1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
15 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
151 * Running Protoize:: Automatically adding or removing function prototypes.
157 @section Option Summary
159 Here is a summary of all the options, grouped by type. Explanations are
160 in the following sections.
163 @item Overall Options
164 @xref{Overall Options,,Options Controlling the Kind of Output}.
165 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{]} --target-help @gol
167 --version @@@var{file}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
193 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
194 -fno-default-inline -fvisibility-inlines-hidden @gol
195 -fvisibility-ms-compat @gol
196 -Wabi -Wctor-dtor-privacy @gol
197 -Wnon-virtual-dtor -Wreorder @gol
198 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
199 -Wno-non-template-friend -Wold-style-cast @gol
200 -Woverloaded-virtual -Wno-pmf-conversions @gol
203 @item Objective-C and Objective-C++ Language Options
204 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
205 Objective-C and Objective-C++ Dialects}.
206 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
207 -fgnu-runtime -fnext-runtime @gol
208 -fno-nil-receivers @gol
209 -fobjc-call-cxx-cdtors @gol
210 -fobjc-direct-dispatch @gol
211 -fobjc-exceptions @gol
213 -freplace-objc-classes @gol
216 -Wassign-intercept @gol
217 -Wno-protocol -Wselector @gol
218 -Wstrict-selector-match @gol
219 -Wundeclared-selector}
221 @item Language Independent Options
222 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
223 @gccoptlist{-fmessage-length=@var{n} @gol
224 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
225 -fdiagnostics-show-option}
227 @item Warning Options
228 @xref{Warning Options,,Options to Request or Suppress Warnings}.
229 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
230 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
231 -Wno-attributes -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
232 -Wchar-subscripts -Wclobbered -Wcomment @gol
233 -Wconversion -Wcoverage-mismatch -Wno-deprecated-declarations @gol
234 -Wdisabled-optimization -Wno-div-by-zero @gol
235 -Wempty-body -Wno-endif-labels @gol
236 -Werror -Werror=* @gol
237 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
238 -Wno-format-extra-args -Wformat-nonliteral @gol
239 -Wformat-security -Wformat-y2k -Wignored-qualifiers @gol
240 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
241 -Wimport -Wno-import -Winit-self -Winline @gol
242 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
243 -Winvalid-pch -Wlarger-than-@var{len} -Wunsafe-loop-optimizations @gol
244 -Wlogical-op -Wlong-long @gol
245 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
246 -Wmissing-format-attribute -Wmissing-include-dirs @gol
247 -Wmissing-noreturn @gol
248 -Wno-multichar -Wnonnull -Wno-overflow @gol
249 -Woverlength-strings -Wpacked -Wpadded @gol
250 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
251 -Wredundant-decls @gol
252 -Wreturn-type -Wsequence-point -Wshadow @gol
253 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
254 -Wstrict-aliasing -Wstrict-aliasing=n @gol
255 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
256 -Wswitch -Wswitch-default -Wswitch-enum @gol
257 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
258 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
259 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
260 -Wunused-value -Wunused-variable @gol
261 -Wvariadic-macros -Wvla @gol
262 -Wvolatile-register-var -Wwrite-strings}
264 @item C and Objective-C-only Warning Options
265 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
266 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
267 -Wold-style-declaration -Wold-style-definition @gol
268 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
269 -Wdeclaration-after-statement -Wpointer-sign}
271 @item Debugging Options
272 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
273 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
274 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
275 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
276 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
277 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
279 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
283 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
284 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
289 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-nrv -fdump-tree-vect @gol
292 -fdump-tree-sink @gol
293 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-salias @gol
295 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
297 -ftree-vectorizer-verbose=@var{n} @gol
298 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
299 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
300 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
301 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
302 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
303 -ftest-coverage -ftime-report -fvar-tracking @gol
304 -g -g@var{level} -gcoff -gdwarf-2 @gol
305 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
306 -fno-merge-debug-strings -fdebug-prefix-map=@var{old}=@var{new} @gol
307 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
308 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
309 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
310 -print-multi-directory -print-multi-lib @gol
311 -print-prog-name=@var{program} -print-search-dirs -Q @gol
312 -print-sysroot-headers-suffix @gol
315 @item Optimization Options
316 @xref{Optimize Options,,Options that Control Optimization}.
318 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
319 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
320 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
321 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
322 -fcheck-data-deps -fcprop-registers -fcrossjumping -fcse-follow-jumps @gol
323 -fcse-skip-blocks -fcx-limited-range -fdata-sections -fdce -fdce @gol
324 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
325 -fearly-inlining -fexpensive-optimizations -ffast-math @gol
326 -ffinite-math-only -ffloat-store -fforward-propagate @gol
327 -ffunction-sections -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
328 -fgcse-sm -fif-conversion -fif-conversion2 -finline-functions @gol
329 -finline-functions-called-once -finline-limit=@var{n} @gol
330 -finline-small-functions -fipa-cp -fipa-marix-reorg -fipa-pta @gol
331 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
332 -fipa-type-escape -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
333 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
334 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
335 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
336 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
337 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
338 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
339 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
340 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
341 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
342 -fprofile-generate -fprofile-use -fprofile-values -freciprocal-math @gol
343 -fregmove -frename-registers -freorder-blocks @gol
344 -freorder-blocks-and-partition -freorder-functions @gol
345 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
346 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
347 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
348 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
349 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
350 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
351 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
352 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer -ftree-ccp @gol
353 -ftree-ch -ftree-copy-prop -ftree-copyrename -ftree-dce @gol
354 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
355 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
356 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc -ftree-salias @gol
357 -ftree-sink -ftree-sra -ftree-store-ccp -ftree-ter @gol
358 -ftree-vect-loop-version -ftree-vectorize -ftree-vrp -funit-at-a-time @gol
359 -funroll-all-loops -funroll-loops -funsafe-loop-optimizations @gol
360 -funsafe-math-optimizations -funswitch-loops @gol
361 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
363 --param @var{name}=@var{value}
364 -O -O0 -O1 -O2 -O3 -Os}
366 @item Preprocessor Options
367 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
368 @gccoptlist{-A@var{question}=@var{answer} @gol
369 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
370 -C -dD -dI -dM -dN @gol
371 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
372 -idirafter @var{dir} @gol
373 -include @var{file} -imacros @var{file} @gol
374 -iprefix @var{file} -iwithprefix @var{dir} @gol
375 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
376 -imultilib @var{dir} -isysroot @var{dir} @gol
377 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
378 -P -fworking-directory -remap @gol
379 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
380 -Xpreprocessor @var{option}}
382 @item Assembler Option
383 @xref{Assembler Options,,Passing Options to the Assembler}.
384 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
387 @xref{Link Options,,Options for Linking}.
388 @gccoptlist{@var{object-file-name} -l@var{library} @gol
389 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
390 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
391 -Wl,@var{option} -Xlinker @var{option} @gol
394 @item Directory Options
395 @xref{Directory Options,,Options for Directory Search}.
396 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
397 -specs=@var{file} -I- --sysroot=@var{dir}}
400 @c I wrote this xref this way to avoid overfull hbox. -- rms
401 @xref{Target Options}.
402 @gccoptlist{-V @var{version} -b @var{machine}}
404 @item Machine Dependent Options
405 @xref{Submodel Options,,Hardware Models and Configurations}.
406 @c This list is ordered alphanumerically by subsection name.
407 @c Try and put the significant identifier (CPU or system) first,
408 @c so users have a clue at guessing where the ones they want will be.
