 The @option{-fstrict-overflow} option is enabled at levels
6573 @option{-O2}, @option{-O3}, @option{-Os}.
6575 @item -falign-functions
6576 @itemx -falign-functions=@var{n}
6577 @opindex falign-functions
6578 Align the start of functions to the next power-of-two greater than
6579 @var{n}, skipping up to @var{n} bytes. For instance,
6580 @option{-falign-functions=32} aligns functions to the next 32-byte
6581 boundary, but @option{-falign-functions=24} would align to the next
6582 32-byte boundary only if this can be done by skipping 23 bytes or less.
6584 @option{-fno-align-functions} and @option{-falign-functions=1} are
6585 equivalent and mean that functions will not be aligned.
6587 Some assemblers only support this flag when @var{n} is a power of two;
6588 in that case, it is rounded up.
6590 If @var{n} is not specified or is zero, use a machine-dependent default.
6592 Enabled at levels @option{-O2}, @option{-O3}.
6594 @item -falign-labels
6595 @itemx -falign-labels=@var{n}
6596 @opindex falign-labels
6597 Align all branch targets to a power-of-two boundary, skipping up to
6598 @var{n} bytes like @option{-falign-functions}. This option can easily
6599 make code slower, because it must insert dummy operations for when the
6600 branch target is reached in the usual flow of the code.
6602 @option{-fno-align-labels} and @option{-falign-labels=1} are
6603 equivalent and mean that labels will not be aligned.
6605 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6606 are greater than this value, then their values are used instead.
6608 If @var{n} is not specified or is zero, use a machine-dependent default
6609 which is very likely to be @samp{1}, meaning no alignment.
6611 Enabled at levels @option{-O2}, @option{-O3}.
6614 @itemx -falign-loops=@var{n}
6615 @opindex falign-loops
6616 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6617 like @option{-falign-functions}. The hope is that the loop will be
6618 executed many times, which will make up for any execution of the dummy
6621 @option{-fno-align-loops} and @option{-falign-loops=1} are
6622 equivalent and mean that loops will not be aligned.
6624 If @var{n} is not specified or is zero, use a machine-dependent default.
6626 Enabled at levels @option{-O2}, @option{-O3}.
6629 @itemx -falign-jumps=@var{n}
6630 @opindex falign-jumps
6631 Align branch targets to a power-of-two boundary, for branch targets
6632 where the targets can only be reached by jumping, skipping up to @var{n}
6633 bytes like @option{-falign-functions}. In this case, no dummy operations
6636 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6637 equivalent and mean that loops will not be aligned.
6639 If @var{n} is not specified or is zero, use a machine-dependent default.
6641 Enabled at levels @option{-O2}, @option{-O3}.
6643 @item -funit-at-a-time
6644 @opindex funit-at-a-time
6645 This option is left for compatibility reasons. @option{-funit-at-a-time}
6646 has no effect, while @option{-fno-unit-at-a-time} implies
6647 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
6651 @item -fno-toplevel-reorder
6652 @opindex fno-toplevel-reorder
6653 Do not reorder top-level functions, variables, and @code{asm}
6654 statements. Output them in the same order that they appear in the
6655 input file. When this option is used, unreferenced static variables
6656 will not be removed. This option is intended to support existing code
6657 which relies on a particular ordering. For new code, it is better to
6660 Enabled at level @option{-O0}. When disabled explicitly, it also imply
6661 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
6666 Constructs webs as commonly used for register allocation purposes and assign
6667 each web individual pseudo register. This allows the register allocation pass
6668 to operate on pseudos directly, but also strengthens several other optimization
6669 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6670 however, make debugging impossible, since variables will no longer stay in a
6673 Enabled by default with @option{-funroll-loops}.
6675 @item -fwhole-program
6676 @opindex fwhole-program
6677 Assume that the current compilation unit represents whole program being
6678 compiled. All public functions and variables with the exception of @code{main}
6679 and those merged by attribute @code{externally_visible} become static functions
6680 and in a affect gets more aggressively optimized by interprocedural optimizers.
6681 While this option is equivalent to proper use of @code{static} keyword for
6682 programs consisting of single file, in combination with option
6683 @option{--combine} this flag can be used to compile most of smaller scale C
6684 programs since the functions and variables become local for the whole combined
6685 compilation unit, not for the single source file itself.
6687 This option is not supported for Fortran programs.
6689 @item -fcprop-registers
6690 @opindex fcprop-registers
6691 After register allocation and post-register allocation instruction splitting,
6692 we perform a copy-propagation pass to try to reduce scheduling dependencies
6693 and occasionally eliminate the copy.
6695 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6697 @item -fprofile-correction
6698 @opindex fprofile-correction
6699 Profiles collected using an instrumented binary for multi-threaded programs may
6700 be inconsistent due to missed counter updates. When this option is specified,
6701 GCC will use heuristics to correct or smooth out such inconsistencies. By
6702 default, GCC will emit an error message when an inconsistent profile is detected.
6704 @item -fprofile-dir=@var{path}
6705 @opindex fprofile-dir
6707 Set the directory to search the profile data files in to @var{path}.
6708 This option affects only the profile data generated by
6709 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
6710 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
6711 and its related options.
6712 By default, GCC will use the current directory as @var{path}
6713 thus the profile data file will appear in the same directory as the object file.
6715 @item -fprofile-generate
6716 @itemx -fprofile-generate=@var{path}
6717 @opindex fprofile-generate
6719 Enable options usually used for instrumenting application to produce
6720 profile useful for later recompilation with profile feedback based
6721 optimization. You must use @option{-fprofile-generate} both when
6722 compiling and when linking your program.
6724 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6726 If @var{path} is specified, GCC will look at the @var{path} to find
6727 the profile feedback data files. See @option{-fprofile-dir}.
6730 @itemx -fprofile-use=@var{path}
6731 @opindex fprofile-use
6732 Enable profile feedback directed optimizations, and optimizations
6733 generally profitable only with profile feedback available.
6735 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6736 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6738 By default, GCC emits an error message if the feedback profiles do not
6739 match the source code. This error can be turned into a warning by using
6740 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6743 If @var{path} is specified, GCC will look at the @var{path} to find
6744 the profile feedback data files. See @option{-fprofile-dir}.
6747 The following options control compiler behavior regarding floating
6748 point arithmetic. These options trade off between speed and
6749 correctness. All must be specifically enabled.
6753 @opindex ffloat-store
6754 Do not store floating point variables in registers, and inhibit other
6755 options that might change whether a floating point value is taken from a
6758 @cindex floating point precision
6759 This option prevents undesirable excess precision on machines such as
6760 the 68000 where the floating registers (of the 68881) keep more
6761 precision than a @code{double} is supposed to have. Similarly for the
6762 x86 architecture. For most programs, the excess precision does only
6763 good, but a few programs rely on the precise definition of IEEE floating
6764 point. Use @option{-ffloat-store} for such programs, after modifying
6765 them to store all pertinent intermediate computations into variables.
6769 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6770 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6771 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6773 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6775 This option is not turned on by any @option{-O} option since
6776 it can result in incorrect output for programs which depend on
6777 an exact implementation of IEEE or ISO rules/specifications for
6778 math functions. It may, however, yield faster code for programs
6779 that do not require the guarantees of these specifications.
6781 @item -fno-math-errno
6782 @opindex fno-math-errno
6783 Do not set ERRNO after calling math functions that are executed
6784 with a single instruction, e.g., sqrt. A program that relies on
6785 IEEE exceptions for math error handling may want to use this flag
6786 for speed while maintaining IEEE arithmetic compatibility.
6788 This option is not turned on by any @option{-O} option since
6789 it can result in incorrect output for programs which depend on
6790 an exact implementation of IEEE or ISO rules/specifications for
6791 math functions. It may, however, yield faster code for programs
6792 that do not require the guarantees of these specifications.
6794 The default is @option{-fmath-errno}.
6796 On Darwin systems, the math library never sets @code{errno}. There is
6797 therefore no reason for the compiler to consider the possibility that
6798 it might, and @option{-fno-math-errno} is the default.
6800 @item -funsafe-math-optimizations
6801 @opindex funsafe-math-optimizations
6803 Allow optimizations for floating-point arithmetic that (a) assume
6804 that arguments and results are valid and (b) may violate IEEE or
6805 ANSI standards. When used at link-time, it may include libraries
6806 or startup files that change the default FPU control word or other
6807 similar optimizations.
6809 This option is not turned on by any @option{-O} option since
6810 it can result in incorrect output for programs which depend on
6811 an exact implementation of IEEE or ISO rules/specifications for
6812 math functions. It may, however, yield faster code for programs
6813 that do not require the guarantees of these specifications.
6814 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6815 @option{-fassociative-math} and @option{-freciprocal-math}.
6817 The default is @option{-fno-unsafe-math-optimizations}.
6819 @item -fassociative-math
6820 @opindex fassociative-math
6822 Allow re-association of operands in series of floating-point operations.
6823 This violates the ISO C and C++ language standard by possibly changing
6824 computation result. NOTE: re-ordering may change the sign of zero as
6825 well as ignore NaNs and inhibit or create underflow or overflow (and
6826 thus cannot be used on a code which relies on rounding behavior like
6827 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6828 and thus may not be used when ordered comparisons are required.
6829 This option requires that both @option{-fno-signed-zeros} and
6830 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6831 much sense with @option{-frounding-math}.
6833 The default is @option{-fno-associative-math}.
6835 @item -freciprocal-math
6836 @opindex freciprocal-math
6838 Allow the reciprocal of a value to be used instead of dividing by
6839 the value if this enables optimizations. For example @code{x / y}
6840 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6841 is subject to common subexpression elimination. Note that this loses
6842 precision and increases the number of flops operating on the value.
6844 The default is @option{-fno-reciprocal-math}.
6846 @item -ffinite-math-only
6847 @opindex ffinite-math-only
6848 Allow optimizations for floating-point arithmetic that assume
6849 that arguments and results are not NaNs or +-Infs.
6851 This option is not turned on by any @option{-O} option since
6852 it can result in incorrect output for programs which depend on
6853 an exact implementation of IEEE or ISO rules/specifications for
6854 math functions. It may, however, yield faster code for programs
6855 that do not require the guarantees of these specifications.
6857 The default is @option{-fno-finite-math-only}.
6859 @item -fno-signed-zeros
6860 @opindex fno-signed-zeros
6861 Allow optimizations for floating point arithmetic that ignore the
6862 signedness of zero. IEEE arithmetic specifies the behavior of
6863 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6864 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6865 This option implies that the sign of a zero result isn't significant.
6867 The default is @option{-fsigned-zeros}.
6869 @item -fno-trapping-math
6870 @opindex fno-trapping-math
6871 Compile code assuming that floating-point operations cannot generate
6872 user-visible traps. These traps include division by zero, overflow,
6873 underflow, inexact result and invalid operation. This option requires
6874 that @option{-fno-signaling-nans} be in effect. Setting this option may
6875 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6877 This option should never be turned on by any @option{-O} option since
6878 it can result in incorrect output for programs which depend on
6879 an exact implementation of IEEE or ISO rules/specifications for
6882 The default is @option{-ftrapping-math}.
6884 @item -frounding-math
6885 @opindex frounding-math
6886 Disable transformations and optimizations that assume default floating
6887 point rounding behavior. This is round-to-zero for all floating point
6888 to integer conversions, and round-to-nearest for all other arithmetic
6889 truncations. This option should be specified for programs that change
6890 the FP rounding mode dynamically, or that may be executed with a
6891 non-default rounding mode. This option disables constant folding of
6892 floating point expressions at compile-time (which may be affected by
6893 rounding mode) and arithmetic transformations that are unsafe in the
6894 presence of sign-dependent rounding modes.
6896 The default is @option{-fno-rounding-math}.
6898 This option is experimental and does not currently guarantee to
6899 disable all GCC optimizations that are affected by rounding mode.
6900 Future versions of GCC may provide finer control of this setting
6901 using C99's @code{FENV_ACCESS} pragma. This command line option
6902 will be used to specify the default state for @code{FENV_ACCESS}.
6904 @item -frtl-abstract-sequences
6905 @opindex frtl-abstract-sequences
6906 It is a size optimization method. This option is to find identical
6907 sequences of code, which can be turned into pseudo-procedures and
6908 then replace all occurrences with calls to the newly created
6909 subroutine. It is kind of an opposite of @option{-finline-functions}.
6910 This optimization runs at RTL level.
6912 @item -fsignaling-nans
6913 @opindex fsignaling-nans
6914 Compile code assuming that IEEE signaling NaNs may generate user-visible
6915 traps during floating-point operations. Setting this option disables
6916 optimizations that may change the number of exceptions visible with
6917 signaling NaNs. This option implies @option{-ftrapping-math}.
6919 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6922 The default is @option{-fno-signaling-nans}.
6924 This option is experimental and does not currently guarantee to
6925 disable all GCC optimizations that affect signaling NaN behavior.
6927 @item -fsingle-precision-constant
6928 @opindex fsingle-precision-constant
6929 Treat floating point constant as single precision constant instead of
6930 implicitly converting it to double precision constant.
6932 @item -fcx-limited-range
6933 @opindex fcx-limited-range
6934 When enabled, this option states that a range reduction step is not
6935 needed when performing complex division. Also, there is no checking
6936 whether the result of a complex multiplication or division is @code{NaN
6937 + I*NaN}, with an attempt to rescue the situation in that case. The
6938 default is @option{-fno-cx-limited-range}, but is enabled by
6939 @option{-ffast-math}.
6941 This option controls the default setting of the ISO C99
6942 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6945 @item -fcx-fortran-rules
6946 @opindex fcx-fortran-rules
6947 Complex multiplication and division follow Fortran rules. Range
6948 reduction is done as part of complex division, but there is no checking
6949 whether the result of a complex multiplication or division is @code{NaN
6950 + I*NaN}, with an attempt to rescue the situation in that case.
6952 The default is @option{-fno-cx-fortran-rules}.
6956 The following options control optimizations that may improve
6957 performance, but are not enabled by any @option{-O} options. This
6958 section includes experimental options that may produce broken code.
6961 @item -fbranch-probabilities
6962 @opindex fbranch-probabilities
6963 After running a program compiled with @option{-fprofile-arcs}
6964 (@pxref{Debugging Options,, Options for Debugging Your Program or
6965 @command{gcc}}), you can compile it a second time using
6966 @option{-fbranch-probabilities}, to improve optimizations based on
6967 the number of times each branch was taken. When the program
6968 compiled with @option{-fprofile-arcs} exits it saves arc execution
6969 counts to a file called @file{@var{sourcename}.gcda} for each source
6970 file. The information in this data file is very dependent on the
6971 structure of the generated code, so you must use the same source code
6972 and the same optimization options for both compilations.
6974 With @option{-fbranch-probabilities}, GCC puts a
6975 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6976 These can be used to improve optimization. Currently, they are only
6977 used in one place: in @file{reorg.c}, instead of guessing which path a
6978 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6979 exactly determine which path is taken more often.
6981 @item -fprofile-values
6982 @opindex fprofile-values
6983 If combined with @option{-fprofile-arcs}, it adds code so that some
6984 data about values of expressions in the program is gathered.
6986 With @option{-fbranch-probabilities}, it reads back the data gathered
6987 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6988 notes to instructions for their later usage in optimizations.
6990 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6994 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6995 a code to gather information about values of expressions.
6997 With @option{-fbranch-probabilities}, it reads back the data gathered
6998 and actually performs the optimizations based on them.
6999 Currently the optimizations include specialization of division operation
7000 using the knowledge about the value of the denominator.
7002 @item -frename-registers
7003 @opindex frename-registers
7004 Attempt to avoid false dependencies in scheduled code by making use
7005 of registers left over after register allocation. This optimization
7006 will most benefit processors with lots of registers. Depending on the
7007 debug information format adopted by the target, however, it can
7008 make debugging impossible, since variables will no longer stay in
7009 a ``home register''.
7011 Enabled by default with @option{-funroll-loops}.
7015 Perform tail duplication to enlarge superblock size. This transformation
7016 simplifies the control flow of the function allowing other optimizations to do
7019 Enabled with @option{-fprofile-use}.
7021 @item -funroll-loops
7022 @opindex funroll-loops
7023 Unroll loops whose number of iterations can be determined at compile time or
7024 upon entry to the loop. @option{-funroll-loops} implies
7025 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7026 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7027 small constant number of iterations). This option makes code larger, and may
7028 or may not make it run faster.
7030 Enabled with @option{-fprofile-use}.
7032 @item -funroll-all-loops
7033 @opindex funroll-all-loops
7034 Unroll all loops, even if their number of iterations is uncertain when
7035 the loop is entered. This usually makes programs run more slowly.
7036 @option{-funroll-all-loops} implies the same options as
7037 @option{-funroll-loops}.
7040 @opindex fpeel-loops
7041 Peels the loops for that there is enough information that they do not
7042 roll much (from profile feedback). It also turns on complete loop peeling
7043 (i.e.@: complete removal of loops with small constant number of iterations).
7045 Enabled with @option{-fprofile-use}.
7047 @item -fmove-loop-invariants
7048 @opindex fmove-loop-invariants
7049 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7050 at level @option{-O1}
7052 @item -funswitch-loops
7053 @opindex funswitch-loops
7054 Move branches with loop invariant conditions out of the loop, with duplicates
7055 of the loop on both branches (modified according to result of the condition).
7057 @item -ffunction-sections
7058 @itemx -fdata-sections
7059 @opindex ffunction-sections
7060 @opindex fdata-sections
7061 Place each function or data item into its own section in the output
7062 file if the target supports arbitrary sections. The name of the
7063 function or the name of the data item determines the section's name
7066 Use these options on systems where the linker can perform optimizations
7067 to improve locality of reference in the instruction space. Most systems
7068 using the ELF object format and SPARC processors running Solaris 2 have
7069 linkers with such optimizations. AIX may have these optimizations in
7072 Only use these options when there are significant benefits from doing
7073 so. When you specify these options, the assembler and linker will
7074 create larger object and executable files and will also be slower.
7075 You will not be able to use @code{gprof} on all systems if you
7076 specify this option and you may have problems with debugging if
7077 you specify both this option and @option{-g}.
7079 @item -fbranch-target-load-optimize
7080 @opindex fbranch-target-load-optimize
7081 Perform branch target register load optimization before prologue / epilogue
7083 The use of target registers can typically be exposed only during reload,
7084 thus hoisting loads out of loops and doing inter-block scheduling needs
7085 a separate optimization pass.
7087 @item -fbranch-target-load-optimize2
7088 @opindex fbranch-target-load-optimize2
7089 Perform branch target register load optimization after prologue / epilogue
7092 @item -fbtr-bb-exclusive
7093 @opindex fbtr-bb-exclusive
7094 When performing branch target register load optimization, don't reuse
7095 branch target registers in within any basic block.
7097 @item -fstack-protector
7098 @opindex fstack-protector
7099 Emit extra code to check for buffer overflows, such as stack smashing
7100 attacks. This is done by adding a guard variable to functions with
7101 vulnerable objects. This includes functions that call alloca, and
7102 functions with buffers larger than 8 bytes. The guards are initialized
7103 when a function is entered and then checked when the function exits.
7104 If a guard check fails, an error message is printed and the program exits.
7106 @item -fstack-protector-all
7107 @opindex fstack-protector-all
7108 Like @option{-fstack-protector} except that all functions are protected.
7110 @item -fsection-anchors
7111 @opindex fsection-anchors
7112 Try to reduce the number of symbolic address calculations by using
7113 shared ``anchor'' symbols to address nearby objects. This transformation
7114 can help to reduce the number of GOT entries and GOT accesses on some
7117 For example, the implementation of the following function @code{foo}:
7121 int foo (void) @{ return a + b + c; @}
7124 would usually calculate the addresses of all three variables, but if you
7125 compile it with @option{-fsection-anchors}, it will access the variables
7126 from a common anchor point instead. The effect is similar to the
7127 following pseudocode (which isn't valid C):
7132 register int *xr = &x;
7133 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
7137 Not all targets support this option.
7139 @item --param @var{name}=@var{value}
7141 In some places, GCC uses various constants to control the amount of
7142 optimization that is done. For example, GCC will not inline functions
7143 that contain more that a certain number of instructions. You can
7144 control some of these constants on the command-line using the
7145 @option{--param} option.
7147 The names of specific parameters, and the meaning of the values, are
7148 tied to the internals of the compiler, and are subject to change
7149 without notice in future releases.
7151 In each case, the @var{value} is an integer. The allowable choices for
7152 @var{name} are given in the following table:
7155 @item sra-max-structure-size
7156 The maximum structure size, in bytes, at which the scalar replacement
7157 of aggregates (SRA) optimization will perform block copies. The
7158 default value, 0, implies that GCC will select the most appropriate
7161 @item sra-field-structure-ratio
7162 The threshold ratio (as a percentage) between instantiated fields and
7163 the complete structure size. We say that if the ratio of the number
7164 of bytes in instantiated fields to the number of bytes in the complete
7165 structure exceeds this parameter, then block copies are not used. The
7168 @item struct-reorg-cold-struct-ratio
7169 The threshold ratio (as a percentage) between a structure frequency
7170 and the frequency of the hottest structure in the program. This parameter
7171 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
7172 We say that if the ratio of a structure frequency, calculated by profiling,
7173 to the hottest structure frequency in the program is less than this
7174 parameter, then structure reorganization is not applied to this structure.
7177 @item predictable-branch-cost-outcome
7178 When branch is predicted to be taken with probability lower than this threshold
7179 (in percent), then it is considered well predictable. The default is 10.
7181 @item max-crossjump-edges
7182 The maximum number of incoming edges to consider for crossjumping.
7183 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
7184 the number of edges incoming to each block. Increasing values mean
7185 more aggressive optimization, making the compile time increase with
7186 probably small improvement in executable size.
7188 @item min-crossjump-insns
7189 The minimum number of instructions which must be matched at the end
7190 of two blocks before crossjumping will be performed on them. This
7191 value is ignored in the case where all instructions in the block being
7192 crossjumped from are matched. The default value is 5.
7194 @item max-grow-copy-bb-insns
7195 The maximum code size expansion factor when copying basic blocks
7196 instead of jumping. The expansion is relative to a jump instruction.
7197 The default value is 8.
7199 @item max-goto-duplication-insns
7200 The maximum number of instructions to duplicate to a block that jumps
7201 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
7202 passes, GCC factors computed gotos early in the compilation process,
7203 and unfactors them as late as possible. Only computed jumps at the
7204 end of a basic blocks with no more than max-goto-duplication-insns are
7205 unfactored. The default value is 8.
7207 @item max-delay-slot-insn-search
7208 The maximum number of instructions to consider when looking for an
7209 instruction to fill a delay slot. If more than this arbitrary number of
7210 instructions is searched, the time savings from filling the delay slot
7211 will be minimal so stop searching. Increasing values mean more
7212 aggressive optimization, making the compile time increase with probably
7213 small improvement in executable run time.
7215 @item max-delay-slot-live-search
7216 When trying to fill delay slots, the maximum number of instructions to
7217 consider when searching for a block with valid live register
7218 information. Increasing this arbitrarily chosen value means more
7219 aggressive optimization, increasing the compile time. This parameter
7220 should be removed when the delay slot code is rewritten to maintain the
7223 @item max-gcse-memory
7224 The approximate maximum amount of memory that will be allocated in
7225 order to perform the global common subexpression elimination
7226 optimization. If more memory than specified is required, the
7227 optimization will not be done.
7229 @item max-gcse-passes
7230 The maximum number of passes of GCSE to run. The default is 1.
7232 @item max-pending-list-length
7233 The maximum number of pending dependencies scheduling will allow
7234 before flushing the current state and starting over. Large functions
7235 with few branches or calls can create excessively large lists which
7236 needlessly consume memory and resources.
7238 @item max-inline-insns-single
7239 Several parameters control the tree inliner used in gcc.
7240 This number sets the maximum number of instructions (counted in GCC's
7241 internal representation) in a single function that the tree inliner
7242 will consider for inlining. This only affects functions declared
7243 inline and methods implemented in a class declaration (C++).
7244 The default value is 450.
7246 @item max-inline-insns-auto
7247 When you use @option{-finline-functions} (included in @option{-O3}),
7248 a lot of functions that would otherwise not be considered for inlining
7249 by the compiler will be investigated. To those functions, a different
7250 (more restrictive) limit compared to functions declared inline can
7252 The default value is 90.
7254 @item large-function-insns
7255 The limit specifying really large functions. For functions larger than this
7256 limit after inlining, inlining is constrained by
7257 @option{--param large-function-growth}. This parameter is useful primarily
7258 to avoid extreme compilation time caused by non-linear algorithms used by the
7260 The default value is 2700.
7262 @item large-function-growth
7263 Specifies maximal growth of large function caused by inlining in percents.
7264 The default value is 100 which limits large function growth to 2.0 times
7267 @item large-unit-insns
7268 The limit specifying large translation unit. Growth caused by inlining of
7269 units larger than this limit is limited by @option{--param inline-unit-growth}.
7270 For small units this might be too tight (consider unit consisting of function A
7271 that is inline and B that just calls A three time. If B is small relative to
7272 A, the growth of unit is 300\% and yet such inlining is very sane. For very
7273 large units consisting of small inlineable functions however the overall unit
7274 growth limit is needed to avoid exponential explosion of code size. Thus for
7275 smaller units, the size is increased to @option{--param large-unit-insns}
7276 before applying @option{--param inline-unit-growth}. The default is 10000
7278 @item inline-unit-growth
7279 Specifies maximal overall growth of the compilation unit caused by inlining.
7280 The default value is 30 which limits unit growth to 1.3 times the original
7283 @item ipcp-unit-growth
7284 Specifies maximal overall growth of the compilation unit caused by
7285 interprocedural constant propagation. The default value is 10 which limits
7286 unit growth to 1.1 times the original size.
7288 @item large-stack-frame
7289 The limit specifying large stack frames. While inlining the algorithm is trying
7290 to not grow past this limit too much. Default value is 256 bytes.
7292 @item large-stack-frame-growth
7293 Specifies maximal growth of large stack frames caused by inlining in percents.
7294 The default value is 1000 which limits large stack frame growth to 11 times
7297 @item max-inline-insns-recursive
7298 @itemx max-inline-insns-recursive-auto
7299 Specifies maximum number of instructions out-of-line copy of self recursive inline
7300 function can grow into by performing recursive inlining.
7302 For functions declared inline @option{--param max-inline-insns-recursive} is
7303 taken into account. For function not declared inline, recursive inlining
7304 happens only when @option{-finline-functions} (included in @option{-O3}) is
7305 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
7306 default value is 450.
7308 @item max-inline-recursive-depth
7309 @itemx max-inline-recursive-depth-auto
7310 Specifies maximum recursion depth used by the recursive inlining.
7312 For functions declared inline @option{--param max-inline-recursive-depth} is
7313 taken into account. For function not declared inline, recursive inlining
7314 happens only when @option{-finline-functions} (included in @option{-O3}) is
7315 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
7318 @item min-inline-recursive-probability
7319 Recursive inlining is profitable only for function having deep recursion
7320 in average and can hurt for function having little recursion depth by
7321 increasing the prologue size or complexity of function body to other
7324 When profile feedback is available (see @option{-fprofile-generate}) the actual
7325 recursion depth can be guessed from probability that function will recurse via
7326 given call expression. This parameter limits inlining only to call expression
7327 whose probability exceeds given threshold (in percents). The default value is
7330 @item inline-call-cost
7331 Specify cost of call instruction relative to simple arithmetics operations
7332 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
7333 functions and at the same time increases size of leaf function that is believed to
7334 reduce function size by being inlined. In effect it increases amount of
7335 inlining for code having large abstraction penalty (many functions that just
7336 pass the arguments to other functions) and decrease inlining for code with low
7337 abstraction penalty. The default value is 12.
7339 @item min-vect-loop-bound
7340 The minimum number of iterations under which a loop will not get vectorized
7341 when @option{-ftree-vectorize} is used. The number of iterations after
7342 vectorization needs to be greater than the value specified by this option
7343 to allow vectorization. The default value is 0.
7345 @item max-unrolled-insns
7346 The maximum number of instructions that a loop should have if that loop
7347 is unrolled, and if the loop is unrolled, it determines how many times
7348 the loop code is unrolled.
7350 @item max-average-unrolled-insns
7351 The maximum number of instructions biased by probabilities of their execution
7352 that a loop should have if that loop is unrolled, and if the loop is unrolled,
7353 it determines how many times the loop code is unrolled.
7355 @item max-unroll-times
7356 The maximum number of unrollings of a single loop.
7358 @item max-peeled-insns
7359 The maximum number of instructions that a loop should have if that loop
7360 is peeled, and if the loop is peeled, it determines how many times
7361 the loop code is peeled.
7363 @item max-peel-times
7364 The maximum number of peelings of a single loop.
7366 @item max-completely-peeled-insns
7367 The maximum number of insns of a completely peeled loop.
7369 @item max-completely-peel-times
7370 The maximum number of iterations of a loop to be suitable for complete peeling.
7372 @item max-unswitch-insns
7373 The maximum number of insns of an unswitched loop.
7375 @item max-unswitch-level
7376 The maximum number of branches unswitched in a single loop.
7379 The minimum cost of an expensive expression in the loop invariant motion.
7381 @item iv-consider-all-candidates-bound
7382 Bound on number of candidates for induction variables below that
7383 all candidates are considered for each use in induction variable
7384 optimizations. Only the most relevant candidates are considered
7385 if there are more candidates, to avoid quadratic time complexity.
7387 @item iv-max-considered-uses
7388 The induction variable optimizations give up on loops that contain more
7389 induction variable uses.
7391 @item iv-always-prune-cand-set-bound
7392 If number of candidates in the set is smaller than this value,
7393 we always try to remove unnecessary ivs from the set during its
7394 optimization when a new iv is added to the set.
7396 @item scev-max-expr-size
7397 Bound on size of expressions used in the scalar evolutions analyzer.
7398 Large expressions slow the analyzer.
7400 @item omega-max-vars
7401 The maximum number of variables in an Omega constraint system.
7402 The default value is 128.
7404 @item omega-max-geqs
7405 The maximum number of inequalities in an Omega constraint system.
7406 The default value is 256.
7409 The maximum number of equalities in an Omega constraint system.
7410 The default value is 128.
7412 @item omega-max-wild-cards
7413 The maximum number of wildcard variables that the Omega solver will
7414 be able to insert. The default value is 18.
7416 @item omega-hash-table-size
7417 The size of the hash table in the Omega solver. The default value is
7420 @item omega-max-keys
7421 The maximal number of keys used by the Omega solver. The default
7424 @item omega-eliminate-redundant-constraints
7425 When set to 1, use expensive methods to eliminate all redundant
7426 constraints. The default value is 0.
7428 @item vect-max-version-for-alignment-checks
7429 The maximum number of runtime checks that can be performed when
7430 doing loop versioning for alignment in the vectorizer. See option
7431 ftree-vect-loop-version for more information.
7433 @item vect-max-version-for-alias-checks
7434 The maximum number of runtime checks that can be performed when
7435 doing loop versioning for alias in the vectorizer. See option
7436 ftree-vect-loop-version for more information.
7438 @item max-iterations-to-track
7440 The maximum number of iterations of a loop the brute force algorithm
7441 for analysis of # of iterations of the loop tries to evaluate.
7443 @item hot-bb-count-fraction
7444 Select fraction of the maximal count of repetitions of basic block in program
7445 given basic block needs to have to be considered hot.
7447 @item hot-bb-frequency-fraction
7448 Select fraction of the maximal frequency of executions of basic block in
7449 function given basic block needs to have to be considered hot
7451 @item max-predicted-iterations
7452 The maximum number of loop iterations we predict statically. This is useful
7453 in cases where function contain single loop with known bound and other loop
7454 with unknown. We predict the known number of iterations correctly, while
7455 the unknown number of iterations average to roughly 10. This means that the
7456 loop without bounds would appear artificially cold relative to the other one.
7458 @item align-threshold
7460 Select fraction of the maximal frequency of executions of basic block in
7461 function given basic block will get aligned.
7463 @item align-loop-iterations
7465 A loop expected to iterate at lest the selected number of iterations will get
7468 @item tracer-dynamic-coverage
7469 @itemx tracer-dynamic-coverage-feedback
7471 This value is used to limit superblock formation once the given percentage of
7472 executed instructions is covered. This limits unnecessary code size
7475 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7476 feedback is available. The real profiles (as opposed to statically estimated
7477 ones) are much less balanced allowing the threshold to be larger value.
7479 @item tracer-max-code-growth
7480 Stop tail duplication once code growth has reached given percentage. This is
7481 rather hokey argument, as most of the duplicates will be eliminated later in
7482 cross jumping, so it may be set to much higher values than is the desired code
7485 @item tracer-min-branch-ratio
7487 Stop reverse growth when the reverse probability of best edge is less than this
7488 threshold (in percent).
7490 @item tracer-min-branch-ratio
7491 @itemx tracer-min-branch-ratio-feedback
7493 Stop forward growth if the best edge do have probability lower than this
7496 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7497 compilation for profile feedback and one for compilation without. The value
7498 for compilation with profile feedback needs to be more conservative (higher) in
7499 order to make tracer effective.
7501 @item max-cse-path-length
7503 Maximum number of basic blocks on path that cse considers. The default is 10.
7506 The maximum instructions CSE process before flushing. The default is 1000.
7508 @item max-aliased-vops
7510 Maximum number of virtual operands per function allowed to represent
7511 aliases before triggering the alias partitioning heuristic. Alias
7512 partitioning reduces compile times and memory consumption needed for
7513 aliasing at the expense of precision loss in alias information. The
7514 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7517 Notice that if a function contains more memory statements than the
7518 value of this parameter, it is not really possible to achieve this
7519 reduction. In this case, the compiler will use the number of memory
7520 statements as the value for @option{max-aliased-vops}.
7522 @item avg-aliased-vops
7524 Average number of virtual operands per statement allowed to represent
7525 aliases before triggering the alias partitioning heuristic. This
7526 works in conjunction with @option{max-aliased-vops}. If a function
7527 contains more than @option{max-aliased-vops} virtual operators, then
7528 memory symbols will be grouped into memory partitions until either the
7529 total number of virtual operators is below @option{max-aliased-vops}
7530 or the average number of virtual operators per memory statement is
7531 below @option{avg-aliased-vops}. The default value for this parameter
7532 is 1 for -O1 and -O2, and 3 for -O3.
7534 @item ggc-min-expand
7536 GCC uses a garbage collector to manage its own memory allocation. This
7537 parameter specifies the minimum percentage by which the garbage
7538 collector's heap should be allowed to expand between collections.
7539 Tuning this may improve compilation speed; it has no effect on code
7542 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7543 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7544 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7545 GCC is not able to calculate RAM on a particular platform, the lower
7546 bound of 30% is used. Setting this parameter and
7547 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7548 every opportunity. This is extremely slow, but can be useful for
7551 @item ggc-min-heapsize
7553 Minimum size of the garbage collector's heap before it begins bothering
7554 to collect garbage. The first collection occurs after the heap expands
7555 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7556 tuning this may improve compilation speed, and has no effect on code
7559 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7560 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7561 with a lower bound of 4096 (four megabytes) and an upper bound of
7562 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7563 particular platform, the lower bound is used. Setting this parameter
7564 very large effectively disables garbage collection. Setting this
7565 parameter and @option{ggc-min-expand} to zero causes a full collection
7566 to occur at every opportunity.
7568 @item max-reload-search-insns
7569 The maximum number of instruction reload should look backward for equivalent
7570 register. Increasing values mean more aggressive optimization, making the
7571 compile time increase with probably slightly better performance. The default
7574 @item max-cselib-memory-locations
7575 The maximum number of memory locations cselib should take into account.
7576 Increasing values mean more aggressive optimization, making the compile time
7577 increase with probably slightly better performance. The default value is 500.
7579 @item reorder-blocks-duplicate
7580 @itemx reorder-blocks-duplicate-feedback
7582 Used by basic block reordering pass to decide whether to use unconditional
7583 branch or duplicate the code on its destination. Code is duplicated when its
7584 estimated size is smaller than this value multiplied by the estimated size of
7585 unconditional jump in the hot spots of the program.
7587 The @option{reorder-block-duplicate-feedback} is used only when profile
7588 feedback is available and may be set to higher values than
7589 @option{reorder-block-duplicate} since information about the hot spots is more
7592 @item max-sched-ready-insns
7593 The maximum number of instructions ready to be issued the scheduler should
7594 consider at any given time during the first scheduling pass. Increasing
7595 values mean more thorough searches, making the compilation time increase
7596 with probably little benefit. The default value is 100.
7598 @item max-sched-region-blocks
7599 The maximum number of blocks in a region to be considered for
7600 interblock scheduling. The default value is 10.
7602 @item max-pipeline-region-blocks
7603 The maximum number of blocks in a region to be considered for
7604 pipelining in the selective scheduler. The default value is 15.
7606 @item max-sched-region-insns
7607 The maximum number of insns in a region to be considered for
7608 interblock scheduling. The default value is 100.
7610 @item max-pipeline-region-insns
7611 The maximum number of insns in a region to be considered for
7612 pipelining in the selective scheduler. The default value is 200.
7615 The minimum probability (in percents) of reaching a source block
7616 for interblock speculative scheduling. The default value is 40.
7618 @item max-sched-extend-regions-iters
7619 The maximum number of iterations through CFG to extend regions.
7620 0 - disable region extension,
7621 N - do at most N iterations.
7622 The default value is 0.
7624 @item max-sched-insn-conflict-delay
7625 The maximum conflict delay for an insn to be considered for speculative motion.
7626 The default value is 3.
7628 @item sched-spec-prob-cutoff
7629 The minimal probability of speculation success (in percents), so that
7630 speculative insn will be scheduled.
7631 The default value is 40.
7633 @item sched-mem-true-dep-cost
7634 Minimal distance (in CPU cycles) between store and load targeting same
7635 memory locations. The default value is 1.
7637 @item selsched-max-lookahead
7638 The maximum size of the lookahead window of selective scheduling. It is a
7639 depth of search for available instructions.
7640 The default value is 50.
7642 @item selsched-max-sched-times
7643 The maximum number of times that an instruction will be scheduled during
7644 selective scheduling. This is the limit on the number of iterations
7645 through which the instruction may be pipelined. The default value is 2.
7647 @item selsched-max-insns-to-rename
7648 The maximum number of best instructions in the ready list that are considered
7649 for renaming in the selective scheduler. The default value is 2.
7651 @item max-last-value-rtl
7652 The maximum size measured as number of RTLs that can be recorded in an expression
7653 in combiner for a pseudo register as last known value of that register. The default
7656 @item integer-share-limit
7657 Small integer constants can use a shared data structure, reducing the
7658 compiler's memory usage and increasing its speed. This sets the maximum
7659 value of a shared integer constant. The default value is 256.
7661 @item min-virtual-mappings
7662 Specifies the minimum number of virtual mappings in the incremental
7663 SSA updater that should be registered to trigger the virtual mappings
7664 heuristic defined by virtual-mappings-ratio. The default value is
7667 @item virtual-mappings-ratio
7668 If the number of virtual mappings is virtual-mappings-ratio bigger
7669 than the number of virtual symbols to be updated, then the incremental
7670 SSA updater switches to a full update for those symbols. The default
7673 @item ssp-buffer-size
7674 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
7675 protection when @option{-fstack-protection} is used.
7677 @item max-jump-thread-duplication-stmts
7678 Maximum number of statements allowed in a block that needs to be
7679 duplicated when threading jumps.
7681 @item max-fields-for-field-sensitive
7682 Maximum number of fields in a structure we will treat in
7683 a field sensitive manner during pointer analysis. The default is zero
7684 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
7686 @item prefetch-latency
7687 Estimate on average number of instructions that are executed before
7688 prefetch finishes. The distance we prefetch ahead is proportional
7689 to this constant. Increasing this number may also lead to less
7690 streams being prefetched (see @option{simultaneous-prefetches}).
7692 @item simultaneous-prefetches
7693 Maximum number of prefetches that can run at the same time.
7695 @item l1-cache-line-size
7696 The size of cache line in L1 cache, in bytes.
7699 The size of L1 cache, in kilobytes.
7702 The size of L2 cache, in kilobytes.
7704 @item use-canonical-types
7705 Whether the compiler should use the ``canonical'' type system. By
7706 default, this should always be 1, which uses a more efficient internal
7707 mechanism for comparing types in C++ and Objective-C++. However, if
7708 bugs in the canonical type system are causing compilation failures,
7709 set this value to 0 to disable canonical types.
7711 @item switch-conversion-max-branch-ratio
7712 Switch initialization conversion will refuse to create arrays that are
7713 bigger than @option{switch-conversion-max-branch-ratio} times the number of
7714 branches in the switch.
7716 @item max-partial-antic-length
7717 Maximum length of the partial antic set computed during the tree
7718 partial redundancy elimination optimization (@option{-ftree-pre}) when
7719 optimizing at @option{-O3} and above. For some sorts of source code
7720 the enhanced partial redundancy elimination optimization can run away,
7721 consuming all of the memory available on the host machine. This
7722 parameter sets a limit on the length of the sets that are computed,
7723 which prevents the runaway behavior. Setting a value of 0 for
7724 this parameter will allow an unlimited set length.
7726 @item sccvn-max-scc-size
7727 Maximum size of a strongly connected component (SCC) during SCCVN
7728 processing. If this limit is hit, SCCVN processing for the whole
7729 function will not be done and optimizations depending on it will
7730 be disabled. The default maximum SCC size is 10000.
7732 @item ira-max-loops-num
7733 IRA uses a regional register allocation by default. If a function
7734 contains loops more than number given by the parameter, only at most
7735 given number of the most frequently executed loops will form regions
7736 for the regional register allocation. The default value of the
7739 @item ira-max-conflict-table-size
7740 Although IRA uses a sophisticated algorithm of compression conflict
7741 table, the table can be still big for huge functions. If the conflict
7742 table for a function could be more than size in MB given by the
7743 parameter, the conflict table is not built and faster, simpler, and
7744 lower quality register allocation algorithm will be used. The
7745 algorithm do not use pseudo-register conflicts. The default value of
7746 the parameter is 2000.
7751 @node Preprocessor Options
7752 @section Options Controlling the Preprocessor
7753 @cindex preprocessor options
7754 @cindex options, preprocessor
7756 These options control the C preprocessor, which is run on each C source
7757 file before actual compilation.
7759 If you use the @option{-E} option, nothing is done except preprocessing.
7760 Some of these options make sense only together with @option{-E} because
7761 they cause the preprocessor output to be unsuitable for actual
7766 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7767 and pass @var{option} directly through to the preprocessor. If
7768 @var{option} contains commas, it is split into multiple options at the
7769 commas. However, many options are modified, translated or interpreted
7770 by the compiler driver before being passed to the preprocessor, and
7771 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7772 interface is undocumented and subject to change, so whenever possible
7773 you should avoid using @option{-Wp} and let the driver handle the
7776 @item -Xpreprocessor @var{option}
7777 @opindex preprocessor
7778 Pass @var{option} as an option to the preprocessor. You can use this to
7779 supply system-specific preprocessor options which GCC does not know how to
7782 If you want to pass an option that takes an argument, you must use
7783 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7786 @include cppopts.texi
7788 @node Assembler Options
7789 @section Passing Options to the Assembler
7791 @c prevent bad page break with this line
7792 You can pass options to the assembler.
7795 @item -Wa,@var{option}
7797 Pass @var{option} as an option to the assembler. If @var{option}
7798 contains commas, it is split into multiple options at the commas.
7800 @item -Xassembler @var{option}
7802 Pass @var{option} as an option to the assembler. You can use this to
7803 supply system-specific assembler options which GCC does not know how to
7806 If you want to pass an option that takes an argument, you must use
7807 @option{-Xassembler} twice, once for the option and once for the argument.
7812 @section Options for Linking
7813 @cindex link options
7814 @cindex options, linking
7816 These options come into play when the compiler links object files into
7817 an executable output file. They are meaningless if the compiler is
7818 not doing a link step.
7822 @item @var{object-file-name}
7823 A file name that does not end in a special recognized suffix is
7824 considered to name an object file or library. (Object files are
7825 distinguished from libraries by the linker according to the file
7826 contents.) If linking is done, these object files are used as input
7835 If any of these options is used, then the linker is not run, and
7836 object file names should not be used as arguments. @xref{Overall
7840 @item -l@var{library}
7841 @itemx -l @var{library}
7843 Search the library named @var{library} when linking. (The second
7844 alternative with the library as a separate argument is only for
7845 POSIX compliance and is not recommended.)
7847 It makes a difference where in the command you write this option; the
7848 linker searches and processes libraries and object files in the order they
7849 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7850 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7851 to functions in @samp{z}, those functions may not be loaded.
7853 The linker searches a standard list of directories for the library,
7854 which is actually a file named @file{lib@var{library}.a}. The linker
7855 then uses this file as if it had been specified precisely by name.
7857 The directories searched include several standard system directories
7858 plus any that you specify with @option{-L}.
7860 Normally the files found this way are library files---archive files
7861 whose members are object files. The linker handles an archive file by
7862 scanning through it for members which define symbols that have so far
7863 been referenced but not defined. But if the file that is found is an
7864 ordinary object file, it is linked in the usual fashion. The only
7865 difference between using an @option{-l} option and specifying a file name
7866 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7867 and searches several directories.
7871 You need this special case of the @option{-l} option in order to
7872 link an Objective-C or Objective-C++ program.
7875 @opindex nostartfiles
7876 Do not use the standard system startup files when linking.
7877 The standard system libraries are used normally, unless @option{-nostdlib}
7878 or @option{-nodefaultlibs} is used.
7880 @item -nodefaultlibs
7881 @opindex nodefaultlibs
7882 Do not use the standard system libraries when linking.
7883 Only the libraries you specify will be passed to the linker.
7884 The standard startup files are used normally, unless @option{-nostartfiles}
7885 is used. The compiler may generate calls to @code{memcmp},
7886 @code{memset}, @code{memcpy} and @code{memmove}.
7887 These entries are usually resolved by entries in
7888 libc. These entry points should be supplied through some other
7889 mechanism when this option is specified.
7893 Do not use the standard system startup files or libraries when linking.
7894 No startup files and only the libraries you specify will be passed to
7895 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7896 @code{memcpy} and @code{memmove}.
7897 These entries are usually resolved by entries in
7898 libc. These entry points should be supplied through some other
7899 mechanism when this option is specified.
7901 @cindex @option{-lgcc}, use with @option{-nostdlib}
7902 @cindex @option{-nostdlib} and unresolved references
7903 @cindex unresolved references and @option{-nostdlib}
7904 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7905 @cindex @option{-nodefaultlibs} and unresolved references
7906 @cindex unresolved references and @option{-nodefaultlibs}
7907 One of the standard libraries bypassed by @option{-nostdlib} and
7908 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7909 that GCC uses to overcome shortcomings of particular machines, or special
7910 needs for some languages.
7911 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7912 Collection (GCC) Internals},
7913 for more discussion of @file{libgcc.a}.)
7914 In most cases, you need @file{libgcc.a} even when you want to avoid
7915 other standard libraries. In other words, when you specify @option{-nostdlib}
7916 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7917 This ensures that you have no unresolved references to internal GCC
7918 library subroutines. (For example, @samp{__main}, used to ensure C++
7919 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7920 GNU Compiler Collection (GCC) Internals}.)
7924 Produce a position independent executable on targets which support it.
7925 For predictable results, you must also specify the same set of options
7926 that were used to generate code (@option{-fpie}, @option{-fPIE},
7927 or model suboptions) when you specify this option.
7931 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7932 that support it. This instructs the linker to add all symbols, not
7933 only used ones, to the dynamic symbol table. This option is needed
7934 for some uses of @code{dlopen} or to allow obtaining backtraces
7935 from within a program.
7939 Remove all symbol table and relocation information from the executable.
7943 On systems that support dynamic linking, this prevents linking with the shared
7944 libraries. On other systems, this option has no effect.
7948 Produce a shared object which can then be linked with other objects to
7949 form an executable. Not all systems support this option. For predictable
7950 results, you must also specify the same set of options that were used to
7951 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7952 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7953 needs to build supplementary stub code for constructors to work. On
7954 multi-libbed systems, @samp{gcc -shared} must select the correct support
7955 libraries to link against. Failing to supply the correct flags may lead
7956 to subtle defects. Supplying them in cases where they are not necessary
7959 @item -shared-libgcc
7960 @itemx -static-libgcc
7961 @opindex shared-libgcc
7962 @opindex static-libgcc
7963 On systems that provide @file{libgcc} as a shared library, these options
7964 force the use of either the shared or static version respectively.
7965 If no shared version of @file{libgcc} was built when the compiler was
7966 configured, these options have no effect.
7968 There are several situations in which an application should use the
7969 shared @file{libgcc} instead of the static version. The most common
7970 of these is when the application wishes to throw and catch exceptions
7971 across different shared libraries. In that case, each of the libraries
7972 as well as the application itself should use the shared @file{libgcc}.
7974 Therefore, the G++ and GCJ drivers automatically add
7975 @option{-shared-libgcc} whenever you build a shared library or a main
7976 executable, because C++ and Java programs typically use exceptions, so
7977 this is the right thing to do.
7979 If, instead, you use the GCC driver to create shared libraries, you may
7980 find that they will not always be linked with the shared @file{libgcc}.
7981 If GCC finds, at its configuration time, that you have a non-GNU linker
7982 or a GNU linker that does not support option @option{--eh-frame-hdr},
7983 it will link the shared version of @file{libgcc} into shared libraries
7984 by default. Otherwise, it will take advantage of the linker and optimize
7985 away the linking with the shared version of @file{libgcc}, linking with
7986 the static version of libgcc by default. This allows exceptions to
7987 propagate through such shared libraries, without incurring relocation
7988 costs at library load time.
7990 However, if a library or main executable is supposed to throw or catch
7991 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7992 for the languages used in the program, or using the option
7993 @option{-shared-libgcc}, such that it is linked with the shared
7998 Bind references to global symbols when building a shared object. Warn
7999 about any unresolved references (unless overridden by the link editor
8000 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8003 @item -T @var{script}
8005 @cindex linker script
8006 Use @var{script} as the linker script. This option is supported by most
8007 systems using the GNU linker. On some targets, such as bare-board
8008 targets without an operating system, the @option{-T} option may be required
8009 when linking to avoid references to undefined symbols.
8011 @item -Xlinker @var{option}
8013 Pass @var{option} as an option to the linker. You can use this to
8014 supply system-specific linker options which GCC does not know how to
8017 If you want to pass an option that takes an argument, you must use
8018 @option{-Xlinker} twice, once for the option and once for the argument.
8019 For example, to pass @option{-assert definitions}, you must write
8020 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8021 @option{-Xlinker "-assert definitions"}, because this passes the entire
8022 string as a single argument, which is not what the linker expects.
8024 @item -Wl,@var{option}
8026 Pass @var{option} as an option to the linker. If @var{option} contains
8027 commas, it is split into multiple options at the commas.
8029 @item -u @var{symbol}
8031 Pretend the symbol @var{symbol} is undefined, to force linking of
8032 library modules to define it. You can use @option{-u} multiple times with
8033 different symbols to force loading of additional library modules.
8036 @node Directory Options
8037 @section Options for Directory Search
8038 @cindex directory options
8039 @cindex options, directory search
8042 These options specify directories to search for header files, for
8043 libraries and for parts of the compiler:
8048 Add the directory @var{dir} to the head of the list of directories to be
8049 searched for header files. This can be used to override a system header
8050 file, substituting your own version, since these directories are
8051 searched before the system header file directories. However, you should
8052 not use this option to add directories that contain vendor-supplied
8053 system header files (use @option{-isystem} for that). If you use more than
8054 one @option{-I} option, the directories are scanned in left-to-right
8055 order; the standard system directories come after.
8057 If a standard system include directory, or a directory specified with
8058 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8059 option will be ignored. The directory will still be searched but as a
8060 system directory at its normal position in the system include chain.
8061 This is to ensure that GCC's procedure to fix buggy system headers and
8062 the ordering for the include_next directive are not inadvertently changed.
8063 If you really need to change the search order for system directories,
8064 use the @option{-nostdinc} and/or @option{-isystem} options.
8066 @item -iquote@var{dir}
8068 Add the directory @var{dir} to the head of the list of directories to
8069 be searched for header files only for the case of @samp{#include
8070 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8071 otherwise just like @option{-I}.
8075 Add directory @var{dir} to the list of directories to be searched
8078 @item -B@var{prefix}
8080 This option specifies where to find the executables, libraries,
8081 include files, and data files of the compiler itself.
8083 The compiler driver program runs one or more of the subprograms
8084 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
8085 @var{prefix} as a prefix for each program it tries to run, both with and
8086 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
8088 For each subprogram to be run, the compiler driver first tries the
8089 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
8090 was not specified, the driver tries two standard prefixes, which are
8091 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
8092 those results in a file name that is found, the unmodified program
8093 name is searched for using the directories specified in your
8094 @env{PATH} environment variable.
8096 The compiler will check to see if the path provided by the @option{-B}
8097 refers to a directory, and if necessary it will add a directory
8098 separator character at the end of the path.
8100 @option{-B} prefixes that effectively specify directory names also apply
8101 to libraries in the linker, because the compiler translates these
8102 options into @option{-L} options for the linker. They also apply to
8103 includes files in the preprocessor, because the compiler translates these
8104 options into @option{-isystem} options for the preprocessor. In this case,
8105 the compiler appends @samp{include} to the prefix.
8107 The run-time support file @file{libgcc.a} can also be searched for using
8108 the @option{-B} prefix, if needed. If it is not found there, the two
8109 standard prefixes above are tried, and that is all. The file is left
8110 out of the link if it is not found by those means.
8112 Another way to specify a prefix much like the @option{-B} prefix is to use
8113 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
8116 As a special kludge, if the path provided by @option{-B} is
8117 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
8118 9, then it will be replaced by @file{[dir/]include}. This is to help
8119 with boot-strapping the compiler.
8121 @item -specs=@var{file}
8123 Process @var{file} after the compiler reads in the standard @file{specs}
8124 file, in order to override the defaults that the @file{gcc} driver
8125 program uses when determining what switches to pass to @file{cc1},
8126 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
8127 @option{-specs=@var{file}} can be specified on the command line, and they
8128 are processed in order, from left to right.
8130 @item --sysroot=@var{dir}
8132 Use @var{dir} as the logical root directory for headers and libraries.
8133 For example, if the compiler would normally search for headers in
8134 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
8135 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
8137 If you use both this option and the @option{-isysroot} option, then
8138 the @option{--sysroot} option will apply to libraries, but the
8139 @option{-isysroot} option will apply to header files.
8141 The GNU linker (beginning with version 2.16) has the necessary support
8142 for this option. If your linker does not support this option, the
8143 header file aspect of @option{--sysroot} will still work, but the
8144 library aspect will not.
8148 This option has been deprecated. Please use @option{-iquote} instead for
8149 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
8150 Any directories you specify with @option{-I} options before the @option{-I-}
8151 option are searched only for the case of @samp{#include "@var{file}"};
8152 they are not searched for @samp{#include <@var{file}>}.
8154 If additional directories are specified with @option{-I} options after
8155 the @option{-I-}, these directories are searched for all @samp{#include}
8156 directives. (Ordinarily @emph{all} @option{-I} directories are used
8159 In addition, the @option{-I-} option inhibits the use of the current
8160 directory (where the current input file came from) as the first search
8161 directory for @samp{#include "@var{file}"}. There is no way to
8162 override this effect of @option{-I-}. With @option{-I.} you can specify
8163 searching the directory which was current when the compiler was
8164 invoked. That is not exactly the same as what the preprocessor does
8165 by default, but it is often satisfactory.
8167 @option{-I-} does not inhibit the use of the standard system directories
8168 for header files. Thus, @option{-I-} and @option{-nostdinc} are
8175 @section Specifying subprocesses and the switches to pass to them
8178 @command{gcc} is a driver program. It performs its job by invoking a
8179 sequence of other programs to do the work of compiling, assembling and
8180 linking. GCC interprets its command-line parameters and uses these to
8181 deduce which programs it should invoke, and which command-line options
8182 it ought to place on their command lines. This behavior is controlled
8183 by @dfn{spec strings}. In most cases there is one spec string for each
8184 program that GCC can invoke, but a few programs have multiple spec
8185 strings to control their behavior. The spec strings built into GCC can
8186 be overridden by using the @option{-specs=} command-line switch to specify
8189 @dfn{Spec files} are plaintext files that are used to construct spec
8190 strings. They consist of a sequence of directives separated by blank
8191 lines. The type of directive is determined by the first non-whitespace
8192 character on the line and it can be one of the following:
8195 @item %@var{command}
8196 Issues a @var{command} to the spec file processor. The commands that can
8200 @item %include <@var{file}>
8202 Search for @var{file} and insert its text at the current point in the
8205 @item %include_noerr <@var{file}>
8206 @cindex %include_noerr
8207 Just like @samp{%include}, but do not generate an error message if the include
8208 file cannot be found.
8210 @item %rename @var{old_name} @var{new_name}
8212 Rename the spec string @var{old_name} to @var{new_name}.
8216 @item *[@var{spec_name}]:
8217 This tells the compiler to create, override or delete the named spec
8218 string. All lines after this directive up to the next directive or
8219 blank line are considered to be the text for the spec string. If this
8220 results in an empty string then the spec will be deleted. (Or, if the
8221 spec did not exist, then nothing will happened.) Otherwise, if the spec
8222 does not currently exist a new spec will be created. If the spec does
8223 exist then its contents will be overridden by the text of this
8224 directive, unless the first character of that text is the @samp{+}
8225 character, in which case the text will be appended to the spec.
8227 @item [@var{suffix}]:
8228 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
8229 and up to the next directive or blank line are considered to make up the
8230 spec string for the indicated suffix. When the compiler encounters an
8231 input file with the named suffix, it will processes the spec string in
8232 order to work out how to compile that file. For example:
8239 This says that any input file whose name ends in @samp{.ZZ} should be
8240 passed to the program @samp{z-compile}, which should be invoked with the
8241 command-line switch @option{-input} and with the result of performing the
8242 @samp{%i} substitution. (See below.)
8244 As an alternative to providing a spec string, the text that follows a
8245 suffix directive can be one of the following:
8248 @item @@@var{language}
8249 This says that the suffix is an alias for a known @var{language}. This is
8250 similar to using the @option{-x} command-line switch to GCC to specify a
8251 language explicitly. For example:
8258 Says that .ZZ files are, in fact, C++ source files.
8261 This causes an error messages saying:
8264 @var{name} compiler not installed on this system.
8268 GCC already has an extensive list of suffixes built into it.
8269 This directive will add an entry to the end of the list of suffixes, but
8270 since the list is searched from the end backwards, it is effectively
8271 possible to override earlier entries using this technique.
8275 GCC has the following spec strings built into it. Spec files can
8276 override these strings or create their own. Note that individual
8277 targets can also add their own spec strings to this list.
8280 asm Options to pass to the assembler
8281 asm_final Options to pass to the assembler post-processor
8282 cpp Options to pass to the C preprocessor
8283 cc1 Options to pass to the C compiler
8284 cc1plus Options to pass to the C++ compiler
8285 endfile Object files to include at the end of the link
8286 link Options to pass to the linker
8287 lib Libraries to include on the command line to the linker
8288 libgcc Decides which GCC support library to pass to the linker
8289 linker Sets the name of the linker
8290 predefines Defines to be passed to the C preprocessor
8291 signed_char Defines to pass to CPP to say whether @code{char} is signed
8293 startfile Object files to include at the start of the link
8296 Here is a small example of a spec file:
8302 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
8305 This example renames the spec called @samp{lib} to @samp{old_lib} and
8306 then overrides the previous definition of @samp{lib} with a new one.
8307 The new definition adds in some extra command-line options before
8308 including the text of the old definition.
8310 @dfn{Spec strings} are a list of command-line options to be passed to their
8311 corresponding program. In addition, the spec strings can contain
8312 @samp{%}-prefixed sequences to substitute variable text or to
8313 conditionally insert text into the command line. Using these constructs
8314 it is possible to generate quite complex command lines.
8316 Here is a table of all defined @samp{%}-sequences for spec
8317 strings. Note that spaces are not generated automatically around the
8318 results of expanding these sequences. Therefore you can concatenate them
8319 together or combine them with constant text in a single argument.
8323 Substitute one @samp{%} into the program name or argument.
8326 Substitute the name of the input file being processed.
8329 Substitute the basename of the input file being processed.
8330 This is the substring up to (and not including) the last period
8331 and not including the directory.
8334 This is the same as @samp{%b}, but include the file suffix (text after
8338 Marks the argument containing or following the @samp{%d} as a
8339 temporary file name, so that that file will be deleted if GCC exits
8340 successfully. Unlike @samp{%g}, this contributes no text to the
8343 @item %g@var{suffix}
8344 Substitute a file name that has suffix @var{suffix} and is chosen
8345 once per compilation, and mark the argument in the same way as
8346 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
8347 name is now chosen in a way that is hard to predict even when previously
8348 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
8349 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
8350 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
8351 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
8352 was simply substituted with a file name chosen once per compilation,
8353 without regard to any appended suffix (which was therefore treated
8354 just like ordinary text), making such attacks more likely to succeed.
8356 @item %u@var{suffix}
8357 Like @samp{%g}, but generates a new temporary file name even if
8358 @samp{%u@var{suffix}} was already seen.
8360 @item %U@var{suffix}
8361 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
8362 new one if there is no such last file name. In the absence of any
8363 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
8364 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
8365 would involve the generation of two distinct file names, one
8366 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
8367 simply substituted with a file name chosen for the previous @samp{%u},
8368 without regard to any appended suffix.
8370 @item %j@var{suffix}
8371 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
8372 writable, and if save-temps is off; otherwise, substitute the name
8373 of a temporary file, just like @samp{%u}. This temporary file is not
8374 meant for communication between processes, but rather as a junk
8377 @item %|@var{suffix}
8378 @itemx %m@var{suffix}
8379 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
8380 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
8381 all. These are the two most common ways to instruct a program that it
8382 should read from standard input or write to standard output. If you
8383 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
8384 construct: see for example @file{f/lang-specs.h}.
8386 @item %.@var{SUFFIX}
8387 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
8388 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
8389 terminated by the next space or %.
8392 Marks the argument containing or following the @samp{%w} as the
8393 designated output file of this compilation. This puts the argument
8394 into the sequence of arguments that @samp{%o} will substitute later.
8397 Substitutes the names of all the output files, with spaces
8398 automatically placed around them. You should write spaces
8399 around the @samp{%o} as well or the results are undefined.
8400 @samp{%o} is for use in the specs for running the linker.
8401 Input files whose names have no recognized suffix are not compiled
8402 at all, but they are included among the output files, so they will
8406 Substitutes the suffix for object files. Note that this is
8407 handled specially when it immediately follows @samp{%g, %u, or %U},
8408 because of the need for those to form complete file names. The
8409 handling is such that @samp{%O} is treated exactly as if it had already
8410 been substituted, except that @samp{%g, %u, and %U} do not currently
8411 support additional @var{suffix} characters following @samp{%O} as they would
8412 following, for example, @samp{.o}.
8415 Substitutes the standard macro predefinitions for the
8416 current target machine. Use this when running @code{cpp}.
8419 Like @samp{%p}, but puts @samp{__} before and after the name of each
8420 predefined macro, except for macros that start with @samp{__} or with
8421 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
8425 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
8426 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
8427 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
8428 and @option{-imultilib} as necessary.
8431 Current argument is the name of a library or startup file of some sort.
8432 Search for that file in a standard list of directories and substitute
8433 the full name found.
8436 Print @var{str} as an error message. @var{str} is terminated by a newline.
8437 Use this when inconsistent options are detected.
8440 Substitute the contents of spec string @var{name} at this point.
8443 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
8445 @item %x@{@var{option}@}
8446 Accumulate an option for @samp{%X}.
8449 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
8453 Output the accumulated assembler options specified by @option{-Wa}.
8456 Output the accumulated preprocessor options specified by @option{-Wp}.
8459 Process the @code{asm} spec. This is used to compute the
8460 switches to be passed to the assembler.
8463 Process the @code{asm_final} spec. This is a spec string for
8464 passing switches to an assembler post-processor, if such a program is
8468 Process the @code{link} spec. This is the spec for computing the
8469 command line passed to the linker. Typically it will make use of the
8470 @samp{%L %G %S %D and %E} sequences.
8473 Dump out a @option{-L} option for each directory that GCC believes might
8474 contain startup files. If the target supports multilibs then the
8475 current multilib directory will be prepended to each of these paths.
8478 Process the @code{lib} spec. This is a spec string for deciding which
8479 libraries should be included on the command line to the linker.
8482 Process the @code{libgcc} spec. This is a spec string for deciding
8483 which GCC support library should be included on the command line to the linker.
8486 Process the @code{startfile} spec. This is a spec for deciding which
8487 object files should be the first ones passed to the linker. Typically
8488 this might be a file named @file{crt0.o}.
8491 Process the @code{endfile} spec. This is a spec string that specifies
8492 the last object files that will be passed to the linker.
8495 Process the @code{cpp} spec. This is used to construct the arguments
8496 to be passed to the C preprocessor.
8499 Process the @code{cc1} spec. This is used to construct the options to be
8500 passed to the actual C compiler (@samp{cc1}).
8503 Process the @code{cc1plus} spec. This is used to construct the options to be
8504 passed to the actual C++ compiler (@samp{cc1plus}).
8507 Substitute the variable part of a matched option. See below.
8508 Note that each comma in the substituted string is replaced by
8512 Remove all occurrences of @code{-S} from the command line. Note---this
8513 command is position dependent. @samp{%} commands in the spec string
8514 before this one will see @code{-S}, @samp{%} commands in the spec string
8515 after this one will not.
8517 @item %:@var{function}(@var{args})
8518 Call the named function @var{function}, passing it @var{args}.
8519 @var{args} is first processed as a nested spec string, then split
8520 into an argument vector in the usual fashion. The function returns
8521 a string which is processed as if it had appeared literally as part
8522 of the current spec.
8524 The following built-in spec functions are provided:
8528 The @code{getenv} spec function takes two arguments: an environment
8529 variable name and a string. If the environment variable is not
8530 defined, a fatal error is issued. Otherwise, the return value is the
8531 value of the environment variable concatenated with the string. For
8532 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8535 %:getenv(TOPDIR /include)
8538 expands to @file{/path/to/top/include}.
8540 @item @code{if-exists}
8541 The @code{if-exists} spec function takes one argument, an absolute
8542 pathname to a file. If the file exists, @code{if-exists} returns the
8543 pathname. Here is a small example of its usage:
8547 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
8550 @item @code{if-exists-else}
8551 The @code{if-exists-else} spec function is similar to the @code{if-exists}
8552 spec function, except that it takes two arguments. The first argument is
8553 an absolute pathname to a file. If the file exists, @code{if-exists-else}
8554 returns the pathname. If it does not exist, it returns the second argument.
8555 This way, @code{if-exists-else} can be used to select one file or another,
8556 based on the existence of the first. Here is a small example of its usage:
8560 crt0%O%s %:if-exists(crti%O%s) \
8561 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8564 @item @code{replace-outfile}
8565 The @code{replace-outfile} spec function takes two arguments. It looks for the
8566 first argument in the outfiles array and replaces it with the second argument. Here
8567 is a small example of its usage:
8570 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8573 @item @code{print-asm-header}
8574 The @code{print-asm-header} function takes no arguments and simply
8575 prints a banner like:
8581 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8584 It is used to separate compiler options from assembler options
8585 in the @option{--target-help} output.
8589 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8590 If that switch was not specified, this substitutes nothing. Note that
8591 the leading dash is omitted when specifying this option, and it is
8592 automatically inserted if the substitution is performed. Thus the spec
8593 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8594 and would output the command line option @option{-foo}.
8596 @item %W@{@code{S}@}
8597 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8600 @item %@{@code{S}*@}
8601 Substitutes all the switches specified to GCC whose names start
8602 with @code{-S}, but which also take an argument. This is used for
8603 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8604 GCC considers @option{-o foo} as being
8605 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
8606 text, including the space. Thus two arguments would be generated.
8608 @item %@{@code{S}*&@code{T}*@}
8609 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8610 (the order of @code{S} and @code{T} in the spec is not significant).
8611 There can be any number of ampersand-separated variables; for each the
8612 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8614 @item %@{@code{S}:@code{X}@}
8615 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8617 @item %@{!@code{S}:@code{X}@}
8618 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8620 @item %@{@code{S}*:@code{X}@}
8621 Substitutes @code{X} if one or more switches whose names start with
8622 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8623 once, no matter how many such switches appeared. However, if @code{%*}
8624 appears somewhere in @code{X}, then @code{X} will be substituted once
8625 for each matching switch, with the @code{%*} replaced by the part of
8626 that switch that matched the @code{*}.
8628 @item %@{.@code{S}:@code{X}@}
8629 Substitutes @code{X}, if processing a file with suffix @code{S}.
8631 @item %@{!.@code{S}:@code{X}@}
8632 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8634 @item %@{,@code{S}:@code{X}@}
8635 Substitutes @code{X}, if processing a file for language @code{S}.
8637 @item %@{!,@code{S}:@code{X}@}
8638 Substitutes @code{X}, if not processing a file for language @code{S}.
8640 @item %@{@code{S}|@code{P}:@code{X}@}
8641 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8642 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8643 @code{*} sequences as well, although they have a stronger binding than
8644 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8645 alternatives must be starred, and only the first matching alternative
8648 For example, a spec string like this:
8651 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8654 will output the following command-line options from the following input
8655 command-line options:
8660 -d fred.c -foo -baz -boggle
8661 -d jim.d -bar -baz -boggle
8664 @item %@{S:X; T:Y; :D@}
8666 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8667 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8668 be as many clauses as you need. This may be combined with @code{.},
8669 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8674 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8675 construct may contain other nested @samp{%} constructs or spaces, or
8676 even newlines. They are processed as usual, as described above.
8677 Trailing white space in @code{X} is ignored. White space may also
8678 appear anywhere on the left side of the colon in these constructs,
8679 except between @code{.} or @code{*} and the corresponding word.
8681 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8682 handled specifically in these constructs. If another value of
8683 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8684 @option{-W} switch is found later in the command line, the earlier
8685 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8686 just one letter, which passes all matching options.
8688 The character @samp{|} at the beginning of the predicate text is used to
8689 indicate that a command should be piped to the following command, but
8690 only if @option{-pipe} is specified.
8692 It is built into GCC which switches take arguments and which do not.
8693 (You might think it would be useful to generalize this to allow each
8694 compiler's spec to say which switches take arguments. But this cannot
8695 be done in a consistent fashion. GCC cannot even decide which input
8696 files have been specified without knowing which switches take arguments,
8697 and it must know which input files to compile in order to tell which
8700 GCC also knows implicitly that arguments starting in @option{-l} are to be
8701 treated as compiler output files, and passed to the linker in their
8702 proper position among the other output files.
8704 @c man begin OPTIONS
8706 @node Target Options
8707 @section Specifying Target Machine and Compiler Version
8708 @cindex target options
8709 @cindex cross compiling
8710 @cindex specifying machine version
8711 @cindex specifying compiler version and target machine
8712 @cindex compiler version, specifying
8713 @cindex target machine, specifying
8715 The usual way to run GCC is to run the executable called @file{gcc}, or
8716 @file{<machine>-gcc} when cross-compiling, or
8717 @file{<machine>-gcc-<version>} to run a version other than the one that
8718 was installed last. Sometimes this is inconvenient, so GCC provides
8719 options that will switch to another cross-compiler or version.
8722 @item -b @var{machine}
8724 The argument @var{machine} specifies the target machine for compilation.
8726 The value to use for @var{machine} is the same as was specified as the
8727 machine type when configuring GCC as a cross-compiler. For
8728 example, if a cross-compiler was configured with @samp{configure
8729 arm-elf}, meaning to compile for an arm processor with elf binaries,
8730 then you would specify @option{-b arm-elf} to run that cross compiler.
8731 Because there are other options beginning with @option{-b}, the
8732 configuration must contain a hyphen, or @option{-b} alone should be one
8733 argument followed by the configuration in the next argument.
8735 @item -V @var{version}
8737 The argument @var{version} specifies which version of GCC to run.
8738 This is useful when multiple versions are installed. For example,
8739 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8742 The @option{-V} and @option{-b} options work by running the
8743 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8744 use them if you can just run that directly.
8746 @node Submodel Options
8747 @section Hardware Models and Configurations
8748 @cindex submodel options
8749 @cindex specifying hardware config
8750 @cindex hardware models and configurations, specifying
8751 @cindex machine dependent options
8753 Earlier we discussed the standard option @option{-b} which chooses among
8754 different installed compilers for completely different target
8755 machines, such as VAX vs.@: 68000 vs.@: 80386.
8757 In addition, each of these target machine types can have its own
8758 special options, starting with @samp{-m}, to choose among various
8759 hardware models or configurations---for example, 68010 vs 68020,
8760 floating coprocessor or none. A single installed version of the
8761 compiler can compile for any model or configuration, according to the
8764 Some configurations of the compiler also support additional special
8765 options, usually for compatibility with other compilers on the same
8768 @c This list is ordered alphanumerically by subsection name.
8769 @c It should be the same order and spelling as these options are listed
8770 @c in Machine Dependent Options
8776 * Blackfin Options::
8780 * DEC Alpha Options::
8781 * DEC Alpha/VMS Options::
8783 * GNU/Linux Options::
8786 * i386 and x86-64 Options::
8797 * picoChip Options::
8799 * RS/6000 and PowerPC Options::
8800 * S/390 and zSeries Options::
8805 * System V Options::
8810 * Xstormy16 Options::
8816 @subsection ARC Options
8819 These options are defined for ARC implementations:
8824 Compile code for little endian mode. This is the default.
8828 Compile code for big endian mode.
8831 @opindex mmangle-cpu
8832 Prepend the name of the cpu to all public symbol names.
8833 In multiple-processor systems, there are many ARC variants with different
8834 instruction and register set characteristics. This flag prevents code
8835 compiled for one cpu to be linked with code compiled for another.
8836 No facility exists for handling variants that are ``almost identical''.
8837 This is an all or nothing option.
8839 @item -mcpu=@var{cpu}
8841 Compile code for ARC variant @var{cpu}.
8842 Which variants are supported depend on the configuration.
8843 All variants support @option{-mcpu=base}, this is the default.
8845 @item -mtext=@var{text-section}
8846 @itemx -mdata=@var{data-section}
8847 @itemx -mrodata=@var{readonly-data-section}
8851 Put functions, data, and readonly data in @var{text-section},
8852 @var{data-section}, and @var{readonly-data-section} respectively
8853 by default. This can be overridden with the @code{section} attribute.
8854 @xref{Variable Attributes}.
8856 @item -mfix-cortex-m3-ldrd
8857 @opindex mfix-cortex-m3-ldrd
8858 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
8859 with overlapping destination and base registers are used. This option avoids
8860 generating these instructions. This option is enabled by default when
8861 @option{-mcpu=cortex-m3} is specified.
8866 @subsection ARM Options
8869 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8873 @item -mabi=@var{name}
8875 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8876 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8879 @opindex mapcs-frame
8880 Generate a stack frame that is compliant with the ARM Procedure Call
8881 Standard for all functions, even if this is not strictly necessary for
8882 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8883 with this option will cause the stack frames not to be generated for
8884 leaf functions. The default is @option{-mno-apcs-frame}.
8888 This is a synonym for @option{-mapcs-frame}.
8891 @c not currently implemented
8892 @item -mapcs-stack-check
8893 @opindex mapcs-stack-check
8894 Generate code to check the amount of stack space available upon entry to
8895 every function (that actually uses some stack space). If there is
8896 insufficient space available then either the function
8897 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8898 called, depending upon the amount of stack space required. The run time
8899 system is required to provide these functions. The default is
8900 @option{-mno-apcs-stack-check}, since this produces smaller code.
8902 @c not currently implemented
8904 @opindex mapcs-float
8905 Pass floating point arguments using the float point registers. This is
8906 one of the variants of the APCS@. This option is recommended if the
8907 target hardware has a floating point unit or if a lot of floating point
8908 arithmetic is going to be performed by the code. The default is
8909 @option{-mno-apcs-float}, since integer only code is slightly increased in
8910 size if @option{-mapcs-float} is used.