411 @gccoptlist{-EB -EL @gol
412 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
413 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
416 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
417 -mabi=@var{name} @gol
418 -mapcs-stack-check -mno-apcs-stack-check @gol
419 -mapcs-float -mno-apcs-float @gol
420 -mapcs-reentrant -mno-apcs-reentrant @gol
421 -msched-prolog -mno-sched-prolog @gol
422 -mlittle-endian -mbig-endian -mwords-little-endian @gol
423 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
424 -mthumb-interwork -mno-thumb-interwork @gol
425 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
426 -mstructure-size-boundary=@var{n} @gol
427 -mabort-on-noreturn @gol
428 -mlong-calls -mno-long-calls @gol
429 -msingle-pic-base -mno-single-pic-base @gol
430 -mpic-register=@var{reg} @gol
431 -mnop-fun-dllimport @gol
432 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
433 -mpoke-function-name @gol
435 -mtpcs-frame -mtpcs-leaf-frame @gol
436 -mcaller-super-interworking -mcallee-super-interworking @gol
440 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
441 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
443 @emph{Blackfin Options}
444 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
445 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
446 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
447 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
448 -mno-id-shared-library -mshared-library-id=@var{n} @gol
449 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
450 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
451 -mfast-fp -minline-plt}
454 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
455 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
456 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
457 -mstack-align -mdata-align -mconst-align @gol
458 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
459 -melf -maout -melinux -mlinux -sim -sim2 @gol
460 -mmul-bug-workaround -mno-mul-bug-workaround}
463 @gccoptlist{-mmac -mpush-args}
465 @emph{Darwin Options}
466 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
467 -arch_only -bind_at_load -bundle -bundle_loader @gol
468 -client_name -compatibility_version -current_version @gol
470 -dependency-file -dylib_file -dylinker_install_name @gol
471 -dynamic -dynamiclib -exported_symbols_list @gol
472 -filelist -flat_namespace -force_cpusubtype_ALL @gol
473 -force_flat_namespace -headerpad_max_install_names @gol
475 -image_base -init -install_name -keep_private_externs @gol
476 -multi_module -multiply_defined -multiply_defined_unused @gol
477 -noall_load -no_dead_strip_inits_and_terms @gol
478 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
479 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
480 -private_bundle -read_only_relocs -sectalign @gol
481 -sectobjectsymbols -whyload -seg1addr @gol
482 -sectcreate -sectobjectsymbols -sectorder @gol
483 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
484 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
485 -segprot -segs_read_only_addr -segs_read_write_addr @gol
486 -single_module -static -sub_library -sub_umbrella @gol
487 -twolevel_namespace -umbrella -undefined @gol
488 -unexported_symbols_list -weak_reference_mismatches @gol
489 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
490 -mkernel -mone-byte-bool}
492 @emph{DEC Alpha Options}
493 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
494 -mieee -mieee-with-inexact -mieee-conformant @gol
495 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
496 -mtrap-precision=@var{mode} -mbuild-constants @gol
497 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
498 -mbwx -mmax -mfix -mcix @gol
499 -mfloat-vax -mfloat-ieee @gol
500 -mexplicit-relocs -msmall-data -mlarge-data @gol
501 -msmall-text -mlarge-text @gol
502 -mmemory-latency=@var{time}}
504 @emph{DEC Alpha/VMS Options}
505 @gccoptlist{-mvms-return-codes}
508 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
509 -mhard-float -msoft-float @gol
510 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
511 -mdouble -mno-double @gol
512 -mmedia -mno-media -mmuladd -mno-muladd @gol
513 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
514 -mlinked-fp -mlong-calls -malign-labels @gol
515 -mlibrary-pic -macc-4 -macc-8 @gol
516 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
517 -moptimize-membar -mno-optimize-membar @gol
518 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
519 -mvliw-branch -mno-vliw-branch @gol
520 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
521 -mno-nested-cond-exec -mtomcat-stats @gol
525 @emph{GNU/Linux Options}
526 @gccoptlist{-muclibc}
528 @emph{H8/300 Options}
529 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
532 @gccoptlist{-march=@var{architecture-type} @gol
533 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
534 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
535 -mfixed-range=@var{register-range} @gol
536 -mjump-in-delay -mlinker-opt -mlong-calls @gol
537 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
538 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
539 -mno-jump-in-delay -mno-long-load-store @gol
540 -mno-portable-runtime -mno-soft-float @gol
541 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
542 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
543 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
544 -munix=@var{unix-std} -nolibdld -static -threads}
546 @emph{i386 and x86-64 Options}
547 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
548 -mfpmath=@var{unit} @gol
549 -masm=@var{dialect} -mno-fancy-math-387 @gol
550 -mno-fp-ret-in-387 -msoft-float @gol
551 -mno-wide-multiply -mrtd -malign-double @gol
552 -mpreferred-stack-boundary=@var{num} -mcx16 -msahf -mrecip @gol
553 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 @gol
554 -msse4a -m3dnow -mpopcnt -mabm -msse5 @gol
555 -mthreads -mno-align-stringops -minline-all-stringops @gol
556 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
557 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
558 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
559 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
560 -mcmodel=@var{code-model} @gol
561 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
562 -mfused-madd -mno-fused-madd}
565 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
566 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
567 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
568 -minline-float-divide-max-throughput @gol
569 -minline-int-divide-min-latency @gol
570 -minline-int-divide-max-throughput @gol
571 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
572 -mno-dwarf2-asm -mearly-stop-bits @gol
573 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
574 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
575 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
576 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
577 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
578 -mno-sched-prefer-non-data-spec-insns @gol
579 -mno-sched-prefer-non-control-spec-insns @gol
580 -mno-sched-count-spec-in-critical-path}
582 @emph{M32R/D Options}
583 @gccoptlist{-m32r2 -m32rx -m32r @gol
585 -malign-loops -mno-align-loops @gol
586 -missue-rate=@var{number} @gol
587 -mbranch-cost=@var{number} @gol
588 -mmodel=@var{code-size-model-type} @gol
589 -msdata=@var{sdata-type} @gol
590 -mno-flush-func -mflush-func=@var{name} @gol
591 -mno-flush-trap -mflush-trap=@var{number} @gol
595 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
597 @emph{M680x0 Options}
598 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
599 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
600 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
601 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
602 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
603 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
604 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
605 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
607 @emph{M68hc1x Options}
608 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
609 -mauto-incdec -minmax -mlong-calls -mshort @gol
610 -msoft-reg-count=@var{count}}
613 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
614 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
615 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
616 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
617 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
620 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
621 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
622 -mips16 -mno-mips16 -mflip-mips16 @gol
623 -minterlink-mips16 -mno-interlink-mips16 @gol
624 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
625 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
626 -mfp32 -mfp64 -mhard-float -msoft-float @gol
627 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
628 -msmartmips -mno-smartmips @gol
629 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
630 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
631 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
632 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
633 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
634 -membedded-data -mno-embedded-data @gol
635 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
636 -mcode-readable=@var{setting} @gol
637 -msplit-addresses -mno-split-addresses @gol
638 -mexplicit-relocs -mno-explicit-relocs @gol
639 -mcheck-zero-division -mno-check-zero-division @gol
640 -mdivide-traps -mdivide-breaks @gol
641 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
642 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
643 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
644 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
645 -mfix-sb1 -mno-fix-sb1 @gol
646 -mflush-func=@var{func} -mno-flush-func @gol
647 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
648 -mfp-exceptions -mno-fp-exceptions @gol
649 -mvr4130-align -mno-vr4130-align}
652 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
653 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
654 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
655 -mno-base-addresses -msingle-exit -mno-single-exit}
657 @emph{MN10300 Options}
658 @gccoptlist{-mmult-bug -mno-mult-bug @gol
659 -mam33 -mno-am33 @gol
660 -mam33-2 -mno-am33-2 @gol
661 -mreturn-pointer-on-d0 @gol
665 @gccoptlist{-mno-crt0 -mbacc -msim @gol
666 -march=@var{cpu-type} }
668 @emph{PDP-11 Options}
669 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
670 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
671 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
672 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
673 -mbranch-expensive -mbranch-cheap @gol
674 -msplit -mno-split -munix-asm -mdec-asm}
676 @emph{PowerPC Options}
677 See RS/6000 and PowerPC Options.
679 @emph{RS/6000 and PowerPC Options}
680 @gccoptlist{-mcpu=@var{cpu-type} @gol
681 -mtune=@var{cpu-type} @gol
682 -mpower -mno-power -mpower2 -mno-power2 @gol
683 -mpowerpc -mpowerpc64 -mno-powerpc @gol
684 -maltivec -mno-altivec @gol
685 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
686 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
687 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
688 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
689 -mnew-mnemonics -mold-mnemonics @gol
690 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
691 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
692 -malign-power -malign-natural @gol
693 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
694 -mstring -mno-string -mupdate -mno-update @gol
695 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
696 -mstrict-align -mno-strict-align -mrelocatable @gol
697 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
698 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
699 -mdynamic-no-pic -maltivec -mswdiv @gol
700 -mprioritize-restricted-insns=@var{priority} @gol
701 -msched-costly-dep=@var{dependence_type} @gol
702 -minsert-sched-nops=@var{scheme} @gol
703 -mcall-sysv -mcall-netbsd @gol
704 -maix-struct-return -msvr4-struct-return @gol
705 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
706 -misel -mno-isel @gol
707 -misel=yes -misel=no @gol
709 -mspe=yes -mspe=no @gol
711 -mvrsave -mno-vrsave @gol
712 -mmulhw -mno-mulhw @gol
713 -mdlmzb -mno-dlmzb @gol
714 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
715 -mprototype -mno-prototype @gol
716 -msim -mmvme -mads -myellowknife -memb -msdata @gol
717 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
719 @emph{S/390 and zSeries Options}
720 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
721 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
722 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
723 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
724 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
725 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
726 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
729 @gccoptlist{-meb -mel @gol
733 -mscore5 -mscore5u -mscore7 -mscore7d}
736 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
737 -m4-nofpu -m4-single-only -m4-single -m4 @gol
738 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
739 -m5-64media -m5-64media-nofpu @gol
740 -m5-32media -m5-32media-nofpu @gol
741 -m5-compact -m5-compact-nofpu @gol
742 -mb -ml -mdalign -mrelax @gol
743 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
744 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
745 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
746 -mdivsi3_libfunc=@var{name} @gol
747 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
751 @gccoptlist{-mcpu=@var{cpu-type} @gol
752 -mtune=@var{cpu-type} @gol
753 -mcmodel=@var{code-model} @gol
754 -m32 -m64 -mapp-regs -mno-app-regs @gol
755 -mfaster-structs -mno-faster-structs @gol
756 -mfpu -mno-fpu -mhard-float -msoft-float @gol
757 -mhard-quad-float -msoft-quad-float @gol
758 -mimpure-text -mno-impure-text -mlittle-endian @gol
759 -mstack-bias -mno-stack-bias @gol
760 -munaligned-doubles -mno-unaligned-doubles @gol
761 -mv8plus -mno-v8plus -mvis -mno-vis
762 -threads -pthreads -pthread}
765 @gccoptlist{-mwarn-reloc -merror-reloc @gol
766 -msafe-dma -munsafe-dma @gol
768 -msmall-mem -mlarge-mem -mstdmain @gol
769 -mfixed-range=@var{register-range}}
771 @emph{System V Options}
772 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
775 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
776 -mprolog-function -mno-prolog-function -mspace @gol
777 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
778 -mapp-regs -mno-app-regs @gol
779 -mdisable-callt -mno-disable-callt @gol
785 @gccoptlist{-mg -mgnu -munix}
787 @emph{VxWorks Options}
788 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
789 -Xbind-lazy -Xbind-now}
791 @emph{x86-64 Options}
792 See i386 and x86-64 Options.
794 @emph{Xstormy16 Options}
797 @emph{Xtensa Options}
798 @gccoptlist{-mconst16 -mno-const16 @gol
799 -mfused-madd -mno-fused-madd @gol
800 -mtext-section-literals -mno-text-section-literals @gol
801 -mtarget-align -mno-target-align @gol
802 -mlongcalls -mno-longcalls}
804 @emph{zSeries Options}
805 See S/390 and zSeries Options.
807 @item Code Generation Options
808 @xref{Code Gen Options,,Options for Code Generation Conventions}.
809 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
810 -ffixed-@var{reg} -fexceptions @gol
811 -fnon-call-exceptions -funwind-tables @gol
812 -fasynchronous-unwind-tables @gol
813 -finhibit-size-directive -finstrument-functions @gol
814 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
815 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
816 -fno-common -fno-ident @gol
817 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
818 -fno-jump-tables @gol
819 -frecord-gcc-switches @gol
820 -freg-struct-return -fshort-enums @gol
821 -fshort-double -fshort-wchar @gol
822 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
823 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
824 -fno-stack-limit -fargument-alias -fargument-noalias @gol
825 -fargument-noalias-global -fargument-noalias-anything @gol
826 -fleading-underscore -ftls-model=@var{model} @gol
827 -ftrapv -fwrapv -fbounds-check @gol
832 * Overall Options:: Controlling the kind of output:
833 an executable, object files, assembler files,
834 or preprocessed source.
835 * C Dialect Options:: Controlling the variant of C language compiled.
836 * C++ Dialect Options:: Variations on C++.
837 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
839 * Language Independent Options:: Controlling how diagnostics should be
841 * Warning Options:: How picky should the compiler be?
842 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
843 * Optimize Options:: How much optimization?
844 * Preprocessor Options:: Controlling header files and macro definitions.
845 Also, getting dependency information for Make.
846 * Assembler Options:: Passing options to the assembler.
847 * Link Options:: Specifying libraries and so on.
848 * Directory Options:: Where to find header files and libraries.
849 Where to find the compiler executable files.
850 * Spec Files:: How to pass switches to sub-processes.
851 * Target Options:: Running a cross-compiler, or an old version of GCC.
854 @node Overall Options
855 @section Options Controlling the Kind of Output
857 Compilation can involve up to four stages: preprocessing, compilation
858 proper, assembly and linking, always in that order. GCC is capable of
859 preprocessing and compiling several files either into several
860 assembler input files, or into one assembler input file; then each
861 assembler input file produces an object file, and linking combines all
862 the object files (those newly compiled, and those specified as input)
863 into an executable file.
865 @cindex file name suffix
866 For any given input file, the file name suffix determines what kind of
871 C source code which must be preprocessed.
874 C source code which should not be preprocessed.
877 C++ source code which should not be preprocessed.
880 Objective-C source code. Note that you must link with the @file{libobjc}
881 library to make an Objective-C program work.
884 Objective-C source code which should not be preprocessed.
888 Objective-C++ source code. Note that you must link with the @file{libobjc}
889 library to make an Objective-C++ program work. Note that @samp{.M} refers
890 to a literal capital M@.
893 Objective-C++ source code which should not be preprocessed.
896 C, C++, Objective-C or Objective-C++ header file to be turned into a
901 @itemx @var{file}.cxx
902 @itemx @var{file}.cpp
903 @itemx @var{file}.CPP
904 @itemx @var{file}.c++
906 C++ source code which must be preprocessed. Note that in @samp{.cxx},
907 the last two letters must both be literally @samp{x}. Likewise,
908 @samp{.C} refers to a literal capital C@.
912 Objective-C++ source code which must be preprocessed.
915 Objective-C++ source code which should not be preprocessed.
920 @itemx @var{file}.hxx
921 @itemx @var{file}.hpp
922 @itemx @var{file}.HPP
923 @itemx @var{file}.h++
924 @itemx @var{file}.tcc
925 C++ header file to be turned into a precompiled header.
928 @itemx @var{file}.for
929 @itemx @var{file}.FOR
930 Fixed form Fortran source code which should not be preprocessed.
933 @itemx @var{file}.fpp
934 @itemx @var{file}.FPP
935 Fixed form Fortran source code which must be preprocessed (with the traditional
939 @itemx @var{file}.f95
940 Free form Fortran source code which should not be preprocessed.
943 @itemx @var{file}.F95
944 Free form Fortran source code which must be preprocessed (with the
945 traditional preprocessor).
947 @c FIXME: Descriptions of Java file types.
954 Ada source code file which contains a library unit declaration (a
955 declaration of a package, subprogram, or generic, or a generic
956 instantiation), or a library unit renaming declaration (a package,
957 generic, or subprogram renaming declaration). Such files are also
960 @itemx @var{file}.adb
961 Ada source code file containing a library unit body (a subprogram or
962 package body). Such files are also called @dfn{bodies}.
964 @c GCC also knows about some suffixes for languages not yet included:
976 Assembler code which must be preprocessed.
979 An object file to be fed straight into linking.
980 Any file name with no recognized suffix is treated this way.
984 You can specify the input language explicitly with the @option{-x} option:
987 @item -x @var{language}
988 Specify explicitly the @var{language} for the following input files
989 (rather than letting the compiler choose a default based on the file
990 name suffix). This option applies to all following input files until
991 the next @option{-x} option. Possible values for @var{language} are:
993 c c-header c-cpp-output
994 c++ c++-header c++-cpp-output
995 objective-c objective-c-header objective-c-cpp-output
996 objective-c++ objective-c++-header objective-c++-cpp-output
997 assembler assembler-with-cpp
1005 Turn off any specification of a language, so that subsequent files are
1006 handled according to their file name suffixes (as they are if @option{-x}
1007 has not been used at all).
1009 @item -pass-exit-codes
1010 @opindex pass-exit-codes
1011 Normally the @command{gcc} program will exit with the code of 1 if any
1012 phase of the compiler returns a non-success return code. If you specify
1013 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1014 numerically highest error produced by any phase that returned an error
1015 indication. The C, C++, and Fortran frontends return 4, if an internal
1016 compiler error is encountered.
1019 If you only want some of the stages of compilation, you can use
1020 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1021 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1022 @command{gcc} is to stop. Note that some combinations (for example,
1023 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1028 Compile or assemble the source files, but do not link. The linking
1029 stage simply is not done. The ultimate output is in the form of an
1030 object file for each source file.
1032 By default, the object file name for a source file is made by replacing
1033 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1035 Unrecognized input files, not requiring compilation or assembly, are
1040 Stop after the stage of compilation proper; do not assemble. The output
1041 is in the form of an assembler code file for each non-assembler input
1044 By default, the assembler file name for a source file is made by
1045 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1047 Input files that don't require compilation are ignored.
1051 Stop after the preprocessing stage; do not run the compiler proper. The
1052 output is in the form of preprocessed source code, which is sent to the
1055 Input files which don't require preprocessing are ignored.
1057 @cindex output file option
1060 Place output in file @var{file}. This applies regardless to whatever
1061 sort of output is being produced, whether it be an executable file,
1062 an object file, an assembler file or preprocessed C code.
1064 If @option{-o} is not specified, the default is to put an executable
1065 file in @file{a.out}, the object file for
1066 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1067 assembler file in @file{@var{source}.s}, a precompiled header file in
1068 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1073 Print (on standard error output) the commands executed to run the stages
1074 of compilation. Also print the version number of the compiler driver
1075 program and of the preprocessor and the compiler proper.
1079 Like @option{-v} except the commands are not executed and all command
1080 arguments are quoted. This is useful for shell scripts to capture the
1081 driver-generated command lines.
1085 Use pipes rather than temporary files for communication between the
1086 various stages of compilation. This fails to work on some systems where
1087 the assembler is unable to read from a pipe; but the GNU assembler has
1092 If you are compiling multiple source files, this option tells the driver
1093 to pass all the source files to the compiler at once (for those
1094 languages for which the compiler can handle this). This will allow
1095 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1096 language for which this is supported is C@. If you pass source files for
1097 multiple languages to the driver, using this option, the driver will invoke
1098 the compiler(s) that support IMA once each, passing each compiler all the
1099 source files appropriate for it. For those languages that do not support
1100 IMA this option will be ignored, and the compiler will be invoked once for
1101 each source file in that language. If you use this option in conjunction
1102 with @option{-save-temps}, the compiler will generate multiple
1104 (one for each source file), but only one (combined) @file{.o} or
1109 Print (on the standard output) a description of the command line options
1110 understood by @command{gcc}. If the @option{-v} option is also specified
1111 then @option{--help} will also be passed on to the various processes
1112 invoked by @command{gcc}, so that they can display the command line options
1113 they accept. If the @option{-Wextra} option has also been specified
1114 (prior to the @option{--help} option), then command line options which
1115 have no documentation associated with them will also be displayed.
1118 @opindex target-help
1119 Print (on the standard output) a description of target-specific command
1120 line options for each tool. For some targets extra target-specific
1121 information may also be printed.
1123 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1124 Print (on the standard output) a description of the command line
1125 options understood by the compiler that fit into a specific class.
1126 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1127 @samp{params}, or @var{language}:
1130 @item @samp{optimizers}
1131 This will display all of the optimization options supported by the
1134 @item @samp{warnings}
1135 This will display all of the options controlling warning messages
1136 produced by the compiler.
1139 This will display target-specific options. Unlike the
1140 @option{--target-help} option however, target-specific options of the
1141 linker and assembler will not be displayed. This is because those
1142 tools do not currently support the extended @option{--help=} syntax.
1145 This will display the values recognized by the @option{--param}
1148 @item @var{language}
1149 This will display the options supported for @var{language}, where
1150 @var{language} is the name of one of the languages supported in this
1154 This will display the options that are common to all languages.
1157 It is possible to further refine the output of the @option{--help=}
1158 option by adding a comma separated list of qualifiers after the
1159 class. These can be any from the following list:
1162 @item @samp{undocumented}
1163 Display only those options which are undocumented.
1166 Display options which take an argument that appears after an equal
1167 sign in the same continuous piece of text, such as:
1168 @samp{--help=target}.
1170 @item @samp{separate}
1171 Display options which take an argument that appears as a separate word
1172 following the original option, such as: @samp{-o output-file}.
1175 Thus for example to display all the undocumented target-specific
1176 switches supported by the compiler the following can be used:
1179 --help=target,undocumented
1182 The sense of a qualifier can be inverted by prefixing it with the
1183 @var{^} character, so for example to display all binary warning
1184 options (i.e. ones that are either on or off and that do not take an
1185 argument), which have a description the following can be used:
1188 --help=warnings,^joined,^undocumented
1191 A class can also be used as a qualifier, although this usually
1192 restricts the output by so much that there is nothing to display. One
1193 case where it does work however is when one of the classes is
1194 @var{target}. So for example to display all the target-specific
1195 optimization options the following can be used:
1198 --help=target,optimizers
1201 The @option{--help=} option can be repeated on the command line. Each
1202 successive use will display its requested class of options, skipping
1203 those that have already been displayed.
1205 If the @option{-Q} option appears on the command line before the
1206 @option{--help=} option, then the descriptive text displayed by
1207 @option{--help=} is changed. Instead of describing the displayed
1208 options, an indication is given as to whether the option is enabled,
1209 disabled or set to a specific value (assuming that the compiler
1210 knows this at the point where the @option{--help=} option is used).
1212 Here is a truncated example from the ARM port of @command{gcc}:
1215 % gcc -Q -mabi=2 --help=target -c
1216 The following options are target specific:
1218 -mabort-on-noreturn [disabled]
1222 The output is sensitive to the effects of previous command line
1223 options, so for example it is possible to find out which optimizations
1224 are enabled at @option{-O2} by using:
1227 -O2 --help=optimizers
1230 Alternatively you can discover which binary optimizations are enabled
1231 by @option{-O3} by using:
1234 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1235 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1236 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1241 Display the version number and copyrights of the invoked GCC@.
1243 @include @value{srcdir}/../libiberty/at-file.texi
1247 @section Compiling C++ Programs
1249 @cindex suffixes for C++ source
1250 @cindex C++ source file suffixes
1251 C++ source files conventionally use one of the suffixes @samp{.C},
1252 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1253 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1254 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1255 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1256 files with these names and compiles them as C++ programs even if you
1257 call the compiler the same way as for compiling C programs (usually
1258 with the name @command{gcc}).
1262 However, the use of @command{gcc} does not add the C++ library.
1263 @command{g++} is a program that calls GCC and treats @samp{.c},
1264 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1265 files unless @option{-x} is used, and automatically specifies linking
1266 against the C++ library. This program is also useful when
1267 precompiling a C header file with a @samp{.h} extension for use in C++
1268 compilations. On many systems, @command{g++} is also installed with
1269 the name @command{c++}.