8912 @c not currently implemented
8913 @item -mapcs-reentrant
8914 @opindex mapcs-reentrant
8915 Generate reentrant, position independent code. The default is
8916 @option{-mno-apcs-reentrant}.
8919 @item -mthumb-interwork
8920 @opindex mthumb-interwork
8921 Generate code which supports calling between the ARM and Thumb
8922 instruction sets. Without this option the two instruction sets cannot
8923 be reliably used inside one program. The default is
8924 @option{-mno-thumb-interwork}, since slightly larger code is generated
8925 when @option{-mthumb-interwork} is specified.
8927 @item -mno-sched-prolog
8928 @opindex mno-sched-prolog
8929 Prevent the reordering of instructions in the function prolog, or the
8930 merging of those instruction with the instructions in the function's
8931 body. This means that all functions will start with a recognizable set
8932 of instructions (or in fact one of a choice from a small set of
8933 different function prologues), and this information can be used to
8934 locate the start if functions inside an executable piece of code. The
8935 default is @option{-msched-prolog}.
8937 @item -mfloat-abi=@var{name}
8939 Specifies which floating-point ABI to use. Permissible values
8940 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8942 Specifying @samp{soft} causes GCC to generate output containing
8943 library calls for floating-point operations.
8944 @samp{softfp} allows the generation of code using hardware floating-point
8945 instructions, but still uses the soft-float calling conventions.
8946 @samp{hard} allows generation of floating-point instructions
8947 and uses FPU-specific calling conventions.
8949 Using @option{-mfloat-abi=hard} with VFP coprocessors is not supported.
8950 Use @option{-mfloat-abi=softfp} with the appropriate @option{-mfpu} option
8951 to allow the compiler to generate code that makes use of the hardware
8952 floating-point capabilities for these CPUs.
8954 The default depends on the specific target configuration. Note that
8955 the hard-float and soft-float ABIs are not link-compatible; you must
8956 compile your entire program with the same ABI, and link with a
8957 compatible set of libraries.
8960 @opindex mhard-float
8961 Equivalent to @option{-mfloat-abi=hard}.
8964 @opindex msoft-float
8965 Equivalent to @option{-mfloat-abi=soft}.
8967 @item -mlittle-endian
8968 @opindex mlittle-endian
8969 Generate code for a processor running in little-endian mode. This is
8970 the default for all standard configurations.
8973 @opindex mbig-endian
8974 Generate code for a processor running in big-endian mode; the default is
8975 to compile code for a little-endian processor.
8977 @item -mwords-little-endian
8978 @opindex mwords-little-endian
8979 This option only applies when generating code for big-endian processors.
8980 Generate code for a little-endian word order but a big-endian byte
8981 order. That is, a byte order of the form @samp{32107654}. Note: this
8982 option should only be used if you require compatibility with code for
8983 big-endian ARM processors generated by versions of the compiler prior to
8986 @item -mcpu=@var{name}
8988 This specifies the name of the target ARM processor. GCC uses this name
8989 to determine what kind of instructions it can emit when generating
8990 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8991 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8992 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8993 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8994 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8996 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8997 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
8998 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8999 @samp{strongarm1110},
9000 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9001 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9002 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9003 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9004 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9005 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9006 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9007 @samp{cortex-a8}, @samp{cortex-a9},
9008 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9010 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9012 @item -mtune=@var{name}
9014 This option is very similar to the @option{-mcpu=} option, except that
9015 instead of specifying the actual target processor type, and hence
9016 restricting which instructions can be used, it specifies that GCC should
9017 tune the performance of the code as if the target were of the type
9018 specified in this option, but still choosing the instructions that it
9019 will generate based on the cpu specified by a @option{-mcpu=} option.
9020 For some ARM implementations better performance can be obtained by using
9023 @item -march=@var{name}
9025 This specifies the name of the target ARM architecture. GCC uses this
9026 name to determine what kind of instructions it can emit when generating
9027 assembly code. This option can be used in conjunction with or instead
9028 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9029 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9030 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9031 @samp{armv6}, @samp{armv6j},
9032 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9033 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9034 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9036 @item -mfpu=@var{name}
9037 @itemx -mfpe=@var{number}
9038 @itemx -mfp=@var{number}
9042 This specifies what floating point hardware (or hardware emulation) is
9043 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9044 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-d16} and
9045 @samp{neon}. @option{-mfp} and @option{-mfpe}
9046 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
9047 with older versions of GCC@.
9049 If @option{-msoft-float} is specified this specifies the format of
9050 floating point values.
9052 @item -mstructure-size-boundary=@var{n}
9053 @opindex mstructure-size-boundary
9054 The size of all structures and unions will be rounded up to a multiple
9055 of the number of bits set by this option. Permissible values are 8, 32
9056 and 64. The default value varies for different toolchains. For the COFF
9057 targeted toolchain the default value is 8. A value of 64 is only allowed
9058 if the underlying ABI supports it.
9060 Specifying the larger number can produce faster, more efficient code, but
9061 can also increase the size of the program. Different values are potentially
9062 incompatible. Code compiled with one value cannot necessarily expect to
9063 work with code or libraries compiled with another value, if they exchange
9064 information using structures or unions.
9066 @item -mabort-on-noreturn
9067 @opindex mabort-on-noreturn
9068 Generate a call to the function @code{abort} at the end of a
9069 @code{noreturn} function. It will be executed if the function tries to
9073 @itemx -mno-long-calls
9074 @opindex mlong-calls
9075 @opindex mno-long-calls
9076 Tells the compiler to perform function calls by first loading the
9077 address of the function into a register and then performing a subroutine
9078 call on this register. This switch is needed if the target function
9079 will lie outside of the 64 megabyte addressing range of the offset based
9080 version of subroutine call instruction.
9082 Even if this switch is enabled, not all function calls will be turned
9083 into long calls. The heuristic is that static functions, functions
9084 which have the @samp{short-call} attribute, functions that are inside
9085 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9086 definitions have already been compiled within the current compilation
9087 unit, will not be turned into long calls. The exception to this rule is
9088 that weak function definitions, functions with the @samp{long-call}
9089 attribute or the @samp{section} attribute, and functions that are within
9090 the scope of a @samp{#pragma long_calls} directive, will always be
9091 turned into long calls.
9093 This feature is not enabled by default. Specifying
9094 @option{-mno-long-calls} will restore the default behavior, as will
9095 placing the function calls within the scope of a @samp{#pragma
9096 long_calls_off} directive. Note these switches have no effect on how
9097 the compiler generates code to handle function calls via function
9100 @item -mnop-fun-dllimport
9101 @opindex mnop-fun-dllimport
9102 Disable support for the @code{dllimport} attribute.
9104 @item -msingle-pic-base
9105 @opindex msingle-pic-base
9106 Treat the register used for PIC addressing as read-only, rather than
9107 loading it in the prologue for each function. The run-time system is
9108 responsible for initializing this register with an appropriate value
9109 before execution begins.
9111 @item -mpic-register=@var{reg}
9112 @opindex mpic-register
9113 Specify the register to be used for PIC addressing. The default is R10
9114 unless stack-checking is enabled, when R9 is used.
9116 @item -mcirrus-fix-invalid-insns
9117 @opindex mcirrus-fix-invalid-insns
9118 @opindex mno-cirrus-fix-invalid-insns
9119 Insert NOPs into the instruction stream to in order to work around
9120 problems with invalid Maverick instruction combinations. This option
9121 is only valid if the @option{-mcpu=ep9312} option has been used to
9122 enable generation of instructions for the Cirrus Maverick floating
9123 point co-processor. This option is not enabled by default, since the
9124 problem is only present in older Maverick implementations. The default
9125 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
9128 @item -mpoke-function-name
9129 @opindex mpoke-function-name
9130 Write the name of each function into the text section, directly
9131 preceding the function prologue. The generated code is similar to this:
9135 .ascii "arm_poke_function_name", 0
9138 .word 0xff000000 + (t1 - t0)
9139 arm_poke_function_name
9141 stmfd sp!, @{fp, ip, lr, pc@}
9145 When performing a stack backtrace, code can inspect the value of
9146 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
9147 location @code{pc - 12} and the top 8 bits are set, then we know that
9148 there is a function name embedded immediately preceding this location
9149 and has length @code{((pc[-3]) & 0xff000000)}.
9153 Generate code for the Thumb instruction set. The default is to
9154 use the 32-bit ARM instruction set.
9155 This option automatically enables either 16-bit Thumb-1 or
9156 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
9157 and @option{-march=@var{name}} options.
9160 @opindex mtpcs-frame
9161 Generate a stack frame that is compliant with the Thumb Procedure Call
9162 Standard for all non-leaf functions. (A leaf function is one that does
9163 not call any other functions.) The default is @option{-mno-tpcs-frame}.
9165 @item -mtpcs-leaf-frame
9166 @opindex mtpcs-leaf-frame
9167 Generate a stack frame that is compliant with the Thumb Procedure Call
9168 Standard for all leaf functions. (A leaf function is one that does
9169 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
9171 @item -mcallee-super-interworking
9172 @opindex mcallee-super-interworking
9173 Gives all externally visible functions in the file being compiled an ARM
9174 instruction set header which switches to Thumb mode before executing the
9175 rest of the function. This allows these functions to be called from
9176 non-interworking code.
9178 @item -mcaller-super-interworking
9179 @opindex mcaller-super-interworking
9180 Allows calls via function pointers (including virtual functions) to
9181 execute correctly regardless of whether the target code has been
9182 compiled for interworking or not. There is a small overhead in the cost
9183 of executing a function pointer if this option is enabled.
9185 @item -mtp=@var{name}
9187 Specify the access model for the thread local storage pointer. The valid
9188 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
9189 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
9190 (supported in the arm6k architecture), and @option{auto}, which uses the
9191 best available method for the selected processor. The default setting is
9194 @item -mword-relocations
9195 @opindex mword-relocations
9196 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
9197 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
9198 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
9204 @subsection AVR Options
9207 These options are defined for AVR implementations:
9210 @item -mmcu=@var{mcu}
9212 Specify ATMEL AVR instruction set or MCU type.
9214 Instruction set avr1 is for the minimal AVR core, not supported by the C
9215 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9216 attiny11, attiny12, attiny15, attiny28).
9218 Instruction set avr2 (default) is for the classic AVR core with up to
9219 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9220 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9221 at90c8534, at90s8535).
9223 Instruction set avr3 is for the classic AVR core with up to 128K program
9224 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9226 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9227 memory space (MCU types: atmega8, atmega83, atmega85).
9229 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9230 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9231 atmega64, atmega128, at43usb355, at94k).
9235 Output instruction sizes to the asm file.
9237 @item -minit-stack=@var{N}
9238 @opindex minit-stack
9239 Specify the initial stack address, which may be a symbol or numeric value,
9240 @samp{__stack} is the default.
9242 @item -mno-interrupts
9243 @opindex mno-interrupts
9244 Generated code is not compatible with hardware interrupts.
9245 Code size will be smaller.
9247 @item -mcall-prologues
9248 @opindex mcall-prologues
9249 Functions prologues/epilogues expanded as call to appropriate
9250 subroutines. Code size will be smaller.
9252 @item -mno-tablejump
9253 @opindex mno-tablejump
9254 Do not generate tablejump insns which sometimes increase code size.
9257 @opindex mtiny-stack
9258 Change only the low 8 bits of the stack pointer.
9262 Assume int to be 8 bit integer. This affects the sizes of all types: A
9263 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
9264 and long long will be 4 bytes. Please note that this option does not
9265 comply to the C standards, but it will provide you with smaller code
9269 @node Blackfin Options
9270 @subsection Blackfin Options
9271 @cindex Blackfin Options
9274 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
9276 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
9277 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
9278 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
9279 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
9280 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
9281 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
9283 The optional @var{sirevision} specifies the silicon revision of the target
9284 Blackfin processor. Any workarounds available for the targeted silicon revision
9285 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
9286 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
9287 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
9288 hexadecimal digits representing the major and minor numbers in the silicon
9289 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
9290 is not defined. If @var{sirevision} is @samp{any}, the
9291 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
9292 If this optional @var{sirevision} is not used, GCC assumes the latest known
9293 silicon revision of the targeted Blackfin processor.
9295 Support for @samp{bf561} is incomplete. For @samp{bf561},
9296 Only the processor macro is defined.
9297 Without this option, @samp{bf532} is used as the processor by default.
9298 The corresponding predefined processor macros for @var{cpu} is to
9299 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
9300 provided by libgloss to be linked in if @option{-msim} is not given.
9304 Specifies that the program will be run on the simulator. This causes
9305 the simulator BSP provided by libgloss to be linked in. This option
9306 has effect only for @samp{bfin-elf} toolchain.
9307 Certain other options, such as @option{-mid-shared-library} and
9308 @option{-mfdpic}, imply @option{-msim}.
9310 @item -momit-leaf-frame-pointer
9311 @opindex momit-leaf-frame-pointer
9312 Don't keep the frame pointer in a register for leaf functions. This
9313 avoids the instructions to save, set up and restore frame pointers and
9314 makes an extra register available in leaf functions. The option
9315 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9316 which might make debugging harder.
9318 @item -mspecld-anomaly
9319 @opindex mspecld-anomaly
9320 When enabled, the compiler will ensure that the generated code does not
9321 contain speculative loads after jump instructions. If this option is used,
9322 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
9324 @item -mno-specld-anomaly
9325 @opindex mno-specld-anomaly
9326 Don't generate extra code to prevent speculative loads from occurring.
9328 @item -mcsync-anomaly
9329 @opindex mcsync-anomaly
9330 When enabled, the compiler will ensure that the generated code does not
9331 contain CSYNC or SSYNC instructions too soon after conditional branches.
9332 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
9334 @item -mno-csync-anomaly
9335 @opindex mno-csync-anomaly
9336 Don't generate extra code to prevent CSYNC or SSYNC instructions from
9337 occurring too soon after a conditional branch.
9341 When enabled, the compiler is free to take advantage of the knowledge that
9342 the entire program fits into the low 64k of memory.
9345 @opindex mno-low-64k
9346 Assume that the program is arbitrarily large. This is the default.
9348 @item -mstack-check-l1
9349 @opindex mstack-check-l1
9350 Do stack checking using information placed into L1 scratchpad memory by the
9353 @item -mid-shared-library
9354 @opindex mid-shared-library
9355 Generate code that supports shared libraries via the library ID method.
9356 This allows for execute in place and shared libraries in an environment
9357 without virtual memory management. This option implies @option{-fPIC}.
9358 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9360 @item -mno-id-shared-library
9361 @opindex mno-id-shared-library
9362 Generate code that doesn't assume ID based shared libraries are being used.
9363 This is the default.
9365 @item -mleaf-id-shared-library
9366 @opindex mleaf-id-shared-library
9367 Generate code that supports shared libraries via the library ID method,
9368 but assumes that this library or executable won't link against any other
9369 ID shared libraries. That allows the compiler to use faster code for jumps
9372 @item -mno-leaf-id-shared-library
9373 @opindex mno-leaf-id-shared-library
9374 Do not assume that the code being compiled won't link against any ID shared
9375 libraries. Slower code will be generated for jump and call insns.
9377 @item -mshared-library-id=n
9378 @opindex mshared-library-id
9379 Specified the identification number of the ID based shared library being
9380 compiled. Specifying a value of 0 will generate more compact code, specifying
9381 other values will force the allocation of that number to the current
9382 library but is no more space or time efficient than omitting this option.
9386 Generate code that allows the data segment to be located in a different
9387 area of memory from the text segment. This allows for execute in place in
9388 an environment without virtual memory management by eliminating relocations
9389 against the text section.
9392 @opindex mno-sep-data
9393 Generate code that assumes that the data segment follows the text segment.
9394 This is the default.
9397 @itemx -mno-long-calls
9398 @opindex mlong-calls
9399 @opindex mno-long-calls
9400 Tells the compiler to perform function calls by first loading the
9401 address of the function into a register and then performing a subroutine
9402 call on this register. This switch is needed if the target function
9403 will lie outside of the 24 bit addressing range of the offset based
9404 version of subroutine call instruction.
9406 This feature is not enabled by default. Specifying
9407 @option{-mno-long-calls} will restore the default behavior. Note these
9408 switches have no effect on how the compiler generates code to handle
9409 function calls via function pointers.
9413 Link with the fast floating-point library. This library relaxes some of
9414 the IEEE floating-point standard's rules for checking inputs against
9415 Not-a-Number (NAN), in the interest of performance.
9418 @opindex minline-plt
9419 Enable inlining of PLT entries in function calls to functions that are
9420 not known to bind locally. It has no effect without @option{-mfdpic}.
9424 Build standalone application for multicore Blackfin processor. Proper
9425 start files and link scripts will be used to support multicore.
9426 This option defines @code{__BFIN_MULTICORE}. It can only be used with
9427 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
9428 @option{-mcorea} or @option{-mcoreb}. If it's used without
9429 @option{-mcorea} or @option{-mcoreb}, single application/dual core
9430 programming model is used. In this model, the main function of Core B
9431 should be named as coreb_main. If it's used with @option{-mcorea} or
9432 @option{-mcoreb}, one application per core programming model is used.
9433 If this option is not used, single core application programming
9438 Build standalone application for Core A of BF561 when using
9439 one application per core programming model. Proper start files
9440 and link scripts will be used to support Core A. This option
9441 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
9445 Build standalone application for Core B of BF561 when using
9446 one application per core programming model. Proper start files
9447 and link scripts will be used to support Core B. This option
9448 defines @code{__BFIN_COREB}. When this option is used, coreb_main
9449 should be used instead of main. It must be used with
9450 @option{-mmulticore}.
9454 Build standalone application for SDRAM. Proper start files and
9455 link scripts will be used to put the application into SDRAM.
9456 Loader should initialize SDRAM before loading the application
9457 into SDRAM. This option defines @code{__BFIN_SDRAM}.
9461 Assume that ICPLBs are enabled at runtime. This has an effect on certain
9462 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
9463 are enabled; for standalone applications the default is off.
9467 @subsection CRIS Options
9468 @cindex CRIS Options
9470 These options are defined specifically for the CRIS ports.
9473 @item -march=@var{architecture-type}
9474 @itemx -mcpu=@var{architecture-type}
9477 Generate code for the specified architecture. The choices for
9478 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9479 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
9480 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9483 @item -mtune=@var{architecture-type}
9485 Tune to @var{architecture-type} everything applicable about the generated
9486 code, except for the ABI and the set of available instructions. The
9487 choices for @var{architecture-type} are the same as for
9488 @option{-march=@var{architecture-type}}.
9490 @item -mmax-stack-frame=@var{n}
9491 @opindex mmax-stack-frame
9492 Warn when the stack frame of a function exceeds @var{n} bytes.
9498 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9499 @option{-march=v3} and @option{-march=v8} respectively.
9501 @item -mmul-bug-workaround
9502 @itemx -mno-mul-bug-workaround
9503 @opindex mmul-bug-workaround
9504 @opindex mno-mul-bug-workaround
9505 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
9506 models where it applies. This option is active by default.
9510 Enable CRIS-specific verbose debug-related information in the assembly
9511 code. This option also has the effect to turn off the @samp{#NO_APP}
9512 formatted-code indicator to the assembler at the beginning of the
9517 Do not use condition-code results from previous instruction; always emit
9518 compare and test instructions before use of condition codes.
9520 @item -mno-side-effects
9521 @opindex mno-side-effects
9522 Do not emit instructions with side-effects in addressing modes other than
9526 @itemx -mno-stack-align
9528 @itemx -mno-data-align
9529 @itemx -mconst-align
9530 @itemx -mno-const-align
9531 @opindex mstack-align
9532 @opindex mno-stack-align
9533 @opindex mdata-align
9534 @opindex mno-data-align
9535 @opindex mconst-align
9536 @opindex mno-const-align
9537 These options (no-options) arranges (eliminate arrangements) for the
9538 stack-frame, individual data and constants to be aligned for the maximum
9539 single data access size for the chosen CPU model. The default is to
9540 arrange for 32-bit alignment. ABI details such as structure layout are
9541 not affected by these options.
9549 Similar to the stack- data- and const-align options above, these options
9550 arrange for stack-frame, writable data and constants to all be 32-bit,
9551 16-bit or 8-bit aligned. The default is 32-bit alignment.
9553 @item -mno-prologue-epilogue
9554 @itemx -mprologue-epilogue
9555 @opindex mno-prologue-epilogue
9556 @opindex mprologue-epilogue
9557 With @option{-mno-prologue-epilogue}, the normal function prologue and
9558 epilogue that sets up the stack-frame are omitted and no return
9559 instructions or return sequences are generated in the code. Use this
9560 option only together with visual inspection of the compiled code: no
9561 warnings or errors are generated when call-saved registers must be saved,
9562 or storage for local variable needs to be allocated.
9568 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9569 instruction sequences that load addresses for functions from the PLT part
9570 of the GOT rather than (traditional on other architectures) calls to the
9571 PLT@. The default is @option{-mgotplt}.
9575 Legacy no-op option only recognized with the cris-axis-elf and
9576 cris-axis-linux-gnu targets.
9580 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9584 This option, recognized for the cris-axis-elf arranges
9585 to link with input-output functions from a simulator library. Code,
9586 initialized data and zero-initialized data are allocated consecutively.
9590 Like @option{-sim}, but pass linker options to locate initialized data at
9591 0x40000000 and zero-initialized data at 0x80000000.
9595 @subsection CRX Options
9598 These options are defined specifically for the CRX ports.
9604 Enable the use of multiply-accumulate instructions. Disabled by default.
9608 Push instructions will be used to pass outgoing arguments when functions
9609 are called. Enabled by default.
9612 @node Darwin Options
9613 @subsection Darwin Options
9614 @cindex Darwin options
9616 These options are defined for all architectures running the Darwin operating
9619 FSF GCC on Darwin does not create ``fat'' object files; it will create
9620 an object file for the single architecture that it was built to
9621 target. Apple's GCC on Darwin does create ``fat'' files if multiple
9622 @option{-arch} options are used; it does so by running the compiler or
9623 linker multiple times and joining the results together with
9626 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9627 @samp{i686}) is determined by the flags that specify the ISA
9628 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
9629 @option{-force_cpusubtype_ALL} option can be used to override this.
9631 The Darwin tools vary in their behavior when presented with an ISA
9632 mismatch. The assembler, @file{as}, will only permit instructions to
9633 be used that are valid for the subtype of the file it is generating,
9634 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9635 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9636 and print an error if asked to create a shared library with a less
9637 restrictive subtype than its input files (for instance, trying to put
9638 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
9639 for executables, @file{ld}, will quietly give the executable the most
9640 restrictive subtype of any of its input files.
9645 Add the framework directory @var{dir} to the head of the list of
9646 directories to be searched for header files. These directories are
9647 interleaved with those specified by @option{-I} options and are
9648 scanned in a left-to-right order.
9650 A framework directory is a directory with frameworks in it. A
9651 framework is a directory with a @samp{"Headers"} and/or
9652 @samp{"PrivateHeaders"} directory contained directly in it that ends
9653 in @samp{".framework"}. The name of a framework is the name of this
9654 directory excluding the @samp{".framework"}. Headers associated with
9655 the framework are found in one of those two directories, with
9656 @samp{"Headers"} being searched first. A subframework is a framework
9657 directory that is in a framework's @samp{"Frameworks"} directory.
9658 Includes of subframework headers can only appear in a header of a
9659 framework that contains the subframework, or in a sibling subframework
9660 header. Two subframeworks are siblings if they occur in the same
9661 framework. A subframework should not have the same name as a
9662 framework, a warning will be issued if this is violated. Currently a
9663 subframework cannot have subframeworks, in the future, the mechanism
9664 may be extended to support this. The standard frameworks can be found
9665 in @samp{"/System/Library/Frameworks"} and
9666 @samp{"/Library/Frameworks"}. An example include looks like
9667 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9668 the name of the framework and header.h is found in the
9669 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9671 @item -iframework@var{dir}
9673 Like @option{-F} except the directory is a treated as a system
9674 directory. The main difference between this @option{-iframework} and
9675 @option{-F} is that with @option{-iframework} the compiler does not
9676 warn about constructs contained within header files found via
9677 @var{dir}. This option is valid only for the C family of languages.
9681 Emit debugging information for symbols that are used. For STABS
9682 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9683 This is by default ON@.
9687 Emit debugging information for all symbols and types.
9689 @item -mmacosx-version-min=@var{version}
9690 The earliest version of MacOS X that this executable will run on
9691 is @var{version}. Typical values of @var{version} include @code{10.1},
9692 @code{10.2}, and @code{10.3.9}.
9694 If the compiler was built to use the system's headers by default,
9695 then the default for this option is the system version on which the
9696 compiler is running, otherwise the default is to make choices which
9697 are compatible with as many systems and code bases as possible.
9701 Enable kernel development mode. The @option{-mkernel} option sets
9702 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9703 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9704 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9705 applicable. This mode also sets @option{-mno-altivec},
9706 @option{-msoft-float}, @option{-fno-builtin} and
9707 @option{-mlong-branch} for PowerPC targets.
9709 @item -mone-byte-bool
9710 @opindex mone-byte-bool
9711 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9712 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9713 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9714 option has no effect on x86.
9716 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9717 to generate code that is not binary compatible with code generated
9718 without that switch. Using this switch may require recompiling all
9719 other modules in a program, including system libraries. Use this
9720 switch to conform to a non-default data model.
9722 @item -mfix-and-continue
9723 @itemx -ffix-and-continue
9724 @itemx -findirect-data
9725 @opindex mfix-and-continue
9726 @opindex ffix-and-continue
9727 @opindex findirect-data
9728 Generate code suitable for fast turn around development. Needed to
9729 enable gdb to dynamically load @code{.o} files into already running
9730 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9731 are provided for backwards compatibility.
9735 Loads all members of static archive libraries.
9736 See man ld(1) for more information.
9738 @item -arch_errors_fatal
9739 @opindex arch_errors_fatal
9740 Cause the errors having to do with files that have the wrong architecture
9744 @opindex bind_at_load
9745 Causes the output file to be marked such that the dynamic linker will
9746 bind all undefined references when the file is loaded or launched.
9750 Produce a Mach-o bundle format file.
9751 See man ld(1) for more information.
9753 @item -bundle_loader @var{executable}
9754 @opindex bundle_loader
9755 This option specifies the @var{executable} that will be loading the build
9756 output file being linked. See man ld(1) for more information.
9760 When passed this option, GCC will produce a dynamic library instead of
9761 an executable when linking, using the Darwin @file{libtool} command.
9763 @item -force_cpusubtype_ALL
9764 @opindex force_cpusubtype_ALL
9765 This causes GCC's output file to have the @var{ALL} subtype, instead of
9766 one controlled by the @option{-mcpu} or @option{-march} option.
9768 @item -allowable_client @var{client_name}
9770 @itemx -compatibility_version
9771 @itemx -current_version
9773 @itemx -dependency-file
9775 @itemx -dylinker_install_name
9777 @itemx -exported_symbols_list
9779 @itemx -flat_namespace
9780 @itemx -force_flat_namespace
9781 @itemx -headerpad_max_install_names
9784 @itemx -install_name
9785 @itemx -keep_private_externs
9786 @itemx -multi_module
9787 @itemx -multiply_defined
9788 @itemx -multiply_defined_unused
9790 @itemx -no_dead_strip_inits_and_terms
9791 @itemx -nofixprebinding
9794 @itemx -noseglinkedit
9795 @itemx -pagezero_size
9797 @itemx -prebind_all_twolevel_modules
9798 @itemx -private_bundle
9799 @itemx -read_only_relocs
9801 @itemx -sectobjectsymbols
9805 @itemx -sectobjectsymbols
9808 @itemx -segs_read_only_addr
9809 @itemx -segs_read_write_addr
9810 @itemx -seg_addr_table
9811 @itemx -seg_addr_table_filename
9814 @itemx -segs_read_only_addr
9815 @itemx -segs_read_write_addr
9816 @itemx -single_module
9819 @itemx -sub_umbrella
9820 @itemx -twolevel_namespace
9823 @itemx -unexported_symbols_list
9824 @itemx -weak_reference_mismatches
9826 @opindex allowable_client
9827 @opindex client_name
9828 @opindex compatibility_version
9829 @opindex current_version
9831 @opindex dependency-file
9833 @opindex dylinker_install_name
9835 @opindex exported_symbols_list
9837 @opindex flat_namespace
9838 @opindex force_flat_namespace
9839 @opindex headerpad_max_install_names
9842 @opindex install_name
9843 @opindex keep_private_externs
9844 @opindex multi_module
9845 @opindex multiply_defined
9846 @opindex multiply_defined_unused
9848 @opindex no_dead_strip_inits_and_terms
9849 @opindex nofixprebinding
9850 @opindex nomultidefs
9852 @opindex noseglinkedit
9853 @opindex pagezero_size
9855 @opindex prebind_all_twolevel_modules
9856 @opindex private_bundle
9857 @opindex read_only_relocs
9859 @opindex sectobjectsymbols
9863 @opindex sectobjectsymbols
9866 @opindex segs_read_only_addr
9867 @opindex segs_read_write_addr
9868 @opindex seg_addr_table
9869 @opindex seg_addr_table_filename
9870 @opindex seglinkedit
9872 @opindex segs_read_only_addr
9873 @opindex segs_read_write_addr
9874 @opindex single_module
9876 @opindex sub_library
9877 @opindex sub_umbrella
9878 @opindex twolevel_namespace
9881 @opindex unexported_symbols_list
9882 @opindex weak_reference_mismatches
9883 @opindex whatsloaded
9884 These options are passed to the Darwin linker. The Darwin linker man page
9885 describes them in detail.
9888 @node DEC Alpha Options
9889 @subsection DEC Alpha Options
9891 These @samp{-m} options are defined for the DEC Alpha implementations:
9894 @item -mno-soft-float
9896 @opindex mno-soft-float
9897 @opindex msoft-float
9898 Use (do not use) the hardware floating-point instructions for
9899 floating-point operations. When @option{-msoft-float} is specified,
9900 functions in @file{libgcc.a} will be used to perform floating-point
9901 operations. Unless they are replaced by routines that emulate the
9902 floating-point operations, or compiled in such a way as to call such
9903 emulations routines, these routines will issue floating-point
9904 operations. If you are compiling for an Alpha without floating-point
9905 operations, you must ensure that the library is built so as not to call
9908 Note that Alpha implementations without floating-point operations are
9909 required to have floating-point registers.
9914 @opindex mno-fp-regs
9915 Generate code that uses (does not use) the floating-point register set.
9916 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9917 register set is not used, floating point operands are passed in integer
9918 registers as if they were integers and floating-point results are passed
9919 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9920 so any function with a floating-point argument or return value called by code
9921 compiled with @option{-mno-fp-regs} must also be compiled with that
9924 A typical use of this option is building a kernel that does not use,
9925 and hence need not save and restore, any floating-point registers.
9929 The Alpha architecture implements floating-point hardware optimized for
9930 maximum performance. It is mostly compliant with the IEEE floating
9931 point standard. However, for full compliance, software assistance is
9932 required. This option generates code fully IEEE compliant code
9933 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9934 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9935 defined during compilation. The resulting code is less efficient but is
9936 able to correctly support denormalized numbers and exceptional IEEE
9937 values such as not-a-number and plus/minus infinity. Other Alpha
9938 compilers call this option @option{-ieee_with_no_inexact}.
9940 @item -mieee-with-inexact
9941 @opindex mieee-with-inexact
9942 This is like @option{-mieee} except the generated code also maintains
9943 the IEEE @var{inexact-flag}. Turning on this option causes the
9944 generated code to implement fully-compliant IEEE math. In addition to
9945 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9946 macro. On some Alpha implementations the resulting code may execute
9947 significantly slower than the code generated by default. Since there is
9948 very little code that depends on the @var{inexact-flag}, you should
9949 normally not specify this option. Other Alpha compilers call this
9950 option @option{-ieee_with_inexact}.
9952 @item -mfp-trap-mode=@var{trap-mode}
9953 @opindex mfp-trap-mode
9954 This option controls what floating-point related traps are enabled.
9955 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9956 The trap mode can be set to one of four values:
9960 This is the default (normal) setting. The only traps that are enabled
9961 are the ones that cannot be disabled in software (e.g., division by zero
9965 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9969 Like @samp{u}, but the instructions are marked to be safe for software
9970 completion (see Alpha architecture manual for details).
9973 Like @samp{su}, but inexact traps are enabled as well.
9976 @item -mfp-rounding-mode=@var{rounding-mode}
9977 @opindex mfp-rounding-mode
9978 Selects the IEEE rounding mode. Other Alpha compilers call this option
9979 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9984 Normal IEEE rounding mode. Floating point numbers are rounded towards
9985 the nearest machine number or towards the even machine number in case
9989 Round towards minus infinity.
9992 Chopped rounding mode. Floating point numbers are rounded towards zero.
9995 Dynamic rounding mode. A field in the floating point control register
9996 (@var{fpcr}, see Alpha architecture reference manual) controls the
9997 rounding mode in effect. The C library initializes this register for
9998 rounding towards plus infinity. Thus, unless your program modifies the
9999 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10002 @item -mtrap-precision=@var{trap-precision}
10003 @opindex mtrap-precision
10004 In the Alpha architecture, floating point traps are imprecise. This
10005 means without software assistance it is impossible to recover from a
10006 floating trap and program execution normally needs to be terminated.
10007 GCC can generate code that can assist operating system trap handlers
10008 in determining the exact location that caused a floating point trap.
10009 Depending on the requirements of an application, different levels of
10010 precisions can be selected:
10014 Program precision. This option is the default and means a trap handler
10015 can only identify which program caused a floating point exception.
10018 Function precision. The trap handler can determine the function that
10019 caused a floating point exception.
10022 Instruction precision. The trap handler can determine the exact
10023 instruction that caused a floating point exception.
10026 Other Alpha compilers provide the equivalent options called
10027 @option{-scope_safe} and @option{-resumption_safe}.
10029 @item -mieee-conformant
10030 @opindex mieee-conformant
10031 This option marks the generated code as IEEE conformant. You must not
10032 use this option unless you also specify @option{-mtrap-precision=i} and either
10033 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10034 is to emit the line @samp{.eflag 48} in the function prologue of the
10035 generated assembly file. Under DEC Unix, this has the effect that
10036 IEEE-conformant math library routines will be linked in.
10038 @item -mbuild-constants
10039 @opindex mbuild-constants
10040 Normally GCC examines a 32- or 64-bit integer constant to
10041 see if it can construct it from smaller constants in two or three
10042 instructions. If it cannot, it will output the constant as a literal and
10043 generate code to load it from the data segment at runtime.
10045 Use this option to require GCC to construct @emph{all} integer constants
10046 using code, even if it takes more instructions (the maximum is six).
10048 You would typically use this option to build a shared library dynamic
10049 loader. Itself a shared library, it must relocate itself in memory
10050 before it can find the variables and constants in its own data segment.
10056 Select whether to generate code to be assembled by the vendor-supplied
10057 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10075 Indicate whether GCC should generate code to use the optional BWX,
10076 CIX, FIX and MAX instruction sets. The default is to use the instruction
10077 sets supported by the CPU type specified via @option{-mcpu=} option or that
10078 of the CPU on which GCC was built if none was specified.
10081 @itemx -mfloat-ieee
10082 @opindex mfloat-vax
10083 @opindex mfloat-ieee
10084 Generate code that uses (does not use) VAX F and G floating point
10085 arithmetic instead of IEEE single and double precision.
10087 @item -mexplicit-relocs
10088 @itemx -mno-explicit-relocs
10089 @opindex mexplicit-relocs
10090 @opindex mno-explicit-relocs
10091 Older Alpha assemblers provided no way to generate symbol relocations
10092 except via assembler macros. Use of these macros does not allow
10093 optimal instruction scheduling. GNU binutils as of version 2.12
10094 supports a new syntax that allows the compiler to explicitly mark
10095 which relocations should apply to which instructions. This option
10096 is mostly useful for debugging, as GCC detects the capabilities of
10097 the assembler when it is built and sets the default accordingly.
10100 @itemx -mlarge-data
10101 @opindex msmall-data
10102 @opindex mlarge-data
10103 When @option{-mexplicit-relocs} is in effect, static data is
10104 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
10105 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
10106 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
10107 16-bit relocations off of the @code{$gp} register. This limits the
10108 size of the small data area to 64KB, but allows the variables to be
10109 directly accessed via a single instruction.
10111 The default is @option{-mlarge-data}. With this option the data area
10112 is limited to just below 2GB@. Programs that require more than 2GB of
10113 data must use @code{malloc} or @code{mmap} to allocate the data in the
10114 heap instead of in the program's data segment.
10116 When generating code for shared libraries, @option{-fpic} implies
10117 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
10120 @itemx -mlarge-text
10121 @opindex msmall-text
10122 @opindex mlarge-text
10123 When @option{-msmall-text} is used, the compiler assumes that the
10124 code of the entire program (or shared library) fits in 4MB, and is
10125 thus reachable with a branch instruction. When @option{-msmall-data}
10126 is used, the compiler can assume that all local symbols share the
10127 same @code{$gp} value, and thus reduce the number of instructions
10128 required for a function call from 4 to 1.
10130 The default is @option{-mlarge-text}.
10132 @item -mcpu=@var{cpu_type}
10134 Set the instruction set and instruction scheduling parameters for
10135 machine type @var{cpu_type}. You can specify either the @samp{EV}
10136 style name or the corresponding chip number. GCC supports scheduling
10137 parameters for the EV4, EV5 and EV6 family of processors and will
10138 choose the default values for the instruction set from the processor
10139 you specify. If you do not specify a processor type, GCC will default
10140 to the processor on which the compiler was built.
10142 Supported values for @var{cpu_type} are
10148 Schedules as an EV4 and has no instruction set extensions.
10152 Schedules as an EV5 and has no instruction set extensions.
10156 Schedules as an EV5 and supports the BWX extension.
10161 Schedules as an EV5 and supports the BWX and MAX extensions.
10165 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
10169 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
10172 @item -mtune=@var{cpu_type}
10174 Set only the instruction scheduling parameters for machine type
10175 @var{cpu_type}. The instruction set is not changed.