1271 @cindex invoking @command{g++}
1272 When you compile C++ programs, you may specify many of the same
1273 command-line options that you use for compiling programs in any
1274 language; or command-line options meaningful for C and related
1275 languages; or options that are meaningful only for C++ programs.
1276 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1277 explanations of options for languages related to C@.
1278 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1279 explanations of options that are meaningful only for C++ programs.
1281 @node C Dialect Options
1282 @section Options Controlling C Dialect
1283 @cindex dialect options
1284 @cindex language dialect options
1285 @cindex options, dialect
1287 The following options control the dialect of C (or languages derived
1288 from C, such as C++, Objective-C and Objective-C++) that the compiler
1292 @cindex ANSI support
1296 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1297 equivalent to @samp{-std=c++98}.
1299 This turns off certain features of GCC that are incompatible with ISO
1300 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1301 such as the @code{asm} and @code{typeof} keywords, and
1302 predefined macros such as @code{unix} and @code{vax} that identify the
1303 type of system you are using. It also enables the undesirable and
1304 rarely used ISO trigraph feature. For the C compiler,
1305 it disables recognition of C++ style @samp{//} comments as well as
1306 the @code{inline} keyword.
1308 The alternate keywords @code{__asm__}, @code{__extension__},
1309 @code{__inline__} and @code{__typeof__} continue to work despite
1310 @option{-ansi}. You would not want to use them in an ISO C program, of
1311 course, but it is useful to put them in header files that might be included
1312 in compilations done with @option{-ansi}. Alternate predefined macros
1313 such as @code{__unix__} and @code{__vax__} are also available, with or
1314 without @option{-ansi}.
1316 The @option{-ansi} option does not cause non-ISO programs to be
1317 rejected gratuitously. For that, @option{-pedantic} is required in
1318 addition to @option{-ansi}. @xref{Warning Options}.
1320 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1321 option is used. Some header files may notice this macro and refrain
1322 from declaring certain functions or defining certain macros that the
1323 ISO standard doesn't call for; this is to avoid interfering with any
1324 programs that might use these names for other things.
1326 Functions that would normally be built in but do not have semantics
1327 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1328 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1329 built-in functions provided by GCC}, for details of the functions
1334 Determine the language standard. @xref{Standards,,Language Standards
1335 Supported by GCC}, for details of these standard versions. This option
1336 is currently only supported when compiling C or C++.
1338 A value for this option must be provided; possible values are
1343 ISO C90 (same as @option{-ansi}).
1345 @item iso9899:199409
1346 ISO C90 as modified in amendment 1.
1352 ISO C99. Note that this standard is not yet fully supported; see
1353 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1354 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1357 Default, ISO C90 plus GNU extensions (including some C99 features).
1361 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1362 this will become the default. The name @samp{gnu9x} is deprecated.
1365 The 1998 ISO C++ standard plus amendments.
1368 The same as @option{-std=c++98} plus GNU extensions. This is the
1369 default for C++ code.
1372 The working draft of the upcoming ISO C++0x standard. This option
1373 enables experimental features that are likely to be included in
1374 C++0x. The working draft is constantly changing, and any feature that is
1375 enabled by this flag may be removed from future versions of GCC if it is
1376 not part of the C++0x standard.
1379 The same as @option{-std=c++0x} plus GNU extensions. As with
1380 @option{-std=c++0x}, this option enables experimental features that may
1381 be removed in future versions of GCC.
1384 Even when this option is not specified, you can still use some of the
1385 features of newer standards in so far as they do not conflict with
1386 previous C standards. For example, you may use @code{__restrict__} even
1387 when @option{-std=c99} is not specified.
1389 The @option{-std} options specifying some version of ISO C have the same
1390 effects as @option{-ansi}, except that features that were not in ISO C90
1391 but are in the specified version (for example, @samp{//} comments and
1392 the @code{inline} keyword in ISO C99) are not disabled.
1394 @item -fgnu89-inline
1395 @opindex fgnu89-inline
1396 The option @option{-fgnu89-inline} tells GCC to use the traditional
1397 GNU semantics for @code{inline} functions when in C99 mode.
1398 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1399 is accepted and ignored by GCC versions 4.1.3 up to but not including
1400 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1401 C99 mode. Using this option is roughly equivalent to adding the
1402 @code{gnu_inline} function attribute to all inline functions
1403 (@pxref{Function Attributes}).
1405 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1406 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1407 specifies the default behavior). This option was first supported in
1408 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1410 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1411 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1412 in effect for @code{inline} functions. @xref{Common Predefined
1413 Macros,,,cpp,The C Preprocessor}.
1415 @item -aux-info @var{filename}
1417 Output to the given filename prototyped declarations for all functions
1418 declared and/or defined in a translation unit, including those in header
1419 files. This option is silently ignored in any language other than C@.
1421 Besides declarations, the file indicates, in comments, the origin of
1422 each declaration (source file and line), whether the declaration was
1423 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1424 @samp{O} for old, respectively, in the first character after the line
1425 number and the colon), and whether it came from a declaration or a
1426 definition (@samp{C} or @samp{F}, respectively, in the following
1427 character). In the case of function definitions, a K&R-style list of
1428 arguments followed by their declarations is also provided, inside
1429 comments, after the declaration.
1433 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1434 keyword, so that code can use these words as identifiers. You can use
1435 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1436 instead. @option{-ansi} implies @option{-fno-asm}.
1438 In C++, this switch only affects the @code{typeof} keyword, since
1439 @code{asm} and @code{inline} are standard keywords. You may want to
1440 use the @option{-fno-gnu-keywords} flag instead, which has the same
1441 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1442 switch only affects the @code{asm} and @code{typeof} keywords, since
1443 @code{inline} is a standard keyword in ISO C99.
1446 @itemx -fno-builtin-@var{function}
1447 @opindex fno-builtin
1448 @cindex built-in functions
1449 Don't recognize built-in functions that do not begin with
1450 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1451 functions provided by GCC}, for details of the functions affected,
1452 including those which are not built-in functions when @option{-ansi} or
1453 @option{-std} options for strict ISO C conformance are used because they
1454 do not have an ISO standard meaning.
1456 GCC normally generates special code to handle certain built-in functions
1457 more efficiently; for instance, calls to @code{alloca} may become single
1458 instructions that adjust the stack directly, and calls to @code{memcpy}
1459 may become inline copy loops. The resulting code is often both smaller
1460 and faster, but since the function calls no longer appear as such, you
1461 cannot set a breakpoint on those calls, nor can you change the behavior
1462 of the functions by linking with a different library. In addition,
1463 when a function is recognized as a built-in function, GCC may use
1464 information about that function to warn about problems with calls to
1465 that function, or to generate more efficient code, even if the
1466 resulting code still contains calls to that function. For example,
1467 warnings are given with @option{-Wformat} for bad calls to
1468 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1469 known not to modify global memory.
1471 With the @option{-fno-builtin-@var{function}} option
1472 only the built-in function @var{function} is
1473 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1474 function is named this is not built-in in this version of GCC, this
1475 option is ignored. There is no corresponding
1476 @option{-fbuiltin-@var{function}} option; if you wish to enable
1477 built-in functions selectively when using @option{-fno-builtin} or
1478 @option{-ffreestanding}, you may define macros such as:
1481 #define abs(n) __builtin_abs ((n))
1482 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1487 @cindex hosted environment
1489 Assert that compilation takes place in a hosted environment. This implies
1490 @option{-fbuiltin}. A hosted environment is one in which the
1491 entire standard library is available, and in which @code{main} has a return
1492 type of @code{int}. Examples are nearly everything except a kernel.
1493 This is equivalent to @option{-fno-freestanding}.
1495 @item -ffreestanding
1496 @opindex ffreestanding
1497 @cindex hosted environment
1499 Assert that compilation takes place in a freestanding environment. This
1500 implies @option{-fno-builtin}. A freestanding environment
1501 is one in which the standard library may not exist, and program startup may
1502 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1503 This is equivalent to @option{-fno-hosted}.
1505 @xref{Standards,,Language Standards Supported by GCC}, for details of
1506 freestanding and hosted environments.
1510 @cindex openmp parallel
1511 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1512 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1513 compiler generates parallel code according to the OpenMP Application
1514 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}. This option
1515 implies @option{-pthread}, and thus is only supported on targets that
1516 have support for @option{-pthread}.
1518 @item -fms-extensions
1519 @opindex fms-extensions
1520 Accept some non-standard constructs used in Microsoft header files.
1522 Some cases of unnamed fields in structures and unions are only
1523 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1524 fields within structs/unions}, for details.
1528 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1529 options for strict ISO C conformance) implies @option{-trigraphs}.
1531 @item -no-integrated-cpp
1532 @opindex no-integrated-cpp
1533 Performs a compilation in two passes: preprocessing and compiling. This
1534 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1535 @option{-B} option. The user supplied compilation step can then add in
1536 an additional preprocessing step after normal preprocessing but before
1537 compiling. The default is to use the integrated cpp (internal cpp)
1539 The semantics of this option will change if "cc1", "cc1plus", and
1540 "cc1obj" are merged.
1542 @cindex traditional C language
1543 @cindex C language, traditional
1545 @itemx -traditional-cpp
1546 @opindex traditional-cpp
1547 @opindex traditional
1548 Formerly, these options caused GCC to attempt to emulate a pre-standard
1549 C compiler. They are now only supported with the @option{-E} switch.
1550 The preprocessor continues to support a pre-standard mode. See the GNU
1551 CPP manual for details.
1553 @item -fcond-mismatch
1554 @opindex fcond-mismatch
1555 Allow conditional expressions with mismatched types in the second and
1556 third arguments. The value of such an expression is void. This option
1557 is not supported for C++.
1559 @item -flax-vector-conversions
1560 @opindex flax-vector-conversions
1561 Allow implicit conversions between vectors with differing numbers of
1562 elements and/or incompatible element types. This option should not be
1565 @item -funsigned-char
1566 @opindex funsigned-char
1567 Let the type @code{char} be unsigned, like @code{unsigned char}.
1569 Each kind of machine has a default for what @code{char} should
1570 be. It is either like @code{unsigned char} by default or like
1571 @code{signed char} by default.
1573 Ideally, a portable program should always use @code{signed char} or
1574 @code{unsigned char} when it depends on the signedness of an object.
1575 But many programs have been written to use plain @code{char} and
1576 expect it to be signed, or expect it to be unsigned, depending on the
1577 machines they were written for. This option, and its inverse, let you
1578 make such a program work with the opposite default.
1580 The type @code{char} is always a distinct type from each of
1581 @code{signed char} or @code{unsigned char}, even though its behavior
1582 is always just like one of those two.
1585 @opindex fsigned-char
1586 Let the type @code{char} be signed, like @code{signed char}.
1588 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1589 the negative form of @option{-funsigned-char}. Likewise, the option
1590 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1592 @item -fsigned-bitfields
1593 @itemx -funsigned-bitfields
1594 @itemx -fno-signed-bitfields
1595 @itemx -fno-unsigned-bitfields
1596 @opindex fsigned-bitfields
1597 @opindex funsigned-bitfields
1598 @opindex fno-signed-bitfields
1599 @opindex fno-unsigned-bitfields
1600 These options control whether a bit-field is signed or unsigned, when the
1601 declaration does not use either @code{signed} or @code{unsigned}. By
1602 default, such a bit-field is signed, because this is consistent: the
1603 basic integer types such as @code{int} are signed types.
1606 @node C++ Dialect Options
1607 @section Options Controlling C++ Dialect
1609 @cindex compiler options, C++
1610 @cindex C++ options, command line
1611 @cindex options, C++
1612 This section describes the command-line options that are only meaningful
1613 for C++ programs; but you can also use most of the GNU compiler options
1614 regardless of what language your program is in. For example, you
1615 might compile a file @code{firstClass.C} like this:
1618 g++ -g -frepo -O -c firstClass.C
1622 In this example, only @option{-frepo} is an option meant
1623 only for C++ programs; you can use the other options with any
1624 language supported by GCC@.
1626 Here is a list of options that are @emph{only} for compiling C++ programs:
1630 @item -fabi-version=@var{n}
1631 @opindex fabi-version
1632 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1633 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1634 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1635 the version that conforms most closely to the C++ ABI specification.
1636 Therefore, the ABI obtained using version 0 will change as ABI bugs
1639 The default is version 2.
1641 @item -fno-access-control
1642 @opindex fno-access-control
1643 Turn off all access checking. This switch is mainly useful for working
1644 around bugs in the access control code.
1648 Check that the pointer returned by @code{operator new} is non-null
1649 before attempting to modify the storage allocated. This check is
1650 normally unnecessary because the C++ standard specifies that
1651 @code{operator new} will only return @code{0} if it is declared
1652 @samp{throw()}, in which case the compiler will always check the
1653 return value even without this option. In all other cases, when
1654 @code{operator new} has a non-empty exception specification, memory
1655 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1656 @samp{new (nothrow)}.
1658 @item -fconserve-space
1659 @opindex fconserve-space
1660 Put uninitialized or runtime-initialized global variables into the
1661 common segment, as C does. This saves space in the executable at the
1662 cost of not diagnosing duplicate definitions. If you compile with this
1663 flag and your program mysteriously crashes after @code{main()} has
1664 completed, you may have an object that is being destroyed twice because
1665 two definitions were merged.
1667 This option is no longer useful on most targets, now that support has
1668 been added for putting variables into BSS without making them common.
1670 @item -ffriend-injection
1671 @opindex ffriend-injection
1672 Inject friend functions into the enclosing namespace, so that they are
1673 visible outside the scope of the class in which they are declared.
1674 Friend functions were documented to work this way in the old Annotated
1675 C++ Reference Manual, and versions of G++ before 4.1 always worked
1676 that way. However, in ISO C++ a friend function which is not declared
1677 in an enclosing scope can only be found using argument dependent
1678 lookup. This option causes friends to be injected as they were in
1681 This option is for compatibility, and may be removed in a future
1684 @item -fno-elide-constructors
1685 @opindex fno-elide-constructors
1686 The C++ standard allows an implementation to omit creating a temporary
1687 which is only used to initialize another object of the same type.
1688 Specifying this option disables that optimization, and forces G++ to
1689 call the copy constructor in all cases.
1691 @item -fno-enforce-eh-specs
1692 @opindex fno-enforce-eh-specs
1693 Don't generate code to check for violation of exception specifications
1694 at runtime. This option violates the C++ standard, but may be useful
1695 for reducing code size in production builds, much like defining
1696 @samp{NDEBUG}. This does not give user code permission to throw
1697 exceptions in violation of the exception specifications; the compiler
1698 will still optimize based on the specifications, so throwing an
1699 unexpected exception will result in undefined behavior.
1702 @itemx -fno-for-scope
1704 @opindex fno-for-scope
1705 If @option{-ffor-scope} is specified, the scope of variables declared in
1706 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1707 as specified by the C++ standard.
1708 If @option{-fno-for-scope} is specified, the scope of variables declared in
1709 a @i{for-init-statement} extends to the end of the enclosing scope,
1710 as was the case in old versions of G++, and other (traditional)
1711 implementations of C++.
1713 The default if neither flag is given to follow the standard,
1714 but to allow and give a warning for old-style code that would
1715 otherwise be invalid, or have different behavior.
1717 @item -fno-gnu-keywords
1718 @opindex fno-gnu-keywords
1719 Do not recognize @code{typeof} as a keyword, so that code can use this
1720 word as an identifier. You can use the keyword @code{__typeof__} instead.
1721 @option{-ansi} implies @option{-fno-gnu-keywords}.
1723 @item -fno-implicit-templates
1724 @opindex fno-implicit-templates
1725 Never emit code for non-inline templates which are instantiated
1726 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1727 @xref{Template Instantiation}, for more information.
1729 @item -fno-implicit-inline-templates
1730 @opindex fno-implicit-inline-templates
1731 Don't emit code for implicit instantiations of inline templates, either.
1732 The default is to handle inlines differently so that compiles with and
1733 without optimization will need the same set of explicit instantiations.
1735 @item -fno-implement-inlines
1736 @opindex fno-implement-inlines
1737 To save space, do not emit out-of-line copies of inline functions
1738 controlled by @samp{#pragma implementation}. This will cause linker
1739 errors if these functions are not inlined everywhere they are called.
1741 @item -fms-extensions
1742 @opindex fms-extensions
1743 Disable pedantic warnings about constructs used in MFC, such as implicit
1744 int and getting a pointer to member function via non-standard syntax.
1746 @item -fno-nonansi-builtins
1747 @opindex fno-nonansi-builtins
1748 Disable built-in declarations of functions that are not mandated by
1749 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1750 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1752 @item -fno-operator-names
1753 @opindex fno-operator-names
1754 Do not treat the operator name keywords @code{and}, @code{bitand},
1755 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1756 synonyms as keywords.
1758 @item -fno-optional-diags
1759 @opindex fno-optional-diags
1760 Disable diagnostics that the standard says a compiler does not need to
1761 issue. Currently, the only such diagnostic issued by G++ is the one for
1762 a name having multiple meanings within a class.
1765 @opindex fpermissive
1766 Downgrade some diagnostics about nonconformant code from errors to
1767 warnings. Thus, using @option{-fpermissive} will allow some
1768 nonconforming code to compile.
1772 Enable automatic template instantiation at link time. This option also
1773 implies @option{-fno-implicit-templates}. @xref{Template
1774 Instantiation}, for more information.
1778 Disable generation of information about every class with virtual
1779 functions for use by the C++ runtime type identification features
1780 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1781 of the language, you can save some space by using this flag. Note that
1782 exception handling uses the same information, but it will generate it as
1783 needed. The @samp{dynamic_cast} operator can still be used for casts that
1784 do not require runtime type information, i.e.@: casts to @code{void *} or to
1785 unambiguous base classes.
1789 Emit statistics about front-end processing at the end of the compilation.
1790 This information is generally only useful to the G++ development team.
1792 @item -ftemplate-depth-@var{n}
1793 @opindex ftemplate-depth
1794 Set the maximum instantiation depth for template classes to @var{n}.
1795 A limit on the template instantiation depth is needed to detect
1796 endless recursions during template class instantiation. ANSI/ISO C++
1797 conforming programs must not rely on a maximum depth greater than 17.
1799 @item -fno-threadsafe-statics
1800 @opindex fno-threadsafe-statics
1801 Do not emit the extra code to use the routines specified in the C++
1802 ABI for thread-safe initialization of local statics. You can use this
1803 option to reduce code size slightly in code that doesn't need to be
1806 @item -fuse-cxa-atexit
1807 @opindex fuse-cxa-atexit
1808 Register destructors for objects with static storage duration with the
1809 @code{__cxa_atexit} function rather than the @code{atexit} function.
1810 This option is required for fully standards-compliant handling of static
1811 destructors, but will only work if your C library supports
1812 @code{__cxa_atexit}.
1814 @item -fno-use-cxa-get-exception-ptr
1815 @opindex fno-use-cxa-get-exception-ptr
1816 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1817 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1818 if the runtime routine is not available.
1820 @item -fvisibility-inlines-hidden
1821 @opindex fvisibility-inlines-hidden
1822 This switch declares that the user does not attempt to compare
1823 pointers to inline methods where the addresses of the two functions
1824 were taken in different shared objects.
1826 The effect of this is that GCC may, effectively, mark inline methods with
1827 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1828 appear in the export table of a DSO and do not require a PLT indirection
1829 when used within the DSO@. Enabling this option can have a dramatic effect
1830 on load and link times of a DSO as it massively reduces the size of the
1831 dynamic export table when the library makes heavy use of templates.
1833 The behavior of this switch is not quite the same as marking the
1834 methods as hidden directly, because it does not affect static variables
1835 local to the function or cause the compiler to deduce that
1836 the function is defined in only one shared object.
1838 You may mark a method as having a visibility explicitly to negate the
1839 effect of the switch for that method. For example, if you do want to
1840 compare pointers to a particular inline method, you might mark it as
1841 having default visibility. Marking the enclosing class with explicit
1842 visibility will have no effect.
1844 Explicitly instantiated inline methods are unaffected by this option
1845 as their linkage might otherwise cross a shared library boundary.
1846 @xref{Template Instantiation}.
1848 @item -fvisibility-ms-compat
1849 @opindex fvisibility-ms-compat
1850 This flag attempts to use visibility settings to make GCC's C++
1851 linkage model compatible with that of Microsoft Visual Studio.
1853 The flag makes these changes to GCC's linkage model:
1857 It sets the default visibility to @code{hidden}, like
1858 @option{-fvisibility=hidden}.
1861 Types, but not their members, are not hidden by default.
1864 The One Definition Rule is relaxed for types without explicit
1865 visibility specifications which are defined in more than one different
1866 shared object: those declarations are permitted if they would have
1867 been permitted when this option was not used.
1870 In new code it is better to use @option{-fvisibility=hidden} and
1871 export those classes which are intended to be externally visible.
1872 Unfortunately it is possible for code to rely, perhaps accidentally,
1873 on the Visual Studio behavior.
1875 Among the consequences of these changes are that static data members
1876 of the same type with the same name but defined in different shared
1877 objects will be different, so changing one will not change the other;
1878 and that pointers to function members defined in different shared
1879 objects may not compare equal. When this flag is given, it is a
1880 violation of the ODR to define types with the same name differently.
1884 Do not use weak symbol support, even if it is provided by the linker.
1885 By default, G++ will use weak symbols if they are available. This
1886 option exists only for testing, and should not be used by end-users;
1887 it will result in inferior code and has no benefits. This option may
1888 be removed in a future release of G++.
1892 Do not search for header files in the standard directories specific to
1893 C++, but do still search the other standard directories. (This option
1894 is used when building the C++ library.)
1897 In addition, these optimization, warning, and code generation options
1898 have meanings only for C++ programs:
1901 @item -fno-default-inline
1902 @opindex fno-default-inline
1903 Do not assume @samp{inline} for functions defined inside a class scope.
1904 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1905 functions will have linkage like inline functions; they just won't be
1908 @item -Wabi @r{(C++ and Objective-C++ only)}
1911 Warn when G++ generates code that is probably not compatible with the
1912 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1913 all such cases, there are probably some cases that are not warned about,
1914 even though G++ is generating incompatible code. There may also be
1915 cases where warnings are emitted even though the code that is generated
1918 You should rewrite your code to avoid these warnings if you are
1919 concerned about the fact that code generated by G++ may not be binary
1920 compatible with code generated by other compilers.
1922 The known incompatibilities at this point include:
1927 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1928 pack data into the same byte as a base class. For example:
1931 struct A @{ virtual void f(); int f1 : 1; @};
1932 struct B : public A @{ int f2 : 1; @};
1936 In this case, G++ will place @code{B::f2} into the same byte
1937 as@code{A::f1}; other compilers will not. You can avoid this problem
1938 by explicitly padding @code{A} so that its size is a multiple of the
1939 byte size on your platform; that will cause G++ and other compilers to
1940 layout @code{B} identically.
1943 Incorrect handling of tail-padding for virtual bases. G++ does not use
1944 tail padding when laying out virtual bases. For example:
1947 struct A @{ virtual void f(); char c1; @};
1948 struct B @{ B(); char c2; @};
1949 struct C : public A, public virtual B @{@};
1953 In this case, G++ will not place @code{B} into the tail-padding for
1954 @code{A}; other compilers will. You can avoid this problem by
1955 explicitly padding @code{A} so that its size is a multiple of its
1956 alignment (ignoring virtual base classes); that will cause G++ and other
1957 compilers to layout @code{C} identically.
1960 Incorrect handling of bit-fields with declared widths greater than that
1961 of their underlying types, when the bit-fields appear in a union. For
1965 union U @{ int i : 4096; @};
1969 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1970 union too small by the number of bits in an @code{int}.
1973 Empty classes can be placed at incorrect offsets. For example:
1983 struct C : public B, public A @{@};
1987 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1988 it should be placed at offset zero. G++ mistakenly believes that the
1989 @code{A} data member of @code{B} is already at offset zero.
1992 Names of template functions whose types involve @code{typename} or
1993 template template parameters can be mangled incorrectly.
1996 template <typename Q>
1997 void f(typename Q::X) @{@}
1999 template <template <typename> class Q>
2000 void f(typename Q<int>::X) @{@}
2004 Instantiations of these templates may be mangled incorrectly.
2008 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2009 @opindex Wctor-dtor-privacy
2010 @opindex Wno-ctor-dtor-privacy
2011 Warn when a class seems unusable because all the constructors or
2012 destructors in that class are private, and it has neither friends nor
2013 public static member functions.
2015 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2016 @opindex Wnon-virtual-dtor
2017 @opindex Wno-non-virtual-dtor
2018 Warn when a class has virtual functions and accessible non-virtual
2019 destructor, in which case it would be possible but unsafe to delete
2020 an instance of a derived class through a pointer to the base class.
2021 This warning is also enabled if -Weffc++ is specified.
2023 @item -Wreorder @r{(C++ and Objective-C++ only)}
2025 @opindex Wno-reorder
2026 @cindex reordering, warning
2027 @cindex warning for reordering of member initializers
2028 Warn when the order of member initializers given in the code does not
2029 match the order in which they must be executed. For instance:
2035 A(): j (0), i (1) @{ @}
2039 The compiler will rearrange the member initializers for @samp{i}
2040 and @samp{j} to match the declaration order of the members, emitting
2041 a warning to that effect. This warning is enabled by @option{-Wall}.
2044 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2047 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2050 Warn about violations of the following style guidelines from Scott Meyers'
2051 @cite{Effective C++} book:
2055 Item 11: Define a copy constructor and an assignment operator for classes
2056 with dynamically allocated memory.
2059 Item 12: Prefer initialization to assignment in constructors.
2062 Item 14: Make destructors virtual in base classes.
2065 Item 15: Have @code{operator=} return a reference to @code{*this}.
2068 Item 23: Don't try to return a reference when you must return an object.
2072 Also warn about violations of the following style guidelines from
2073 Scott Meyers' @cite{More Effective C++} book:
2077 Item 6: Distinguish between prefix and postfix forms of increment and
2078 decrement operators.
2081 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2085 When selecting this option, be aware that the standard library
2086 headers do not obey all of these guidelines; use @samp{grep -v}
2087 to filter out those warnings.
2089 @item -Wno-deprecated @r{(C++ and Objective-C++ only)}
2090 @opindex Wno-deprecated
2091 @opindex Wdeprecated
2092 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
2094 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2095 @opindex Wstrict-null-sentinel
2096 @opindex Wno-strict-null-sentinel
2097 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2098 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2099 to @code{__null}. Although it is a null pointer constant not a null pointer,
2100 it is guaranteed to of the same size as a pointer. But this use is
2101 not portable across different compilers.
2103 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2104 @opindex Wno-non-template-friend
2105 @opindex Wnon-template-friend
2106 Disable warnings when non-templatized friend functions are declared
2107 within a template. Since the advent of explicit template specification
2108 support in G++, if the name of the friend is an unqualified-id (i.e.,
2109 @samp{friend foo(int)}), the C++ language specification demands that the
2110 friend declare or define an ordinary, nontemplate function. (Section
2111 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2112 could be interpreted as a particular specialization of a templatized
2113 function. Because this non-conforming behavior is no longer the default
2114 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2115 check existing code for potential trouble spots and is on by default.
2116 This new compiler behavior can be turned off with
2117 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2118 but disables the helpful warning.
2120 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2121 @opindex Wold-style-cast
2122 @opindex Wno-old-style-cast
2123 Warn if an old-style (C-style) cast to a non-void type is used within
2124 a C++ program. The new-style casts (@samp{dynamic_cast},
2125 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2126 less vulnerable to unintended effects and much easier to search for.
2128 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2129 @opindex Woverloaded-virtual
2130 @opindex Wno-overloaded-virtual
2131 @cindex overloaded virtual fn, warning
2132 @cindex warning for overloaded virtual fn
2133 Warn when a function declaration hides virtual functions from a
2134 base class. For example, in:
2141 struct B: public A @{
2146 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2154 will fail to compile.
2156 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2157 @opindex Wno-pmf-conversions
2158 @opindex Wpmf-conversions
2159 Disable the diagnostic for converting a bound pointer to member function
2162 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2163 @opindex Wsign-promo
2164 @opindex Wno-sign-promo
2165 Warn when overload resolution chooses a promotion from unsigned or
2166 enumerated type to a signed type, over a conversion to an unsigned type of
2167 the same size. Previous versions of G++ would try to preserve
2168 unsignedness, but the standard mandates the current behavior.
2173 A& operator = (int);
2183 In this example, G++ will synthesize a default @samp{A& operator =
2184 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2187 @node Objective-C and Objective-C++ Dialect Options
2188 @section Options Controlling Objective-C and Objective-C++ Dialects
2190 @cindex compiler options, Objective-C and Objective-C++
2191 @cindex Objective-C and Objective-C++ options, command line
2192 @cindex options, Objective-C and Objective-C++
2193 (NOTE: This manual does not describe the Objective-C and Objective-C++
2194 languages themselves. See @xref{Standards,,Language Standards
2195 Supported by GCC}, for references.)
2197 This section describes the command-line options that are only meaningful
2198 for Objective-C and Objective-C++ programs, but you can also use most of
2199 the language-independent GNU compiler options.
2200 For example, you might compile a file @code{some_class.m} like this:
2203 gcc -g -fgnu-runtime -O -c some_class.m
2207 In this example, @option{-fgnu-runtime} is an option meant only for
2208 Objective-C and Objective-C++ programs; you can use the other options with
2209 any language supported by GCC@.
2211 Note that since Objective-C is an extension of the C language, Objective-C
2212 compilations may also use options specific to the C front-end (e.g.,
2213 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2214 C++-specific options (e.g., @option{-Wabi}).
2216 Here is a list of options that are @emph{only} for compiling Objective-C
2217 and Objective-C++ programs:
2220 @item -fconstant-string-class=@var{class-name}
2221 @opindex fconstant-string-class
2222 Use @var{class-name} as the name of the class to instantiate for each
2223 literal string specified with the syntax @code{@@"@dots{}"}. The default
2224 class name is @code{NXConstantString} if the GNU runtime is being used, and
2225 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2226 @option{-fconstant-cfstrings} option, if also present, will override the
2227 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2228 to be laid out as constant CoreFoundation strings.
2231 @opindex fgnu-runtime
2232 Generate object code compatible with the standard GNU Objective-C
2233 runtime. This is the default for most types of systems.
2235 @item -fnext-runtime
2236 @opindex fnext-runtime
2237 Generate output compatible with the NeXT runtime. This is the default
2238 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2239 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2242 @item -fno-nil-receivers
2243 @opindex fno-nil-receivers
2244 Assume that all Objective-C message dispatches (e.g.,
2245 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2246 is not @code{nil}. This allows for more efficient entry points in the runtime
2247 to be used. Currently, this option is only available in conjunction with
2248 the NeXT runtime on Mac OS X 10.3 and later.
2250 @item -fobjc-call-cxx-cdtors
2251 @opindex fobjc-call-cxx-cdtors
2252 For each Objective-C class, check if any of its instance variables is a
2253 C++ object with a non-trivial default constructor. If so, synthesize a
2254 special @code{- (id) .cxx_construct} instance method that will run
2255 non-trivial default constructors on any such instance variables, in order,
2256 and then return @code{self}. Similarly, check if any instance variable
2257 is a C++ object with a non-trivial destructor, and if so, synthesize a
2258 special @code{- (void) .cxx_destruct} method that will run
2259 all such default destructors, in reverse order.
2261 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2262 thusly generated will only operate on instance variables declared in the
2263 current Objective-C class, and not those inherited from superclasses. It
2264 is the responsibility of the Objective-C runtime to invoke all such methods
2265 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2266 will be invoked by the runtime immediately after a new object
2267 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2268 be invoked immediately before the runtime deallocates an object instance.