10177 @item -mmemory-latency=@var{time}
10178 @opindex mmemory-latency
10179 Sets the latency the scheduler should assume for typical memory
10180 references as seen by the application. This number is highly
10181 dependent on the memory access patterns used by the application
10182 and the size of the external cache on the machine.
10184 Valid options for @var{time} are
10188 A decimal number representing clock cycles.
10194 The compiler contains estimates of the number of clock cycles for
10195 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
10196 (also called Dcache, Scache, and Bcache), as well as to main memory.
10197 Note that L3 is only valid for EV5.
10202 @node DEC Alpha/VMS Options
10203 @subsection DEC Alpha/VMS Options
10205 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
10208 @item -mvms-return-codes
10209 @opindex mvms-return-codes
10210 Return VMS condition codes from main. The default is to return POSIX
10211 style condition (e.g.@: error) codes.
10215 @subsection FRV Options
10216 @cindex FRV Options
10222 Only use the first 32 general purpose registers.
10227 Use all 64 general purpose registers.
10232 Use only the first 32 floating point registers.
10237 Use all 64 floating point registers
10240 @opindex mhard-float
10242 Use hardware instructions for floating point operations.
10245 @opindex msoft-float
10247 Use library routines for floating point operations.
10252 Dynamically allocate condition code registers.
10257 Do not try to dynamically allocate condition code registers, only
10258 use @code{icc0} and @code{fcc0}.
10263 Change ABI to use double word insns.
10268 Do not use double word instructions.
10273 Use floating point double instructions.
10276 @opindex mno-double
10278 Do not use floating point double instructions.
10283 Use media instructions.
10288 Do not use media instructions.
10293 Use multiply and add/subtract instructions.
10296 @opindex mno-muladd
10298 Do not use multiply and add/subtract instructions.
10303 Select the FDPIC ABI, that uses function descriptors to represent
10304 pointers to functions. Without any PIC/PIE-related options, it
10305 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
10306 assumes GOT entries and small data are within a 12-bit range from the
10307 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
10308 are computed with 32 bits.
10309 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10312 @opindex minline-plt
10314 Enable inlining of PLT entries in function calls to functions that are
10315 not known to bind locally. It has no effect without @option{-mfdpic}.
10316 It's enabled by default if optimizing for speed and compiling for
10317 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
10318 optimization option such as @option{-O3} or above is present in the
10324 Assume a large TLS segment when generating thread-local code.
10329 Do not assume a large TLS segment when generating thread-local code.
10334 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
10335 that is known to be in read-only sections. It's enabled by default,
10336 except for @option{-fpic} or @option{-fpie}: even though it may help
10337 make the global offset table smaller, it trades 1 instruction for 4.
10338 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
10339 one of which may be shared by multiple symbols, and it avoids the need
10340 for a GOT entry for the referenced symbol, so it's more likely to be a
10341 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
10343 @item -multilib-library-pic
10344 @opindex multilib-library-pic
10346 Link with the (library, not FD) pic libraries. It's implied by
10347 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
10348 @option{-fpic} without @option{-mfdpic}. You should never have to use
10352 @opindex mlinked-fp
10354 Follow the EABI requirement of always creating a frame pointer whenever
10355 a stack frame is allocated. This option is enabled by default and can
10356 be disabled with @option{-mno-linked-fp}.
10359 @opindex mlong-calls
10361 Use indirect addressing to call functions outside the current
10362 compilation unit. This allows the functions to be placed anywhere
10363 within the 32-bit address space.
10365 @item -malign-labels
10366 @opindex malign-labels
10368 Try to align labels to an 8-byte boundary by inserting nops into the
10369 previous packet. This option only has an effect when VLIW packing
10370 is enabled. It doesn't create new packets; it merely adds nops to
10373 @item -mlibrary-pic
10374 @opindex mlibrary-pic
10376 Generate position-independent EABI code.
10381 Use only the first four media accumulator registers.
10386 Use all eight media accumulator registers.
10391 Pack VLIW instructions.
10396 Do not pack VLIW instructions.
10399 @opindex mno-eflags
10401 Do not mark ABI switches in e_flags.
10404 @opindex mcond-move
10406 Enable the use of conditional-move instructions (default).
10408 This switch is mainly for debugging the compiler and will likely be removed
10409 in a future version.
10411 @item -mno-cond-move
10412 @opindex mno-cond-move
10414 Disable the use of conditional-move instructions.
10416 This switch is mainly for debugging the compiler and will likely be removed
10417 in a future version.
10422 Enable the use of conditional set instructions (default).
10424 This switch is mainly for debugging the compiler and will likely be removed
10425 in a future version.
10430 Disable the use of conditional set instructions.
10432 This switch is mainly for debugging the compiler and will likely be removed
10433 in a future version.
10436 @opindex mcond-exec
10438 Enable the use of conditional execution (default).
10440 This switch is mainly for debugging the compiler and will likely be removed
10441 in a future version.
10443 @item -mno-cond-exec
10444 @opindex mno-cond-exec
10446 Disable the use of conditional execution.
10448 This switch is mainly for debugging the compiler and will likely be removed
10449 in a future version.
10451 @item -mvliw-branch
10452 @opindex mvliw-branch
10454 Run a pass to pack branches into VLIW instructions (default).
10456 This switch is mainly for debugging the compiler and will likely be removed
10457 in a future version.
10459 @item -mno-vliw-branch
10460 @opindex mno-vliw-branch
10462 Do not run a pass to pack branches into VLIW instructions.
10464 This switch is mainly for debugging the compiler and will likely be removed
10465 in a future version.
10467 @item -mmulti-cond-exec
10468 @opindex mmulti-cond-exec
10470 Enable optimization of @code{&&} and @code{||} in conditional execution
10473 This switch is mainly for debugging the compiler and will likely be removed
10474 in a future version.
10476 @item -mno-multi-cond-exec
10477 @opindex mno-multi-cond-exec
10479 Disable optimization of @code{&&} and @code{||} in conditional execution.
10481 This switch is mainly for debugging the compiler and will likely be removed
10482 in a future version.
10484 @item -mnested-cond-exec
10485 @opindex mnested-cond-exec
10487 Enable nested conditional execution optimizations (default).
10489 This switch is mainly for debugging the compiler and will likely be removed
10490 in a future version.
10492 @item -mno-nested-cond-exec
10493 @opindex mno-nested-cond-exec
10495 Disable nested conditional execution optimizations.
10497 This switch is mainly for debugging the compiler and will likely be removed
10498 in a future version.
10500 @item -moptimize-membar
10501 @opindex moptimize-membar
10503 This switch removes redundant @code{membar} instructions from the
10504 compiler generated code. It is enabled by default.
10506 @item -mno-optimize-membar
10507 @opindex mno-optimize-membar
10509 This switch disables the automatic removal of redundant @code{membar}
10510 instructions from the generated code.
10512 @item -mtomcat-stats
10513 @opindex mtomcat-stats
10515 Cause gas to print out tomcat statistics.
10517 @item -mcpu=@var{cpu}
10520 Select the processor type for which to generate code. Possible values are
10521 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
10522 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
10526 @node GNU/Linux Options
10527 @subsection GNU/Linux Options
10529 These @samp{-m} options are defined for GNU/Linux targets:
10534 Use the GNU C library instead of uClibc. This is the default except
10535 on @samp{*-*-linux-*uclibc*} targets.
10539 Use uClibc instead of the GNU C library. This is the default on
10540 @samp{*-*-linux-*uclibc*} targets.
10543 @node H8/300 Options
10544 @subsection H8/300 Options
10546 These @samp{-m} options are defined for the H8/300 implementations:
10551 Shorten some address references at link time, when possible; uses the
10552 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
10553 ld, Using ld}, for a fuller description.
10557 Generate code for the H8/300H@.
10561 Generate code for the H8S@.
10565 Generate code for the H8S and H8/300H in the normal mode. This switch
10566 must be used either with @option{-mh} or @option{-ms}.
10570 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
10574 Make @code{int} data 32 bits by default.
10577 @opindex malign-300
10578 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
10579 The default for the H8/300H and H8S is to align longs and floats on 4
10581 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
10582 This option has no effect on the H8/300.
10586 @subsection HPPA Options
10587 @cindex HPPA Options
10589 These @samp{-m} options are defined for the HPPA family of computers:
10592 @item -march=@var{architecture-type}
10594 Generate code for the specified architecture. The choices for
10595 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
10596 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
10597 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
10598 architecture option for your machine. Code compiled for lower numbered
10599 architectures will run on higher numbered architectures, but not the
10602 @item -mpa-risc-1-0
10603 @itemx -mpa-risc-1-1
10604 @itemx -mpa-risc-2-0
10605 @opindex mpa-risc-1-0
10606 @opindex mpa-risc-1-1
10607 @opindex mpa-risc-2-0
10608 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10611 @opindex mbig-switch
10612 Generate code suitable for big switch tables. Use this option only if
10613 the assembler/linker complain about out of range branches within a switch
10616 @item -mjump-in-delay
10617 @opindex mjump-in-delay
10618 Fill delay slots of function calls with unconditional jump instructions
10619 by modifying the return pointer for the function call to be the target
10620 of the conditional jump.
10622 @item -mdisable-fpregs
10623 @opindex mdisable-fpregs
10624 Prevent floating point registers from being used in any manner. This is
10625 necessary for compiling kernels which perform lazy context switching of
10626 floating point registers. If you use this option and attempt to perform
10627 floating point operations, the compiler will abort.
10629 @item -mdisable-indexing
10630 @opindex mdisable-indexing
10631 Prevent the compiler from using indexing address modes. This avoids some
10632 rather obscure problems when compiling MIG generated code under MACH@.
10634 @item -mno-space-regs
10635 @opindex mno-space-regs
10636 Generate code that assumes the target has no space registers. This allows
10637 GCC to generate faster indirect calls and use unscaled index address modes.
10639 Such code is suitable for level 0 PA systems and kernels.
10641 @item -mfast-indirect-calls
10642 @opindex mfast-indirect-calls
10643 Generate code that assumes calls never cross space boundaries. This
10644 allows GCC to emit code which performs faster indirect calls.
10646 This option will not work in the presence of shared libraries or nested
10649 @item -mfixed-range=@var{register-range}
10650 @opindex mfixed-range
10651 Generate code treating the given register range as fixed registers.
10652 A fixed register is one that the register allocator can not use. This is
10653 useful when compiling kernel code. A register range is specified as
10654 two registers separated by a dash. Multiple register ranges can be
10655 specified separated by a comma.
10657 @item -mlong-load-store
10658 @opindex mlong-load-store
10659 Generate 3-instruction load and store sequences as sometimes required by
10660 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
10663 @item -mportable-runtime
10664 @opindex mportable-runtime
10665 Use the portable calling conventions proposed by HP for ELF systems.
10669 Enable the use of assembler directives only GAS understands.
10671 @item -mschedule=@var{cpu-type}
10673 Schedule code according to the constraints for the machine type
10674 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10675 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10676 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10677 proper scheduling option for your machine. The default scheduling is
10681 @opindex mlinker-opt
10682 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10683 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10684 linkers in which they give bogus error messages when linking some programs.
10687 @opindex msoft-float
10688 Generate output containing library calls for floating point.
10689 @strong{Warning:} the requisite libraries are not available for all HPPA
10690 targets. Normally the facilities of the machine's usual C compiler are
10691 used, but this cannot be done directly in cross-compilation. You must make
10692 your own arrangements to provide suitable library functions for
10695 @option{-msoft-float} changes the calling convention in the output file;
10696 therefore, it is only useful if you compile @emph{all} of a program with
10697 this option. In particular, you need to compile @file{libgcc.a}, the
10698 library that comes with GCC, with @option{-msoft-float} in order for
10703 Generate the predefine, @code{_SIO}, for server IO@. The default is
10704 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10705 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10706 options are available under HP-UX and HI-UX@.
10710 Use GNU ld specific options. This passes @option{-shared} to ld when
10711 building a shared library. It is the default when GCC is configured,
10712 explicitly or implicitly, with the GNU linker. This option does not
10713 have any affect on which ld is called, it only changes what parameters
10714 are passed to that ld. The ld that is called is determined by the
10715 @option{--with-ld} configure option, GCC's program search path, and
10716 finally by the user's @env{PATH}. The linker used by GCC can be printed
10717 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10718 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10722 Use HP ld specific options. This passes @option{-b} to ld when building
10723 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10724 links. It is the default when GCC is configured, explicitly or
10725 implicitly, with the HP linker. This option does not have any affect on
10726 which ld is called, it only changes what parameters are passed to that
10727 ld. The ld that is called is determined by the @option{--with-ld}
10728 configure option, GCC's program search path, and finally by the user's
10729 @env{PATH}. The linker used by GCC can be printed using @samp{which
10730 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10731 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10734 @opindex mno-long-calls
10735 Generate code that uses long call sequences. This ensures that a call
10736 is always able to reach linker generated stubs. The default is to generate
10737 long calls only when the distance from the call site to the beginning
10738 of the function or translation unit, as the case may be, exceeds a
10739 predefined limit set by the branch type being used. The limits for
10740 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10741 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10744 Distances are measured from the beginning of functions when using the
10745 @option{-ffunction-sections} option, or when using the @option{-mgas}
10746 and @option{-mno-portable-runtime} options together under HP-UX with
10749 It is normally not desirable to use this option as it will degrade
10750 performance. However, it may be useful in large applications,
10751 particularly when partial linking is used to build the application.
10753 The types of long calls used depends on the capabilities of the
10754 assembler and linker, and the type of code being generated. The
10755 impact on systems that support long absolute calls, and long pic
10756 symbol-difference or pc-relative calls should be relatively small.
10757 However, an indirect call is used on 32-bit ELF systems in pic code
10758 and it is quite long.
10760 @item -munix=@var{unix-std}
10762 Generate compiler predefines and select a startfile for the specified
10763 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10764 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10765 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10766 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10767 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10770 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10771 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10772 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10773 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10774 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10775 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10777 It is @emph{important} to note that this option changes the interfaces
10778 for various library routines. It also affects the operational behavior
10779 of the C library. Thus, @emph{extreme} care is needed in using this
10782 Library code that is intended to operate with more than one UNIX
10783 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10784 as appropriate. Most GNU software doesn't provide this capability.
10788 Suppress the generation of link options to search libdld.sl when the
10789 @option{-static} option is specified on HP-UX 10 and later.
10793 The HP-UX implementation of setlocale in libc has a dependency on
10794 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10795 when the @option{-static} option is specified, special link options
10796 are needed to resolve this dependency.
10798 On HP-UX 10 and later, the GCC driver adds the necessary options to
10799 link with libdld.sl when the @option{-static} option is specified.
10800 This causes the resulting binary to be dynamic. On the 64-bit port,
10801 the linkers generate dynamic binaries by default in any case. The
10802 @option{-nolibdld} option can be used to prevent the GCC driver from
10803 adding these link options.
10807 Add support for multithreading with the @dfn{dce thread} library
10808 under HP-UX@. This option sets flags for both the preprocessor and
10812 @node i386 and x86-64 Options
10813 @subsection Intel 386 and AMD x86-64 Options
10814 @cindex i386 Options
10815 @cindex x86-64 Options
10816 @cindex Intel 386 Options
10817 @cindex AMD x86-64 Options
10819 These @samp{-m} options are defined for the i386 and x86-64 family of
10823 @item -mtune=@var{cpu-type}
10825 Tune to @var{cpu-type} everything applicable about the generated code, except
10826 for the ABI and the set of available instructions. The choices for
10827 @var{cpu-type} are:
10830 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10831 If you know the CPU on which your code will run, then you should use
10832 the corresponding @option{-mtune} option instead of
10833 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10834 of your application will have, then you should use this option.
10836 As new processors are deployed in the marketplace, the behavior of this
10837 option will change. Therefore, if you upgrade to a newer version of
10838 GCC, the code generated option will change to reflect the processors
10839 that were most common when that version of GCC was released.
10841 There is no @option{-march=generic} option because @option{-march}
10842 indicates the instruction set the compiler can use, and there is no
10843 generic instruction set applicable to all processors. In contrast,
10844 @option{-mtune} indicates the processor (or, in this case, collection of
10845 processors) for which the code is optimized.
10847 This selects the CPU to tune for at compilation time by determining
10848 the processor type of the compiling machine. Using @option{-mtune=native}
10849 will produce code optimized for the local machine under the constraints
10850 of the selected instruction set. Using @option{-march=native} will
10851 enable all instruction subsets supported by the local machine (hence
10852 the result might not run on different machines).
10854 Original Intel's i386 CPU@.
10856 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10857 @item i586, pentium
10858 Intel Pentium CPU with no MMX support.
10860 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10862 Intel PentiumPro CPU@.
10864 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10865 instruction set will be used, so the code will run on all i686 family chips.
10867 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10868 @item pentium3, pentium3m
10869 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10872 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10873 support. Used by Centrino notebooks.
10874 @item pentium4, pentium4m
10875 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10877 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10880 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10881 SSE2 and SSE3 instruction set support.
10883 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10884 instruction set support.
10886 AMD K6 CPU with MMX instruction set support.
10888 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
10889 @item athlon, athlon-tbird
10890 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
10892 @item athlon-4, athlon-xp, athlon-mp
10893 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
10894 instruction set support.
10895 @item k8, opteron, athlon64, athlon-fx
10896 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10897 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
10898 @item k8-sse3, opteron-sse3, athlon64-sse3
10899 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10900 @item amdfam10, barcelona
10901 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10902 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10903 instruction set extensions.)
10905 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10908 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
10909 instruction set support.
10911 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
10912 implemented for this chip.)
10914 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10915 implemented for this chip.)
10917 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10920 While picking a specific @var{cpu-type} will schedule things appropriately
10921 for that particular chip, the compiler will not generate any code that
10922 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10925 @item -march=@var{cpu-type}
10927 Generate instructions for the machine type @var{cpu-type}. The choices
10928 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10929 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10931 @item -mcpu=@var{cpu-type}
10933 A deprecated synonym for @option{-mtune}.
10935 @item -mfpmath=@var{unit}
10937 Generate floating point arithmetics for selected unit @var{unit}. The choices
10938 for @var{unit} are:
10942 Use the standard 387 floating point coprocessor present majority of chips and
10943 emulated otherwise. Code compiled with this option will run almost everywhere.
10944 The temporary results are computed in 80bit precision instead of precision
10945 specified by the type resulting in slightly different results compared to most
10946 of other chips. See @option{-ffloat-store} for more detailed description.
10948 This is the default choice for i386 compiler.
10951 Use scalar floating point instructions present in the SSE instruction set.
10952 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10953 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10954 instruction set supports only single precision arithmetics, thus the double and
10955 extended precision arithmetics is still done using 387. Later version, present
10956 only in Pentium4 and the future AMD x86-64 chips supports double precision
10959 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10960 or @option{-msse2} switches to enable SSE extensions and make this option
10961 effective. For the x86-64 compiler, these extensions are enabled by default.
10963 The resulting code should be considerably faster in the majority of cases and avoid
10964 the numerical instability problems of 387 code, but may break some existing
10965 code that expects temporaries to be 80bit.
10967 This is the default choice for the x86-64 compiler.
10972 Attempt to utilize both instruction sets at once. This effectively double the
10973 amount of available registers and on chips with separate execution units for
10974 387 and SSE the execution resources too. Use this option with care, as it is
10975 still experimental, because the GCC register allocator does not model separate
10976 functional units well resulting in instable performance.
10979 @item -masm=@var{dialect}
10980 @opindex masm=@var{dialect}
10981 Output asm instructions using selected @var{dialect}. Supported
10982 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10983 not support @samp{intel}.
10986 @itemx -mno-ieee-fp
10988 @opindex mno-ieee-fp
10989 Control whether or not the compiler uses IEEE floating point
10990 comparisons. These handle correctly the case where the result of a
10991 comparison is unordered.
10994 @opindex msoft-float
10995 Generate output containing library calls for floating point.
10996 @strong{Warning:} the requisite libraries are not part of GCC@.
10997 Normally the facilities of the machine's usual C compiler are used, but
10998 this can't be done directly in cross-compilation. You must make your
10999 own arrangements to provide suitable library functions for
11002 On machines where a function returns floating point results in the 80387
11003 register stack, some floating point opcodes may be emitted even if
11004 @option{-msoft-float} is used.
11006 @item -mno-fp-ret-in-387
11007 @opindex mno-fp-ret-in-387
11008 Do not use the FPU registers for return values of functions.
11010 The usual calling convention has functions return values of types
11011 @code{float} and @code{double} in an FPU register, even if there
11012 is no FPU@. The idea is that the operating system should emulate
11015 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11016 in ordinary CPU registers instead.
11018 @item -mno-fancy-math-387
11019 @opindex mno-fancy-math-387
11020 Some 387 emulators do not support the @code{sin}, @code{cos} and
11021 @code{sqrt} instructions for the 387. Specify this option to avoid
11022 generating those instructions. This option is the default on FreeBSD,
11023 OpenBSD and NetBSD@. This option is overridden when @option{-march}
11024 indicates that the target cpu will always have an FPU and so the
11025 instruction will not need emulation. As of revision 2.6.1, these
11026 instructions are not generated unless you also use the
11027 @option{-funsafe-math-optimizations} switch.
11029 @item -malign-double
11030 @itemx -mno-align-double
11031 @opindex malign-double
11032 @opindex mno-align-double
11033 Control whether GCC aligns @code{double}, @code{long double}, and
11034 @code{long long} variables on a two word boundary or a one word
11035 boundary. Aligning @code{double} variables on a two word boundary will
11036 produce code that runs somewhat faster on a @samp{Pentium} at the
11037 expense of more memory.
11039 On x86-64, @option{-malign-double} is enabled by default.
11041 @strong{Warning:} if you use the @option{-malign-double} switch,
11042 structures containing the above types will be aligned differently than
11043 the published application binary interface specifications for the 386
11044 and will not be binary compatible with structures in code compiled
11045 without that switch.
11047 @item -m96bit-long-double
11048 @itemx -m128bit-long-double
11049 @opindex m96bit-long-double
11050 @opindex m128bit-long-double
11051 These switches control the size of @code{long double} type. The i386
11052 application binary interface specifies the size to be 96 bits,
11053 so @option{-m96bit-long-double} is the default in 32 bit mode.
11055 Modern architectures (Pentium and newer) would prefer @code{long double}
11056 to be aligned to an 8 or 16 byte boundary. In arrays or structures
11057 conforming to the ABI, this would not be possible. So specifying a
11058 @option{-m128bit-long-double} will align @code{long double}
11059 to a 16 byte boundary by padding the @code{long double} with an additional
11062 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
11063 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
11065 Notice that neither of these options enable any extra precision over the x87
11066 standard of 80 bits for a @code{long double}.
11068 @strong{Warning:} if you override the default value for your target ABI, the
11069 structures and arrays containing @code{long double} variables will change
11070 their size as well as function calling convention for function taking
11071 @code{long double} will be modified. Hence they will not be binary
11072 compatible with arrays or structures in code compiled without that switch.
11074 @item -mlarge-data-threshold=@var{number}
11075 @opindex mlarge-data-threshold=@var{number}
11076 When @option{-mcmodel=medium} is specified, the data greater than
11077 @var{threshold} are placed in large data section. This value must be the
11078 same across all object linked into the binary and defaults to 65535.
11082 Use a different function-calling convention, in which functions that
11083 take a fixed number of arguments return with the @code{ret} @var{num}
11084 instruction, which pops their arguments while returning. This saves one
11085 instruction in the caller since there is no need to pop the arguments
11088 You can specify that an individual function is called with this calling
11089 sequence with the function attribute @samp{stdcall}. You can also
11090 override the @option{-mrtd} option by using the function attribute
11091 @samp{cdecl}. @xref{Function Attributes}.
11093 @strong{Warning:} this calling convention is incompatible with the one
11094 normally used on Unix, so you cannot use it if you need to call
11095 libraries compiled with the Unix compiler.
11097 Also, you must provide function prototypes for all functions that
11098 take variable numbers of arguments (including @code{printf});
11099 otherwise incorrect code will be generated for calls to those
11102 In addition, seriously incorrect code will result if you call a
11103 function with too many arguments. (Normally, extra arguments are
11104 harmlessly ignored.)
11106 @item -mregparm=@var{num}
11108 Control how many registers are used to pass integer arguments. By
11109 default, no registers are used to pass arguments, and at most 3
11110 registers can be used. You can control this behavior for a specific
11111 function by using the function attribute @samp{regparm}.
11112 @xref{Function Attributes}.
11114 @strong{Warning:} if you use this switch, and
11115 @var{num} is nonzero, then you must build all modules with the same
11116 value, including any libraries. This includes the system libraries and
11120 @opindex msseregparm
11121 Use SSE register passing conventions for float and double arguments
11122 and return values. You can control this behavior for a specific
11123 function by using the function attribute @samp{sseregparm}.
11124 @xref{Function Attributes}.
11126 @strong{Warning:} if you use this switch then you must build all
11127 modules with the same value, including any libraries. This includes
11128 the system libraries and startup modules.
11137 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
11138 is specified, the significands of results of floating-point operations are
11139 rounded to 24 bits (single precision); @option{-mpc64} rounds the
11140 significands of results of floating-point operations to 53 bits (double
11141 precision) and @option{-mpc80} rounds the significands of results of
11142 floating-point operations to 64 bits (extended double precision), which is
11143 the default. When this option is used, floating-point operations in higher
11144 precisions are not available to the programmer without setting the FPU
11145 control word explicitly.
11147 Setting the rounding of floating-point operations to less than the default
11148 80 bits can speed some programs by 2% or more. Note that some mathematical
11149 libraries assume that extended precision (80 bit) floating-point operations
11150 are enabled by default; routines in such libraries could suffer significant
11151 loss of accuracy, typically through so-called "catastrophic cancellation",
11152 when this option is used to set the precision to less than extended precision.
11154 @item -mstackrealign
11155 @opindex mstackrealign
11156 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
11157 option will generate an alternate prologue and epilogue that realigns the
11158 runtime stack if necessary. This supports mixing legacy codes that keep
11159 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
11160 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
11161 applicable to individual functions.
11163 @item -mpreferred-stack-boundary=@var{num}
11164 @opindex mpreferred-stack-boundary
11165 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
11166 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
11167 the default is 4 (16 bytes or 128 bits).
11169 @item -mincoming-stack-boundary=@var{num}
11170 @opindex mincoming-stack-boundary
11171 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
11172 boundary. If @option{-mincoming-stack-boundary} is not specified,
11173 the one specified by @option{-mpreferred-stack-boundary} will be used.
11175 On Pentium and PentiumPro, @code{double} and @code{long double} values
11176 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
11177 suffer significant run time performance penalties. On Pentium III, the
11178 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
11179 properly if it is not 16 byte aligned.
11181 To ensure proper alignment of this values on the stack, the stack boundary
11182 must be as aligned as that required by any value stored on the stack.
11183 Further, every function must be generated such that it keeps the stack
11184 aligned. Thus calling a function compiled with a higher preferred
11185 stack boundary from a function compiled with a lower preferred stack
11186 boundary will most likely misalign the stack. It is recommended that
11187 libraries that use callbacks always use the default setting.
11189 This extra alignment does consume extra stack space, and generally
11190 increases code size. Code that is sensitive to stack space usage, such
11191 as embedded systems and operating system kernels, may want to reduce the
11192 preferred alignment to @option{-mpreferred-stack-boundary=2}.
11232 These switches enable or disable the use of instructions in the MMX,
11233 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, SSE5, ABM or
11234 3DNow!@: extended instruction sets.
11235 These extensions are also available as built-in functions: see
11236 @ref{X86 Built-in Functions}, for details of the functions enabled and
11237 disabled by these switches.
11239 To have SSE/SSE2 instructions generated automatically from floating-point
11240 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
11242 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
11243 generates new AVX instructions or AVX equivalence for all SSEx instructions
11246 These options will enable GCC to use these extended instructions in
11247 generated code, even without @option{-mfpmath=sse}. Applications which
11248 perform runtime CPU detection must compile separate files for each
11249 supported architecture, using the appropriate flags. In particular,
11250 the file containing the CPU detection code should be compiled without
11255 This option instructs GCC to emit a @code{cld} instruction in the prologue
11256 of functions that use string instructions. String instructions depend on
11257 the DF flag to select between autoincrement or autodecrement mode. While the
11258 ABI specifies the DF flag to be cleared on function entry, some operating
11259 systems violate this specification by not clearing the DF flag in their
11260 exception dispatchers. The exception handler can be invoked with the DF flag
11261 set which leads to wrong direction mode, when string instructions are used.
11262 This option can be enabled by default on 32-bit x86 targets by configuring
11263 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
11264 instructions can be suppressed with the @option{-mno-cld} compiler option
11269 This option will enable GCC to use CMPXCHG16B instruction in generated code.
11270 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
11271 data types. This is useful for high resolution counters that could be updated
11272 by multiple processors (or cores). This instruction is generated as part of
11273 atomic built-in functions: see @ref{Atomic Builtins} for details.
11277 This option will enable GCC to use SAHF instruction in generated 64-bit code.
11278 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
11279 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
11280 SAHF are load and store instructions, respectively, for certain status flags.
11281 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
11282 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
11286 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
11287 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Rhapson step
11288 to increase precision instead of DIVSS and SQRTSS (and their vectorized
11289 variants) for single precision floating point arguments. These instructions
11290 are generated only when @option{-funsafe-math-optimizations} is enabled
11291 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
11292 Note that while the throughput of the sequence is higher than the throughput
11293 of the non-reciprocal instruction, the precision of the sequence can be
11294 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
11296 @item -mveclibabi=@var{type}
11297 @opindex mveclibabi
11298 Specifies the ABI type to use for vectorizing intrinsics using an
11299 external library. Supported types are @code{svml} for the Intel short
11300 vector math library and @code{acml} for the AMD math core library style
11301 of interfacing. GCC will currently emit calls to @code{vmldExp2},
11302 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
11303 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
11304 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
11305 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
11306 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
11307 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
11308 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
11309 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
11310 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
11311 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
11312 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
11313 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
11314 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
11315 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
11316 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
11317 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
11318 compatible library will have to be specified at link time.
11321 @itemx -mno-push-args
11322 @opindex mpush-args
11323 @opindex mno-push-args
11324 Use PUSH operations to store outgoing parameters. This method is shorter
11325 and usually equally fast as method using SUB/MOV operations and is enabled
11326 by default. In some cases disabling it may improve performance because of
11327 improved scheduling and reduced dependencies.
11329 @item -maccumulate-outgoing-args
11330 @opindex maccumulate-outgoing-args
11331 If enabled, the maximum amount of space required for outgoing arguments will be
11332 computed in the function prologue. This is faster on most modern CPUs
11333 because of reduced dependencies, improved scheduling and reduced stack usage
11334 when preferred stack boundary is not equal to 2. The drawback is a notable
11335 increase in code size. This switch implies @option{-mno-push-args}.
11339 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
11340 on thread-safe exception handling must compile and link all code with the
11341 @option{-mthreads} option. When compiling, @option{-mthreads} defines
11342 @option{-D_MT}; when linking, it links in a special thread helper library
11343 @option{-lmingwthrd} which cleans up per thread exception handling data.
11345 @item -mno-align-stringops
11346 @opindex mno-align-stringops
11347 Do not align destination of inlined string operations. This switch reduces
11348 code size and improves performance in case the destination is already aligned,
11349 but GCC doesn't know about it.
11351 @item -minline-all-stringops
11352 @opindex minline-all-stringops
11353 By default GCC inlines string operations only when destination is known to be
11354 aligned at least to 4 byte boundary. This enables more inlining, increase code
11355 size, but may improve performance of code that depends on fast memcpy, strlen
11356 and memset for short lengths.
11358 @item -minline-stringops-dynamically
11359 @opindex minline-stringops-dynamically
11360 For string operation of unknown size, inline runtime checks so for small
11361 blocks inline code is used, while for large blocks library call is used.
11363 @item -mstringop-strategy=@var{alg}
11364 @opindex mstringop-strategy=@var{alg}
11365 Overwrite internal decision heuristic about particular algorithm to inline
11366 string operation with. The allowed values are @code{rep_byte},
11367 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
11368 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
11369 expanding inline loop, @code{libcall} for always expanding library call.
11371 @item -momit-leaf-frame-pointer
11372 @opindex momit-leaf-frame-pointer
11373 Don't keep the frame pointer in a register for leaf functions. This
11374 avoids the instructions to save, set up and restore frame pointers and
11375 makes an extra register available in leaf functions. The option
11376 @option{-fomit-frame-pointer} removes the frame pointer for all functions
11377 which might make debugging harder.
11379 @item -mtls-direct-seg-refs
11380 @itemx -mno-tls-direct-seg-refs
11381 @opindex mtls-direct-seg-refs
11382 Controls whether TLS variables may be accessed with offsets from the
11383 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
11384 or whether the thread base pointer must be added. Whether or not this
11385 is legal depends on the operating system, and whether it maps the
11386 segment to cover the entire TLS area.
11388 For systems that use GNU libc, the default is on.
11391 @itemx -mno-fused-madd
11392 @opindex mfused-madd
11393 Enable automatic generation of fused floating point multiply-add instructions
11394 if the ISA supports such instructions. The -mfused-madd option is on by
11395 default. The fused multiply-add instructions have a different
11396 rounding behavior compared to executing a multiply followed by an add.
11399 @itemx -mno-sse2avx
11401 Specify that the assembler should encode SSE instructions with VEX
11402 prefix. The option @option{-mavx} turns this on by default.
11405 These @samp{-m} switches are supported in addition to the above
11406 on AMD x86-64 processors in 64-bit environments.
11413 Generate code for a 32-bit or 64-bit environment.
11414 The 32-bit environment sets int, long and pointer to 32 bits and
11415 generates code that runs on any i386 system.
11416 The 64-bit environment sets int to 32 bits and long and pointer
11417 to 64 bits and generates code for AMD's x86-64 architecture. For
11418 darwin only the -m64 option turns off the @option{-fno-pic} and
11419 @option{-mdynamic-no-pic} options.
11421 @item -mno-red-zone
11422 @opindex no-red-zone
11423 Do not use a so called red zone for x86-64 code. The red zone is mandated
11424 by the x86-64 ABI, it is a 128-byte area beyond the location of the
11425 stack pointer that will not be modified by signal or interrupt handlers
11426 and therefore can be used for temporary data without adjusting the stack
11427 pointer. The flag @option{-mno-red-zone} disables this red zone.
11429 @item -mcmodel=small
11430 @opindex mcmodel=small
11431 Generate code for the small code model: the program and its symbols must
11432 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
11433 Programs can be statically or dynamically linked. This is the default
11436 @item -mcmodel=kernel
11437 @opindex mcmodel=kernel
11438 Generate code for the kernel code model. The kernel runs in the
11439 negative 2 GB of the address space.
11440 This model has to be used for Linux kernel code.
11442 @item -mcmodel=medium
11443 @opindex mcmodel=medium
11444 Generate code for the medium model: The program is linked in the lower 2
11445 GB of the address space. Small symbols are also placed there. Symbols
11446 with sizes larger than @option{-mlarge-data-threshold} are put into
11447 large data or bss sections and can be located above 2GB. Programs can
11448 be statically or dynamically linked.
11450 @item -mcmodel=large
11451 @opindex mcmodel=large
11452 Generate code for the large model: This model makes no assumptions
11453 about addresses and sizes of sections.
11456 @node IA-64 Options
11457 @subsection IA-64 Options
11458 @cindex IA-64 Options
11460 These are the @samp{-m} options defined for the Intel IA-64 architecture.
11464 @opindex mbig-endian
11465 Generate code for a big endian target. This is the default for HP-UX@.
11467 @item -mlittle-endian
11468 @opindex mlittle-endian
11469 Generate code for a little endian target. This is the default for AIX5
11475 @opindex mno-gnu-as
11476 Generate (or don't) code for the GNU assembler. This is the default.
11477 @c Also, this is the default if the configure option @option{--with-gnu-as}
11483 @opindex mno-gnu-ld
11484 Generate (or don't) code for the GNU linker. This is the default.
11485 @c Also, this is the default if the configure option @option{--with-gnu-ld}
11490 Generate code that does not use a global pointer register. The result
11491 is not position independent code, and violates the IA-64 ABI@.
11493 @item -mvolatile-asm-stop
11494 @itemx -mno-volatile-asm-stop
11495 @opindex mvolatile-asm-stop
11496 @opindex mno-volatile-asm-stop
11497 Generate (or don't) a stop bit immediately before and after volatile asm
11500 @item -mregister-names
11501 @itemx -mno-register-names
11502 @opindex mregister-names
11503 @opindex mno-register-names
11504 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
11505 the stacked registers. This may make assembler output more readable.
11511 Disable (or enable) optimizations that use the small data section. This may
11512 be useful for working around optimizer bugs.
11514 @item -mconstant-gp
11515 @opindex mconstant-gp
11516 Generate code that uses a single constant global pointer value. This is
11517 useful when compiling kernel code.
11521 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
11522 This is useful when compiling firmware code.
11524 @item -minline-float-divide-min-latency
11525 @opindex minline-float-divide-min-latency
11526 Generate code for inline divides of floating point values
11527 using the minimum latency algorithm.
11529 @item -minline-float-divide-max-throughput
11530 @opindex minline-float-divide-max-throughput
11531 Generate code for inline divides of floating point values
11532 using the maximum throughput algorithm.
11534 @item -minline-int-divide-min-latency
11535 @opindex minline-int-divide-min-latency
11536 Generate code for inline divides of integer values
11537 using the minimum latency algorithm.
11539 @item -minline-int-divide-max-throughput
11540 @opindex minline-int-divide-max-throughput
11541 Generate code for inline divides of integer values
11542 using the maximum throughput algorithm.
11544 @item -minline-sqrt-min-latency
11545 @opindex minline-sqrt-min-latency
11546 Generate code for inline square roots
11547 using the minimum latency algorithm.
11549 @item -minline-sqrt-max-throughput
11550 @opindex minline-sqrt-max-throughput
11551 Generate code for inline square roots
11552 using the maximum throughput algorithm.