2270 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2271 support for invoking the @code{- (id) .cxx_construct} and
2272 @code{- (void) .cxx_destruct} methods.
2274 @item -fobjc-direct-dispatch
2275 @opindex fobjc-direct-dispatch
2276 Allow fast jumps to the message dispatcher. On Darwin this is
2277 accomplished via the comm page.
2279 @item -fobjc-exceptions
2280 @opindex fobjc-exceptions
2281 Enable syntactic support for structured exception handling in Objective-C,
2282 similar to what is offered by C++ and Java. This option is
2283 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2292 @@catch (AnObjCClass *exc) @{
2299 @@catch (AnotherClass *exc) @{
2302 @@catch (id allOthers) @{
2312 The @code{@@throw} statement may appear anywhere in an Objective-C or
2313 Objective-C++ program; when used inside of a @code{@@catch} block, the
2314 @code{@@throw} may appear without an argument (as shown above), in which case
2315 the object caught by the @code{@@catch} will be rethrown.
2317 Note that only (pointers to) Objective-C objects may be thrown and
2318 caught using this scheme. When an object is thrown, it will be caught
2319 by the nearest @code{@@catch} clause capable of handling objects of that type,
2320 analogously to how @code{catch} blocks work in C++ and Java. A
2321 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2322 any and all Objective-C exceptions not caught by previous @code{@@catch}
2325 The @code{@@finally} clause, if present, will be executed upon exit from the
2326 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2327 regardless of whether any exceptions are thrown, caught or rethrown
2328 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2329 of the @code{finally} clause in Java.
2331 There are several caveats to using the new exception mechanism:
2335 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2336 idioms provided by the @code{NSException} class, the new
2337 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2338 systems, due to additional functionality needed in the (NeXT) Objective-C
2342 As mentioned above, the new exceptions do not support handling
2343 types other than Objective-C objects. Furthermore, when used from
2344 Objective-C++, the Objective-C exception model does not interoperate with C++
2345 exceptions at this time. This means you cannot @code{@@throw} an exception
2346 from Objective-C and @code{catch} it in C++, or vice versa
2347 (i.e., @code{throw @dots{} @@catch}).
2350 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2351 blocks for thread-safe execution:
2354 @@synchronized (ObjCClass *guard) @{
2359 Upon entering the @code{@@synchronized} block, a thread of execution shall
2360 first check whether a lock has been placed on the corresponding @code{guard}
2361 object by another thread. If it has, the current thread shall wait until
2362 the other thread relinquishes its lock. Once @code{guard} becomes available,
2363 the current thread will place its own lock on it, execute the code contained in
2364 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2365 making @code{guard} available to other threads).
2367 Unlike Java, Objective-C does not allow for entire methods to be marked
2368 @code{@@synchronized}. Note that throwing exceptions out of
2369 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2370 to be unlocked properly.
2374 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2376 @item -freplace-objc-classes
2377 @opindex freplace-objc-classes
2378 Emit a special marker instructing @command{ld(1)} not to statically link in
2379 the resulting object file, and allow @command{dyld(1)} to load it in at
2380 run time instead. This is used in conjunction with the Fix-and-Continue
2381 debugging mode, where the object file in question may be recompiled and
2382 dynamically reloaded in the course of program execution, without the need
2383 to restart the program itself. Currently, Fix-and-Continue functionality
2384 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2389 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2390 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2391 compile time) with static class references that get initialized at load time,
2392 which improves run-time performance. Specifying the @option{-fzero-link} flag
2393 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2394 to be retained. This is useful in Zero-Link debugging mode, since it allows
2395 for individual class implementations to be modified during program execution.
2399 Dump interface declarations for all classes seen in the source file to a
2400 file named @file{@var{sourcename}.decl}.
2402 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2403 @opindex Wassign-intercept
2404 @opindex Wno-assign-intercept
2405 Warn whenever an Objective-C assignment is being intercepted by the
2408 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2409 @opindex Wno-protocol
2411 If a class is declared to implement a protocol, a warning is issued for
2412 every method in the protocol that is not implemented by the class. The
2413 default behavior is to issue a warning for every method not explicitly
2414 implemented in the class, even if a method implementation is inherited
2415 from the superclass. If you use the @option{-Wno-protocol} option, then
2416 methods inherited from the superclass are considered to be implemented,
2417 and no warning is issued for them.
2419 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2421 @opindex Wno-selector
2422 Warn if multiple methods of different types for the same selector are
2423 found during compilation. The check is performed on the list of methods
2424 in the final stage of compilation. Additionally, a check is performed
2425 for each selector appearing in a @code{@@selector(@dots{})}
2426 expression, and a corresponding method for that selector has been found
2427 during compilation. Because these checks scan the method table only at
2428 the end of compilation, these warnings are not produced if the final
2429 stage of compilation is not reached, for example because an error is
2430 found during compilation, or because the @option{-fsyntax-only} option is
2433 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2434 @opindex Wstrict-selector-match
2435 @opindex Wno-strict-selector-match
2436 Warn if multiple methods with differing argument and/or return types are
2437 found for a given selector when attempting to send a message using this
2438 selector to a receiver of type @code{id} or @code{Class}. When this flag
2439 is off (which is the default behavior), the compiler will omit such warnings
2440 if any differences found are confined to types which share the same size
2443 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2444 @opindex Wundeclared-selector
2445 @opindex Wno-undeclared-selector
2446 Warn if a @code{@@selector(@dots{})} expression referring to an
2447 undeclared selector is found. A selector is considered undeclared if no
2448 method with that name has been declared before the
2449 @code{@@selector(@dots{})} expression, either explicitly in an
2450 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2451 an @code{@@implementation} section. This option always performs its
2452 checks as soon as a @code{@@selector(@dots{})} expression is found,
2453 while @option{-Wselector} only performs its checks in the final stage of
2454 compilation. This also enforces the coding style convention
2455 that methods and selectors must be declared before being used.
2457 @item -print-objc-runtime-info
2458 @opindex print-objc-runtime-info
2459 Generate C header describing the largest structure that is passed by
2464 @node Language Independent Options
2465 @section Options to Control Diagnostic Messages Formatting
2466 @cindex options to control diagnostics formatting
2467 @cindex diagnostic messages
2468 @cindex message formatting
2470 Traditionally, diagnostic messages have been formatted irrespective of
2471 the output device's aspect (e.g.@: its width, @dots{}). The options described
2472 below can be used to control the diagnostic messages formatting
2473 algorithm, e.g.@: how many characters per line, how often source location
2474 information should be reported. Right now, only the C++ front end can
2475 honor these options. However it is expected, in the near future, that
2476 the remaining front ends would be able to digest them correctly.
2479 @item -fmessage-length=@var{n}
2480 @opindex fmessage-length
2481 Try to format error messages so that they fit on lines of about @var{n}
2482 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2483 the front ends supported by GCC@. If @var{n} is zero, then no
2484 line-wrapping will be done; each error message will appear on a single
2487 @opindex fdiagnostics-show-location
2488 @item -fdiagnostics-show-location=once
2489 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2490 reporter to emit @emph{once} source location information; that is, in
2491 case the message is too long to fit on a single physical line and has to
2492 be wrapped, the source location won't be emitted (as prefix) again,
2493 over and over, in subsequent continuation lines. This is the default
2496 @item -fdiagnostics-show-location=every-line
2497 Only meaningful in line-wrapping mode. Instructs the diagnostic
2498 messages reporter to emit the same source location information (as
2499 prefix) for physical lines that result from the process of breaking
2500 a message which is too long to fit on a single line.
2502 @item -fdiagnostics-show-option
2503 @opindex fdiagnostics-show-option
2504 This option instructs the diagnostic machinery to add text to each
2505 diagnostic emitted, which indicates which command line option directly
2506 controls that diagnostic, when such an option is known to the
2507 diagnostic machinery.
2509 @item -Wcoverage-mismatch
2510 @opindex Wcoverage-mismatch
2511 Warn if feedback profiles do not match when using the
2512 @option{-fprofile-use} option.
2513 If a source file was changed between @option{-fprofile-gen} and
2514 @option{-fprofile-use}, the files with the profile feedback can fail
2515 to match the source file and GCC can not use the profile feedback
2516 information. By default, GCC emits an error message in this case.
2517 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2518 error. GCC does not use appropriate feedback profiles, so using this
2519 option can result in poorly optimized code. This option is useful
2520 only in the case of very minor changes such as bug fixes to an
2525 @node Warning Options
2526 @section Options to Request or Suppress Warnings
2527 @cindex options to control warnings
2528 @cindex warning messages
2529 @cindex messages, warning
2530 @cindex suppressing warnings
2532 Warnings are diagnostic messages that report constructions which
2533 are not inherently erroneous but which are risky or suggest there
2534 may have been an error.
2536 The following language-independent options do not enable specific
2537 warnings but control the kinds of diagnostics produced by GCC.
2540 @cindex syntax checking
2542 @opindex fsyntax-only
2543 Check the code for syntax errors, but don't do anything beyond that.
2547 Inhibit all warning messages.
2552 Make all warnings into errors.
2557 Make the specified warning into an error. The specifier for a warning
2558 is appended, for example @option{-Werror=switch} turns the warnings
2559 controlled by @option{-Wswitch} into errors. This switch takes a
2560 negative form, to be used to negate @option{-Werror} for specific
2561 warnings, for example @option{-Wno-error=switch} makes
2562 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2563 is in effect. You can use the @option{-fdiagnostics-show-option}
2564 option to have each controllable warning amended with the option which
2565 controls it, to determine what to use with this option.
2567 Note that specifying @option{-Werror=}@var{foo} automatically implies
2568 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2571 @item -Wfatal-errors
2572 @opindex Wfatal-errors
2573 @opindex Wno-fatal-errors
2574 This option causes the compiler to abort compilation on the first error
2575 occurred rather than trying to keep going and printing further error
2580 You can request many specific warnings with options beginning
2581 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2582 implicit declarations. Each of these specific warning options also
2583 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2584 example, @option{-Wno-implicit}. This manual lists only one of the
2585 two forms, whichever is not the default. For further,
2586 language-specific options also refer to @ref{C++ Dialect Options} and
2587 @ref{Objective-C and Objective-C++ Dialect Options}.
2592 Issue all the warnings demanded by strict ISO C and ISO C++;
2593 reject all programs that use forbidden extensions, and some other
2594 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2595 version of the ISO C standard specified by any @option{-std} option used.
2597 Valid ISO C and ISO C++ programs should compile properly with or without
2598 this option (though a rare few will require @option{-ansi} or a
2599 @option{-std} option specifying the required version of ISO C)@. However,
2600 without this option, certain GNU extensions and traditional C and C++
2601 features are supported as well. With this option, they are rejected.
2603 @option{-pedantic} does not cause warning messages for use of the
2604 alternate keywords whose names begin and end with @samp{__}. Pedantic
2605 warnings are also disabled in the expression that follows
2606 @code{__extension__}. However, only system header files should use
2607 these escape routes; application programs should avoid them.
2608 @xref{Alternate Keywords}.
2610 Some users try to use @option{-pedantic} to check programs for strict ISO
2611 C conformance. They soon find that it does not do quite what they want:
2612 it finds some non-ISO practices, but not all---only those for which
2613 ISO C @emph{requires} a diagnostic, and some others for which
2614 diagnostics have been added.
2616 A feature to report any failure to conform to ISO C might be useful in
2617 some instances, but would require considerable additional work and would
2618 be quite different from @option{-pedantic}. We don't have plans to
2619 support such a feature in the near future.
2621 Where the standard specified with @option{-std} represents a GNU
2622 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2623 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2624 extended dialect is based. Warnings from @option{-pedantic} are given
2625 where they are required by the base standard. (It would not make sense
2626 for such warnings to be given only for features not in the specified GNU
2627 C dialect, since by definition the GNU dialects of C include all
2628 features the compiler supports with the given option, and there would be
2629 nothing to warn about.)
2631 @item -pedantic-errors
2632 @opindex pedantic-errors
2633 Like @option{-pedantic}, except that errors are produced rather than
2639 This enables all the warnings about constructions that some users
2640 consider questionable, and that are easy to avoid (or modify to
2641 prevent the warning), even in conjunction with macros. This also
2642 enables some language-specific warnings described in @ref{C++ Dialect
2643 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2645 @option{-Wall} turns on the following warning flags:
2647 @gccoptlist{-Waddress @gol
2648 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2650 -Wchar-subscripts @gol
2652 -Wimplicit-function-declaration @gol
2655 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2656 -Wmissing-braces @gol
2662 -Wsequence-point @gol
2663 -Wsign-compare @r{(only in C++)} @gol
2664 -Wstrict-aliasing @gol
2665 -Wstrict-overflow=1 @gol
2668 -Wuninitialized @r{(only with} @option{-O1} @r{and above)} @gol
2669 -Wunknown-pragmas @gol
2670 -Wunused-function @gol
2673 -Wunused-variable @gol
2676 Note that some warning flags are not implied by @option{-Wall}. Some of
2677 them warn about constructions that users generally do not consider
2678 questionable, but which occasionally you might wish to check for;
2679 others warn about constructions that are necessary or hard to avoid in
2680 some cases, and there is no simple way to modify the code to suppress
2681 the warning. Some of them are enabled by @option{-Wextra} but many of
2682 them must be enabled individually.
2688 This enables some extra warning flags that are not enabled by
2689 @option{-Wall}. (This option used to be called @option{-W}. The older
2690 name is still supported, but the newer name is more descriptive.)
2692 @gccoptlist{-Wclobbered @gol
2694 -Wignored-qualifiers @gol
2695 -Wmissing-field-initializers @gol
2696 -Wmissing-parameter-type @r{(C only)} @gol
2697 -Wold-style-declaration @r{(C only)} @gol
2698 -Woverride-init @gol
2701 -Wuninitialized @r{(only with} @option{-O1} @r{and above)} @gol
2702 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2705 The option @option{-Wextra} also prints warning messages for the
2711 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2712 @samp{>}, or @samp{>=}.
2715 (C++ only) An enumerator and a non-enumerator both appear in a
2716 conditional expression.
2719 (C++ only) A non-static reference or non-static @samp{const} member
2720 appears in a class without constructors.
2723 (C++ only) Ambiguous virtual bases.
2726 (C++ only) Subscripting an array which has been declared @samp{register}.
2729 (C++ only) Taking the address of a variable which has been declared
2733 (C++ only) A base class is not initialized in a derived class' copy
2741 Inhibit warning messages about the use of @samp{#import}.
2743 @item -Wchar-subscripts
2744 @opindex Wchar-subscripts
2745 @opindex Wno-char-subscripts
2746 Warn if an array subscript has type @code{char}. This is a common cause
2747 of error, as programmers often forget that this type is signed on some
2749 This warning is enabled by @option{-Wall}.
2753 @opindex Wno-comment
2754 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2755 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2756 This warning is enabled by @option{-Wall}.
2761 @opindex ffreestanding
2762 @opindex fno-builtin
2763 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2764 the arguments supplied have types appropriate to the format string
2765 specified, and that the conversions specified in the format string make
2766 sense. This includes standard functions, and others specified by format
2767 attributes (@pxref{Function Attributes}), in the @code{printf},
2768 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2769 not in the C standard) families (or other target-specific families).
2770 Which functions are checked without format attributes having been
2771 specified depends on the standard version selected, and such checks of
2772 functions without the attribute specified are disabled by
2773 @option{-ffreestanding} or @option{-fno-builtin}.
2775 The formats are checked against the format features supported by GNU
2776 libc version 2.2. These include all ISO C90 and C99 features, as well
2777 as features from the Single Unix Specification and some BSD and GNU
2778 extensions. Other library implementations may not support all these
2779 features; GCC does not support warning about features that go beyond a
2780 particular library's limitations. However, if @option{-pedantic} is used
2781 with @option{-Wformat}, warnings will be given about format features not
2782 in the selected standard version (but not for @code{strfmon} formats,
2783 since those are not in any version of the C standard). @xref{C Dialect
2784 Options,,Options Controlling C Dialect}.
2786 Since @option{-Wformat} also checks for null format arguments for
2787 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2789 @option{-Wformat} is included in @option{-Wall}. For more control over some
2790 aspects of format checking, the options @option{-Wformat-y2k},
2791 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2792 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2793 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2796 @opindex Wformat-y2k
2797 @opindex Wno-format-y2k
2798 If @option{-Wformat} is specified, also warn about @code{strftime}
2799 formats which may yield only a two-digit year.