11554 @item -mno-dwarf2-asm
11555 @itemx -mdwarf2-asm
11556 @opindex mno-dwarf2-asm
11557 @opindex mdwarf2-asm
11558 Don't (or do) generate assembler code for the DWARF2 line number debugging
11559 info. This may be useful when not using the GNU assembler.
11561 @item -mearly-stop-bits
11562 @itemx -mno-early-stop-bits
11563 @opindex mearly-stop-bits
11564 @opindex mno-early-stop-bits
11565 Allow stop bits to be placed earlier than immediately preceding the
11566 instruction that triggered the stop bit. This can improve instruction
11567 scheduling, but does not always do so.
11569 @item -mfixed-range=@var{register-range}
11570 @opindex mfixed-range
11571 Generate code treating the given register range as fixed registers.
11572 A fixed register is one that the register allocator can not use. This is
11573 useful when compiling kernel code. A register range is specified as
11574 two registers separated by a dash. Multiple register ranges can be
11575 specified separated by a comma.
11577 @item -mtls-size=@var{tls-size}
11579 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
11582 @item -mtune=@var{cpu-type}
11584 Tune the instruction scheduling for a particular CPU, Valid values are
11585 itanium, itanium1, merced, itanium2, and mckinley.
11591 Add support for multithreading using the POSIX threads library. This
11592 option sets flags for both the preprocessor and linker. It does
11593 not affect the thread safety of object code produced by the compiler or
11594 that of libraries supplied with it. These are HP-UX specific flags.
11600 Generate code for a 32-bit or 64-bit environment.
11601 The 32-bit environment sets int, long and pointer to 32 bits.
11602 The 64-bit environment sets int to 32 bits and long and pointer
11603 to 64 bits. These are HP-UX specific flags.
11605 @item -mno-sched-br-data-spec
11606 @itemx -msched-br-data-spec
11607 @opindex mno-sched-br-data-spec
11608 @opindex msched-br-data-spec
11609 (Dis/En)able data speculative scheduling before reload.
11610 This will result in generation of the ld.a instructions and
11611 the corresponding check instructions (ld.c / chk.a).
11612 The default is 'disable'.
11614 @item -msched-ar-data-spec
11615 @itemx -mno-sched-ar-data-spec
11616 @opindex msched-ar-data-spec
11617 @opindex mno-sched-ar-data-spec
11618 (En/Dis)able data speculative scheduling after reload.
11619 This will result in generation of the ld.a instructions and
11620 the corresponding check instructions (ld.c / chk.a).
11621 The default is 'enable'.
11623 @item -mno-sched-control-spec
11624 @itemx -msched-control-spec
11625 @opindex mno-sched-control-spec
11626 @opindex msched-control-spec
11627 (Dis/En)able control speculative scheduling. This feature is
11628 available only during region scheduling (i.e.@: before reload).
11629 This will result in generation of the ld.s instructions and
11630 the corresponding check instructions chk.s .
11631 The default is 'disable'.
11633 @item -msched-br-in-data-spec
11634 @itemx -mno-sched-br-in-data-spec
11635 @opindex msched-br-in-data-spec
11636 @opindex mno-sched-br-in-data-spec
11637 (En/Dis)able speculative scheduling of the instructions that
11638 are dependent on the data speculative loads before reload.
11639 This is effective only with @option{-msched-br-data-spec} enabled.
11640 The default is 'enable'.
11642 @item -msched-ar-in-data-spec
11643 @itemx -mno-sched-ar-in-data-spec
11644 @opindex msched-ar-in-data-spec
11645 @opindex mno-sched-ar-in-data-spec
11646 (En/Dis)able speculative scheduling of the instructions that
11647 are dependent on the data speculative loads after reload.
11648 This is effective only with @option{-msched-ar-data-spec} enabled.
11649 The default is 'enable'.
11651 @item -msched-in-control-spec
11652 @itemx -mno-sched-in-control-spec
11653 @opindex msched-in-control-spec
11654 @opindex mno-sched-in-control-spec
11655 (En/Dis)able speculative scheduling of the instructions that
11656 are dependent on the control speculative loads.
11657 This is effective only with @option{-msched-control-spec} enabled.
11658 The default is 'enable'.
11661 @itemx -mno-sched-ldc
11662 @opindex msched-ldc
11663 @opindex mno-sched-ldc
11664 (En/Dis)able use of simple data speculation checks ld.c .
11665 If disabled, only chk.a instructions will be emitted to check
11666 data speculative loads.
11667 The default is 'enable'.
11669 @item -mno-sched-control-ldc
11670 @itemx -msched-control-ldc
11671 @opindex mno-sched-control-ldc
11672 @opindex msched-control-ldc
11673 (Dis/En)able use of ld.c instructions to check control speculative loads.
11674 If enabled, in case of control speculative load with no speculatively
11675 scheduled dependent instructions this load will be emitted as ld.sa and
11676 ld.c will be used to check it.
11677 The default is 'disable'.
11679 @item -mno-sched-spec-verbose
11680 @itemx -msched-spec-verbose
11681 @opindex mno-sched-spec-verbose
11682 @opindex msched-spec-verbose
11683 (Dis/En)able printing of the information about speculative motions.
11685 @item -mno-sched-prefer-non-data-spec-insns
11686 @itemx -msched-prefer-non-data-spec-insns
11687 @opindex mno-sched-prefer-non-data-spec-insns
11688 @opindex msched-prefer-non-data-spec-insns
11689 If enabled, data speculative instructions will be chosen for schedule
11690 only if there are no other choices at the moment. This will make
11691 the use of the data speculation much more conservative.
11692 The default is 'disable'.
11694 @item -mno-sched-prefer-non-control-spec-insns
11695 @itemx -msched-prefer-non-control-spec-insns
11696 @opindex mno-sched-prefer-non-control-spec-insns
11697 @opindex msched-prefer-non-control-spec-insns
11698 If enabled, control speculative instructions will be chosen for schedule
11699 only if there are no other choices at the moment. This will make
11700 the use of the control speculation much more conservative.
11701 The default is 'disable'.
11703 @item -mno-sched-count-spec-in-critical-path
11704 @itemx -msched-count-spec-in-critical-path
11705 @opindex mno-sched-count-spec-in-critical-path
11706 @opindex msched-count-spec-in-critical-path
11707 If enabled, speculative dependencies will be considered during
11708 computation of the instructions priorities. This will make the use of the
11709 speculation a bit more conservative.
11710 The default is 'disable'.
11715 @subsection M32C Options
11716 @cindex M32C options
11719 @item -mcpu=@var{name}
11721 Select the CPU for which code is generated. @var{name} may be one of
11722 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11723 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11724 the M32C/80 series.
11728 Specifies that the program will be run on the simulator. This causes
11729 an alternate runtime library to be linked in which supports, for
11730 example, file I/O@. You must not use this option when generating
11731 programs that will run on real hardware; you must provide your own
11732 runtime library for whatever I/O functions are needed.
11734 @item -memregs=@var{number}
11736 Specifies the number of memory-based pseudo-registers GCC will use
11737 during code generation. These pseudo-registers will be used like real
11738 registers, so there is a tradeoff between GCC's ability to fit the
11739 code into available registers, and the performance penalty of using
11740 memory instead of registers. Note that all modules in a program must
11741 be compiled with the same value for this option. Because of that, you
11742 must not use this option with the default runtime libraries gcc
11747 @node M32R/D Options
11748 @subsection M32R/D Options
11749 @cindex M32R/D options
11751 These @option{-m} options are defined for Renesas M32R/D architectures:
11756 Generate code for the M32R/2@.
11760 Generate code for the M32R/X@.
11764 Generate code for the M32R@. This is the default.
11766 @item -mmodel=small
11767 @opindex mmodel=small
11768 Assume all objects live in the lower 16MB of memory (so that their addresses
11769 can be loaded with the @code{ld24} instruction), and assume all subroutines
11770 are reachable with the @code{bl} instruction.
11771 This is the default.
11773 The addressability of a particular object can be set with the
11774 @code{model} attribute.
11776 @item -mmodel=medium
11777 @opindex mmodel=medium
11778 Assume objects may be anywhere in the 32-bit address space (the compiler
11779 will generate @code{seth/add3} instructions to load their addresses), and
11780 assume all subroutines are reachable with the @code{bl} instruction.
11782 @item -mmodel=large
11783 @opindex mmodel=large
11784 Assume objects may be anywhere in the 32-bit address space (the compiler
11785 will generate @code{seth/add3} instructions to load their addresses), and
11786 assume subroutines may not be reachable with the @code{bl} instruction
11787 (the compiler will generate the much slower @code{seth/add3/jl}
11788 instruction sequence).
11791 @opindex msdata=none
11792 Disable use of the small data area. Variables will be put into
11793 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11794 @code{section} attribute has been specified).
11795 This is the default.
11797 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11798 Objects may be explicitly put in the small data area with the
11799 @code{section} attribute using one of these sections.
11801 @item -msdata=sdata
11802 @opindex msdata=sdata
11803 Put small global and static data in the small data area, but do not
11804 generate special code to reference them.
11807 @opindex msdata=use
11808 Put small global and static data in the small data area, and generate
11809 special instructions to reference them.
11813 @cindex smaller data references
11814 Put global and static objects less than or equal to @var{num} bytes
11815 into the small data or bss sections instead of the normal data or bss
11816 sections. The default value of @var{num} is 8.
11817 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11818 for this option to have any effect.
11820 All modules should be compiled with the same @option{-G @var{num}} value.
11821 Compiling with different values of @var{num} may or may not work; if it
11822 doesn't the linker will give an error message---incorrect code will not be
11827 Makes the M32R specific code in the compiler display some statistics
11828 that might help in debugging programs.
11830 @item -malign-loops
11831 @opindex malign-loops
11832 Align all loops to a 32-byte boundary.
11834 @item -mno-align-loops
11835 @opindex mno-align-loops
11836 Do not enforce a 32-byte alignment for loops. This is the default.
11838 @item -missue-rate=@var{number}
11839 @opindex missue-rate=@var{number}
11840 Issue @var{number} instructions per cycle. @var{number} can only be 1
11843 @item -mbranch-cost=@var{number}
11844 @opindex mbranch-cost=@var{number}
11845 @var{number} can only be 1 or 2. If it is 1 then branches will be
11846 preferred over conditional code, if it is 2, then the opposite will
11849 @item -mflush-trap=@var{number}
11850 @opindex mflush-trap=@var{number}
11851 Specifies the trap number to use to flush the cache. The default is
11852 12. Valid numbers are between 0 and 15 inclusive.
11854 @item -mno-flush-trap
11855 @opindex mno-flush-trap
11856 Specifies that the cache cannot be flushed by using a trap.
11858 @item -mflush-func=@var{name}
11859 @opindex mflush-func=@var{name}
11860 Specifies the name of the operating system function to call to flush
11861 the cache. The default is @emph{_flush_cache}, but a function call
11862 will only be used if a trap is not available.
11864 @item -mno-flush-func
11865 @opindex mno-flush-func
11866 Indicates that there is no OS function for flushing the cache.
11870 @node M680x0 Options
11871 @subsection M680x0 Options
11872 @cindex M680x0 options
11874 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11875 The default settings depend on which architecture was selected when
11876 the compiler was configured; the defaults for the most common choices
11880 @item -march=@var{arch}
11882 Generate code for a specific M680x0 or ColdFire instruction set
11883 architecture. Permissible values of @var{arch} for M680x0
11884 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11885 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11886 architectures are selected according to Freescale's ISA classification
11887 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11888 @samp{isab} and @samp{isac}.
11890 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11891 code for a ColdFire target. The @var{arch} in this macro is one of the
11892 @option{-march} arguments given above.
11894 When used together, @option{-march} and @option{-mtune} select code
11895 that runs on a family of similar processors but that is optimized
11896 for a particular microarchitecture.
11898 @item -mcpu=@var{cpu}
11900 Generate code for a specific M680x0 or ColdFire processor.
11901 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11902 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11903 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11904 below, which also classifies the CPUs into families:
11906 @multitable @columnfractions 0.20 0.80
11907 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11908 @item @samp{51qe} @tab @samp{51qe}
11909 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11910 @item @samp{5206e} @tab @samp{5206e}
11911 @item @samp{5208} @tab @samp{5207} @samp{5208}
11912 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11913 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11914 @item @samp{5216} @tab @samp{5214} @samp{5216}
11915 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11916 @item @samp{5225} @tab @samp{5224} @samp{5225}
11917 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11918 @item @samp{5249} @tab @samp{5249}
11919 @item @samp{5250} @tab @samp{5250}
11920 @item @samp{5271} @tab @samp{5270} @samp{5271}
11921 @item @samp{5272} @tab @samp{5272}
11922 @item @samp{5275} @tab @samp{5274} @samp{5275}
11923 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11924 @item @samp{5307} @tab @samp{5307}
11925 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11926 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11927 @item @samp{5407} @tab @samp{5407}
11928 @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}
11931 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11932 @var{arch} is compatible with @var{cpu}. Other combinations of
11933 @option{-mcpu} and @option{-march} are rejected.
11935 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11936 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11937 where the value of @var{family} is given by the table above.
11939 @item -mtune=@var{tune}
11941 Tune the code for a particular microarchitecture, within the
11942 constraints set by @option{-march} and @option{-mcpu}.
11943 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11944 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11945 and @samp{cpu32}. The ColdFire microarchitectures
11946 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11948 You can also use @option{-mtune=68020-40} for code that needs
11949 to run relatively well on 68020, 68030 and 68040 targets.
11950 @option{-mtune=68020-60} is similar but includes 68060 targets
11951 as well. These two options select the same tuning decisions as
11952 @option{-m68020-40} and @option{-m68020-60} respectively.
11954 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11955 when tuning for 680x0 architecture @var{arch}. It also defines
11956 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11957 option is used. If gcc is tuning for a range of architectures,
11958 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11959 it defines the macros for every architecture in the range.
11961 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11962 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11963 of the arguments given above.
11969 Generate output for a 68000. This is the default
11970 when the compiler is configured for 68000-based systems.
11971 It is equivalent to @option{-march=68000}.
11973 Use this option for microcontrollers with a 68000 or EC000 core,
11974 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11978 Generate output for a 68010. This is the default
11979 when the compiler is configured for 68010-based systems.
11980 It is equivalent to @option{-march=68010}.
11986 Generate output for a 68020. This is the default
11987 when the compiler is configured for 68020-based systems.
11988 It is equivalent to @option{-march=68020}.
11992 Generate output for a 68030. This is the default when the compiler is
11993 configured for 68030-based systems. It is equivalent to
11994 @option{-march=68030}.
11998 Generate output for a 68040. This is the default when the compiler is
11999 configured for 68040-based systems. It is equivalent to
12000 @option{-march=68040}.
12002 This option inhibits the use of 68881/68882 instructions that have to be
12003 emulated by software on the 68040. Use this option if your 68040 does not
12004 have code to emulate those instructions.
12008 Generate output for a 68060. This is the default when the compiler is
12009 configured for 68060-based systems. It is equivalent to
12010 @option{-march=68060}.
12012 This option inhibits the use of 68020 and 68881/68882 instructions that
12013 have to be emulated by software on the 68060. Use this option if your 68060
12014 does not have code to emulate those instructions.
12018 Generate output for a CPU32. This is the default
12019 when the compiler is configured for CPU32-based systems.
12020 It is equivalent to @option{-march=cpu32}.
12022 Use this option for microcontrollers with a
12023 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
12024 68336, 68340, 68341, 68349 and 68360.
12028 Generate output for a 520X ColdFire CPU@. This is the default
12029 when the compiler is configured for 520X-based systems.
12030 It is equivalent to @option{-mcpu=5206}, and is now deprecated
12031 in favor of that option.
12033 Use this option for microcontroller with a 5200 core, including
12034 the MCF5202, MCF5203, MCF5204 and MCF5206.
12038 Generate output for a 5206e ColdFire CPU@. The option is now
12039 deprecated in favor of the equivalent @option{-mcpu=5206e}.
12043 Generate output for a member of the ColdFire 528X family.
12044 The option is now deprecated in favor of the equivalent
12045 @option{-mcpu=528x}.
12049 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
12050 in favor of the equivalent @option{-mcpu=5307}.
12054 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
12055 in favor of the equivalent @option{-mcpu=5407}.
12059 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
12060 This includes use of hardware floating point instructions.
12061 The option is equivalent to @option{-mcpu=547x}, and is now
12062 deprecated in favor of that option.
12066 Generate output for a 68040, without using any of the new instructions.
12067 This results in code which can run relatively efficiently on either a
12068 68020/68881 or a 68030 or a 68040. The generated code does use the
12069 68881 instructions that are emulated on the 68040.
12071 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
12075 Generate output for a 68060, without using any of the new instructions.
12076 This results in code which can run relatively efficiently on either a
12077 68020/68881 or a 68030 or a 68040. The generated code does use the
12078 68881 instructions that are emulated on the 68060.
12080 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
12084 @opindex mhard-float
12086 Generate floating-point instructions. This is the default for 68020
12087 and above, and for ColdFire devices that have an FPU@. It defines the
12088 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
12089 on ColdFire targets.
12092 @opindex msoft-float
12093 Do not generate floating-point instructions; use library calls instead.
12094 This is the default for 68000, 68010, and 68832 targets. It is also
12095 the default for ColdFire devices that have no FPU.
12101 Generate (do not generate) ColdFire hardware divide and remainder
12102 instructions. If @option{-march} is used without @option{-mcpu},
12103 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
12104 architectures. Otherwise, the default is taken from the target CPU
12105 (either the default CPU, or the one specified by @option{-mcpu}). For
12106 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
12107 @option{-mcpu=5206e}.
12109 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
12113 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12114 Additionally, parameters passed on the stack are also aligned to a
12115 16-bit boundary even on targets whose API mandates promotion to 32-bit.
12119 Do not consider type @code{int} to be 16 bits wide. This is the default.
12122 @itemx -mno-bitfield
12123 @opindex mnobitfield
12124 @opindex mno-bitfield
12125 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
12126 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
12130 Do use the bit-field instructions. The @option{-m68020} option implies
12131 @option{-mbitfield}. This is the default if you use a configuration
12132 designed for a 68020.
12136 Use a different function-calling convention, in which functions
12137 that take a fixed number of arguments return with the @code{rtd}
12138 instruction, which pops their arguments while returning. This
12139 saves one instruction in the caller since there is no need to pop
12140 the arguments there.
12142 This calling convention is incompatible with the one normally
12143 used on Unix, so you cannot use it if you need to call libraries
12144 compiled with the Unix compiler.
12146 Also, you must provide function prototypes for all functions that
12147 take variable numbers of arguments (including @code{printf});
12148 otherwise incorrect code will be generated for calls to those
12151 In addition, seriously incorrect code will result if you call a
12152 function with too many arguments. (Normally, extra arguments are
12153 harmlessly ignored.)
12155 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
12156 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
12160 Do not use the calling conventions selected by @option{-mrtd}.
12161 This is the default.
12164 @itemx -mno-align-int
12165 @opindex malign-int
12166 @opindex mno-align-int
12167 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
12168 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
12169 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
12170 Aligning variables on 32-bit boundaries produces code that runs somewhat
12171 faster on processors with 32-bit busses at the expense of more memory.
12173 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
12174 align structures containing the above types differently than
12175 most published application binary interface specifications for the m68k.
12179 Use the pc-relative addressing mode of the 68000 directly, instead of
12180 using a global offset table. At present, this option implies @option{-fpic},
12181 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
12182 not presently supported with @option{-mpcrel}, though this could be supported for
12183 68020 and higher processors.
12185 @item -mno-strict-align
12186 @itemx -mstrict-align
12187 @opindex mno-strict-align
12188 @opindex mstrict-align
12189 Do not (do) assume that unaligned memory references will be handled by
12193 Generate code that allows the data segment to be located in a different
12194 area of memory from the text segment. This allows for execute in place in
12195 an environment without virtual memory management. This option implies
12198 @item -mno-sep-data
12199 Generate code that assumes that the data segment follows the text segment.
12200 This is the default.
12202 @item -mid-shared-library
12203 Generate code that supports shared libraries via the library ID method.
12204 This allows for execute in place and shared libraries in an environment
12205 without virtual memory management. This option implies @option{-fPIC}.
12207 @item -mno-id-shared-library
12208 Generate code that doesn't assume ID based shared libraries are being used.
12209 This is the default.
12211 @item -mshared-library-id=n
12212 Specified the identification number of the ID based shared library being
12213 compiled. Specifying a value of 0 will generate more compact code, specifying
12214 other values will force the allocation of that number to the current
12215 library but is no more space or time efficient than omitting this option.
12221 When generating position-independent code for ColdFire, generate code
12222 that works if the GOT has more than 8192 entries. This code is
12223 larger and slower than code generated without this option. On M680x0
12224 processors, this option is not needed; @option{-fPIC} suffices.
12226 GCC normally uses a single instruction to load values from the GOT@.
12227 While this is relatively efficient, it only works if the GOT
12228 is smaller than about 64k. Anything larger causes the linker
12229 to report an error such as:
12231 @cindex relocation truncated to fit (ColdFire)
12233 relocation truncated to fit: R_68K_GOT16O foobar
12236 If this happens, you should recompile your code with @option{-mxgot}.
12237 It should then work with very large GOTs. However, code generated with
12238 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
12239 the value of a global symbol.
12241 Note that some linkers, including newer versions of the GNU linker,
12242 can create multiple GOTs and sort GOT entries. If you have such a linker,
12243 you should only need to use @option{-mxgot} when compiling a single
12244 object file that accesses more than 8192 GOT entries. Very few do.
12246 These options have no effect unless GCC is generating
12247 position-independent code.
12251 @node M68hc1x Options
12252 @subsection M68hc1x Options
12253 @cindex M68hc1x options
12255 These are the @samp{-m} options defined for the 68hc11 and 68hc12
12256 microcontrollers. The default values for these options depends on
12257 which style of microcontroller was selected when the compiler was configured;
12258 the defaults for the most common choices are given below.
12265 Generate output for a 68HC11. This is the default
12266 when the compiler is configured for 68HC11-based systems.
12272 Generate output for a 68HC12. This is the default
12273 when the compiler is configured for 68HC12-based systems.
12279 Generate output for a 68HCS12.
12281 @item -mauto-incdec
12282 @opindex mauto-incdec
12283 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
12290 Enable the use of 68HC12 min and max instructions.
12293 @itemx -mno-long-calls
12294 @opindex mlong-calls
12295 @opindex mno-long-calls
12296 Treat all calls as being far away (near). If calls are assumed to be
12297 far away, the compiler will use the @code{call} instruction to
12298 call a function and the @code{rtc} instruction for returning.
12302 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12304 @item -msoft-reg-count=@var{count}
12305 @opindex msoft-reg-count
12306 Specify the number of pseudo-soft registers which are used for the
12307 code generation. The maximum number is 32. Using more pseudo-soft
12308 register may or may not result in better code depending on the program.
12309 The default is 4 for 68HC11 and 2 for 68HC12.
12313 @node MCore Options
12314 @subsection MCore Options
12315 @cindex MCore options
12317 These are the @samp{-m} options defined for the Motorola M*Core
12323 @itemx -mno-hardlit
12325 @opindex mno-hardlit
12326 Inline constants into the code stream if it can be done in two
12327 instructions or less.
12333 Use the divide instruction. (Enabled by default).
12335 @item -mrelax-immediate
12336 @itemx -mno-relax-immediate
12337 @opindex mrelax-immediate
12338 @opindex mno-relax-immediate
12339 Allow arbitrary sized immediates in bit operations.
12341 @item -mwide-bitfields
12342 @itemx -mno-wide-bitfields
12343 @opindex mwide-bitfields
12344 @opindex mno-wide-bitfields
12345 Always treat bit-fields as int-sized.
12347 @item -m4byte-functions
12348 @itemx -mno-4byte-functions
12349 @opindex m4byte-functions
12350 @opindex mno-4byte-functions
12351 Force all functions to be aligned to a four byte boundary.
12353 @item -mcallgraph-data
12354 @itemx -mno-callgraph-data
12355 @opindex mcallgraph-data
12356 @opindex mno-callgraph-data
12357 Emit callgraph information.
12360 @itemx -mno-slow-bytes
12361 @opindex mslow-bytes
12362 @opindex mno-slow-bytes
12363 Prefer word access when reading byte quantities.
12365 @item -mlittle-endian
12366 @itemx -mbig-endian
12367 @opindex mlittle-endian
12368 @opindex mbig-endian
12369 Generate code for a little endian target.
12375 Generate code for the 210 processor.
12379 @subsection MIPS Options
12380 @cindex MIPS options
12386 Generate big-endian code.
12390 Generate little-endian code. This is the default for @samp{mips*el-*-*}
12393 @item -march=@var{arch}
12395 Generate code that will run on @var{arch}, which can be the name of a
12396 generic MIPS ISA, or the name of a particular processor.
12398 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
12399 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
12400 The processor names are:
12401 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
12402 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
12403 @samp{5kc}, @samp{5kf},
12405 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
12406 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
12407 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
12408 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
12409 @samp{loongson2e}, @samp{loongson2f},
12413 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
12414 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
12415 @samp{rm7000}, @samp{rm9000},
12416 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
12419 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
12420 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
12422 The special value @samp{from-abi} selects the
12423 most compatible architecture for the selected ABI (that is,
12424 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
12426 Native Linux/GNU toolchains also support the value @samp{native},
12427 which selects the best architecture option for the host processor.
12428 @option{-march=native} has no effect if GCC does not recognize
12431 In processor names, a final @samp{000} can be abbreviated as @samp{k}
12432 (for example, @samp{-march=r2k}). Prefixes are optional, and
12433 @samp{vr} may be written @samp{r}.
12435 Names of the form @samp{@var{n}f2_1} refer to processors with
12436 FPUs clocked at half the rate of the core, names of the form
12437 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
12438 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
12439 processors with FPUs clocked a ratio of 3:2 with respect to the core.
12440 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
12441 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
12442 accepted as synonyms for @samp{@var{n}f1_1}.
12444 GCC defines two macros based on the value of this option. The first
12445 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
12446 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
12447 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
12448 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
12449 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
12451 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
12452 above. In other words, it will have the full prefix and will not
12453 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
12454 the macro names the resolved architecture (either @samp{"mips1"} or
12455 @samp{"mips3"}). It names the default architecture when no
12456 @option{-march} option is given.
12458 @item -mtune=@var{arch}
12460 Optimize for @var{arch}. Among other things, this option controls
12461 the way instructions are scheduled, and the perceived cost of arithmetic
12462 operations. The list of @var{arch} values is the same as for
12465 When this option is not used, GCC will optimize for the processor
12466 specified by @option{-march}. By using @option{-march} and
12467 @option{-mtune} together, it is possible to generate code that will
12468 run on a family of processors, but optimize the code for one
12469 particular member of that family.
12471 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
12472 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
12473 @samp{-march} ones described above.
12477 Equivalent to @samp{-march=mips1}.
12481 Equivalent to @samp{-march=mips2}.
12485 Equivalent to @samp{-march=mips3}.
12489 Equivalent to @samp{-march=mips4}.
12493 Equivalent to @samp{-march=mips32}.
12497 Equivalent to @samp{-march=mips32r2}.
12501 Equivalent to @samp{-march=mips64}.
12505 Equivalent to @samp{-march=mips64r2}.
12510 @opindex mno-mips16
12511 Generate (do not generate) MIPS16 code. If GCC is targetting a
12512 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
12514 MIPS16 code generation can also be controlled on a per-function basis
12515 by means of @code{mips16} and @code{nomips16} attributes.
12516 @xref{Function Attributes}, for more information.
12518 @item -mflip-mips16
12519 @opindex mflip-mips16
12520 Generate MIPS16 code on alternating functions. This option is provided
12521 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
12522 not intended for ordinary use in compiling user code.
12524 @item -minterlink-mips16
12525 @itemx -mno-interlink-mips16
12526 @opindex minterlink-mips16
12527 @opindex mno-interlink-mips16
12528 Require (do not require) that non-MIPS16 code be link-compatible with
12531 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
12532 it must either use a call or an indirect jump. @option{-minterlink-mips16}
12533 therefore disables direct jumps unless GCC knows that the target of the
12534 jump is not MIPS16.
12546 Generate code for the given ABI@.
12548 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
12549 generates 64-bit code when you select a 64-bit architecture, but you
12550 can use @option{-mgp32} to get 32-bit code instead.
12552 For information about the O64 ABI, see
12553 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
12555 GCC supports a variant of the o32 ABI in which floating-point registers
12556 are 64 rather than 32 bits wide. You can select this combination with
12557 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
12558 and @samp{mfhc1} instructions and is therefore only supported for
12559 MIPS32R2 processors.
12561 The register assignments for arguments and return values remain the
12562 same, but each scalar value is passed in a single 64-bit register
12563 rather than a pair of 32-bit registers. For example, scalar
12564 floating-point values are returned in @samp{$f0} only, not a
12565 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
12566 remains the same, but all 64 bits are saved.
12569 @itemx -mno-abicalls
12571 @opindex mno-abicalls
12572 Generate (do not generate) code that is suitable for SVR4-style
12573 dynamic objects. @option{-mabicalls} is the default for SVR4-based
12578 Generate (do not generate) code that is fully position-independent,
12579 and that can therefore be linked into shared libraries. This option
12580 only affects @option{-mabicalls}.
12582 All @option{-mabicalls} code has traditionally been position-independent,
12583 regardless of options like @option{-fPIC} and @option{-fpic}. However,
12584 as an extension, the GNU toolchain allows executables to use absolute
12585 accesses for locally-binding symbols. It can also use shorter GP
12586 initialization sequences and generate direct calls to locally-defined
12587 functions. This mode is selected by @option{-mno-shared}.
12589 @option{-mno-shared} depends on binutils 2.16 or higher and generates
12590 objects that can only be linked by the GNU linker. However, the option
12591 does not affect the ABI of the final executable; it only affects the ABI
12592 of relocatable objects. Using @option{-mno-shared} will generally make
12593 executables both smaller and quicker.
12595 @option{-mshared} is the default.
12601 Assume (do not assume) that the static and dynamic linkers
12602 support PLTs and copy relocations. This option only affects
12603 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
12604 has no effect without @samp{-msym32}.
12606 You can make @option{-mplt} the default by configuring
12607 GCC with @option{--with-mips-plt}. The default is
12608 @option{-mno-plt} otherwise.
12614 Lift (do not lift) the usual restrictions on the size of the global
12617 GCC normally uses a single instruction to load values from the GOT@.
12618 While this is relatively efficient, it will only work if the GOT
12619 is smaller than about 64k. Anything larger will cause the linker
12620 to report an error such as:
12622 @cindex relocation truncated to fit (MIPS)
12624 relocation truncated to fit: R_MIPS_GOT16 foobar
12627 If this happens, you should recompile your code with @option{-mxgot}.
12628 It should then work with very large GOTs, although it will also be
12629 less efficient, since it will take three instructions to fetch the
12630 value of a global symbol.
12632 Note that some linkers can create multiple GOTs. If you have such a
12633 linker, you should only need to use @option{-mxgot} when a single object
12634 file accesses more than 64k's worth of GOT entries. Very few do.
12636 These options have no effect unless GCC is generating position
12641 Assume that general-purpose registers are 32 bits wide.
12645 Assume that general-purpose registers are 64 bits wide.
12649 Assume that floating-point registers are 32 bits wide.
12653 Assume that floating-point registers are 64 bits wide.
12656 @opindex mhard-float
12657 Use floating-point coprocessor instructions.
12660 @opindex msoft-float
12661 Do not use floating-point coprocessor instructions. Implement
12662 floating-point calculations using library calls instead.
12664 @item -msingle-float
12665 @opindex msingle-float
12666 Assume that the floating-point coprocessor only supports single-precision
12669 @item -mdouble-float
12670 @opindex mdouble-float
12671 Assume that the floating-point coprocessor supports double-precision
12672 operations. This is the default.
12678 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
12679 implement atomic memory built-in functions. When neither option is
12680 specified, GCC will use the instructions if the target architecture
12683 @option{-mllsc} is useful if the runtime environment can emulate the
12684 instructions and @option{-mno-llsc} can be useful when compiling for
12685 nonstandard ISAs. You can make either option the default by
12686 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
12687 respectively. @option{--with-llsc} is the default for some
12688 configurations; see the installation documentation for details.
12694 Use (do not use) revision 1 of the MIPS DSP ASE@.
12695 @xref{MIPS DSP Built-in Functions}. This option defines the
12696 preprocessor macro @samp{__mips_dsp}. It also defines
12697 @samp{__mips_dsp_rev} to 1.
12703 Use (do not use) revision 2 of the MIPS DSP ASE@.
12704 @xref{MIPS DSP Built-in Functions}. This option defines the
12705 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
12706 It also defines @samp{__mips_dsp_rev} to 2.
12709 @itemx -mno-smartmips
12710 @opindex msmartmips
12711 @opindex mno-smartmips
12712 Use (do not use) the MIPS SmartMIPS ASE.
12714 @item -mpaired-single
12715 @itemx -mno-paired-single
12716 @opindex mpaired-single
12717 @opindex mno-paired-single
12718 Use (do not use) paired-single floating-point instructions.
12719 @xref{MIPS Paired-Single Support}. This option requires
12720 hardware floating-point support to be enabled.
12726 Use (do not use) MIPS Digital Media Extension instructions.
12727 This option can only be used when generating 64-bit code and requires
12728 hardware floating-point support to be enabled.
12733 @opindex mno-mips3d
12734 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
12735 The option @option{-mips3d} implies @option{-mpaired-single}.
12741 Use (do not use) MT Multithreading instructions.
12745 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
12746 an explanation of the default and the way that the pointer size is
12751 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12753 The default size of @code{int}s, @code{long}s and pointers depends on
12754 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
12755 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12756 32-bit @code{long}s. Pointers are the same size as @code{long}s,
12757 or the same size as integer registers, whichever is smaller.
12763 Assume (do not assume) that all symbols have 32-bit values, regardless
12764 of the selected ABI@. This option is useful in combination with
12765 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12766 to generate shorter and faster references to symbolic addresses.
12770 Put definitions of externally-visible data in a small data section
12771 if that data is no bigger than @var{num} bytes. GCC can then access
12772 the data more efficiently; see @option{-mgpopt} for details.
12774 The default @option{-G} option depends on the configuration.
12776 @item -mlocal-sdata
12777 @itemx -mno-local-sdata
12778 @opindex mlocal-sdata
12779 @opindex mno-local-sdata
12780 Extend (do not extend) the @option{-G} behavior to local data too,
12781 such as to static variables in C@. @option{-mlocal-sdata} is the
12782 default for all configurations.
12784 If the linker complains that an application is using too much small data,
12785 you might want to try rebuilding the less performance-critical parts with
12786 @option{-mno-local-sdata}. You might also want to build large
12787 libraries with @option{-mno-local-sdata}, so that the libraries leave
12788 more room for the main program.
12790 @item -mextern-sdata
12791 @itemx -mno-extern-sdata
12792 @opindex mextern-sdata
12793 @opindex mno-extern-sdata
12794 Assume (do not assume) that externally-defined data will be in
12795 a small data section if that data is within the @option{-G} limit.
12796 @option{-mextern-sdata} is the default for all configurations.
12798 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12799 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12800 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12801 is placed in a small data section. If @var{Var} is defined by another
12802 module, you must either compile that module with a high-enough
12803 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12804 definition. If @var{Var} is common, you must link the application
12805 with a high-enough @option{-G} setting.
12807 The easiest way of satisfying these restrictions is to compile
12808 and link every module with the same @option{-G} option. However,
12809 you may wish to build a library that supports several different
12810 small data limits. You can do this by compiling the library with
12811 the highest supported @option{-G} setting and additionally using
12812 @option{-mno-extern-sdata} to stop the library from making assumptions
12813 about externally-defined data.
12819 Use (do not use) GP-relative accesses for symbols that are known to be
12820 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12821 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12824 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12825 might not hold the value of @code{_gp}. For example, if the code is
12826 part of a library that might be used in a boot monitor, programs that
12827 call boot monitor routines will pass an unknown value in @code{$gp}.
12828 (In such situations, the boot monitor itself would usually be compiled
12829 with @option{-G0}.)
12831 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12832 @option{-mno-extern-sdata}.
12834 @item -membedded-data
12835 @itemx -mno-embedded-data
12836 @opindex membedded-data
12837 @opindex mno-embedded-data
12838 Allocate variables to the read-only data section first if possible, then
12839 next in the small data section if possible, otherwise in data. This gives
12840 slightly slower code than the default, but reduces the amount of RAM required
12841 when executing, and thus may be preferred for some embedded systems.
12843 @item -muninit-const-in-rodata
12844 @itemx -mno-uninit-const-in-rodata
12845 @opindex muninit-const-in-rodata
12846 @opindex mno-uninit-const-in-rodata
12847 Put uninitialized @code{const} variables in the read-only data section.
12848 This option is only meaningful in conjunction with @option{-membedded-data}.
12850 @item -mcode-readable=@var{setting}
12851 @opindex mcode-readable
12852 Specify whether GCC may generate code that reads from executable sections.
12853 There are three possible settings:
12856 @item -mcode-readable=yes
12857 Instructions may freely access executable sections. This is the
12860 @item -mcode-readable=pcrel
12861 MIPS16 PC-relative load instructions can access executable sections,
12862 but other instructions must not do so. This option is useful on 4KSc
12863 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12864 It is also useful on processors that can be configured to have a dual
12865 instruction/data SRAM interface and that, like the M4K, automatically
12866 redirect PC-relative loads to the instruction RAM.
12868 @item -mcode-readable=no
12869 Instructions must not access executable sections. This option can be
12870 useful on targets that are configured to have a dual instruction/data
12871 SRAM interface but that (unlike the M4K) do not automatically redirect
12872 PC-relative loads to the instruction RAM.
12875 @item -msplit-addresses
12876 @itemx -mno-split-addresses
12877 @opindex msplit-addresses
12878 @opindex mno-split-addresses
12879 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12880 relocation operators. This option has been superseded by
12881 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12883 @item -mexplicit-relocs
12884 @itemx -mno-explicit-relocs
12885 @opindex mexplicit-relocs
12886 @opindex mno-explicit-relocs
12887 Use (do not use) assembler relocation operators when dealing with symbolic
12888 addresses. The alternative, selected by @option{-mno-explicit-relocs},
12889 is to use assembler macros instead.
12891 @option{-mexplicit-relocs} is the default if GCC was configured
12892 to use an assembler that supports relocation operators.
12894 @item -mcheck-zero-division
12895 @itemx -mno-check-zero-division
12896 @opindex mcheck-zero-division
12897 @opindex mno-check-zero-division
12898 Trap (do not trap) on integer division by zero.
12900 The default is @option{-mcheck-zero-division}.
12902 @item -mdivide-traps
12903 @itemx -mdivide-breaks
12904 @opindex mdivide-traps
12905 @opindex mdivide-breaks
12906 MIPS systems check for division by zero by generating either a
12907 conditional trap or a break instruction. Using traps results in
12908 smaller code, but is only supported on MIPS II and later. Also, some
12909 versions of the Linux kernel have a bug that prevents trap from
12910 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
12911 allow conditional traps on architectures that support them and
12912 @option{-mdivide-breaks} to force the use of breaks.
12914 The default is usually @option{-mdivide-traps}, but this can be
12915 overridden at configure time using @option{--with-divide=breaks}.
12916 Divide-by-zero checks can be completely disabled using
12917 @option{-mno-check-zero-division}.
12922 @opindex mno-memcpy
12923 Force (do not force) the use of @code{memcpy()} for non-trivial block
12924 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
12925 most constant-sized copies.
12928 @itemx -mno-long-calls
12929 @opindex mlong-calls
12930 @opindex mno-long-calls
12931 Disable (do not disable) use of the @code{jal} instruction. Calling
12932 functions using @code{jal} is more efficient but requires the caller
12933 and callee to be in the same 256 megabyte segment.
12935 This option has no effect on abicalls code. The default is
12936 @option{-mno-long-calls}.
12942 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
12943 instructions, as provided by the R4650 ISA@.
12946 @itemx -mno-fused-madd
12947 @opindex mfused-madd
12948 @opindex mno-fused-madd
12949 Enable (disable) use of the floating point multiply-accumulate
12950 instructions, when they are available. The default is
12951 @option{-mfused-madd}.
12953 When multiply-accumulate instructions are used, the intermediate
12954 product is calculated to infinite precision and is not subject to
12955 the FCSR Flush to Zero bit. This may be undesirable in some
12960 Tell the MIPS assembler to not run its preprocessor over user
12961 assembler files (with a @samp{.s} suffix) when assembling them.
12964 @itemx -mno-fix-r4000
12965 @opindex mfix-r4000
12966 @opindex mno-fix-r4000
12967 Work around certain R4000 CPU errata:
12970 A double-word or a variable shift may give an incorrect result if executed
12971 immediately after starting an integer division.
12973 A double-word or a variable shift may give an incorrect result if executed
12974 while an integer multiplication is in progress.
12976 An integer division may give an incorrect result if started in a delay slot
12977 of a taken branch or a jump.
12981 @itemx -mno-fix-r4400
12982 @opindex mfix-r4400
12983 @opindex mno-fix-r4400
12984 Work around certain R4400 CPU errata:
12987 A double-word or a variable shift may give an incorrect result if executed
12988 immediately after starting an integer division.
12992 @itemx -mno-fix-r10000
12993 @opindex mfix-r10000
12994 @opindex mno-fix-r10000
12995 Work around certain R10000 errata:
12998 @code{ll}/@code{sc} sequences may not behave atomically on revisions
12999 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
13002 This option can only be used if the target architecture supports
13003 branch-likely instructions. @option{-mfix-r10000} is the default when
13004 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
13008 @itemx -mno-fix-vr4120
13009 @opindex mfix-vr4120
13010 Work around certain VR4120 errata:
13013 @code{dmultu} does not always produce the correct result.
13015 @code{div} and @code{ddiv} do not always produce the correct result if one
13016 of the operands is negative.
13018 The workarounds for the division errata rely on special functions in
13019 @file{libgcc.a}. At present, these functions are only provided by
13020 the @code{mips64vr*-elf} configurations.
13022 Other VR4120 errata require a nop to be inserted between certain pairs of
13023 instructions. These errata are handled by the assembler, not by GCC itself.
13026 @opindex mfix-vr4130
13027 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
13028 workarounds are implemented by the assembler rather than by GCC,
13029 although GCC will avoid using @code{mflo} and @code{mfhi} if the
13030 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
13031 instructions are available instead.
13034 @itemx -mno-fix-sb1
13036 Work around certain SB-1 CPU core errata.
13037 (This flag currently works around the SB-1 revision 2
13038 ``F1'' and ``F2'' floating point errata.)
13040 @item -mr10k-cache-barrier=@var{setting}
13041 @opindex mr10k-cache-barrier
13042 Specify whether GCC should insert cache barriers to avoid the
13043 side-effects of speculation on R10K processors.
13045 In common with many processors, the R10K tries to predict the outcome
13046 of a conditional branch and speculatively executes instructions from
13047 the ``taken'' branch. It later aborts these instructions if the
13048 predicted outcome was wrong. However, on the R10K, even aborted
13049 instructions can have side effects.
13051 This problem only affects kernel stores and, depending on the system,
13052 kernel loads. As an example, a speculatively-executed store may load
13053 the target memory into cache and mark the cache line as dirty, even if
13054 the store itself is later aborted. If a DMA operation writes to the
13055 same area of memory before the ``dirty'' line is flushed, the cached
13056 data will overwrite the DMA-ed data. See the R10K processor manual
13057 for a full description, including other potential problems.
13059 One workaround is to insert cache barrier instructions before every memory
13060 access that might be speculatively executed and that might have side
13061 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
13062 controls GCC's implementation of this workaround. It assumes that
13063 aborted accesses to any byte in the following regions will not have
13068 the memory occupied by the current function's stack frame;
13071 the memory occupied by an incoming stack argument;
13074 the memory occupied by an object with a link-time-constant address.
13077 It is the kernel's responsibility to ensure that speculative
13078 accesses to these regions are indeed safe.
13080 If the input program contains a function declaration such as:
13086 then the implementation of @code{foo} must allow @code{j foo} and
13087 @code{jal foo} to be executed speculatively. GCC honors this
13088 restriction for functions it compiles itself. It expects non-GCC
13089 functions (such as hand-written assembly code) to do the same.
13091 The option has three forms:
13094 @item -mr10k-cache-barrier=load-store
13095 Insert a cache barrier before a load or store that might be
13096 speculatively executed and that might have side effects even
13099 @item -mr10k-cache-barrier=store
13100 Insert a cache barrier before a store that might be speculatively
13101 executed and that might have side effects even if aborted.
13103 @item -mr10k-cache-barrier=none
13104 Disable the insertion of cache barriers. This is the default setting.
13107 @item -mflush-func=@var{func}
13108 @itemx -mno-flush-func
13109 @opindex mflush-func
13110 Specifies the function to call to flush the I and D caches, or to not
13111 call any such function. If called, the function must take the same
13112 arguments as the common @code{_flush_func()}, that is, the address of the
13113 memory range for which the cache is being flushed, the size of the
13114 memory range, and the number 3 (to flush both caches). The default
13115 depends on the target GCC was configured for, but commonly is either
13116 @samp{_flush_func} or @samp{__cpu_flush}.
13118 @item mbranch-cost=@var{num}
13119 @opindex mbranch-cost
13120 Set the cost of branches to roughly @var{num} ``simple'' instructions.
13121 This cost is only a heuristic and is not guaranteed to produce
13122 consistent results across releases. A zero cost redundantly selects
13123 the default, which is based on the @option{-mtune} setting.
13125 @item -mbranch-likely
13126 @itemx -mno-branch-likely
13127 @opindex mbranch-likely
13128 @opindex mno-branch-likely
13129 Enable or disable use of Branch Likely instructions, regardless of the
13130 default for the selected architecture. By default, Branch Likely
13131 instructions may be generated if they are supported by the selected
13132 architecture. An exception is for the MIPS32 and MIPS64 architectures
13133 and processors which implement those architectures; for those, Branch
13134 Likely instructions will not be generated by default because the MIPS32
13135 and MIPS64 architectures specifically deprecate their use.
13137 @item -mfp-exceptions
13138 @itemx -mno-fp-exceptions
13139 @opindex mfp-exceptions
13140 Specifies whether FP exceptions are enabled. This affects how we schedule
13141 FP instructions for some processors. The default is that FP exceptions are
13144 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
13145 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
13148 @item -mvr4130-align
13149 @itemx -mno-vr4130-align
13150 @opindex mvr4130-align
13151 The VR4130 pipeline is two-way superscalar, but can only issue two
13152 instructions together if the first one is 8-byte aligned. When this
13153 option is enabled, GCC will align pairs of instructions that it
13154 thinks should execute in parallel.
13156 This option only has an effect when optimizing for the VR4130.
13157 It normally makes code faster, but at the expense of making it bigger.
13158 It is enabled by default at optimization level @option{-O3}.
13162 @subsection MMIX Options
13163 @cindex MMIX Options
13165 These options are defined for the MMIX:
13169 @itemx -mno-libfuncs
13171 @opindex mno-libfuncs
13172 Specify that intrinsic library functions are being compiled, passing all
13173 values in registers, no matter the size.
13176 @itemx -mno-epsilon
13178 @opindex mno-epsilon
13179 Generate floating-point comparison instructions that compare with respect
13180 to the @code{rE} epsilon register.
13182 @item -mabi=mmixware
13184 @opindex mabi-mmixware
13186 Generate code that passes function parameters and return values that (in
13187 the called function) are seen as registers @code{$0} and up, as opposed to
13188 the GNU ABI which uses global registers @code{$231} and up.
13190 @item -mzero-extend
13191 @itemx -mno-zero-extend
13192 @opindex mzero-extend
13193 @opindex mno-zero-extend
13194 When reading data from memory in sizes shorter than 64 bits, use (do not
13195 use) zero-extending load instructions by default, rather than
13196 sign-extending ones.
13199 @itemx -mno-knuthdiv
13201 @opindex mno-knuthdiv
13202 Make the result of a division yielding a remainder have the same sign as
13203 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
13204 remainder follows the sign of the dividend. Both methods are
13205 arithmetically valid, the latter being almost exclusively used.
13207 @item -mtoplevel-symbols
13208 @itemx -mno-toplevel-symbols
13209 @opindex mtoplevel-symbols
13210 @opindex mno-toplevel-symbols
13211 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
13212 code can be used with the @code{PREFIX} assembly directive.
13216 Generate an executable in the ELF format, rather than the default
13217 @samp{mmo} format used by the @command{mmix} simulator.
13219 @item -mbranch-predict
13220 @itemx -mno-branch-predict
13221 @opindex mbranch-predict
13222 @opindex mno-branch-predict
13223 Use (do not use) the probable-branch instructions, when static branch
13224 prediction indicates a probable branch.
13226 @item -mbase-addresses
13227 @itemx -mno-base-addresses
13228 @opindex mbase-addresses
13229 @opindex mno-base-addresses
13230 Generate (do not generate) code that uses @emph{base addresses}. Using a
13231 base address automatically generates a request (handled by the assembler
13232 and the linker) for a constant to be set up in a global register. The
13233 register is used for one or more base address requests within the range 0
13234 to 255 from the value held in the register. The generally leads to short
13235 and fast code, but the number of different data items that can be
13236 addressed is limited. This means that a program that uses lots of static
13237 data may require @option{-mno-base-addresses}.
13239 @item -msingle-exit
13240 @itemx -mno-single-exit
13241 @opindex msingle-exit
13242 @opindex mno-single-exit
13243 Force (do not force) generated code to have a single exit point in each
13247 @node MN10300 Options
13248 @subsection MN10300 Options
13249 @cindex MN10300 options
13251 These @option{-m} options are defined for Matsushita MN10300 architectures:
13256 Generate code to avoid bugs in the multiply instructions for the MN10300
13257 processors. This is the default.
13259 @item -mno-mult-bug
13260 @opindex mno-mult-bug
13261 Do not generate code to avoid bugs in the multiply instructions for the
13262 MN10300 processors.
13266 Generate code which uses features specific to the AM33 processor.
13270 Do not generate code which uses features specific to the AM33 processor. This
13273 @item -mreturn-pointer-on-d0
13274 @opindex mreturn-pointer-on-d0
13275 When generating a function which returns a pointer, return the pointer
13276 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
13277 only in a0, and attempts to call such functions without a prototype
13278 would result in errors. Note that this option is on by default; use
13279 @option{-mno-return-pointer-on-d0} to disable it.
13283 Do not link in the C run-time initialization object file.
13287 Indicate to the linker that it should perform a relaxation optimization pass
13288 to shorten branches, calls and absolute memory addresses. This option only
13289 has an effect when used on the command line for the final link step.
13291 This option makes symbolic debugging impossible.
13294 @node PDP-11 Options
13295 @subsection PDP-11 Options
13296 @cindex PDP-11 Options
13298 These options are defined for the PDP-11:
13303 Use hardware FPP floating point. This is the default. (FIS floating
13304 point on the PDP-11/40 is not supported.)
13307 @opindex msoft-float
13308 Do not use hardware floating point.
13312 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
13316 Return floating-point results in memory. This is the default.
13320 Generate code for a PDP-11/40.
13324 Generate code for a PDP-11/45. This is the default.
13328 Generate code for a PDP-11/10.
13330 @item -mbcopy-builtin
13331 @opindex bcopy-builtin
13332 Use inline @code{movmemhi} patterns for copying memory. This is the
13337 Do not use inline @code{movmemhi} patterns for copying memory.
13343 Use 16-bit @code{int}. This is the default.
13349 Use 32-bit @code{int}.
13352 @itemx -mno-float32
13354 @opindex mno-float32
13355 Use 64-bit @code{float}. This is the default.
13358 @itemx -mno-float64
13360 @opindex mno-float64
13361 Use 32-bit @code{float}.
13365 Use @code{abshi2} pattern. This is the default.
13369 Do not use @code{abshi2} pattern.
13371 @item -mbranch-expensive
13372 @opindex mbranch-expensive
13373 Pretend that branches are expensive. This is for experimenting with
13374 code generation only.
13376 @item -mbranch-cheap
13377 @opindex mbranch-cheap
13378 Do not pretend that branches are expensive. This is the default.
13382 Generate code for a system with split I&D@.
13386 Generate code for a system without split I&D@. This is the default.
13390 Use Unix assembler syntax. This is the default when configured for
13391 @samp{pdp11-*-bsd}.
13395 Use DEC assembler syntax. This is the default when configured for any
13396 PDP-11 target other than @samp{pdp11-*-bsd}.
13399 @node picoChip Options
13400 @subsection picoChip Options
13401 @cindex picoChip options
13403 These @samp{-m} options are defined for picoChip implementations:
13407 @item -mae=@var{ae_type}
13409 Set the instruction set, register set, and instruction scheduling
13410 parameters for array element type @var{ae_type}. Supported values
13411 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
13413 @option{-mae=ANY} selects a completely generic AE type. Code
13414 generated with this option will run on any of the other AE types. The
13415 code will not be as efficient as it would be if compiled for a specific
13416 AE type, and some types of operation (e.g., multiplication) will not
13417 work properly on all types of AE.
13419 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
13420 for compiled code, and is the default.
13422 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
13423 option may suffer from poor performance of byte (char) manipulation,
13424 since the DSP AE does not provide hardware support for byte load/stores.
13426 @item -msymbol-as-address
13427 Enable the compiler to directly use a symbol name as an address in a
13428 load/store instruction, without first loading it into a
13429 register. Typically, the use of this option will generate larger
13430 programs, which run faster than when the option isn't used. However, the
13431 results vary from program to program, so it is left as a user option,
13432 rather than being permanently enabled.
13434 @item -mno-inefficient-warnings
13435 Disables warnings about the generation of inefficient code. These
13436 warnings can be generated, for example, when compiling code which
13437 performs byte-level memory operations on the MAC AE type. The MAC AE has
13438 no hardware support for byte-level memory operations, so all byte
13439 load/stores must be synthesized from word load/store operations. This is
13440 inefficient and a warning will be generated indicating to the programmer
13441 that they should rewrite the code to avoid byte operations, or to target
13442 an AE type which has the necessary hardware support. This option enables
13443 the warning to be turned off.
13447 @node PowerPC Options
13448 @subsection PowerPC Options
13449 @cindex PowerPC options
13451 These are listed under @xref{RS/6000 and PowerPC Options}.
13453 @node RS/6000 and PowerPC Options
13454 @subsection IBM RS/6000 and PowerPC Options
13455 @cindex RS/6000 and PowerPC Options
13456 @cindex IBM RS/6000 and PowerPC Options
13458 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
13465 @itemx -mno-powerpc
13466 @itemx -mpowerpc-gpopt
13467 @itemx -mno-powerpc-gpopt
13468 @itemx -mpowerpc-gfxopt
13469 @itemx -mno-powerpc-gfxopt
13471 @itemx -mno-powerpc64
13475 @itemx -mno-popcntb
13483 @itemx -mno-hard-dfp
13487 @opindex mno-power2
13489 @opindex mno-powerpc
13490 @opindex mpowerpc-gpopt
13491 @opindex mno-powerpc-gpopt
13492 @opindex mpowerpc-gfxopt
13493 @opindex mno-powerpc-gfxopt
13494 @opindex mpowerpc64
13495 @opindex mno-powerpc64
13499 @opindex mno-popcntb
13505 @opindex mno-mfpgpr
13507 @opindex mno-hard-dfp
13508 GCC supports two related instruction set architectures for the
13509 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
13510 instructions supported by the @samp{rios} chip set used in the original
13511 RS/6000 systems and the @dfn{PowerPC} instruction set is the
13512 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
13513 the IBM 4xx, 6xx, and follow-on microprocessors.
13515 Neither architecture is a subset of the other. However there is a
13516 large common subset of instructions supported by both. An MQ
13517 register is included in processors supporting the POWER architecture.
13519 You use these options to specify which instructions are available on the
13520 processor you are using. The default value of these options is
13521 determined when configuring GCC@. Specifying the
13522 @option{-mcpu=@var{cpu_type}} overrides the specification of these
13523 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
13524 rather than the options listed above.
13526 The @option{-mpower} option allows GCC to generate instructions that
13527 are found only in the POWER architecture and to use the MQ register.
13528 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
13529 to generate instructions that are present in the POWER2 architecture but
13530 not the original POWER architecture.
13532 The @option{-mpowerpc} option allows GCC to generate instructions that
13533 are found only in the 32-bit subset of the PowerPC architecture.
13534 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
13535 GCC to use the optional PowerPC architecture instructions in the
13536 General Purpose group, including floating-point square root. Specifying
13537 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
13538 use the optional PowerPC architecture instructions in the Graphics
13539 group, including floating-point select.
13541 The @option{-mmfcrf} option allows GCC to generate the move from
13542 condition register field instruction implemented on the POWER4
13543 processor and other processors that support the PowerPC V2.01
13545 The @option{-mpopcntb} option allows GCC to generate the popcount and
13546 double precision FP reciprocal estimate instruction implemented on the
13547 POWER5 processor and other processors that support the PowerPC V2.02
13549 The @option{-mfprnd} option allows GCC to generate the FP round to
13550 integer instructions implemented on the POWER5+ processor and other
13551 processors that support the PowerPC V2.03 architecture.
13552 The @option{-mcmpb} option allows GCC to generate the compare bytes
13553 instruction implemented on the POWER6 processor and other processors
13554 that support the PowerPC V2.05 architecture.
13555 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
13556 general purpose register instructions implemented on the POWER6X
13557 processor and other processors that support the extended PowerPC V2.05
13559 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
13560 point instructions implemented on some POWER processors.
13562 The @option{-mpowerpc64} option allows GCC to generate the additional
13563 64-bit instructions that are found in the full PowerPC64 architecture
13564 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
13565 @option{-mno-powerpc64}.
13567 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
13568 will use only the instructions in the common subset of both
13569 architectures plus some special AIX common-mode calls, and will not use
13570 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
13571 permits GCC to use any instruction from either architecture and to
13572 allow use of the MQ register; specify this for the Motorola MPC601.
13574 @item -mnew-mnemonics
13575 @itemx -mold-mnemonics
13576 @opindex mnew-mnemonics
13577 @opindex mold-mnemonics
13578 Select which mnemonics to use in the generated assembler code. With
13579 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
13580 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
13581 assembler mnemonics defined for the POWER architecture. Instructions
13582 defined in only one architecture have only one mnemonic; GCC uses that
13583 mnemonic irrespective of which of these options is specified.
13585 GCC defaults to the mnemonics appropriate for the architecture in
13586 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
13587 value of these option. Unless you are building a cross-compiler, you
13588 should normally not specify either @option{-mnew-mnemonics} or
13589 @option{-mold-mnemonics}, but should instead accept the default.
13591 @item -mcpu=@var{cpu_type}
13593 Set architecture type, register usage, choice of mnemonics, and
13594 instruction scheduling parameters for machine type @var{cpu_type}.
13595 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
13596 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
13597 @samp{505}, @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
13598 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
13599 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
13600 @samp{860}, @samp{970}, @samp{8540}, @samp{e300c2}, @samp{e300c3},
13601 @samp{e500mc}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
13602 @samp{power}, @samp{power2}, @samp{power3}, @samp{power4},
13603 @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x}, @samp{power7}
13604 @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
13605 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
13607 @option{-mcpu=common} selects a completely generic processor. Code
13608 generated under this option will run on any POWER or PowerPC processor.
13609 GCC will use only the instructions in the common subset of both
13610 architectures, and will not use the MQ register. GCC assumes a generic
13611 processor model for scheduling purposes.
13613 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
13614 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
13615 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
13616 types, with an appropriate, generic processor model assumed for
13617 scheduling purposes.
13619 The other options specify a specific processor. Code generated under
13620 those options will run best on that processor, and may not run at all on
13623 The @option{-mcpu} options automatically enable or disable the
13626 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
13627 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
13628 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
13629 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr}
13631 The particular options set for any particular CPU will vary between
13632 compiler versions, depending on what setting seems to produce optimal
13633 code for that CPU; it doesn't necessarily reflect the actual hardware's
13634 capabilities. If you wish to set an individual option to a particular
13635 value, you may specify it after the @option{-mcpu} option, like
13636 @samp{-mcpu=970 -mno-altivec}.
13638 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
13639 not enabled or disabled by the @option{-mcpu} option at present because
13640 AIX does not have full support for these options. You may still
13641 enable or disable them individually if you're sure it'll work in your
13644 @item -mtune=@var{cpu_type}
13646 Set the instruction scheduling parameters for machine type
13647 @var{cpu_type}, but do not set the architecture type, register usage, or
13648 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
13649 values for @var{cpu_type} are used for @option{-mtune} as for
13650 @option{-mcpu}. If both are specified, the code generated will use the
13651 architecture, registers, and mnemonics set by @option{-mcpu}, but the
13652 scheduling parameters set by @option{-mtune}.
13658 Generate code to compute division as reciprocal estimate and iterative
13659 refinement, creating opportunities for increased throughput. This
13660 feature requires: optional PowerPC Graphics instruction set for single
13661 precision and FRE instruction for double precision, assuming divides
13662 cannot generate user-visible traps, and the domain values not include
13663 Infinities, denormals or zero denominator.
13666 @itemx -mno-altivec
13668 @opindex mno-altivec
13669 Generate code that uses (does not use) AltiVec instructions, and also
13670 enable the use of built-in functions that allow more direct access to
13671 the AltiVec instruction set. You may also need to set
13672 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
13678 @opindex mno-vrsave
13679 Generate VRSAVE instructions when generating AltiVec code.
13681 @item -mgen-cell-microcode
13682 @opindex mgen-cell-microcode
13683 Generate Cell microcode instructions
13685 @item -mwarn-cell-microcode
13686 @opindex mwarn-cell-microcode
13687 Warning when a Cell microcode instruction is going to emitted. An example
13688 of a Cell microcode instruction is a variable shift.
13691 @opindex msecure-plt
13692 Generate code that allows ld and ld.so to build executables and shared
13693 libraries with non-exec .plt and .got sections. This is a PowerPC
13694 32-bit SYSV ABI option.
13698 Generate code that uses a BSS .plt section that ld.so fills in, and
13699 requires .plt and .got sections that are both writable and executable.
13700 This is a PowerPC 32-bit SYSV ABI option.
13706 This switch enables or disables the generation of ISEL instructions.
13708 @item -misel=@var{yes/no}
13709 This switch has been deprecated. Use @option{-misel} and
13710 @option{-mno-isel} instead.
13716 This switch enables or disables the generation of SPE simd
13722 @opindex mno-paired
13723 This switch enables or disables the generation of PAIRED simd
13726 @item -mspe=@var{yes/no}
13727 This option has been deprecated. Use @option{-mspe} and
13728 @option{-mno-spe} instead.
13730 @item -mfloat-gprs=@var{yes/single/double/no}
13731 @itemx -mfloat-gprs
13732 @opindex mfloat-gprs
13733 This switch enables or disables the generation of floating point
13734 operations on the general purpose registers for architectures that
13737 The argument @var{yes} or @var{single} enables the use of
13738 single-precision floating point operations.
13740 The argument @var{double} enables the use of single and
13741 double-precision floating point operations.
13743 The argument @var{no} disables floating point operations on the
13744 general purpose registers.
13746 This option is currently only available on the MPC854x.
13752 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
13753 targets (including GNU/Linux). The 32-bit environment sets int, long
13754 and pointer to 32 bits and generates code that runs on any PowerPC
13755 variant. The 64-bit environment sets int to 32 bits and long and
13756 pointer to 64 bits, and generates code for PowerPC64, as for
13757 @option{-mpowerpc64}.
13760 @itemx -mno-fp-in-toc
13761 @itemx -mno-sum-in-toc
13762 @itemx -mminimal-toc
13764 @opindex mno-fp-in-toc
13765 @opindex mno-sum-in-toc
13766 @opindex mminimal-toc
13767 Modify generation of the TOC (Table Of Contents), which is created for
13768 every executable file. The @option{-mfull-toc} option is selected by
13769 default. In that case, GCC will allocate at least one TOC entry for
13770 each unique non-automatic variable reference in your program. GCC
13771 will also place floating-point constants in the TOC@. However, only
13772 16,384 entries are available in the TOC@.
13774 If you receive a linker error message that saying you have overflowed
13775 the available TOC space, you can reduce the amount of TOC space used
13776 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
13777 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
13778 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
13779 generate code to calculate the sum of an address and a constant at
13780 run-time instead of putting that sum into the TOC@. You may specify one
13781 or both of these options. Each causes GCC to produce very slightly
13782 slower and larger code at the expense of conserving TOC space.
13784 If you still run out of space in the TOC even when you specify both of
13785 these options, specify @option{-mminimal-toc} instead. This option causes
13786 GCC to make only one TOC entry for every file. When you specify this
13787 option, GCC will produce code that is slower and larger but which
13788 uses extremely little TOC space. You may wish to use this option
13789 only on files that contain less frequently executed code.
13795 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
13796 @code{long} type, and the infrastructure needed to support them.
13797 Specifying @option{-maix64} implies @option{-mpowerpc64} and
13798 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
13799 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
13802 @itemx -mno-xl-compat
13803 @opindex mxl-compat
13804 @opindex mno-xl-compat
13805 Produce code that conforms more closely to IBM XL compiler semantics
13806 when using AIX-compatible ABI@. Pass floating-point arguments to
13807 prototyped functions beyond the register save area (RSA) on the stack
13808 in addition to argument FPRs. Do not assume that most significant
13809 double in 128-bit long double value is properly rounded when comparing
13810 values and converting to double. Use XL symbol names for long double
13813 The AIX calling convention was extended but not initially documented to
13814 handle an obscure K&R C case of calling a function that takes the
13815 address of its arguments with fewer arguments than declared. IBM XL
13816 compilers access floating point arguments which do not fit in the
13817 RSA from the stack when a subroutine is compiled without
13818 optimization. Because always storing floating-point arguments on the
13819 stack is inefficient and rarely needed, this option is not enabled by
13820 default and only is necessary when calling subroutines compiled by IBM
13821 XL compilers without optimization.
13825 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
13826 application written to use message passing with special startup code to
13827 enable the application to run. The system must have PE installed in the
13828 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13829 must be overridden with the @option{-specs=} option to specify the
13830 appropriate directory location. The Parallel Environment does not
13831 support threads, so the @option{-mpe} option and the @option{-pthread}
13832 option are incompatible.
13834 @item -malign-natural
13835 @itemx -malign-power
13836 @opindex malign-natural
13837 @opindex malign-power
13838 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13839 @option{-malign-natural} overrides the ABI-defined alignment of larger
13840 types, such as floating-point doubles, on their natural size-based boundary.
13841 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13842 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
13844 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13848 @itemx -mhard-float
13849 @opindex msoft-float
13850 @opindex mhard-float
13851 Generate code that does not use (uses) the floating-point register set.
13852 Software floating point emulation is provided if you use the
13853 @option{-msoft-float} option, and pass the option to GCC when linking.
13855 @item -msingle-float
13856 @itemx -mdouble-float
13857 @opindex msingle-float
13858 @opindex mdouble-float
13859 Generate code for single or double-precision floating point operations.
13860 @option{-mdouble-float} implies @option{-msingle-float}.
13863 @opindex msimple-fpu
13864 Do not generate sqrt and div instructions for hardware floating point unit.
13868 Specify type of floating point unit. Valid values are @var{sp_lite}
13869 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
13870 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
13871 and @var{dp_full} (equivalent to -mdouble-float).
13874 @opindex mxilinx-fpu
13875 Perform optimizations for floating point unit on Xilinx PPC 405/440.
13878 @itemx -mno-multiple
13880 @opindex mno-multiple
13881 Generate code that uses (does not use) the load multiple word
13882 instructions and the store multiple word instructions. These
13883 instructions are generated by default on POWER systems, and not
13884 generated on PowerPC systems. Do not use @option{-mmultiple} on little
13885 endian PowerPC systems, since those instructions do not work when the
13886 processor is in little endian mode. The exceptions are PPC740 and
13887 PPC750 which permit the instructions usage in little endian mode.
13892 @opindex mno-string
13893 Generate code that uses (does not use) the load string instructions
13894 and the store string word instructions to save multiple registers and
13895 do small block moves. These instructions are generated by default on
13896 POWER systems, and not generated on PowerPC systems. Do not use
13897 @option{-mstring} on little endian PowerPC systems, since those
13898 instructions do not work when the processor is in little endian mode.
13899 The exceptions are PPC740 and PPC750 which permit the instructions
13900 usage in little endian mode.
13905 @opindex mno-update
13906 Generate code that uses (does not use) the load or store instructions
13907 that update the base register to the address of the calculated memory
13908 location. These instructions are generated by default. If you use
13909 @option{-mno-update}, there is a small window between the time that the
13910 stack pointer is updated and the address of the previous frame is
13911 stored, which means code that walks the stack frame across interrupts or
13912 signals may get corrupted data.
13914 @item -mavoid-indexed-addresses
13915 @item -mno-avoid-indexed-addresses
13916 @opindex mavoid-indexed-addresses
13917 @opindex mno-avoid-indexed-addresses
13918 Generate code that tries to avoid (not avoid) the use of indexed load
13919 or store instructions. These instructions can incur a performance
13920 penalty on Power6 processors in certain situations, such as when
13921 stepping through large arrays that cross a 16M boundary. This option
13922 is enabled by default when targetting Power6 and disabled otherwise.
13925 @itemx -mno-fused-madd
13926 @opindex mfused-madd
13927 @opindex mno-fused-madd
13928 Generate code that uses (does not use) the floating point multiply and
13929 accumulate instructions. These instructions are generated by default if
13930 hardware floating is used.
13936 Generate code that uses (does not use) the half-word multiply and
13937 multiply-accumulate instructions on the IBM 405, 440 and 464 processors.
13938 These instructions are generated by default when targetting those
13945 Generate code that uses (does not use) the string-search @samp{dlmzb}
13946 instruction on the IBM 405, 440 and 464 processors. This instruction is
13947 generated by default when targetting those processors.
13949 @item -mno-bit-align
13951 @opindex mno-bit-align
13952 @opindex mbit-align
13953 On System V.4 and embedded PowerPC systems do not (do) force structures
13954 and unions that contain bit-fields to be aligned to the base type of the
13957 For example, by default a structure containing nothing but 8
13958 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
13959 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
13960 the structure would be aligned to a 1 byte boundary and be one byte in
13963 @item -mno-strict-align
13964 @itemx -mstrict-align
13965 @opindex mno-strict-align
13966 @opindex mstrict-align
13967 On System V.4 and embedded PowerPC systems do not (do) assume that
13968 unaligned memory references will be handled by the system.
13970 @item -mrelocatable
13971 @itemx -mno-relocatable
13972 @opindex mrelocatable
13973 @opindex mno-relocatable
13974 On embedded PowerPC systems generate code that allows (does not allow)
13975 the program to be relocated to a different address at runtime. If you
13976 use @option{-mrelocatable} on any module, all objects linked together must
13977 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
13979 @item -mrelocatable-lib
13980 @itemx -mno-relocatable-lib
13981 @opindex mrelocatable-lib
13982 @opindex mno-relocatable-lib
13983 On embedded PowerPC systems generate code that allows (does not allow)
13984 the program to be relocated to a different address at runtime. Modules
13985 compiled with @option{-mrelocatable-lib} can be linked with either modules
13986 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
13987 with modules compiled with the @option{-mrelocatable} options.
13993 On System V.4 and embedded PowerPC systems do not (do) assume that
13994 register 2 contains a pointer to a global area pointing to the addresses
13995 used in the program.
13998 @itemx -mlittle-endian
14000 @opindex mlittle-endian
14001 On System V.4 and embedded PowerPC systems compile code for the
14002 processor in little endian mode. The @option{-mlittle-endian} option is
14003 the same as @option{-mlittle}.
14006 @itemx -mbig-endian
14008 @opindex mbig-endian
14009 On System V.4 and embedded PowerPC systems compile code for the
14010 processor in big endian mode. The @option{-mbig-endian} option is
14011 the same as @option{-mbig}.
14013 @item -mdynamic-no-pic
14014 @opindex mdynamic-no-pic
14015 On Darwin and Mac OS X systems, compile code so that it is not
14016 relocatable, but that its external references are relocatable. The
14017 resulting code is suitable for applications, but not shared
14020 @item -mprioritize-restricted-insns=@var{priority}
14021 @opindex mprioritize-restricted-insns
14022 This option controls the priority that is assigned to
14023 dispatch-slot restricted instructions during the second scheduling
14024 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
14025 @var{no/highest/second-highest} priority to dispatch slot restricted
14028 @item -msched-costly-dep=@var{dependence_type}
14029 @opindex msched-costly-dep
14030 This option controls which dependences are considered costly
14031 by the target during instruction scheduling. The argument
14032 @var{dependence_type} takes one of the following values:
14033 @var{no}: no dependence is costly,
14034 @var{all}: all dependences are costly,
14035 @var{true_store_to_load}: a true dependence from store to load is costly,
14036 @var{store_to_load}: any dependence from store to load is costly,
14037 @var{number}: any dependence which latency >= @var{number} is costly.
14039 @item -minsert-sched-nops=@var{scheme}
14040 @opindex minsert-sched-nops
14041 This option controls which nop insertion scheme will be used during
14042 the second scheduling pass. The argument @var{scheme} takes one of the
14044 @var{no}: Don't insert nops.
14045 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
14046 according to the scheduler's grouping.
14047 @var{regroup_exact}: Insert nops to force costly dependent insns into
14048 separate groups. Insert exactly as many nops as needed to force an insn
14049 to a new group, according to the estimated processor grouping.
14050 @var{number}: Insert nops to force costly dependent insns into
14051 separate groups. Insert @var{number} nops to force an insn to a new group.
14054 @opindex mcall-sysv
14055 On System V.4 and embedded PowerPC systems compile code using calling
14056 conventions that adheres to the March 1995 draft of the System V
14057 Application Binary Interface, PowerPC processor supplement. This is the
14058 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
14060 @item -mcall-sysv-eabi
14061 @opindex mcall-sysv-eabi
14062 Specify both @option{-mcall-sysv} and @option{-meabi} options.
14064 @item -mcall-sysv-noeabi
14065 @opindex mcall-sysv-noeabi
14066 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
14068 @item -mcall-solaris
14069 @opindex mcall-solaris
14070 On System V.4 and embedded PowerPC systems compile code for the Solaris
14074 @opindex mcall-linux
14075 On System V.4 and embedded PowerPC systems compile code for the
14076 Linux-based GNU system.
14080 On System V.4 and embedded PowerPC systems compile code for the
14081 Hurd-based GNU system.
14083 @item -mcall-netbsd
14084 @opindex mcall-netbsd
14085 On System V.4 and embedded PowerPC systems compile code for the
14086 NetBSD operating system.
14088 @item -maix-struct-return
14089 @opindex maix-struct-return
14090 Return all structures in memory (as specified by the AIX ABI)@.
14092 @item -msvr4-struct-return
14093 @opindex msvr4-struct-return
14094 Return structures smaller than 8 bytes in registers (as specified by the
14097 @item -mabi=@var{abi-type}
14099 Extend the current ABI with a particular extension, or remove such extension.
14100 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
14101 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
14105 Extend the current ABI with SPE ABI extensions. This does not change
14106 the default ABI, instead it adds the SPE ABI extensions to the current
14110 @opindex mabi=no-spe
14111 Disable Booke SPE ABI extensions for the current ABI@.
14113 @item -mabi=ibmlongdouble
14114 @opindex mabi=ibmlongdouble
14115 Change the current ABI to use IBM extended precision long double.
14116 This is a PowerPC 32-bit SYSV ABI option.
14118 @item -mabi=ieeelongdouble
14119 @opindex mabi=ieeelongdouble
14120 Change the current ABI to use IEEE extended precision long double.
14121 This is a PowerPC 32-bit Linux ABI option.
14124 @itemx -mno-prototype
14125 @opindex mprototype
14126 @opindex mno-prototype
14127 On System V.4 and embedded PowerPC systems assume that all calls to
14128 variable argument functions are properly prototyped. Otherwise, the
14129 compiler must insert an instruction before every non prototyped call to
14130 set or clear bit 6 of the condition code register (@var{CR}) to
14131 indicate whether floating point values were passed in the floating point
14132 registers in case the function takes a variable arguments. With
14133 @option{-mprototype}, only calls to prototyped variable argument functions
14134 will set or clear the bit.