2801 @item -Wno-format-extra-args
2802 @opindex Wno-format-extra-args
2803 @opindex Wformat-extra-args
2804 If @option{-Wformat} is specified, do not warn about excess arguments to a
2805 @code{printf} or @code{scanf} format function. The C standard specifies
2806 that such arguments are ignored.
2808 Where the unused arguments lie between used arguments that are
2809 specified with @samp{$} operand number specifications, normally
2810 warnings are still given, since the implementation could not know what
2811 type to pass to @code{va_arg} to skip the unused arguments. However,
2812 in the case of @code{scanf} formats, this option will suppress the
2813 warning if the unused arguments are all pointers, since the Single
2814 Unix Specification says that such unused arguments are allowed.
2816 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2817 @opindex Wno-format-zero-length
2818 @opindex Wformat-zero-length
2819 If @option{-Wformat} is specified, do not warn about zero-length formats.
2820 The C standard specifies that zero-length formats are allowed.
2822 @item -Wformat-nonliteral
2823 @opindex Wformat-nonliteral
2824 @opindex Wno-format-nonliteral
2825 If @option{-Wformat} is specified, also warn if the format string is not a
2826 string literal and so cannot be checked, unless the format function
2827 takes its format arguments as a @code{va_list}.
2829 @item -Wformat-security
2830 @opindex Wformat-security
2831 @opindex Wno-format-security
2832 If @option{-Wformat} is specified, also warn about uses of format
2833 functions that represent possible security problems. At present, this
2834 warns about calls to @code{printf} and @code{scanf} functions where the
2835 format string is not a string literal and there are no format arguments,
2836 as in @code{printf (foo);}. This may be a security hole if the format
2837 string came from untrusted input and contains @samp{%n}. (This is
2838 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2839 in future warnings may be added to @option{-Wformat-security} that are not
2840 included in @option{-Wformat-nonliteral}.)
2844 @opindex Wno-format=2
2845 Enable @option{-Wformat} plus format checks not included in
2846 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2847 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2849 @item -Wnonnull @r{(C and Objective-C only)}
2851 @opindex Wno-nonnull
2852 Warn about passing a null pointer for arguments marked as
2853 requiring a non-null value by the @code{nonnull} function attribute.
2855 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2856 can be disabled with the @option{-Wno-nonnull} option.
2858 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2860 @opindex Wno-init-self
2861 Warn about uninitialized variables which are initialized with themselves.
2862 Note this option can only be used with the @option{-Wuninitialized} option,
2863 which in turn only works with @option{-O1} and above.
2865 For example, GCC will warn about @code{i} being uninitialized in the
2866 following snippet only when @option{-Winit-self} has been specified:
2877 @item -Wimplicit-int @r{(C and Objective-C only)}
2878 @opindex Wimplicit-int
2879 @opindex Wno-implicit-int
2880 Warn when a declaration does not specify a type.
2881 This warning is enabled by @option{-Wall}.
2883 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
2884 @opindex Wimplicit-function-declaration
2885 @opindex Wno-implicit-function-declaration
2886 Give a warning whenever a function is used before being declared. In
2887 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2888 enabled by default and it is made into an error by
2889 @option{-pedantic-errors}. This warning is also enabled by
2894 @opindex Wno-implicit
2895 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2896 This warning is enabled by @option{-Wall}.
2898 @item -Wignored-qualifiers @r{(C and C++ only)}
2899 @opindex Wignored-qualifiers
2900 @opindex Wno-ignored-qualifiers
2901 Warn if the return type of a function has a type qualifier
2902 such as @code{const}. For ISO C such a type qualifier has no effect,
2903 since the value returned by a function is not an lvalue.
2904 For C++, the warning is only emitted for scalar types or @code{void}.
2905 ISO C prohibits qualified @code{void} return types on function
2906 definitions, so such return types always receive a warning
2907 even without this option.
2909 This warning is also enabled by @option{-Wextra}.
2914 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2915 function with external linkage, returning int, taking either zero
2916 arguments, two, or three arguments of appropriate types.
2917 This warning is enabled by @option{-Wall}.
2919 @item -Wmissing-braces
2920 @opindex Wmissing-braces
2921 @opindex Wno-missing-braces
2922 Warn if an aggregate or union initializer is not fully bracketed. In
2923 the following example, the initializer for @samp{a} is not fully
2924 bracketed, but that for @samp{b} is fully bracketed.
2927 int a[2][2] = @{ 0, 1, 2, 3 @};
2928 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2931 This warning is enabled by @option{-Wall}.
2933 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2934 @opindex Wmissing-include-dirs
2935 @opindex Wno-missing-include-dirs
2936 Warn if a user-supplied include directory does not exist.
2939 @opindex Wparentheses
2940 @opindex Wno-parentheses
2941 Warn if parentheses are omitted in certain contexts, such
2942 as when there is an assignment in a context where a truth value
2943 is expected, or when operators are nested whose precedence people
2944 often get confused about.
2946 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2947 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2948 interpretation from that of ordinary mathematical notation.
2950 Also warn about constructions where there may be confusion to which
2951 @code{if} statement an @code{else} branch belongs. Here is an example of
2966 In C/C++, every @code{else} branch belongs to the innermost possible
2967 @code{if} statement, which in this example is @code{if (b)}. This is
2968 often not what the programmer expected, as illustrated in the above
2969 example by indentation the programmer chose. When there is the
2970 potential for this confusion, GCC will issue a warning when this flag
2971 is specified. To eliminate the warning, add explicit braces around
2972 the innermost @code{if} statement so there is no way the @code{else}
2973 could belong to the enclosing @code{if}. The resulting code would
2990 This warning is enabled by @option{-Wall}.
2992 @item -Wsequence-point
2993 @opindex Wsequence-point
2994 @opindex Wno-sequence-point
2995 Warn about code that may have undefined semantics because of violations
2996 of sequence point rules in the C and C++ standards.
2998 The C and C++ standards defines the order in which expressions in a C/C++
2999 program are evaluated in terms of @dfn{sequence points}, which represent
3000 a partial ordering between the execution of parts of the program: those
3001 executed before the sequence point, and those executed after it. These
3002 occur after the evaluation of a full expression (one which is not part
3003 of a larger expression), after the evaluation of the first operand of a
3004 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3005 function is called (but after the evaluation of its arguments and the
3006 expression denoting the called function), and in certain other places.
3007 Other than as expressed by the sequence point rules, the order of
3008 evaluation of subexpressions of an expression is not specified. All
3009 these rules describe only a partial order rather than a total order,
3010 since, for example, if two functions are called within one expression
3011 with no sequence point between them, the order in which the functions
3012 are called is not specified. However, the standards committee have
3013 ruled that function calls do not overlap.
3015 It is not specified when between sequence points modifications to the
3016 values of objects take effect. Programs whose behavior depends on this
3017 have undefined behavior; the C and C++ standards specify that ``Between
3018 the previous and next sequence point an object shall have its stored
3019 value modified at most once by the evaluation of an expression.
3020 Furthermore, the prior value shall be read only to determine the value
3021 to be stored.''. If a program breaks these rules, the results on any
3022 particular implementation are entirely unpredictable.
3024 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3025 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3026 diagnosed by this option, and it may give an occasional false positive
3027 result, but in general it has been found fairly effective at detecting
3028 this sort of problem in programs.
3030 The standard is worded confusingly, therefore there is some debate
3031 over the precise meaning of the sequence point rules in subtle cases.
3032 Links to discussions of the problem, including proposed formal
3033 definitions, may be found on the GCC readings page, at
3034 @w{@uref{http://gcc.gnu.org/readings.html}}.
3036 This warning is enabled by @option{-Wall} for C and C++.
3039 @opindex Wreturn-type
3040 @opindex Wno-return-type
3041 Warn whenever a function is defined with a return-type that defaults
3042 to @code{int}. Also warn about any @code{return} statement with no
3043 return-value in a function whose return-type is not @code{void}
3044 (falling off the end of the function body is considered returning
3045 without a value), and about a @code{return} statement with a
3046 expression in a function whose return-type is @code{void}.
3048 For C++, a function without return type always produces a diagnostic
3049 message, even when @option{-Wno-return-type} is specified. The only
3050 exceptions are @samp{main} and functions defined in system headers.
3052 This warning is enabled by @option{-Wall}.
3057 Warn whenever a @code{switch} statement has an index of enumerated type
3058 and lacks a @code{case} for one or more of the named codes of that
3059 enumeration. (The presence of a @code{default} label prevents this
3060 warning.) @code{case} labels outside the enumeration range also
3061 provoke warnings when this option is used.
3062 This warning is enabled by @option{-Wall}.
3064 @item -Wswitch-default
3065 @opindex Wswitch-default
3066 @opindex Wno-switch-default
3067 Warn whenever a @code{switch} statement does not have a @code{default}
3071 @opindex Wswitch-enum
3072 @opindex Wno-switch-enum
3073 Warn whenever a @code{switch} statement has an index of enumerated type
3074 and lacks a @code{case} for one or more of the named codes of that
3075 enumeration. @code{case} labels outside the enumeration range also
3076 provoke warnings when this option is used.
3080 @opindex Wno-trigraphs
3081 Warn if any trigraphs are encountered that might change the meaning of
3082 the program (trigraphs within comments are not warned about).
3083 This warning is enabled by @option{-Wall}.
3085 @item -Wunused-function
3086 @opindex Wunused-function
3087 @opindex Wno-unused-function
3088 Warn whenever a static function is declared but not defined or a
3089 non-inline static function is unused.
3090 This warning is enabled by @option{-Wall}.
3092 @item -Wunused-label
3093 @opindex Wunused-label
3094 @opindex Wno-unused-label
3095 Warn whenever a label is declared but not used.
3096 This warning is enabled by @option{-Wall}.
3098 To suppress this warning use the @samp{unused} attribute
3099 (@pxref{Variable Attributes}).
3101 @item -Wunused-parameter
3102 @opindex Wunused-parameter
3103 @opindex Wno-unused-parameter
3104 Warn whenever a function parameter is unused aside from its declaration.
3106 To suppress this warning use the @samp{unused} attribute
3107 (@pxref{Variable Attributes}).
3109 @item -Wunused-variable
3110 @opindex Wunused-variable
3111 @opindex Wno-unused-variable
3112 Warn whenever a local variable or non-constant static variable is unused
3113 aside from its declaration.
3114 This warning is enabled by @option{-Wall}.
3116 To suppress this warning use the @samp{unused} attribute
3117 (@pxref{Variable Attributes}).
3119 @item -Wunused-value
3120 @opindex Wunused-value
3121 @opindex Wno-unused-value
3122 Warn whenever a statement computes a result that is explicitly not
3123 used. To suppress this warning cast the unused expression to
3124 @samp{void}. This includes an expression-statement or the left-hand
3125 side of a comma expression that contains no side effects. For example,
3126 an expression such as @samp{x[i,j]} will cause a warning, while
3127 @samp{x[(void)i,j]} will not.
3129 This warning is enabled by @option{-Wall}.
3134 All the above @option{-Wunused} options combined.
3136 In order to get a warning about an unused function parameter, you must
3137 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3138 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3140 @item -Wuninitialized
3141 @opindex Wuninitialized
3142 @opindex Wno-uninitialized
3143 Warn if an automatic variable is used without first being initialized or
3144 if a variable may be clobbered by a @code{setjmp} call.
3146 These warnings are possible only in optimizing compilation,
3147 because they require data flow information that is computed only
3148 when optimizing. If you do not specify @option{-O}, you will not get
3149 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
3150 requiring @option{-O}.
3152 If you want to warn about code which uses the uninitialized value of the
3153 variable in its own initializer, use the @option{-Winit-self} option.
3155 These warnings occur for individual uninitialized or clobbered
3156 elements of structure, union or array variables as well as for
3157 variables which are uninitialized or clobbered as a whole. They do
3158 not occur for variables or elements declared @code{volatile}. Because
3159 these warnings depend on optimization, the exact variables or elements
3160 for which there are warnings will depend on the precise optimization
3161 options and version of GCC used.
3163 Note that there may be no warning about a variable that is used only
3164 to compute a value that itself is never used, because such
3165 computations may be deleted by data flow analysis before the warnings
3168 These warnings are made optional because GCC is not smart
3169 enough to see all the reasons why the code might be correct
3170 despite appearing to have an error. Here is one example of how
3191 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3192 always initialized, but GCC doesn't know this. Here is
3193 another common case:
3198 if (change_y) save_y = y, y = new_y;
3200 if (change_y) y = save_y;
3205 This has no bug because @code{save_y} is used only if it is set.
3207 @cindex @code{longjmp} warnings
3208 This option also warns when a non-volatile automatic variable might be
3209 changed by a call to @code{longjmp}. These warnings as well are possible
3210 only in optimizing compilation.
3212 The compiler sees only the calls to @code{setjmp}. It cannot know
3213 where @code{longjmp} will be called; in fact, a signal handler could
3214 call it at any point in the code. As a result, you may get a warning
3215 even when there is in fact no problem because @code{longjmp} cannot
3216 in fact be called at the place which would cause a problem.
3218 Some spurious warnings can be avoided if you declare all the functions
3219 you use that never return as @code{noreturn}. @xref{Function
3222 This warning is enabled by @option{-Wall} or @option{-Wextra} in
3223 optimizing compilations (@option{-O1} and above).
3225 @item -Wunknown-pragmas
3226 @opindex Wunknown-pragmas
3227 @opindex Wno-unknown-pragmas
3228 @cindex warning for unknown pragmas
3229 @cindex unknown pragmas, warning
3230 @cindex pragmas, warning of unknown
3231 Warn when a #pragma directive is encountered which is not understood by
3232 GCC@. If this command line option is used, warnings will even be issued
3233 for unknown pragmas in system header files. This is not the case if
3234 the warnings were only enabled by the @option{-Wall} command line option.
3237 @opindex Wno-pragmas
3239 Do not warn about misuses of pragmas, such as incorrect parameters,
3240 invalid syntax, or conflicts between pragmas. See also
3241 @samp{-Wunknown-pragmas}.
3243 @item -Wstrict-aliasing
3244 @opindex Wstrict-aliasing
3245 @opindex Wno-strict-aliasing
3246 This option is only active when @option{-fstrict-aliasing} is active.
3247 It warns about code which might break the strict aliasing rules that the
3248 compiler is using for optimization. The warning does not catch all
3249 cases, but does attempt to catch the more common pitfalls. It is
3250 included in @option{-Wall}.
3251 It is equivalent to @option{-Wstrict-aliasing=3}
3253 @item -Wstrict-aliasing=n
3254 @opindex Wstrict-aliasing=n
3255 @opindex Wno-strict-aliasing=n
3256 This option is only active when @option{-fstrict-aliasing} is active.
3257 It warns about code which might break the strict aliasing rules that the
3258 compiler is using for optimization.
3259 Higher levels correspond to higher accuracy (fewer false positives).
3260 Higher levels also correspond to more effort, similar to the way -O works.
3261 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3264 Level 1: Most aggressive, quick, least accurate.
3265 Possibly useful when higher levels
3266 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3267 false negatives. However, it has many false positives.
3268 Warns for all pointer conversions between possibly incompatible types,
3269 even if never dereferenced. Runs in the frontend only.
3271 Level 2: Aggressive, quick, not too precise.
3272 May still have many false positives (not as many as level 1 though),
3273 and few false negatives (but possibly more than level 1).
3274 Unlike level 1, it only warns when an address is taken. Warns about
3275 incomplete types. Runs in the frontend only.