14138 On embedded PowerPC systems, assume that the startup module is called
14139 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
14140 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
14145 On embedded PowerPC systems, assume that the startup module is called
14146 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
14151 On embedded PowerPC systems, assume that the startup module is called
14152 @file{crt0.o} and the standard C libraries are @file{libads.a} and
14155 @item -myellowknife
14156 @opindex myellowknife
14157 On embedded PowerPC systems, assume that the startup module is called
14158 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
14163 On System V.4 and embedded PowerPC systems, specify that you are
14164 compiling for a VxWorks system.
14168 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
14169 header to indicate that @samp{eabi} extended relocations are used.
14175 On System V.4 and embedded PowerPC systems do (do not) adhere to the
14176 Embedded Applications Binary Interface (eabi) which is a set of
14177 modifications to the System V.4 specifications. Selecting @option{-meabi}
14178 means that the stack is aligned to an 8 byte boundary, a function
14179 @code{__eabi} is called to from @code{main} to set up the eabi
14180 environment, and the @option{-msdata} option can use both @code{r2} and
14181 @code{r13} to point to two separate small data areas. Selecting
14182 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
14183 do not call an initialization function from @code{main}, and the
14184 @option{-msdata} option will only use @code{r13} to point to a single
14185 small data area. The @option{-meabi} option is on by default if you
14186 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
14189 @opindex msdata=eabi
14190 On System V.4 and embedded PowerPC systems, put small initialized
14191 @code{const} global and static data in the @samp{.sdata2} section, which
14192 is pointed to by register @code{r2}. Put small initialized
14193 non-@code{const} global and static data in the @samp{.sdata} section,
14194 which is pointed to by register @code{r13}. Put small uninitialized
14195 global and static data in the @samp{.sbss} section, which is adjacent to
14196 the @samp{.sdata} section. The @option{-msdata=eabi} option is
14197 incompatible with the @option{-mrelocatable} option. The
14198 @option{-msdata=eabi} option also sets the @option{-memb} option.
14201 @opindex msdata=sysv
14202 On System V.4 and embedded PowerPC systems, put small global and static
14203 data in the @samp{.sdata} section, which is pointed to by register
14204 @code{r13}. Put small uninitialized global and static data in the
14205 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
14206 The @option{-msdata=sysv} option is incompatible with the
14207 @option{-mrelocatable} option.
14209 @item -msdata=default
14211 @opindex msdata=default
14213 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
14214 compile code the same as @option{-msdata=eabi}, otherwise compile code the
14215 same as @option{-msdata=sysv}.
14218 @opindex msdata-data
14219 On System V.4 and embedded PowerPC systems, put small global
14220 data in the @samp{.sdata} section. Put small uninitialized global
14221 data in the @samp{.sbss} section. Do not use register @code{r13}
14222 to address small data however. This is the default behavior unless
14223 other @option{-msdata} options are used.
14227 @opindex msdata=none
14229 On embedded PowerPC systems, put all initialized global and static data
14230 in the @samp{.data} section, and all uninitialized data in the
14231 @samp{.bss} section.
14235 @cindex smaller data references (PowerPC)
14236 @cindex .sdata/.sdata2 references (PowerPC)
14237 On embedded PowerPC systems, put global and static items less than or
14238 equal to @var{num} bytes into the small data or bss sections instead of
14239 the normal data or bss section. By default, @var{num} is 8. The
14240 @option{-G @var{num}} switch is also passed to the linker.
14241 All modules should be compiled with the same @option{-G @var{num}} value.
14244 @itemx -mno-regnames
14246 @opindex mno-regnames
14247 On System V.4 and embedded PowerPC systems do (do not) emit register
14248 names in the assembly language output using symbolic forms.
14251 @itemx -mno-longcall
14253 @opindex mno-longcall
14254 By default assume that all calls are far away so that a longer more
14255 expensive calling sequence is required. This is required for calls
14256 further than 32 megabytes (33,554,432 bytes) from the current location.
14257 A short call will be generated if the compiler knows
14258 the call cannot be that far away. This setting can be overridden by
14259 the @code{shortcall} function attribute, or by @code{#pragma
14262 Some linkers are capable of detecting out-of-range calls and generating
14263 glue code on the fly. On these systems, long calls are unnecessary and
14264 generate slower code. As of this writing, the AIX linker can do this,
14265 as can the GNU linker for PowerPC/64. It is planned to add this feature
14266 to the GNU linker for 32-bit PowerPC systems as well.
14268 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
14269 callee, L42'', plus a ``branch island'' (glue code). The two target
14270 addresses represent the callee and the ``branch island''. The
14271 Darwin/PPC linker will prefer the first address and generate a ``bl
14272 callee'' if the PPC ``bl'' instruction will reach the callee directly;
14273 otherwise, the linker will generate ``bl L42'' to call the ``branch
14274 island''. The ``branch island'' is appended to the body of the
14275 calling function; it computes the full 32-bit address of the callee
14278 On Mach-O (Darwin) systems, this option directs the compiler emit to
14279 the glue for every direct call, and the Darwin linker decides whether
14280 to use or discard it.
14282 In the future, we may cause GCC to ignore all longcall specifications
14283 when the linker is known to generate glue.
14287 Adds support for multithreading with the @dfn{pthreads} library.
14288 This option sets flags for both the preprocessor and linker.
14292 @node S/390 and zSeries Options
14293 @subsection S/390 and zSeries Options
14294 @cindex S/390 and zSeries Options
14296 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
14300 @itemx -msoft-float
14301 @opindex mhard-float
14302 @opindex msoft-float
14303 Use (do not use) the hardware floating-point instructions and registers
14304 for floating-point operations. When @option{-msoft-float} is specified,
14305 functions in @file{libgcc.a} will be used to perform floating-point
14306 operations. When @option{-mhard-float} is specified, the compiler
14307 generates IEEE floating-point instructions. This is the default.
14310 @itemx -mno-hard-dfp
14312 @opindex mno-hard-dfp
14313 Use (do not use) the hardware decimal-floating-point instructions for
14314 decimal-floating-point operations. When @option{-mno-hard-dfp} is
14315 specified, functions in @file{libgcc.a} will be used to perform
14316 decimal-floating-point operations. When @option{-mhard-dfp} is
14317 specified, the compiler generates decimal-floating-point hardware
14318 instructions. This is the default for @option{-march=z9-ec} or higher.
14320 @item -mlong-double-64
14321 @itemx -mlong-double-128
14322 @opindex mlong-double-64
14323 @opindex mlong-double-128
14324 These switches control the size of @code{long double} type. A size
14325 of 64bit makes the @code{long double} type equivalent to the @code{double}
14326 type. This is the default.
14329 @itemx -mno-backchain
14330 @opindex mbackchain
14331 @opindex mno-backchain
14332 Store (do not store) the address of the caller's frame as backchain pointer
14333 into the callee's stack frame.
14334 A backchain may be needed to allow debugging using tools that do not understand
14335 DWARF-2 call frame information.
14336 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
14337 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
14338 the backchain is placed into the topmost word of the 96/160 byte register
14341 In general, code compiled with @option{-mbackchain} is call-compatible with
14342 code compiled with @option{-mmo-backchain}; however, use of the backchain
14343 for debugging purposes usually requires that the whole binary is built with
14344 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
14345 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
14346 to build a linux kernel use @option{-msoft-float}.
14348 The default is to not maintain the backchain.
14350 @item -mpacked-stack
14351 @itemx -mno-packed-stack
14352 @opindex mpacked-stack
14353 @opindex mno-packed-stack
14354 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
14355 specified, the compiler uses the all fields of the 96/160 byte register save
14356 area only for their default purpose; unused fields still take up stack space.
14357 When @option{-mpacked-stack} is specified, register save slots are densely
14358 packed at the top of the register save area; unused space is reused for other
14359 purposes, allowing for more efficient use of the available stack space.
14360 However, when @option{-mbackchain} is also in effect, the topmost word of
14361 the save area is always used to store the backchain, and the return address
14362 register is always saved two words below the backchain.
14364 As long as the stack frame backchain is not used, code generated with
14365 @option{-mpacked-stack} is call-compatible with code generated with
14366 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
14367 S/390 or zSeries generated code that uses the stack frame backchain at run
14368 time, not just for debugging purposes. Such code is not call-compatible
14369 with code compiled with @option{-mpacked-stack}. Also, note that the
14370 combination of @option{-mbackchain},
14371 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
14372 to build a linux kernel use @option{-msoft-float}.
14374 The default is to not use the packed stack layout.
14377 @itemx -mno-small-exec
14378 @opindex msmall-exec
14379 @opindex mno-small-exec
14380 Generate (or do not generate) code using the @code{bras} instruction
14381 to do subroutine calls.
14382 This only works reliably if the total executable size does not
14383 exceed 64k. The default is to use the @code{basr} instruction instead,
14384 which does not have this limitation.
14390 When @option{-m31} is specified, generate code compliant to the
14391 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
14392 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
14393 particular to generate 64-bit instructions. For the @samp{s390}
14394 targets, the default is @option{-m31}, while the @samp{s390x}
14395 targets default to @option{-m64}.
14401 When @option{-mzarch} is specified, generate code using the
14402 instructions available on z/Architecture.
14403 When @option{-mesa} is specified, generate code using the
14404 instructions available on ESA/390. Note that @option{-mesa} is
14405 not possible with @option{-m64}.
14406 When generating code compliant to the GNU/Linux for S/390 ABI,
14407 the default is @option{-mesa}. When generating code compliant
14408 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
14414 Generate (or do not generate) code using the @code{mvcle} instruction
14415 to perform block moves. When @option{-mno-mvcle} is specified,
14416 use a @code{mvc} loop instead. This is the default unless optimizing for
14423 Print (or do not print) additional debug information when compiling.
14424 The default is to not print debug information.
14426 @item -march=@var{cpu-type}
14428 Generate code that will run on @var{cpu-type}, which is the name of a system
14429 representing a certain processor type. Possible values for
14430 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
14431 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
14432 When generating code using the instructions available on z/Architecture,
14433 the default is @option{-march=z900}. Otherwise, the default is
14434 @option{-march=g5}.
14436 @item -mtune=@var{cpu-type}
14438 Tune to @var{cpu-type} everything applicable about the generated code,
14439 except for the ABI and the set of available instructions.
14440 The list of @var{cpu-type} values is the same as for @option{-march}.
14441 The default is the value used for @option{-march}.
14444 @itemx -mno-tpf-trace
14445 @opindex mtpf-trace
14446 @opindex mno-tpf-trace
14447 Generate code that adds (does not add) in TPF OS specific branches to trace
14448 routines in the operating system. This option is off by default, even
14449 when compiling for the TPF OS@.
14452 @itemx -mno-fused-madd
14453 @opindex mfused-madd
14454 @opindex mno-fused-madd
14455 Generate code that uses (does not use) the floating point multiply and
14456 accumulate instructions. These instructions are generated by default if
14457 hardware floating point is used.
14459 @item -mwarn-framesize=@var{framesize}
14460 @opindex mwarn-framesize
14461 Emit a warning if the current function exceeds the given frame size. Because
14462 this is a compile time check it doesn't need to be a real problem when the program
14463 runs. It is intended to identify functions which most probably cause
14464 a stack overflow. It is useful to be used in an environment with limited stack
14465 size e.g.@: the linux kernel.
14467 @item -mwarn-dynamicstack
14468 @opindex mwarn-dynamicstack
14469 Emit a warning if the function calls alloca or uses dynamically
14470 sized arrays. This is generally a bad idea with a limited stack size.
14472 @item -mstack-guard=@var{stack-guard}
14473 @itemx -mstack-size=@var{stack-size}
14474 @opindex mstack-guard
14475 @opindex mstack-size
14476 If these options are provided the s390 back end emits additional instructions in
14477 the function prologue which trigger a trap if the stack size is @var{stack-guard}
14478 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
14479 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
14480 the frame size of the compiled function is chosen.
14481 These options are intended to be used to help debugging stack overflow problems.
14482 The additionally emitted code causes only little overhead and hence can also be
14483 used in production like systems without greater performance degradation. The given
14484 values have to be exact powers of 2 and @var{stack-size} has to be greater than
14485 @var{stack-guard} without exceeding 64k.
14486 In order to be efficient the extra code makes the assumption that the stack starts
14487 at an address aligned to the value given by @var{stack-size}.
14488 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
14491 @node Score Options
14492 @subsection Score Options
14493 @cindex Score Options
14495 These options are defined for Score implementations:
14500 Compile code for big endian mode. This is the default.
14504 Compile code for little endian mode.
14508 Disable generate bcnz instruction.
14512 Enable generate unaligned load and store instruction.
14516 Enable the use of multiply-accumulate instructions. Disabled by default.
14520 Specify the SCORE5 as the target architecture.
14524 Specify the SCORE5U of the target architecture.
14528 Specify the SCORE7 as the target architecture. This is the default.
14532 Specify the SCORE7D as the target architecture.
14536 @subsection SH Options
14538 These @samp{-m} options are defined for the SH implementations:
14543 Generate code for the SH1.
14547 Generate code for the SH2.
14550 Generate code for the SH2e.
14554 Generate code for the SH3.
14558 Generate code for the SH3e.
14562 Generate code for the SH4 without a floating-point unit.
14564 @item -m4-single-only
14565 @opindex m4-single-only
14566 Generate code for the SH4 with a floating-point unit that only
14567 supports single-precision arithmetic.
14571 Generate code for the SH4 assuming the floating-point unit is in
14572 single-precision mode by default.
14576 Generate code for the SH4.
14580 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
14581 floating-point unit is not used.
14583 @item -m4a-single-only
14584 @opindex m4a-single-only
14585 Generate code for the SH4a, in such a way that no double-precision
14586 floating point operations are used.
14589 @opindex m4a-single
14590 Generate code for the SH4a assuming the floating-point unit is in
14591 single-precision mode by default.
14595 Generate code for the SH4a.
14599 Same as @option{-m4a-nofpu}, except that it implicitly passes
14600 @option{-dsp} to the assembler. GCC doesn't generate any DSP
14601 instructions at the moment.
14605 Compile code for the processor in big endian mode.
14609 Compile code for the processor in little endian mode.
14613 Align doubles at 64-bit boundaries. Note that this changes the calling
14614 conventions, and thus some functions from the standard C library will
14615 not work unless you recompile it first with @option{-mdalign}.
14619 Shorten some address references at link time, when possible; uses the
14620 linker option @option{-relax}.
14624 Use 32-bit offsets in @code{switch} tables. The default is to use
14629 Enable the use of bit manipulation instructions on SH2A.
14633 Enable the use of the instruction @code{fmovd}.
14637 Comply with the calling conventions defined by Renesas.
14641 Comply with the calling conventions defined by Renesas.
14645 Comply with the calling conventions defined for GCC before the Renesas
14646 conventions were available. This option is the default for all
14647 targets of the SH toolchain except for @samp{sh-symbianelf}.
14650 @opindex mnomacsave
14651 Mark the @code{MAC} register as call-clobbered, even if
14652 @option{-mhitachi} is given.
14656 Increase IEEE-compliance of floating-point code.
14657 At the moment, this is equivalent to @option{-fno-finite-math-only}.
14658 When generating 16 bit SH opcodes, getting IEEE-conforming results for
14659 comparisons of NANs / infinities incurs extra overhead in every
14660 floating point comparison, therefore the default is set to
14661 @option{-ffinite-math-only}.
14663 @item -minline-ic_invalidate
14664 @opindex minline-ic_invalidate
14665 Inline code to invalidate instruction cache entries after setting up
14666 nested function trampolines.
14667 This option has no effect if -musermode is in effect and the selected
14668 code generation option (e.g. -m4) does not allow the use of the icbi
14670 If the selected code generation option does not allow the use of the icbi
14671 instruction, and -musermode is not in effect, the inlined code will
14672 manipulate the instruction cache address array directly with an associative
14673 write. This not only requires privileged mode, but it will also
14674 fail if the cache line had been mapped via the TLB and has become unmapped.
14678 Dump instruction size and location in the assembly code.
14681 @opindex mpadstruct
14682 This option is deprecated. It pads structures to multiple of 4 bytes,
14683 which is incompatible with the SH ABI@.
14687 Optimize for space instead of speed. Implied by @option{-Os}.
14690 @opindex mprefergot
14691 When generating position-independent code, emit function calls using
14692 the Global Offset Table instead of the Procedure Linkage Table.
14696 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
14697 if the inlined code would not work in user mode.
14698 This is the default when the target is @code{sh-*-linux*}.
14700 @item -multcost=@var{number}
14701 @opindex multcost=@var{number}
14702 Set the cost to assume for a multiply insn.
14704 @item -mdiv=@var{strategy}
14705 @opindex mdiv=@var{strategy}
14706 Set the division strategy to use for SHmedia code. @var{strategy} must be
14707 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
14708 inv:call2, inv:fp .
14709 "fp" performs the operation in floating point. This has a very high latency,
14710 but needs only a few instructions, so it might be a good choice if
14711 your code has enough easily exploitable ILP to allow the compiler to
14712 schedule the floating point instructions together with other instructions.
14713 Division by zero causes a floating point exception.
14714 "inv" uses integer operations to calculate the inverse of the divisor,
14715 and then multiplies the dividend with the inverse. This strategy allows
14716 cse and hoisting of the inverse calculation. Division by zero calculates
14717 an unspecified result, but does not trap.
14718 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
14719 have been found, or if the entire operation has been hoisted to the same
14720 place, the last stages of the inverse calculation are intertwined with the
14721 final multiply to reduce the overall latency, at the expense of using a few
14722 more instructions, and thus offering fewer scheduling opportunities with
14724 "call" calls a library function that usually implements the inv:minlat
14726 This gives high code density for m5-*media-nofpu compilations.
14727 "call2" uses a different entry point of the same library function, where it
14728 assumes that a pointer to a lookup table has already been set up, which
14729 exposes the pointer load to cse / code hoisting optimizations.
14730 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
14731 code generation, but if the code stays unoptimized, revert to the "call",
14732 "call2", or "fp" strategies, respectively. Note that the
14733 potentially-trapping side effect of division by zero is carried by a
14734 separate instruction, so it is possible that all the integer instructions
14735 are hoisted out, but the marker for the side effect stays where it is.
14736 A recombination to fp operations or a call is not possible in that case.
14737 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
14738 that the inverse calculation was nor separated from the multiply, they speed
14739 up division where the dividend fits into 20 bits (plus sign where applicable),
14740 by inserting a test to skip a number of operations in this case; this test
14741 slows down the case of larger dividends. inv20u assumes the case of a such
14742 a small dividend to be unlikely, and inv20l assumes it to be likely.
14744 @item -mdivsi3_libfunc=@var{name}
14745 @opindex mdivsi3_libfunc=@var{name}
14746 Set the name of the library function used for 32 bit signed division to
14747 @var{name}. This only affect the name used in the call and inv:call
14748 division strategies, and the compiler will still expect the same
14749 sets of input/output/clobbered registers as if this option was not present.
14751 @item -mfixed-range=@var{register-range}
14752 @opindex mfixed-range
14753 Generate code treating the given register range as fixed registers.
14754 A fixed register is one that the register allocator can not use. This is
14755 useful when compiling kernel code. A register range is specified as
14756 two registers separated by a dash. Multiple register ranges can be
14757 specified separated by a comma.
14759 @item -madjust-unroll
14760 @opindex madjust-unroll
14761 Throttle unrolling to avoid thrashing target registers.
14762 This option only has an effect if the gcc code base supports the
14763 TARGET_ADJUST_UNROLL_MAX target hook.
14765 @item -mindexed-addressing
14766 @opindex mindexed-addressing
14767 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
14768 This is only safe if the hardware and/or OS implement 32 bit wrap-around
14769 semantics for the indexed addressing mode. The architecture allows the
14770 implementation of processors with 64 bit MMU, which the OS could use to
14771 get 32 bit addressing, but since no current hardware implementation supports
14772 this or any other way to make the indexed addressing mode safe to use in
14773 the 32 bit ABI, the default is -mno-indexed-addressing.
14775 @item -mgettrcost=@var{number}
14776 @opindex mgettrcost=@var{number}
14777 Set the cost assumed for the gettr instruction to @var{number}.
14778 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
14782 Assume pt* instructions won't trap. This will generally generate better
14783 scheduled code, but is unsafe on current hardware. The current architecture
14784 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
14785 This has the unintentional effect of making it unsafe to schedule ptabs /
14786 ptrel before a branch, or hoist it out of a loop. For example,
14787 __do_global_ctors, a part of libgcc that runs constructors at program
14788 startup, calls functions in a list which is delimited by @minus{}1. With the
14789 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
14790 That means that all the constructors will be run a bit quicker, but when
14791 the loop comes to the end of the list, the program crashes because ptabs
14792 loads @minus{}1 into a target register. Since this option is unsafe for any
14793 hardware implementing the current architecture specification, the default
14794 is -mno-pt-fixed. Unless the user specifies a specific cost with
14795 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
14796 this deters register allocation using target registers for storing
14799 @item -minvalid-symbols
14800 @opindex minvalid-symbols
14801 Assume symbols might be invalid. Ordinary function symbols generated by
14802 the compiler will always be valid to load with movi/shori/ptabs or
14803 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
14804 to generate symbols that will cause ptabs / ptrel to trap.
14805 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
14806 It will then prevent cross-basic-block cse, hoisting and most scheduling
14807 of symbol loads. The default is @option{-mno-invalid-symbols}.
14810 @node SPARC Options
14811 @subsection SPARC Options
14812 @cindex SPARC options
14814 These @samp{-m} options are supported on the SPARC:
14817 @item -mno-app-regs
14819 @opindex mno-app-regs
14821 Specify @option{-mapp-regs} to generate output using the global registers
14822 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
14825 To be fully SVR4 ABI compliant at the cost of some performance loss,
14826 specify @option{-mno-app-regs}. You should compile libraries and system
14827 software with this option.
14830 @itemx -mhard-float
14832 @opindex mhard-float
14833 Generate output containing floating point instructions. This is the
14837 @itemx -msoft-float
14839 @opindex msoft-float
14840 Generate output containing library calls for floating point.
14841 @strong{Warning:} the requisite libraries are not available for all SPARC
14842 targets. Normally the facilities of the machine's usual C compiler are
14843 used, but this cannot be done directly in cross-compilation. You must make
14844 your own arrangements to provide suitable library functions for
14845 cross-compilation. The embedded targets @samp{sparc-*-aout} and
14846 @samp{sparclite-*-*} do provide software floating point support.
14848 @option{-msoft-float} changes the calling convention in the output file;
14849 therefore, it is only useful if you compile @emph{all} of a program with
14850 this option. In particular, you need to compile @file{libgcc.a}, the
14851 library that comes with GCC, with @option{-msoft-float} in order for
14854 @item -mhard-quad-float
14855 @opindex mhard-quad-float
14856 Generate output containing quad-word (long double) floating point
14859 @item -msoft-quad-float
14860 @opindex msoft-quad-float
14861 Generate output containing library calls for quad-word (long double)
14862 floating point instructions. The functions called are those specified
14863 in the SPARC ABI@. This is the default.
14865 As of this writing, there are no SPARC implementations that have hardware
14866 support for the quad-word floating point instructions. They all invoke
14867 a trap handler for one of these instructions, and then the trap handler
14868 emulates the effect of the instruction. Because of the trap handler overhead,
14869 this is much slower than calling the ABI library routines. Thus the
14870 @option{-msoft-quad-float} option is the default.
14872 @item -mno-unaligned-doubles
14873 @itemx -munaligned-doubles
14874 @opindex mno-unaligned-doubles
14875 @opindex munaligned-doubles
14876 Assume that doubles have 8 byte alignment. This is the default.
14878 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
14879 alignment only if they are contained in another type, or if they have an
14880 absolute address. Otherwise, it assumes they have 4 byte alignment.
14881 Specifying this option avoids some rare compatibility problems with code
14882 generated by other compilers. It is not the default because it results
14883 in a performance loss, especially for floating point code.
14885 @item -mno-faster-structs
14886 @itemx -mfaster-structs
14887 @opindex mno-faster-structs
14888 @opindex mfaster-structs
14889 With @option{-mfaster-structs}, the compiler assumes that structures
14890 should have 8 byte alignment. This enables the use of pairs of
14891 @code{ldd} and @code{std} instructions for copies in structure
14892 assignment, in place of twice as many @code{ld} and @code{st} pairs.
14893 However, the use of this changed alignment directly violates the SPARC
14894 ABI@. Thus, it's intended only for use on targets where the developer
14895 acknowledges that their resulting code will not be directly in line with
14896 the rules of the ABI@.
14898 @item -mimpure-text
14899 @opindex mimpure-text
14900 @option{-mimpure-text}, used in addition to @option{-shared}, tells
14901 the compiler to not pass @option{-z text} to the linker when linking a
14902 shared object. Using this option, you can link position-dependent
14903 code into a shared object.
14905 @option{-mimpure-text} suppresses the ``relocations remain against
14906 allocatable but non-writable sections'' linker error message.
14907 However, the necessary relocations will trigger copy-on-write, and the
14908 shared object is not actually shared across processes. Instead of
14909 using @option{-mimpure-text}, you should compile all source code with
14910 @option{-fpic} or @option{-fPIC}.
14912 This option is only available on SunOS and Solaris.
14914 @item -mcpu=@var{cpu_type}
14916 Set the instruction set, register set, and instruction scheduling parameters
14917 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
14918 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
14919 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
14920 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
14921 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
14923 Default instruction scheduling parameters are used for values that select
14924 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
14925 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
14927 Here is a list of each supported architecture and their supported
14932 v8: supersparc, hypersparc
14933 sparclite: f930, f934, sparclite86x
14935 v9: ultrasparc, ultrasparc3, niagara, niagara2
14938 By default (unless configured otherwise), GCC generates code for the V7
14939 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
14940 additionally optimizes it for the Cypress CY7C602 chip, as used in the
14941 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
14942 SPARCStation 1, 2, IPX etc.
14944 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
14945 architecture. The only difference from V7 code is that the compiler emits
14946 the integer multiply and integer divide instructions which exist in SPARC-V8
14947 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
14948 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
14951 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
14952 the SPARC architecture. This adds the integer multiply, integer divide step
14953 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
14954 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
14955 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
14956 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
14957 MB86934 chip, which is the more recent SPARClite with FPU@.
14959 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
14960 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
14961 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
14962 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
14963 optimizes it for the TEMIC SPARClet chip.
14965 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
14966 architecture. This adds 64-bit integer and floating-point move instructions,
14967 3 additional floating-point condition code registers and conditional move
14968 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
14969 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
14970 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
14971 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
14972 @option{-mcpu=niagara}, the compiler additionally optimizes it for
14973 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
14974 additionally optimizes it for Sun UltraSPARC T2 chips.
14976 @item -mtune=@var{cpu_type}
14978 Set the instruction scheduling parameters for machine type
14979 @var{cpu_type}, but do not set the instruction set or register set that the
14980 option @option{-mcpu=@var{cpu_type}} would.
14982 The same values for @option{-mcpu=@var{cpu_type}} can be used for
14983 @option{-mtune=@var{cpu_type}}, but the only useful values are those
14984 that select a particular cpu implementation. Those are @samp{cypress},
14985 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
14986 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
14987 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
14992 @opindex mno-v8plus
14993 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
14994 difference from the V8 ABI is that the global and out registers are
14995 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
14996 mode for all SPARC-V9 processors.
15002 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
15003 Visual Instruction Set extensions. The default is @option{-mno-vis}.
15006 These @samp{-m} options are supported in addition to the above
15007 on SPARC-V9 processors in 64-bit environments:
15010 @item -mlittle-endian
15011 @opindex mlittle-endian
15012 Generate code for a processor running in little-endian mode. It is only
15013 available for a few configurations and most notably not on Solaris and Linux.
15019 Generate code for a 32-bit or 64-bit environment.
15020 The 32-bit environment sets int, long and pointer to 32 bits.
15021 The 64-bit environment sets int to 32 bits and long and pointer
15024 @item -mcmodel=medlow
15025 @opindex mcmodel=medlow
15026 Generate code for the Medium/Low code model: 64-bit addresses, programs
15027 must be linked in the low 32 bits of memory. Programs can be statically
15028 or dynamically linked.
15030 @item -mcmodel=medmid
15031 @opindex mcmodel=medmid
15032 Generate code for the Medium/Middle code model: 64-bit addresses, programs
15033 must be linked in the low 44 bits of memory, the text and data segments must
15034 be less than 2GB in size and the data segment must be located within 2GB of
15037 @item -mcmodel=medany
15038 @opindex mcmodel=medany
15039 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
15040 may be linked anywhere in memory, the text and data segments must be less
15041 than 2GB in size and the data segment must be located within 2GB of the
15044 @item -mcmodel=embmedany
15045 @opindex mcmodel=embmedany
15046 Generate code for the Medium/Anywhere code model for embedded systems:
15047 64-bit addresses, the text and data segments must be less than 2GB in
15048 size, both starting anywhere in memory (determined at link time). The
15049 global register %g4 points to the base of the data segment. Programs
15050 are statically linked and PIC is not supported.
15053 @itemx -mno-stack-bias
15054 @opindex mstack-bias
15055 @opindex mno-stack-bias
15056 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
15057 frame pointer if present, are offset by @minus{}2047 which must be added back
15058 when making stack frame references. This is the default in 64-bit mode.
15059 Otherwise, assume no such offset is present.
15062 These switches are supported in addition to the above on Solaris:
15067 Add support for multithreading using the Solaris threads library. This
15068 option sets flags for both the preprocessor and linker. This option does
15069 not affect the thread safety of object code produced by the compiler or
15070 that of libraries supplied with it.
15074 Add support for multithreading using the POSIX threads library. This
15075 option sets flags for both the preprocessor and linker. This option does
15076 not affect the thread safety of object code produced by the compiler or
15077 that of libraries supplied with it.
15081 This is a synonym for @option{-pthreads}.
15085 @subsection SPU Options
15086 @cindex SPU options
15088 These @samp{-m} options are supported on the SPU:
15092 @itemx -merror-reloc
15093 @opindex mwarn-reloc
15094 @opindex merror-reloc
15096 The loader for SPU does not handle dynamic relocations. By default, GCC
15097 will give an error when it generates code that requires a dynamic
15098 relocation. @option{-mno-error-reloc} disables the error,
15099 @option{-mwarn-reloc} will generate a warning instead.
15102 @itemx -munsafe-dma
15104 @opindex munsafe-dma
15106 Instructions which initiate or test completion of DMA must not be
15107 reordered with respect to loads and stores of the memory which is being
15108 accessed. Users typically address this problem using the volatile
15109 keyword, but that can lead to inefficient code in places where the
15110 memory is known to not change. Rather than mark the memory as volatile
15111 we treat the DMA instructions as potentially effecting all memory. With
15112 @option{-munsafe-dma} users must use the volatile keyword to protect
15115 @item -mbranch-hints
15116 @opindex mbranch-hints
15118 By default, GCC will generate a branch hint instruction to avoid
15119 pipeline stalls for always taken or probably taken branches. A hint
15120 will not be generated closer than 8 instructions away from its branch.
15121 There is little reason to disable them, except for debugging purposes,
15122 or to make an object a little bit smaller.
15126 @opindex msmall-mem
15127 @opindex mlarge-mem
15129 By default, GCC generates code assuming that addresses are never larger
15130 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
15131 a full 32 bit address.
15136 By default, GCC links against startup code that assumes the SPU-style
15137 main function interface (which has an unconventional parameter list).
15138 With @option{-mstdmain}, GCC will link your program against startup
15139 code that assumes a C99-style interface to @code{main}, including a
15140 local copy of @code{argv} strings.
15142 @item -mfixed-range=@var{register-range}
15143 @opindex mfixed-range
15144 Generate code treating the given register range as fixed registers.
15145 A fixed register is one that the register allocator can not use. This is
15146 useful when compiling kernel code. A register range is specified as
15147 two registers separated by a dash. Multiple register ranges can be
15148 specified separated by a comma.
15151 @itemx -mdual-nops=@var{n}
15152 @opindex mdual-nops
15153 By default, GCC will insert nops to increase dual issue when it expects
15154 it to increase performance. @var{n} can be a value from 0 to 10. A
15155 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
15156 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
15158 @item -mhint-max-nops=@var{n}
15159 @opindex mhint-max-nops
15160 Maximum number of nops to insert for a branch hint. A branch hint must
15161 be at least 8 instructions away from the branch it is effecting. GCC
15162 will insert up to @var{n} nops to enforce this, otherwise it will not
15163 generate the branch hint.
15165 @item -mhint-max-distance=@var{n}
15166 @opindex mhint-max-distance
15167 The encoding of the branch hint instruction limits the hint to be within
15168 256 instructions of the branch it is effecting. By default, GCC makes
15169 sure it is within 125.
15172 @opindex msafe-hints
15173 Work around a hardware bug which causes the SPU to stall indefinitely.
15174 By default, GCC will insert the @code{hbrp} instruction to make sure
15175 this stall won't happen.
15179 @node System V Options
15180 @subsection Options for System V
15182 These additional options are available on System V Release 4 for
15183 compatibility with other compilers on those systems:
15188 Create a shared object.
15189 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
15193 Identify the versions of each tool used by the compiler, in a
15194 @code{.ident} assembler directive in the output.
15198 Refrain from adding @code{.ident} directives to the output file (this is
15201 @item -YP,@var{dirs}
15203 Search the directories @var{dirs}, and no others, for libraries
15204 specified with @option{-l}.
15206 @item -Ym,@var{dir}
15208 Look in the directory @var{dir} to find the M4 preprocessor.
15209 The assembler uses this option.
15210 @c This is supposed to go with a -Yd for predefined M4 macro files, but
15211 @c the generic assembler that comes with Solaris takes just -Ym.
15215 @subsection V850 Options
15216 @cindex V850 Options
15218 These @samp{-m} options are defined for V850 implementations:
15222 @itemx -mno-long-calls
15223 @opindex mlong-calls
15224 @opindex mno-long-calls
15225 Treat all calls as being far away (near). If calls are assumed to be
15226 far away, the compiler will always load the functions address up into a
15227 register, and call indirect through the pointer.
15233 Do not optimize (do optimize) basic blocks that use the same index
15234 pointer 4 or more times to copy pointer into the @code{ep} register, and
15235 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
15236 option is on by default if you optimize.
15238 @item -mno-prolog-function
15239 @itemx -mprolog-function
15240 @opindex mno-prolog-function
15241 @opindex mprolog-function
15242 Do not use (do use) external functions to save and restore registers
15243 at the prologue and epilogue of a function. The external functions
15244 are slower, but use less code space if more than one function saves
15245 the same number of registers. The @option{-mprolog-function} option
15246 is on by default if you optimize.
15250 Try to make the code as small as possible. At present, this just turns
15251 on the @option{-mep} and @option{-mprolog-function} options.
15253 @item -mtda=@var{n}
15255 Put static or global variables whose size is @var{n} bytes or less into
15256 the tiny data area that register @code{ep} points to. The tiny data
15257 area can hold up to 256 bytes in total (128 bytes for byte references).
15259 @item -msda=@var{n}
15261 Put static or global variables whose size is @var{n} bytes or less into
15262 the small data area that register @code{gp} points to. The small data
15263 area can hold up to 64 kilobytes.
15265 @item -mzda=@var{n}
15267 Put static or global variables whose size is @var{n} bytes or less into
15268 the first 32 kilobytes of memory.
15272 Specify that the target processor is the V850.
15275 @opindex mbig-switch
15276 Generate code suitable for big switch tables. Use this option only if
15277 the assembler/linker complain about out of range branches within a switch
15282 This option will cause r2 and r5 to be used in the code generated by
15283 the compiler. This setting is the default.
15285 @item -mno-app-regs
15286 @opindex mno-app-regs
15287 This option will cause r2 and r5 to be treated as fixed registers.
15291 Specify that the target processor is the V850E1. The preprocessor
15292 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
15293 this option is used.
15297 Specify that the target processor is the V850E@. The preprocessor
15298 constant @samp{__v850e__} will be defined if this option is used.
15300 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
15301 are defined then a default target processor will be chosen and the
15302 relevant @samp{__v850*__} preprocessor constant will be defined.
15304 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
15305 defined, regardless of which processor variant is the target.
15307 @item -mdisable-callt
15308 @opindex mdisable-callt
15309 This option will suppress generation of the CALLT instruction for the
15310 v850e and v850e1 flavors of the v850 architecture. The default is
15311 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
15316 @subsection VAX Options
15317 @cindex VAX options
15319 These @samp{-m} options are defined for the VAX:
15324 Do not output certain jump instructions (@code{aobleq} and so on)
15325 that the Unix assembler for the VAX cannot handle across long
15330 Do output those jump instructions, on the assumption that you
15331 will assemble with the GNU assembler.
15335 Output code for g-format floating point numbers instead of d-format.
15338 @node VxWorks Options
15339 @subsection VxWorks Options
15340 @cindex VxWorks Options
15342 The options in this section are defined for all VxWorks targets.
15343 Options specific to the target hardware are listed with the other
15344 options for that target.
15349 GCC can generate code for both VxWorks kernels and real time processes
15350 (RTPs). This option switches from the former to the latter. It also
15351 defines the preprocessor macro @code{__RTP__}.
15354 @opindex non-static
15355 Link an RTP executable against shared libraries rather than static
15356 libraries. The options @option{-static} and @option{-shared} can
15357 also be used for RTPs (@pxref{Link Options}); @option{-static}
15364 These options are passed down to the linker. They are defined for
15365 compatibility with Diab.
15368 @opindex Xbind-lazy
15369 Enable lazy binding of function calls. This option is equivalent to
15370 @option{-Wl,-z,now} and is defined for compatibility with Diab.
15374 Disable lazy binding of function calls. This option is the default and
15375 is defined for compatibility with Diab.