3277 Level 3 (default for @option{-Wstrict-aliasing}):
3278 Should have very few false positives and few false
3279 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3280 Takes care of the common punn+dereference pattern in the frontend:
3281 @code{*(int*)&some_float}.
3282 If optimization is enabled, it also runs in the backend, where it deals
3283 with multiple statement cases using flow-sensitive points-to information.
3284 Only warns when the converted pointer is dereferenced.
3285 Does not warn about incomplete types.
3287 @item -Wstrict-overflow
3288 @itemx -Wstrict-overflow=@var{n}
3289 @opindex Wstrict-overflow
3290 @opindex Wno-strict-overflow
3291 This option is only active when @option{-fstrict-overflow} is active.
3292 It warns about cases where the compiler optimizes based on the
3293 assumption that signed overflow does not occur. Note that it does not
3294 warn about all cases where the code might overflow: it only warns
3295 about cases where the compiler implements some optimization. Thus
3296 this warning depends on the optimization level.
3298 An optimization which assumes that signed overflow does not occur is
3299 perfectly safe if the values of the variables involved are such that
3300 overflow never does, in fact, occur. Therefore this warning can
3301 easily give a false positive: a warning about code which is not
3302 actually a problem. To help focus on important issues, several
3303 warning levels are defined. No warnings are issued for the use of
3304 undefined signed overflow when estimating how many iterations a loop
3305 will require, in particular when determining whether a loop will be
3309 @item -Wstrict-overflow=1
3310 Warn about cases which are both questionable and easy to avoid. For
3311 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3312 compiler will simplify this to @code{1}. This level of
3313 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3314 are not, and must be explicitly requested.
3316 @item -Wstrict-overflow=2
3317 Also warn about other cases where a comparison is simplified to a
3318 constant. For example: @code{abs (x) >= 0}. This can only be
3319 simplified when @option{-fstrict-overflow} is in effect, because
3320 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3321 zero. @option{-Wstrict-overflow} (with no level) is the same as
3322 @option{-Wstrict-overflow=2}.
3324 @item -Wstrict-overflow=3
3325 Also warn about other cases where a comparison is simplified. For
3326 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3328 @item -Wstrict-overflow=4
3329 Also warn about other simplifications not covered by the above cases.
3330 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3332 @item -Wstrict-overflow=5
3333 Also warn about cases where the compiler reduces the magnitude of a
3334 constant involved in a comparison. For example: @code{x + 2 > y} will
3335 be simplified to @code{x + 1 >= y}. This is reported only at the
3336 highest warning level because this simplification applies to many
3337 comparisons, so this warning level will give a very large number of
3341 @item -Warray-bounds
3342 @opindex Wno-array-bounds
3343 @opindex Warray-bounds
3344 This option is only active when @option{-ftree-vrp} is active
3345 (default for -O2 and above). It warns about subscripts to arrays
3346 that are always out of bounds. This warning is enabled by @option{-Wall}.
3348 @item -Wno-div-by-zero
3349 @opindex Wno-div-by-zero
3350 @opindex Wdiv-by-zero
3351 Do not warn about compile-time integer division by zero. Floating point
3352 division by zero is not warned about, as it can be a legitimate way of
3353 obtaining infinities and NaNs.
3355 @item -Wsystem-headers
3356 @opindex Wsystem-headers
3357 @opindex Wno-system-headers
3358 @cindex warnings from system headers
3359 @cindex system headers, warnings from
3360 Print warning messages for constructs found in system header files.
3361 Warnings from system headers are normally suppressed, on the assumption
3362 that they usually do not indicate real problems and would only make the
3363 compiler output harder to read. Using this command line option tells
3364 GCC to emit warnings from system headers as if they occurred in user
3365 code. However, note that using @option{-Wall} in conjunction with this
3366 option will @emph{not} warn about unknown pragmas in system
3367 headers---for that, @option{-Wunknown-pragmas} must also be used.
3370 @opindex Wfloat-equal
3371 @opindex Wno-float-equal
3372 Warn if floating point values are used in equality comparisons.
3374 The idea behind this is that sometimes it is convenient (for the
3375 programmer) to consider floating-point values as approximations to
3376 infinitely precise real numbers. If you are doing this, then you need
3377 to compute (by analyzing the code, or in some other way) the maximum or
3378 likely maximum error that the computation introduces, and allow for it
3379 when performing comparisons (and when producing output, but that's a
3380 different problem). In particular, instead of testing for equality, you
3381 would check to see whether the two values have ranges that overlap; and
3382 this is done with the relational operators, so equality comparisons are
3385 @item -Wtraditional @r{(C and Objective-C only)}
3386 @opindex Wtraditional
3387 @opindex Wno-traditional
3388 Warn about certain constructs that behave differently in traditional and
3389 ISO C@. Also warn about ISO C constructs that have no traditional C
3390 equivalent, and/or problematic constructs which should be avoided.
3394 Macro parameters that appear within string literals in the macro body.
3395 In traditional C macro replacement takes place within string literals,
3396 but does not in ISO C@.
3399 In traditional C, some preprocessor directives did not exist.
3400 Traditional preprocessors would only consider a line to be a directive
3401 if the @samp{#} appeared in column 1 on the line. Therefore
3402 @option{-Wtraditional} warns about directives that traditional C
3403 understands but would ignore because the @samp{#} does not appear as the
3404 first character on the line. It also suggests you hide directives like
3405 @samp{#pragma} not understood by traditional C by indenting them. Some
3406 traditional implementations would not recognize @samp{#elif}, so it
3407 suggests avoiding it altogether.
3410 A function-like macro that appears without arguments.
3413 The unary plus operator.
3416 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3417 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3418 constants.) Note, these suffixes appear in macros defined in the system
3419 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3420 Use of these macros in user code might normally lead to spurious
3421 warnings, however GCC's integrated preprocessor has enough context to
3422 avoid warning in these cases.
3425 A function declared external in one block and then used after the end of
3429 A @code{switch} statement has an operand of type @code{long}.
3432 A non-@code{static} function declaration follows a @code{static} one.
3433 This construct is not accepted by some traditional C compilers.
3436 The ISO type of an integer constant has a different width or
3437 signedness from its traditional type. This warning is only issued if
3438 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3439 typically represent bit patterns, are not warned about.
3442 Usage of ISO string concatenation is detected.
3445 Initialization of automatic aggregates.
3448 Identifier conflicts with labels. Traditional C lacks a separate
3449 namespace for labels.
3452 Initialization of unions. If the initializer is zero, the warning is
3453 omitted. This is done under the assumption that the zero initializer in
3454 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3455 initializer warnings and relies on default initialization to zero in the
3459 Conversions by prototypes between fixed/floating point values and vice
3460 versa. The absence of these prototypes when compiling with traditional
3461 C would cause serious problems. This is a subset of the possible
3462 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3465 Use of ISO C style function definitions. This warning intentionally is
3466 @emph{not} issued for prototype declarations or variadic functions
3467 because these ISO C features will appear in your code when using
3468 libiberty's traditional C compatibility macros, @code{PARAMS} and
3469 @code{VPARAMS}. This warning is also bypassed for nested functions
3470 because that feature is already a GCC extension and thus not relevant to
3471 traditional C compatibility.
3474 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3475 @opindex Wtraditional-conversion
3476 @opindex Wno-traditional-conversion
3477 Warn if a prototype causes a type conversion that is different from what
3478 would happen to the same argument in the absence of a prototype. This
3479 includes conversions of fixed point to floating and vice versa, and
3480 conversions changing the width or signedness of a fixed point argument
3481 except when the same as the default promotion.
3483 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3484 @opindex Wdeclaration-after-statement
3485 @opindex Wno-declaration-after-statement
3486 Warn when a declaration is found after a statement in a block. This
3487 construct, known from C++, was introduced with ISO C99 and is by default
3488 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3489 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3494 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3496 @item -Wno-endif-labels
3497 @opindex Wno-endif-labels
3498 @opindex Wendif-labels
3499 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3504 Warn whenever a local variable shadows another local variable, parameter or
3505 global variable or whenever a built-in function is shadowed.
3507 @item -Wlarger-than-@var{len}
3508 @opindex Wlarger-than-@var{len}
3509 Warn whenever an object of larger than @var{len} bytes is defined.
3511 @item -Wunsafe-loop-optimizations
3512 @opindex Wunsafe-loop-optimizations
3513 @opindex Wno-unsafe-loop-optimizations
3514 Warn if the loop cannot be optimized because the compiler could not
3515 assume anything on the bounds of the loop indices. With
3516 @option{-funsafe-loop-optimizations} warn if the compiler made
3519 @item -Wpointer-arith
3520 @opindex Wpointer-arith
3521 @opindex Wno-pointer-arith
3522 Warn about anything that depends on the ``size of'' a function type or
3523 of @code{void}. GNU C assigns these types a size of 1, for
3524 convenience in calculations with @code{void *} pointers and pointers
3525 to functions. In C++, warn also when an arithmetic operation involves
3526 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3529 @opindex Wtype-limits
3530 @opindex Wno-type-limits
3531 Warn if a comparison is always true or always false due to the limited
3532 range of the data type, but do not warn for constant expressions. For
3533 example, warn if an unsigned variable is compared against zero with
3534 @samp{<} or @samp{>=}. This warning is also enabled by
3537 @item -Wbad-function-cast @r{(C and Objective-C only)}
3538 @opindex Wbad-function-cast
3539 @opindex Wno-bad-function-cast
3540 Warn whenever a function call is cast to a non-matching type.
3541 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3543 @item -Wc++-compat @r{(C and Objective-C only)}
3544 Warn about ISO C constructs that are outside of the common subset of
3545 ISO C and ISO C++, e.g.@: request for implicit conversion from
3546 @code{void *} to a pointer to non-@code{void} type.
3548 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3549 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3550 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3551 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3555 @opindex Wno-cast-qual
3556 Warn whenever a pointer is cast so as to remove a type qualifier from
3557 the target type. For example, warn if a @code{const char *} is cast
3558 to an ordinary @code{char *}.
3561 @opindex Wcast-align
3562 @opindex Wno-cast-align
3563 Warn whenever a pointer is cast such that the required alignment of the
3564 target is increased. For example, warn if a @code{char *} is cast to
3565 an @code{int *} on machines where integers can only be accessed at
3566 two- or four-byte boundaries.
3568 @item -Wwrite-strings
3569 @opindex Wwrite-strings
3570 @opindex Wno-write-strings
3571 When compiling C, give string constants the type @code{const
3572 char[@var{length}]} so that
3573 copying the address of one into a non-@code{const} @code{char *}
3574 pointer will get a warning; when compiling C++, warn about the
3575 deprecated conversion from string literals to @code{char *}. This
3576 warning, by default, is enabled for C++ programs.
3577 These warnings will help you find at
3578 compile time code that can try to write into a string constant, but
3579 only if you have been very careful about using @code{const} in
3580 declarations and prototypes. Otherwise, it will just be a nuisance;
3581 this is why we did not make @option{-Wall} request these warnings.
3585 @opindex Wno-clobbered
3586 Warn for variables that might be changed by @samp{longjmp} or
3587 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3590 @opindex Wconversion
3591 @opindex Wno-conversion
3592 Warn for implicit conversions that may alter a value. This includes
3593 conversions between real and integer, like @code{abs (x)} when
3594 @code{x} is @code{double}; conversions between signed and unsigned,
3595 like @code{unsigned ui = -1}; and conversions to smaller types, like
3596 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3597 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3598 changed by the conversion like in @code{abs (2.0)}. Warnings about
3599 conversions between signed and unsigned integers can be disabled by
3600 using @option{-Wno-sign-conversion}.
3602 For C++, also warn for conversions between @code{NULL} and non-pointer
3603 types; confusing overload resolution for user-defined conversions; and
3604 conversions that will never use a type conversion operator:
3605 conversions to @code{void}, the same type, a base class or a reference
3606 to them. Warnings about conversions between signed and unsigned
3607 integers are disabled by default in C++ unless
3608 @option{-Wsign-conversion} is explicitly enabled.
3611 @opindex Wempty-body
3612 @opindex Wno-empty-body
3613 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3614 while} statement. Additionally, in C++, warn when an empty body occurs
3615 in a @samp{while} or @samp{for} statement with no whitespacing before
3616 the semicolon. This warning is also enabled by @option{-Wextra}.
3618 @item -Wsign-compare
3619 @opindex Wsign-compare
3620 @opindex Wno-sign-compare
3621 @cindex warning for comparison of signed and unsigned values
3622 @cindex comparison of signed and unsigned values, warning
3623 @cindex signed and unsigned values, comparison warning
3624 Warn when a comparison between signed and unsigned values could produce
3625 an incorrect result when the signed value is converted to unsigned.
3626 This warning is also enabled by @option{-Wextra}; to get the other warnings
3627 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3629 @item -Wsign-conversion
3630 @opindex Wsign-conversion
3631 @opindex Wno-sign-conversion
3632 Warn for implicit conversions that may change the sign of an integer
3633 value, like assigning a signed integer expression to an unsigned
3634 integer variable. An explicit cast silences the warning. In C, this
3635 option is enabled also by @option{-Wconversion}.
3639 @opindex Wno-address
3640 Warn about suspicious uses of memory addresses. These include using
3641 the address of a function in a conditional expression, such as
3642 @code{void func(void); if (func)}, and comparisons against the memory
3643 address of a string literal, such as @code{if (x == "abc")}. Such
3644 uses typically indicate a programmer error: the address of a function
3645 always evaluates to true, so their use in a conditional usually
3646 indicate that the programmer forgot the parentheses in a function
3647 call; and comparisons against string literals result in unspecified
3648 behavior and are not portable in C, so they usually indicate that the
3649 programmer intended to use @code{strcmp}. This warning is enabled by
3653 @opindex Wlogical-op
3654 @opindex Wno-logical-op
3655 Warn about suspicious uses of logical operators in expressions.
3656 This includes using logical operators in contexts where a
3657 bit-wise operator is likely to be expected.
3659 @item -Waggregate-return
3660 @opindex Waggregate-return
3661 @opindex Wno-aggregate-return
3662 Warn if any functions that return structures or unions are defined or
3663 called. (In languages where you can return an array, this also elicits
3666 @item -Wno-attributes
3667 @opindex Wno-attributes
3668 @opindex Wattributes
3669 Do not warn if an unexpected @code{__attribute__} is used, such as
3670 unrecognized attributes, function attributes applied to variables,
3671 etc. This will not stop errors for incorrect use of supported
3674 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3675 @opindex Wstrict-prototypes
3676 @opindex Wno-strict-prototypes
3677 Warn if a function is declared or defined without specifying the
3678 argument types. (An old-style function definition is permitted without
3679 a warning if preceded by a declaration which specifies the argument
3682 @item -Wold-style-declaration @r{(C and Objective-C only)}
3683 @opindex Wold-style-declaration
3684 @opindex Wno-old-style-declaration
3685 Warn for obsolescent usages, according to the C Standard, in a
3686 declaration. For example, warn if storage-class specifiers like
3687 @code{static} are not the first things in a declaration. This warning
3688 is also enabled by @option{-Wextra}.
3690 @item -Wold-style-definition @r{(C and Objective-C only)}
3691 @opindex Wold-style-definition
3692 @opindex Wno-old-style-definition
3693 Warn if an old-style function definition is used. A warning is given
3694 even if there is a previous prototype.
3696 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3697 @opindex Wmissing-parameter-type
3698 @opindex Wno-missing-parameter-type
3699 A function parameter is declared without a type specifier in K&R-style
3706 This warning is also enabled by @option{-Wextra}.
3708 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3709 @opindex Wmissing-prototypes
3710 @opindex Wno-missing-prototypes
3711 Warn if a global function is defined without a previous prototype
3712 declaration. This warning is issued even if the definition itself
3713 provides a prototype. The aim is to detect global functions that fail
3714 to be declared in header files.