15378 @node x86-64 Options
15379 @subsection x86-64 Options
15380 @cindex x86-64 options
15382 These are listed under @xref{i386 and x86-64 Options}.
15384 @node Xstormy16 Options
15385 @subsection Xstormy16 Options
15386 @cindex Xstormy16 Options
15388 These options are defined for Xstormy16:
15393 Choose startup files and linker script suitable for the simulator.
15396 @node Xtensa Options
15397 @subsection Xtensa Options
15398 @cindex Xtensa Options
15400 These options are supported for Xtensa targets:
15404 @itemx -mno-const16
15406 @opindex mno-const16
15407 Enable or disable use of @code{CONST16} instructions for loading
15408 constant values. The @code{CONST16} instruction is currently not a
15409 standard option from Tensilica. When enabled, @code{CONST16}
15410 instructions are always used in place of the standard @code{L32R}
15411 instructions. The use of @code{CONST16} is enabled by default only if
15412 the @code{L32R} instruction is not available.
15415 @itemx -mno-fused-madd
15416 @opindex mfused-madd
15417 @opindex mno-fused-madd
15418 Enable or disable use of fused multiply/add and multiply/subtract
15419 instructions in the floating-point option. This has no effect if the
15420 floating-point option is not also enabled. Disabling fused multiply/add
15421 and multiply/subtract instructions forces the compiler to use separate
15422 instructions for the multiply and add/subtract operations. This may be
15423 desirable in some cases where strict IEEE 754-compliant results are
15424 required: the fused multiply add/subtract instructions do not round the
15425 intermediate result, thereby producing results with @emph{more} bits of
15426 precision than specified by the IEEE standard. Disabling fused multiply
15427 add/subtract instructions also ensures that the program output is not
15428 sensitive to the compiler's ability to combine multiply and add/subtract
15431 @item -mserialize-volatile
15432 @itemx -mno-serialize-volatile
15433 @opindex mserialize-volatile
15434 @opindex mno-serialize-volatile
15435 When this option is enabled, GCC inserts @code{MEMW} instructions before
15436 @code{volatile} memory references to guarantee sequential consistency.
15437 The default is @option{-mserialize-volatile}. Use
15438 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
15440 @item -mtext-section-literals
15441 @itemx -mno-text-section-literals
15442 @opindex mtext-section-literals
15443 @opindex mno-text-section-literals
15444 Control the treatment of literal pools. The default is
15445 @option{-mno-text-section-literals}, which places literals in a separate
15446 section in the output file. This allows the literal pool to be placed
15447 in a data RAM/ROM, and it also allows the linker to combine literal
15448 pools from separate object files to remove redundant literals and
15449 improve code size. With @option{-mtext-section-literals}, the literals
15450 are interspersed in the text section in order to keep them as close as
15451 possible to their references. This may be necessary for large assembly
15454 @item -mtarget-align
15455 @itemx -mno-target-align
15456 @opindex mtarget-align
15457 @opindex mno-target-align
15458 When this option is enabled, GCC instructs the assembler to
15459 automatically align instructions to reduce branch penalties at the
15460 expense of some code density. The assembler attempts to widen density
15461 instructions to align branch targets and the instructions following call
15462 instructions. If there are not enough preceding safe density
15463 instructions to align a target, no widening will be performed. The
15464 default is @option{-mtarget-align}. These options do not affect the
15465 treatment of auto-aligned instructions like @code{LOOP}, which the
15466 assembler will always align, either by widening density instructions or
15467 by inserting no-op instructions.
15470 @itemx -mno-longcalls
15471 @opindex mlongcalls
15472 @opindex mno-longcalls
15473 When this option is enabled, GCC instructs the assembler to translate
15474 direct calls to indirect calls unless it can determine that the target
15475 of a direct call is in the range allowed by the call instruction. This
15476 translation typically occurs for calls to functions in other source
15477 files. Specifically, the assembler translates a direct @code{CALL}
15478 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
15479 The default is @option{-mno-longcalls}. This option should be used in
15480 programs where the call target can potentially be out of range. This
15481 option is implemented in the assembler, not the compiler, so the
15482 assembly code generated by GCC will still show direct call
15483 instructions---look at the disassembled object code to see the actual
15484 instructions. Note that the assembler will use an indirect call for
15485 every cross-file call, not just those that really will be out of range.
15488 @node zSeries Options
15489 @subsection zSeries Options
15490 @cindex zSeries options
15492 These are listed under @xref{S/390 and zSeries Options}.
15494 @node Code Gen Options
15495 @section Options for Code Generation Conventions
15496 @cindex code generation conventions
15497 @cindex options, code generation
15498 @cindex run-time options
15500 These machine-independent options control the interface conventions
15501 used in code generation.
15503 Most of them have both positive and negative forms; the negative form
15504 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
15505 one of the forms is listed---the one which is not the default. You
15506 can figure out the other form by either removing @samp{no-} or adding
15510 @item -fbounds-check
15511 @opindex fbounds-check
15512 For front-ends that support it, generate additional code to check that
15513 indices used to access arrays are within the declared range. This is
15514 currently only supported by the Java and Fortran front-ends, where
15515 this option defaults to true and false respectively.
15519 This option generates traps for signed overflow on addition, subtraction,
15520 multiplication operations.
15524 This option instructs the compiler to assume that signed arithmetic
15525 overflow of addition, subtraction and multiplication wraps around
15526 using twos-complement representation. This flag enables some optimizations
15527 and disables others. This option is enabled by default for the Java
15528 front-end, as required by the Java language specification.
15531 @opindex fexceptions
15532 Enable exception handling. Generates extra code needed to propagate
15533 exceptions. For some targets, this implies GCC will generate frame
15534 unwind information for all functions, which can produce significant data
15535 size overhead, although it does not affect execution. If you do not
15536 specify this option, GCC will enable it by default for languages like
15537 C++ which normally require exception handling, and disable it for
15538 languages like C that do not normally require it. However, you may need
15539 to enable this option when compiling C code that needs to interoperate
15540 properly with exception handlers written in C++. You may also wish to
15541 disable this option if you are compiling older C++ programs that don't
15542 use exception handling.
15544 @item -fnon-call-exceptions
15545 @opindex fnon-call-exceptions
15546 Generate code that allows trapping instructions to throw exceptions.
15547 Note that this requires platform-specific runtime support that does
15548 not exist everywhere. Moreover, it only allows @emph{trapping}
15549 instructions to throw exceptions, i.e.@: memory references or floating
15550 point instructions. It does not allow exceptions to be thrown from
15551 arbitrary signal handlers such as @code{SIGALRM}.
15553 @item -funwind-tables
15554 @opindex funwind-tables
15555 Similar to @option{-fexceptions}, except that it will just generate any needed
15556 static data, but will not affect the generated code in any other way.
15557 You will normally not enable this option; instead, a language processor
15558 that needs this handling would enable it on your behalf.
15560 @item -fasynchronous-unwind-tables
15561 @opindex fasynchronous-unwind-tables
15562 Generate unwind table in dwarf2 format, if supported by target machine. The
15563 table is exact at each instruction boundary, so it can be used for stack
15564 unwinding from asynchronous events (such as debugger or garbage collector).
15566 @item -fpcc-struct-return
15567 @opindex fpcc-struct-return
15568 Return ``short'' @code{struct} and @code{union} values in memory like
15569 longer ones, rather than in registers. This convention is less
15570 efficient, but it has the advantage of allowing intercallability between
15571 GCC-compiled files and files compiled with other compilers, particularly
15572 the Portable C Compiler (pcc).
15574 The precise convention for returning structures in memory depends
15575 on the target configuration macros.
15577 Short structures and unions are those whose size and alignment match
15578 that of some integer type.
15580 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
15581 switch is not binary compatible with code compiled with the
15582 @option{-freg-struct-return} switch.
15583 Use it to conform to a non-default application binary interface.
15585 @item -freg-struct-return
15586 @opindex freg-struct-return
15587 Return @code{struct} and @code{union} values in registers when possible.
15588 This is more efficient for small structures than
15589 @option{-fpcc-struct-return}.
15591 If you specify neither @option{-fpcc-struct-return} nor
15592 @option{-freg-struct-return}, GCC defaults to whichever convention is
15593 standard for the target. If there is no standard convention, GCC
15594 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
15595 the principal compiler. In those cases, we can choose the standard, and
15596 we chose the more efficient register return alternative.
15598 @strong{Warning:} code compiled with the @option{-freg-struct-return}
15599 switch is not binary compatible with code compiled with the
15600 @option{-fpcc-struct-return} switch.
15601 Use it to conform to a non-default application binary interface.
15603 @item -fshort-enums
15604 @opindex fshort-enums
15605 Allocate to an @code{enum} type only as many bytes as it needs for the
15606 declared range of possible values. Specifically, the @code{enum} type
15607 will be equivalent to the smallest integer type which has enough room.
15609 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
15610 code that is not binary compatible with code generated without that switch.
15611 Use it to conform to a non-default application binary interface.
15613 @item -fshort-double
15614 @opindex fshort-double
15615 Use the same size for @code{double} as for @code{float}.
15617 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
15618 code that is not binary compatible with code generated without that switch.
15619 Use it to conform to a non-default application binary interface.
15621 @item -fshort-wchar
15622 @opindex fshort-wchar
15623 Override the underlying type for @samp{wchar_t} to be @samp{short
15624 unsigned int} instead of the default for the target. This option is
15625 useful for building programs to run under WINE@.
15627 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
15628 code that is not binary compatible with code generated without that switch.
15629 Use it to conform to a non-default application binary interface.
15632 @opindex fno-common
15633 In C, allocate even uninitialized global variables in the data section of the
15634 object file, rather than generating them as common blocks. This has the
15635 effect that if the same variable is declared (without @code{extern}) in
15636 two different compilations, you will get an error when you link them.
15637 The only reason this might be useful is if you wish to verify that the
15638 program will work on other systems which always work this way.
15642 Ignore the @samp{#ident} directive.
15644 @item -finhibit-size-directive
15645 @opindex finhibit-size-directive
15646 Don't output a @code{.size} assembler directive, or anything else that
15647 would cause trouble if the function is split in the middle, and the
15648 two halves are placed at locations far apart in memory. This option is
15649 used when compiling @file{crtstuff.c}; you should not need to use it
15652 @item -fverbose-asm
15653 @opindex fverbose-asm
15654 Put extra commentary information in the generated assembly code to
15655 make it more readable. This option is generally only of use to those
15656 who actually need to read the generated assembly code (perhaps while
15657 debugging the compiler itself).
15659 @option{-fno-verbose-asm}, the default, causes the
15660 extra information to be omitted and is useful when comparing two assembler
15663 @item -frecord-gcc-switches
15664 @opindex frecord-gcc-switches
15665 This switch causes the command line that was used to invoke the
15666 compiler to be recorded into the object file that is being created.
15667 This switch is only implemented on some targets and the exact format
15668 of the recording is target and binary file format dependent, but it
15669 usually takes the form of a section containing ASCII text. This
15670 switch is related to the @option{-fverbose-asm} switch, but that
15671 switch only records information in the assembler output file as
15672 comments, so it never reaches the object file.
15676 @cindex global offset table
15678 Generate position-independent code (PIC) suitable for use in a shared
15679 library, if supported for the target machine. Such code accesses all
15680 constant addresses through a global offset table (GOT)@. The dynamic
15681 loader resolves the GOT entries when the program starts (the dynamic
15682 loader is not part of GCC; it is part of the operating system). If
15683 the GOT size for the linked executable exceeds a machine-specific
15684 maximum size, you get an error message from the linker indicating that
15685 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
15686 instead. (These maximums are 8k on the SPARC and 32k
15687 on the m68k and RS/6000. The 386 has no such limit.)
15689 Position-independent code requires special support, and therefore works
15690 only on certain machines. For the 386, GCC supports PIC for System V
15691 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
15692 position-independent.
15694 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15699 If supported for the target machine, emit position-independent code,
15700 suitable for dynamic linking and avoiding any limit on the size of the
15701 global offset table. This option makes a difference on the m68k,
15702 PowerPC and SPARC@.
15704 Position-independent code requires special support, and therefore works
15705 only on certain machines.
15707 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15714 These options are similar to @option{-fpic} and @option{-fPIC}, but
15715 generated position independent code can be only linked into executables.
15716 Usually these options are used when @option{-pie} GCC option will be
15717 used during linking.
15719 @option{-fpie} and @option{-fPIE} both define the macros
15720 @code{__pie__} and @code{__PIE__}. The macros have the value 1
15721 for @option{-fpie} and 2 for @option{-fPIE}.
15723 @item -fno-jump-tables
15724 @opindex fno-jump-tables
15725 Do not use jump tables for switch statements even where it would be
15726 more efficient than other code generation strategies. This option is
15727 of use in conjunction with @option{-fpic} or @option{-fPIC} for
15728 building code which forms part of a dynamic linker and cannot
15729 reference the address of a jump table. On some targets, jump tables
15730 do not require a GOT and this option is not needed.
15732 @item -ffixed-@var{reg}
15734 Treat the register named @var{reg} as a fixed register; generated code
15735 should never refer to it (except perhaps as a stack pointer, frame
15736 pointer or in some other fixed role).
15738 @var{reg} must be the name of a register. The register names accepted
15739 are machine-specific and are defined in the @code{REGISTER_NAMES}
15740 macro in the machine description macro file.
15742 This flag does not have a negative form, because it specifies a
15745 @item -fcall-used-@var{reg}
15746 @opindex fcall-used
15747 Treat the register named @var{reg} as an allocable register that is
15748 clobbered by function calls. It may be allocated for temporaries or
15749 variables that do not live across a call. Functions compiled this way
15750 will not save and restore the register @var{reg}.
15752 It is an error to used this flag with the frame pointer or stack pointer.
15753 Use of this flag for other registers that have fixed pervasive roles in
15754 the machine's execution model will produce disastrous results.
15756 This flag does not have a negative form, because it specifies a
15759 @item -fcall-saved-@var{reg}
15760 @opindex fcall-saved
15761 Treat the register named @var{reg} as an allocable register saved by
15762 functions. It may be allocated even for temporaries or variables that
15763 live across a call. Functions compiled this way will save and restore
15764 the register @var{reg} if they use it.
15766 It is an error to used this flag with the frame pointer or stack pointer.
15767 Use of this flag for other registers that have fixed pervasive roles in
15768 the machine's execution model will produce disastrous results.
15770 A different sort of disaster will result from the use of this flag for
15771 a register in which function values may be returned.
15773 This flag does not have a negative form, because it specifies a
15776 @item -fpack-struct[=@var{n}]
15777 @opindex fpack-struct
15778 Without a value specified, pack all structure members together without
15779 holes. When a value is specified (which must be a small power of two), pack
15780 structure members according to this value, representing the maximum
15781 alignment (that is, objects with default alignment requirements larger than
15782 this will be output potentially unaligned at the next fitting location.
15784 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
15785 code that is not binary compatible with code generated without that switch.
15786 Additionally, it makes the code suboptimal.
15787 Use it to conform to a non-default application binary interface.
15789 @item -finstrument-functions
15790 @opindex finstrument-functions
15791 Generate instrumentation calls for entry and exit to functions. Just
15792 after function entry and just before function exit, the following
15793 profiling functions will be called with the address of the current
15794 function and its call site. (On some platforms,
15795 @code{__builtin_return_address} does not work beyond the current
15796 function, so the call site information may not be available to the
15797 profiling functions otherwise.)
15800 void __cyg_profile_func_enter (void *this_fn,
15802 void __cyg_profile_func_exit (void *this_fn,
15806 The first argument is the address of the start of the current function,
15807 which may be looked up exactly in the symbol table.
15809 This instrumentation is also done for functions expanded inline in other
15810 functions. The profiling calls will indicate where, conceptually, the
15811 inline function is entered and exited. This means that addressable
15812 versions of such functions must be available. If all your uses of a
15813 function are expanded inline, this may mean an additional expansion of
15814 code size. If you use @samp{extern inline} in your C code, an
15815 addressable version of such functions must be provided. (This is
15816 normally the case anyways, but if you get lucky and the optimizer always
15817 expands the functions inline, you might have gotten away without
15818 providing static copies.)
15820 A function may be given the attribute @code{no_instrument_function}, in
15821 which case this instrumentation will not be done. This can be used, for
15822 example, for the profiling functions listed above, high-priority
15823 interrupt routines, and any functions from which the profiling functions
15824 cannot safely be called (perhaps signal handlers, if the profiling
15825 routines generate output or allocate memory).
15827 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
15828 @opindex finstrument-functions-exclude-file-list
15830 Set the list of functions that are excluded from instrumentation (see
15831 the description of @code{-finstrument-functions}). If the file that
15832 contains a function definition matches with one of @var{file}, then
15833 that function is not instrumented. The match is done on substrings:
15834 if the @var{file} parameter is a substring of the file name, it is
15835 considered to be a match.
15838 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
15839 will exclude any inline function defined in files whose pathnames
15840 contain @code{/bits/stl} or @code{include/sys}.
15842 If, for some reason, you want to include letter @code{','} in one of
15843 @var{sym}, write @code{'\,'}. For example,
15844 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
15845 (note the single quote surrounding the option).
15847 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
15848 @opindex finstrument-functions-exclude-function-list
15850 This is similar to @code{-finstrument-functions-exclude-file-list},
15851 but this option sets the list of function names to be excluded from
15852 instrumentation. The function name to be matched is its user-visible
15853 name, such as @code{vector<int> blah(const vector<int> &)}, not the
15854 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
15855 match is done on substrings: if the @var{sym} parameter is a substring
15856 of the function name, it is considered to be a match.
15858 @item -fstack-check
15859 @opindex fstack-check
15860 Generate code to verify that you do not go beyond the boundary of the
15861 stack. You should specify this flag if you are running in an
15862 environment with multiple threads, but only rarely need to specify it in
15863 a single-threaded environment since stack overflow is automatically
15864 detected on nearly all systems if there is only one stack.
15866 Note that this switch does not actually cause checking to be done; the
15867 operating system or the language runtime must do that. The switch causes
15868 generation of code to ensure that they see the stack being extended.
15870 You can additionally specify a string parameter: @code{no} means no
15871 checking, @code{generic} means force the use of old-style checking,
15872 @code{specific} means use the best checking method and is equivalent
15873 to bare @option{-fstack-check}.
15875 Old-style checking is a generic mechanism that requires no specific
15876 target support in the compiler but comes with the following drawbacks:
15880 Modified allocation strategy for large objects: they will always be
15881 allocated dynamically if their size exceeds a fixed threshold.
15884 Fixed limit on the size of the static frame of functions: when it is
15885 topped by a particular function, stack checking is not reliable and
15886 a warning is issued by the compiler.
15889 Inefficiency: because of both the modified allocation strategy and the
15890 generic implementation, the performances of the code are hampered.
15893 Note that old-style stack checking is also the fallback method for
15894 @code{specific} if no target support has been added in the compiler.
15896 @item -fstack-limit-register=@var{reg}
15897 @itemx -fstack-limit-symbol=@var{sym}
15898 @itemx -fno-stack-limit
15899 @opindex fstack-limit-register
15900 @opindex fstack-limit-symbol
15901 @opindex fno-stack-limit
15902 Generate code to ensure that the stack does not grow beyond a certain value,
15903 either the value of a register or the address of a symbol. If the stack
15904 would grow beyond the value, a signal is raised. For most targets,
15905 the signal is raised before the stack overruns the boundary, so
15906 it is possible to catch the signal without taking special precautions.
15908 For instance, if the stack starts at absolute address @samp{0x80000000}
15909 and grows downwards, you can use the flags
15910 @option{-fstack-limit-symbol=__stack_limit} and
15911 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
15912 of 128KB@. Note that this may only work with the GNU linker.
15914 @cindex aliasing of parameters
15915 @cindex parameters, aliased
15916 @item -fargument-alias
15917 @itemx -fargument-noalias
15918 @itemx -fargument-noalias-global
15919 @itemx -fargument-noalias-anything
15920 @opindex fargument-alias
15921 @opindex fargument-noalias
15922 @opindex fargument-noalias-global
15923 @opindex fargument-noalias-anything
15924 Specify the possible relationships among parameters and between
15925 parameters and global data.
15927 @option{-fargument-alias} specifies that arguments (parameters) may
15928 alias each other and may alias global storage.@*
15929 @option{-fargument-noalias} specifies that arguments do not alias
15930 each other, but may alias global storage.@*
15931 @option{-fargument-noalias-global} specifies that arguments do not
15932 alias each other and do not alias global storage.
15933 @option{-fargument-noalias-anything} specifies that arguments do not
15934 alias any other storage.
15936 Each language will automatically use whatever option is required by
15937 the language standard. You should not need to use these options yourself.
15939 @item -fleading-underscore
15940 @opindex fleading-underscore
15941 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
15942 change the way C symbols are represented in the object file. One use
15943 is to help link with legacy assembly code.
15945 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
15946 generate code that is not binary compatible with code generated without that
15947 switch. Use it to conform to a non-default application binary interface.
15948 Not all targets provide complete support for this switch.
15950 @item -ftls-model=@var{model}
15951 @opindex ftls-model
15952 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
15953 The @var{model} argument should be one of @code{global-dynamic},
15954 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
15956 The default without @option{-fpic} is @code{initial-exec}; with
15957 @option{-fpic} the default is @code{global-dynamic}.
15959 @item -fvisibility=@var{default|internal|hidden|protected}
15960 @opindex fvisibility
15961 Set the default ELF image symbol visibility to the specified option---all
15962 symbols will be marked with this unless overridden within the code.
15963 Using this feature can very substantially improve linking and
15964 load times of shared object libraries, produce more optimized
15965 code, provide near-perfect API export and prevent symbol clashes.
15966 It is @strong{strongly} recommended that you use this in any shared objects
15969 Despite the nomenclature, @code{default} always means public ie;
15970 available to be linked against from outside the shared object.
15971 @code{protected} and @code{internal} are pretty useless in real-world
15972 usage so the only other commonly used option will be @code{hidden}.
15973 The default if @option{-fvisibility} isn't specified is
15974 @code{default}, i.e., make every
15975 symbol public---this causes the same behavior as previous versions of
15978 A good explanation of the benefits offered by ensuring ELF
15979 symbols have the correct visibility is given by ``How To Write
15980 Shared Libraries'' by Ulrich Drepper (which can be found at
15981 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
15982 solution made possible by this option to marking things hidden when
15983 the default is public is to make the default hidden and mark things
15984 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
15985 and @code{__attribute__ ((visibility("default")))} instead of
15986 @code{__declspec(dllexport)} you get almost identical semantics with
15987 identical syntax. This is a great boon to those working with
15988 cross-platform projects.
15990 For those adding visibility support to existing code, you may find
15991 @samp{#pragma GCC visibility} of use. This works by you enclosing
15992 the declarations you wish to set visibility for with (for example)
15993 @samp{#pragma GCC visibility push(hidden)} and
15994 @samp{#pragma GCC visibility pop}.
15995 Bear in mind that symbol visibility should be viewed @strong{as
15996 part of the API interface contract} and thus all new code should
15997 always specify visibility when it is not the default ie; declarations
15998 only for use within the local DSO should @strong{always} be marked explicitly
15999 as hidden as so to avoid PLT indirection overheads---making this
16000 abundantly clear also aids readability and self-documentation of the code.
16001 Note that due to ISO C++ specification requirements, operator new and
16002 operator delete must always be of default visibility.
16004 Be aware that headers from outside your project, in particular system
16005 headers and headers from any other library you use, may not be
16006 expecting to be compiled with visibility other than the default. You
16007 may need to explicitly say @samp{#pragma GCC visibility push(default)}
16008 before including any such headers.
16010 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
16011 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
16012 no modifications. However, this means that calls to @samp{extern}
16013 functions with no explicit visibility will use the PLT, so it is more
16014 effective to use @samp{__attribute ((visibility))} and/or
16015 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
16016 declarations should be treated as hidden.
16018 Note that @samp{-fvisibility} does affect C++ vague linkage
16019 entities. This means that, for instance, an exception class that will
16020 be thrown between DSOs must be explicitly marked with default
16021 visibility so that the @samp{type_info} nodes will be unified between
16024 An overview of these techniques, their benefits and how to use them
16025 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
16031 @node Environment Variables
16032 @section Environment Variables Affecting GCC
16033 @cindex environment variables
16035 @c man begin ENVIRONMENT
16036 This section describes several environment variables that affect how GCC
16037 operates. Some of them work by specifying directories or prefixes to use
16038 when searching for various kinds of files. Some are used to specify other
16039 aspects of the compilation environment.
16041 Note that you can also specify places to search using options such as
16042 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
16043 take precedence over places specified using environment variables, which
16044 in turn take precedence over those specified by the configuration of GCC@.
16045 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
16046 GNU Compiler Collection (GCC) Internals}.
16051 @c @itemx LC_COLLATE
16053 @c @itemx LC_MONETARY
16054 @c @itemx LC_NUMERIC
16059 @c @findex LC_COLLATE
16060 @findex LC_MESSAGES
16061 @c @findex LC_MONETARY
16062 @c @findex LC_NUMERIC
16066 These environment variables control the way that GCC uses
16067 localization information that allow GCC to work with different
16068 national conventions. GCC inspects the locale categories
16069 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
16070 so. These locale categories can be set to any value supported by your
16071 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
16072 Kingdom encoded in UTF-8.
16074 The @env{LC_CTYPE} environment variable specifies character
16075 classification. GCC uses it to determine the character boundaries in
16076 a string; this is needed for some multibyte encodings that contain quote
16077 and escape characters that would otherwise be interpreted as a string
16080 The @env{LC_MESSAGES} environment variable specifies the language to
16081 use in diagnostic messages.
16083 If the @env{LC_ALL} environment variable is set, it overrides the value
16084 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
16085 and @env{LC_MESSAGES} default to the value of the @env{LANG}
16086 environment variable. If none of these variables are set, GCC
16087 defaults to traditional C English behavior.
16091 If @env{TMPDIR} is set, it specifies the directory to use for temporary
16092 files. GCC uses temporary files to hold the output of one stage of
16093 compilation which is to be used as input to the next stage: for example,
16094 the output of the preprocessor, which is the input to the compiler
16097 @item GCC_EXEC_PREFIX
16098 @findex GCC_EXEC_PREFIX
16099 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
16100 names of the subprograms executed by the compiler. No slash is added
16101 when this prefix is combined with the name of a subprogram, but you can
16102 specify a prefix that ends with a slash if you wish.
16104 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
16105 an appropriate prefix to use based on the pathname it was invoked with.
16107 If GCC cannot find the subprogram using the specified prefix, it
16108 tries looking in the usual places for the subprogram.
16110 The default value of @env{GCC_EXEC_PREFIX} is
16111 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
16112 the installed compiler. In many cases @var{prefix} is the value
16113 of @code{prefix} when you ran the @file{configure} script.
16115 Other prefixes specified with @option{-B} take precedence over this prefix.
16117 This prefix is also used for finding files such as @file{crt0.o} that are
16120 In addition, the prefix is used in an unusual way in finding the
16121 directories to search for header files. For each of the standard
16122 directories whose name normally begins with @samp{/usr/local/lib/gcc}
16123 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
16124 replacing that beginning with the specified prefix to produce an
16125 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
16126 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
16127 These alternate directories are searched first; the standard directories
16128 come next. If a standard directory begins with the configured
16129 @var{prefix} then the value of @var{prefix} is replaced by
16130 @env{GCC_EXEC_PREFIX} when looking for header files.
16132 @item COMPILER_PATH
16133 @findex COMPILER_PATH
16134 The value of @env{COMPILER_PATH} is a colon-separated list of
16135 directories, much like @env{PATH}. GCC tries the directories thus
16136 specified when searching for subprograms, if it can't find the
16137 subprograms using @env{GCC_EXEC_PREFIX}.
16140 @findex LIBRARY_PATH
16141 The value of @env{LIBRARY_PATH} is a colon-separated list of
16142 directories, much like @env{PATH}. When configured as a native compiler,
16143 GCC tries the directories thus specified when searching for special
16144 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
16145 using GCC also uses these directories when searching for ordinary
16146 libraries for the @option{-l} option (but directories specified with
16147 @option{-L} come first).
16151 @cindex locale definition
16152 This variable is used to pass locale information to the compiler. One way in
16153 which this information is used is to determine the character set to be used
16154 when character literals, string literals and comments are parsed in C and C++.
16155 When the compiler is configured to allow multibyte characters,
16156 the following values for @env{LANG} are recognized:
16160 Recognize JIS characters.
16162 Recognize SJIS characters.
16164 Recognize EUCJP characters.
16167 If @env{LANG} is not defined, or if it has some other value, then the
16168 compiler will use mblen and mbtowc as defined by the default locale to
16169 recognize and translate multibyte characters.
16173 Some additional environments variables affect the behavior of the
16176 @include cppenv.texi
16180 @node Precompiled Headers
16181 @section Using Precompiled Headers
16182 @cindex precompiled headers
16183 @cindex speed of compilation
16185 Often large projects have many header files that are included in every
16186 source file. The time the compiler takes to process these header files
16187 over and over again can account for nearly all of the time required to
16188 build the project. To make builds faster, GCC allows users to
16189 `precompile' a header file; then, if builds can use the precompiled
16190 header file they will be much faster.
16192 To create a precompiled header file, simply compile it as you would any
16193 other file, if necessary using the @option{-x} option to make the driver
16194 treat it as a C or C++ header file. You will probably want to use a
16195 tool like @command{make} to keep the precompiled header up-to-date when
16196 the headers it contains change.
16198 A precompiled header file will be searched for when @code{#include} is
16199 seen in the compilation. As it searches for the included file
16200 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
16201 compiler looks for a precompiled header in each directory just before it
16202 looks for the include file in that directory. The name searched for is
16203 the name specified in the @code{#include} with @samp{.gch} appended. If
16204 the precompiled header file can't be used, it is ignored.
16206 For instance, if you have @code{#include "all.h"}, and you have
16207 @file{all.h.gch} in the same directory as @file{all.h}, then the
16208 precompiled header file will be used if possible, and the original
16209 header will be used otherwise.
16211 Alternatively, you might decide to put the precompiled header file in a
16212 directory and use @option{-I} to ensure that directory is searched
16213 before (or instead of) the directory containing the original header.
16214 Then, if you want to check that the precompiled header file is always
16215 used, you can put a file of the same name as the original header in this
16216 directory containing an @code{#error} command.
16218 This also works with @option{-include}. So yet another way to use
16219 precompiled headers, good for projects not designed with precompiled
16220 header files in mind, is to simply take most of the header files used by
16221 a project, include them from another header file, precompile that header
16222 file, and @option{-include} the precompiled header. If the header files
16223 have guards against multiple inclusion, they will be skipped because
16224 they've already been included (in the precompiled header).
16226 If you need to precompile the same header file for different
16227 languages, targets, or compiler options, you can instead make a
16228 @emph{directory} named like @file{all.h.gch}, and put each precompiled
16229 header in the directory, perhaps using @option{-o}. It doesn't matter
16230 what you call the files in the directory, every precompiled header in
16231 the directory will be considered. The first precompiled header
16232 encountered in the directory that is valid for this compilation will
16233 be used; they're searched in no particular order.
16235 There are many other possibilities, limited only by your imagination,
16236 good sense, and the constraints of your build system.
16238 A precompiled header file can be used only when these conditions apply:
16242 Only one precompiled header can be used in a particular compilation.
16245 A precompiled header can't be used once the first C token is seen. You
16246 can have preprocessor directives before a precompiled header; you can
16247 even include a precompiled header from inside another header, so long as
16248 there are no C tokens before the @code{#include}.
16251 The precompiled header file must be produced for the same language as
16252 the current compilation. You can't use a C precompiled header for a C++
16256 The precompiled header file must have been produced by the same compiler
16257 binary as the current compilation is using.
16260 Any macros defined before the precompiled header is included must
16261 either be defined in the same way as when the precompiled header was
16262 generated, or must not affect the precompiled header, which usually
16263 means that they don't appear in the precompiled header at all.
16265 The @option{-D} option is one way to define a macro before a
16266 precompiled header is included; using a @code{#define} can also do it.
16267 There are also some options that define macros implicitly, like
16268 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
16271 @item If debugging information is output when using the precompiled
16272 header, using @option{-g} or similar, the same kind of debugging information
16273 must have been output when building the precompiled header. However,
16274 a precompiled header built using @option{-g} can be used in a compilation
16275 when no debugging information is being output.
16277 @item The same @option{-m} options must generally be used when building
16278 and using the precompiled header. @xref{Submodel Options},
16279 for any cases where this rule is relaxed.
16281 @item Each of the following options must be the same when building and using
16282 the precompiled header:
16284 @gccoptlist{-fexceptions}
16287 Some other command-line options starting with @option{-f},
16288 @option{-p}, or @option{-O} must be defined in the same way as when
16289 the precompiled header was generated. At present, it's not clear
16290 which options are safe to change and which are not; the safest choice
16291 is to use exactly the same options when generating and using the
16292 precompiled header. The following are known to be safe:
16294 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
16295 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
16296 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
16301 For all of these except the last, the compiler will automatically
16302 ignore the precompiled header if the conditions aren't met. If you
16303 find an option combination that doesn't work and doesn't cause the
16304 precompiled header to be ignored, please consider filing a bug report,
16307 If you do use differing options when generating and using the
16308 precompiled header, the actual behavior will be a mixture of the
16309 behavior for the options. For instance, if you use @option{-g} to
16310 generate the precompiled header but not when using it, you may or may
16311 not get debugging information for routines in the precompiled header.
16313 @node Running Protoize
16314 @section Running Protoize
16316 The program @code{protoize} is an optional part of GCC@. You can use
16317 it to add prototypes to a program, thus converting the program to ISO
16318 C in one respect. The companion program @code{unprotoize} does the
16319 reverse: it removes argument types from any prototypes that are found.
16321 When you run these programs, you must specify a set of source files as
16322 command line arguments. The conversion programs start out by compiling
16323 these files to see what functions they define. The information gathered
16324 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
16326 After scanning comes actual conversion. The specified files are all
16327 eligible to be converted; any files they include (whether sources or
16328 just headers) are eligible as well.
16330 But not all the eligible files are converted. By default,
16331 @code{protoize} and @code{unprotoize} convert only source and header
16332 files in the current directory. You can specify additional directories
16333 whose files should be converted with the @option{-d @var{directory}}
16334 option. You can also specify particular files to exclude with the
16335 @option{-x @var{file}} option. A file is converted if it is eligible, its
16336 directory name matches one of the specified directory names, and its
16337 name within the directory has not been excluded.
16339 Basic conversion with @code{protoize} consists of rewriting most
16340 function definitions and function declarations to specify the types of
16341 the arguments. The only ones not rewritten are those for varargs
16344 @code{protoize} optionally inserts prototype declarations at the
16345 beginning of the source file, to make them available for any calls that
16346 precede the function's definition. Or it can insert prototype
16347 declarations with block scope in the blocks where undeclared functions
16350 Basic conversion with @code{unprotoize} consists of rewriting most
16351 function declarations to remove any argument types, and rewriting
16352 function definitions to the old-style pre-ISO form.
16354 Both conversion programs print a warning for any function declaration or
16355 definition that they can't convert. You can suppress these warnings
16358 The output from @code{protoize} or @code{unprotoize} replaces the
16359 original source file. The original file is renamed to a name ending
16360 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
16361 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
16362 for DOS) file already exists, then the source file is simply discarded.
16364 @code{protoize} and @code{unprotoize} both depend on GCC itself to
16365 scan the program and collect information about the functions it uses.
16366 So neither of these programs will work until GCC is installed.
16368 Here is a table of the options you can use with @code{protoize} and
16369 @code{unprotoize}. Each option works with both programs unless
16373 @item -B @var{directory}
16374 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
16375 usual directory (normally @file{/usr/local/lib}). This file contains
16376 prototype information about standard system functions. This option
16377 applies only to @code{protoize}.
16379 @item -c @var{compilation-options}
16380 Use @var{compilation-options} as the options when running @command{gcc} to
16381 produce the @samp{.X} files. The special option @option{-aux-info} is
16382 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
16384 Note that the compilation options must be given as a single argument to
16385 @code{protoize} or @code{unprotoize}. If you want to specify several
16386 @command{gcc} options, you must quote the entire set of compilation options
16387 to make them a single word in the shell.
16389 There are certain @command{gcc} arguments that you cannot use, because they
16390 would produce the wrong kind of output. These include @option{-g},
16391 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
16392 the @var{compilation-options}, they are ignored.
16395 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
16396 systems) instead of @samp{.c}. This is convenient if you are converting
16397 a C program to C++. This option applies only to @code{protoize}.
16400 Add explicit global declarations. This means inserting explicit
16401 declarations at the beginning of each source file for each function
16402 that is called in the file and was not declared. These declarations
16403 precede the first function definition that contains a call to an
16404 undeclared function. This option applies only to @code{protoize}.
16406 @item -i @var{string}
16407 Indent old-style parameter declarations with the string @var{string}.
16408 This option applies only to @code{protoize}.
16410 @code{unprotoize} converts prototyped function definitions to old-style
16411 function definitions, where the arguments are declared between the
16412 argument list and the initial @samp{@{}. By default, @code{unprotoize}
16413 uses five spaces as the indentation. If you want to indent with just
16414 one space instead, use @option{-i " "}.
16417 Keep the @samp{.X} files. Normally, they are deleted after conversion
16421 Add explicit local declarations. @code{protoize} with @option{-l} inserts
16422 a prototype declaration for each function in each block which calls the
16423 function without any declaration. This option applies only to
16427 Make no real changes. This mode just prints information about the conversions
16428 that would have been done without @option{-n}.
16431 Make no @samp{.save} files. The original files are simply deleted.
16432 Use this option with caution.
16434 @item -p @var{program}
16435 Use the program @var{program} as the compiler. Normally, the name
16436 @file{gcc} is used.
16439 Work quietly. Most warnings are suppressed.
16442 Print the version number, just like @option{-v} for @command{gcc}.
16445 If you need special compiler options to compile one of your program's
16446 source files, then you should generate that file's @samp{.X} file
16447 specially, by running @command{gcc} on that source file with the
16448 appropriate options and the option @option{-aux-info}. Then run
16449 @code{protoize} on the entire set of files. @code{protoize} will use
16450 the existing @samp{.X} file because it is newer than the source file.
16454 gcc -Dfoo=bar file1.c -aux-info file1.X
16459 You need to include the special files along with the rest in the
16460 @code{protoize} command, even though their @samp{.X} files already
16461 exist, because otherwise they won't get converted.
16463 @xref{Protoize Caveats}, for more information on how to use
16464 @code{protoize} successfully.