3716 @item -Wmissing-declarations
3717 @opindex Wmissing-declarations
3718 @opindex Wno-missing-declarations
3719 Warn if a global function is defined without a previous declaration.
3720 Do so even if the definition itself provides a prototype.
3721 Use this option to detect global functions that are not declared in
3722 header files. In C++, no warnings are issued for function templates,
3723 or for inline functions, or for functions in anonymous namespaces.
3725 @item -Wmissing-field-initializers
3726 @opindex Wmissing-field-initializers
3727 @opindex Wno-missing-field-initializers
3731 Warn if a structure's initializer has some fields missing. For
3732 example, the following code would cause such a warning, because
3733 @code{x.h} is implicitly zero:
3736 struct s @{ int f, g, h; @};
3737 struct s x = @{ 3, 4 @};
3740 This option does not warn about designated initializers, so the following
3741 modification would not trigger a warning:
3744 struct s @{ int f, g, h; @};
3745 struct s x = @{ .f = 3, .g = 4 @};
3748 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3749 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3751 @item -Wmissing-noreturn
3752 @opindex Wmissing-noreturn
3753 @opindex Wno-missing-noreturn
3754 Warn about functions which might be candidates for attribute @code{noreturn}.
3755 Note these are only possible candidates, not absolute ones. Care should
3756 be taken to manually verify functions actually do not ever return before
3757 adding the @code{noreturn} attribute, otherwise subtle code generation
3758 bugs could be introduced. You will not get a warning for @code{main} in
3759 hosted C environments.
3761 @item -Wmissing-format-attribute
3762 @opindex Wmissing-format-attribute
3763 @opindex Wno-missing-format-attribute
3766 Warn about function pointers which might be candidates for @code{format}
3767 attributes. Note these are only possible candidates, not absolute ones.
3768 GCC will guess that function pointers with @code{format} attributes that
3769 are used in assignment, initialization, parameter passing or return
3770 statements should have a corresponding @code{format} attribute in the
3771 resulting type. I.e.@: the left-hand side of the assignment or
3772 initialization, the type of the parameter variable, or the return type
3773 of the containing function respectively should also have a @code{format}
3774 attribute to avoid the warning.
3776 GCC will also warn about function definitions which might be
3777 candidates for @code{format} attributes. Again, these are only
3778 possible candidates. GCC will guess that @code{format} attributes
3779 might be appropriate for any function that calls a function like
3780 @code{vprintf} or @code{vscanf}, but this might not always be the
3781 case, and some functions for which @code{format} attributes are
3782 appropriate may not be detected.
3784 @item -Wno-multichar
3785 @opindex Wno-multichar
3787 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3788 Usually they indicate a typo in the user's code, as they have
3789 implementation-defined values, and should not be used in portable code.
3791 @item -Wnormalized=<none|id|nfc|nfkc>
3792 @opindex Wnormalized=
3795 @cindex character set, input normalization
3796 In ISO C and ISO C++, two identifiers are different if they are
3797 different sequences of characters. However, sometimes when characters
3798 outside the basic ASCII character set are used, you can have two
3799 different character sequences that look the same. To avoid confusion,
3800 the ISO 10646 standard sets out some @dfn{normalization rules} which
3801 when applied ensure that two sequences that look the same are turned into
3802 the same sequence. GCC can warn you if you are using identifiers which
3803 have not been normalized; this option controls that warning.
3805 There are four levels of warning that GCC supports. The default is
3806 @option{-Wnormalized=nfc}, which warns about any identifier which is
3807 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3808 recommended form for most uses.
3810 Unfortunately, there are some characters which ISO C and ISO C++ allow
3811 in identifiers that when turned into NFC aren't allowable as
3812 identifiers. That is, there's no way to use these symbols in portable
3813 ISO C or C++ and have all your identifiers in NFC.
3814 @option{-Wnormalized=id} suppresses the warning for these characters.
3815 It is hoped that future versions of the standards involved will correct
3816 this, which is why this option is not the default.
3818 You can switch the warning off for all characters by writing
3819 @option{-Wnormalized=none}. You would only want to do this if you
3820 were using some other normalization scheme (like ``D''), because
3821 otherwise you can easily create bugs that are literally impossible to see.
3823 Some characters in ISO 10646 have distinct meanings but look identical
3824 in some fonts or display methodologies, especially once formatting has
3825 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3826 LETTER N'', will display just like a regular @code{n} which has been
3827 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3828 normalization scheme to convert all these into a standard form as
3829 well, and GCC will warn if your code is not in NFKC if you use
3830 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3831 about every identifier that contains the letter O because it might be
3832 confused with the digit 0, and so is not the default, but may be
3833 useful as a local coding convention if the programming environment is
3834 unable to be fixed to display these characters distinctly.
3836 @item -Wno-deprecated-declarations
3837 @opindex Wno-deprecated-declarations
3838 @opindex Wdeprecated-declarations
3839 Do not warn about uses of functions (@pxref{Function Attributes}),
3840 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3841 Attributes}) marked as deprecated by using the @code{deprecated}
3845 @opindex Wno-overflow
3847 Do not warn about compile-time overflow in constant expressions.
3849 @item -Woverride-init @r{(C and Objective-C only)}
3850 @opindex Woverride-init
3851 @opindex Wno-override-init
3855 Warn if an initialized field without side effects is overridden when
3856 using designated initializers (@pxref{Designated Inits, , Designated
3859 This warning is included in @option{-Wextra}. To get other
3860 @option{-Wextra} warnings without this one, use @samp{-Wextra
3861 -Wno-override-init}.
3866 Warn if a structure is given the packed attribute, but the packed
3867 attribute has no effect on the layout or size of the structure.
3868 Such structures may be mis-aligned for little benefit. For
3869 instance, in this code, the variable @code{f.x} in @code{struct bar}
3870 will be misaligned even though @code{struct bar} does not itself
3871 have the packed attribute:
3878 @} __attribute__((packed));
3889 Warn if padding is included in a structure, either to align an element
3890 of the structure or to align the whole structure. Sometimes when this
3891 happens it is possible to rearrange the fields of the structure to
3892 reduce the padding and so make the structure smaller.
3894 @item -Wredundant-decls
3895 @opindex Wredundant-decls
3896 @opindex Wno-redundant-decls
3897 Warn if anything is declared more than once in the same scope, even in
3898 cases where multiple declaration is valid and changes nothing.
3900 @item -Wnested-externs @r{(C and Objective-C only)}
3901 @opindex Wnested-externs
3902 @opindex Wno-nested-externs
3903 Warn if an @code{extern} declaration is encountered within a function.
3905 @item -Wunreachable-code
3906 @opindex Wunreachable-code
3907 @opindex Wno-unreachable-code
3908 Warn if the compiler detects that code will never be executed.
3910 This option is intended to warn when the compiler detects that at
3911 least a whole line of source code will never be executed, because
3912 some condition is never satisfied or because it is after a
3913 procedure that never returns.
3915 It is possible for this option to produce a warning even though there
3916 are circumstances under which part of the affected line can be executed,
3917 so care should be taken when removing apparently-unreachable code.
3919 For instance, when a function is inlined, a warning may mean that the
3920 line is unreachable in only one inlined copy of the function.
3922 This option is not made part of @option{-Wall} because in a debugging
3923 version of a program there is often substantial code which checks
3924 correct functioning of the program and is, hopefully, unreachable
3925 because the program does work. Another common use of unreachable
3926 code is to provide behavior which is selectable at compile-time.
3931 Warn if a function can not be inlined and it was declared as inline.
3932 Even with this option, the compiler will not warn about failures to
3933 inline functions declared in system headers.
3935 The compiler uses a variety of heuristics to determine whether or not
3936 to inline a function. For example, the compiler takes into account
3937 the size of the function being inlined and the amount of inlining
3938 that has already been done in the current function. Therefore,
3939 seemingly insignificant changes in the source program can cause the
3940 warnings produced by @option{-Winline} to appear or disappear.
3942 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
3943 @opindex Wno-invalid-offsetof
3944 @opindex Winvalid-offsetof
3945 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3946 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3947 to a non-POD type is undefined. In existing C++ implementations,
3948 however, @samp{offsetof} typically gives meaningful results even when
3949 applied to certain kinds of non-POD types. (Such as a simple
3950 @samp{struct} that fails to be a POD type only by virtue of having a
3951 constructor.) This flag is for users who are aware that they are
3952 writing nonportable code and who have deliberately chosen to ignore the
3955 The restrictions on @samp{offsetof} may be relaxed in a future version
3956 of the C++ standard.
3958 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
3959 @opindex Wno-int-to-pointer-cast
3960 @opindex Wint-to-pointer-cast
3961 Suppress warnings from casts to pointer type of an integer of a
3964 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
3965 @opindex Wno-pointer-to-int-cast
3966 @opindex Wpointer-to-int-cast
3967 Suppress warnings from casts from a pointer to an integer type of a
3971 @opindex Winvalid-pch
3972 @opindex Wno-invalid-pch
3973 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3974 the search path but can't be used.
3978 @opindex Wno-long-long
3979 Warn if @samp{long long} type is used. This is default. To inhibit
3980 the warning messages, use @option{-Wno-long-long}. Flags
3981 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3982 only when @option{-pedantic} flag is used.
3984 @item -Wvariadic-macros
3985 @opindex Wvariadic-macros
3986 @opindex Wno-variadic-macros
3987 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3988 alternate syntax when in pedantic ISO C99 mode. This is default.
3989 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3994 Warn if variable length array is used in the code.
3995 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
3996 the variable length array.
3998 @item -Wvolatile-register-var
3999 @opindex Wvolatile-register-var
4000 @opindex Wno-volatile-register-var
4001 Warn if a register variable is declared volatile. The volatile
4002 modifier does not inhibit all optimizations that may eliminate reads
4003 and/or writes to register variables.
4005 @item -Wdisabled-optimization
4006 @opindex Wdisabled-optimization
4007 @opindex Wno-disabled-optimization
4008 Warn if a requested optimization pass is disabled. This warning does
4009 not generally indicate that there is anything wrong with your code; it
4010 merely indicates that GCC's optimizers were unable to handle the code
4011 effectively. Often, the problem is that your code is too big or too
4012 complex; GCC will refuse to optimize programs when the optimization
4013 itself is likely to take inordinate amounts of time.
4015 @item -Wpointer-sign @r{(C and Objective-C only)}
4016 @opindex Wpointer-sign
4017 @opindex Wno-pointer-sign
4018 Warn for pointer argument passing or assignment with different signedness.
4019 This option is only supported for C and Objective-C@. It is implied by
4020 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4021 @option{-Wno-pointer-sign}.
4023 @item -Wstack-protector
4024 @opindex Wstack-protector
4025 @opindex Wno-stack-protector
4026 This option is only active when @option{-fstack-protector} is active. It
4027 warns about functions that will not be protected against stack smashing.
4029 @item -Woverlength-strings
4030 @opindex Woverlength-strings
4031 @opindex Wno-overlength-strings
4032 Warn about string constants which are longer than the ``minimum
4033 maximum'' length specified in the C standard. Modern compilers
4034 generally allow string constants which are much longer than the
4035 standard's minimum limit, but very portable programs should avoid
4036 using longer strings.
4038 The limit applies @emph{after} string constant concatenation, and does
4039 not count the trailing NUL@. In C89, the limit was 509 characters; in
4040 C99, it was raised to 4095. C++98 does not specify a normative
4041 minimum maximum, so we do not diagnose overlength strings in C++@.
4043 This option is implied by @option{-pedantic}, and can be disabled with
4044 @option{-Wno-overlength-strings}.
4047 @node Debugging Options
4048 @section Options for Debugging Your Program or GCC
4049 @cindex options, debugging
4050 @cindex debugging information options
4052 GCC has various special options that are used for debugging
4053 either your program or GCC:
4058 Produce debugging information in the operating system's native format
4059 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4062 On most systems that use stabs format, @option{-g} enables use of extra
4063 debugging information that only GDB can use; this extra information
4064 makes debugging work better in GDB but will probably make other debuggers
4066 refuse to read the program. If you want to control for certain whether
4067 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4068 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4070 GCC allows you to use @option{-g} with
4071 @option{-O}. The shortcuts taken by optimized code may occasionally
4072 produce surprising results: some variables you declared may not exist
4073 at all; flow of control may briefly move where you did not expect it;
4074 some statements may not be executed because they compute constant
4075 results or their values were already at hand; some statements may
4076 execute in different places because they were moved out of loops.
4078 Nevertheless it proves possible to debug optimized output. This makes
4079 it reasonable to use the optimizer for programs that might have bugs.
4081 The following options are useful when GCC is generated with the
4082 capability for more than one debugging format.
4086 Produce debugging information for use by GDB@. This means to use the
4087 most expressive format available (DWARF 2, stabs, or the native format
4088 if neither of those are supported), including GDB extensions if at all
4093 Produce debugging information in stabs format (if that is supported),
4094 without GDB extensions. This is the format used by DBX on most BSD
4095 systems. On MIPS, Alpha and System V Release 4 systems this option
4096 produces stabs debugging output which is not understood by DBX or SDB@.
4097 On System V Release 4 systems this option requires the GNU assembler.
4099 @item -feliminate-unused-debug-symbols
4100 @opindex feliminate-unused-debug-symbols
4101 Produce debugging information in stabs format (if that is supported),
4102 for only symbols that are actually used.
4104 @item -femit-class-debug-always
4105 Instead of emitting debugging information for a C++ class in only one
4106 object file, emit it in all object files using the class. This option
4107 should be used only with debuggers that are unable to handle the way GCC
4108 normally emits debugging information for classes because using this
4109 option will increase the size of debugging information by as much as a
4114 Produce debugging information in stabs format (if that is supported),
4115 using GNU extensions understood only by the GNU debugger (GDB)@. The
4116 use of these extensions is likely to make other debuggers crash or
4117 refuse to read the program.
4121 Produce debugging information in COFF format (if that is supported).
4122 This is the format used by SDB on most System V systems prior to
4127 Produce debugging information in XCOFF format (if that is supported).
4128 This is the format used by the DBX debugger on IBM RS/6000 systems.
4132 Produce debugging information in XCOFF format (if that is supported),
4133 using GNU extensions understood only by the GNU debugger (GDB)@. The
4134 use of these extensions is likely to make other debuggers crash or
4135 refuse to read the program, and may cause assemblers other than the GNU
4136 assembler (GAS) to fail with an error.
4140 Produce debugging information in DWARF version 2 format (if that is
4141 supported). This is the format used by DBX on IRIX 6. With this
4142 option, GCC uses features of DWARF version 3 when they are useful;
4143 version 3 is upward compatible with version 2, but may still cause
4144 problems for older debuggers.
4148 Produce debugging information in VMS debug format (if that is
4149 supported). This is the format used by DEBUG on VMS systems.
4152 @itemx -ggdb@var{level}
4153 @itemx -gstabs@var{level}
4154 @itemx -gcoff@var{level}
4155 @itemx -gxcoff@var{level}
4156 @itemx -gvms@var{level}
4157 Request debugging information and also use @var{level} to specify how
4158 much information. The default level is 2.
4160 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4163 Level 1 produces minimal information, enough for making backtraces in
4164 parts of the program that you don't plan to debug. This includes
4165 descriptions of functions and external variables, but no information
4166 about local variables and no line numbers.
4168 Level 3 includes extra information, such as all the macro definitions
4169 present in the program. Some debuggers support macro expansion when
4170 you use @option{-g3}.
4172 @option{-gdwarf-2} does not accept a concatenated debug level, because
4173 GCC used to support an option @option{-gdwarf} that meant to generate
4174 debug information in version 1 of the DWARF format (which is very
4175 different from version 2), and it would have been too confusing. That
4176 debug format is long obsolete, but the option cannot be changed now.
4177 Instead use an additional @option{-g@var{level}} option to change the
4178 debug level for DWARF2.
4180 @item -feliminate-dwarf2-dups
4181 @opindex feliminate-dwarf2-dups
4182 Compress DWARF2 debugging information by eliminating duplicated
4183 information about each symbol. This option only makes sense when
4184 generating DWARF2 debugging information with @option{-gdwarf-2}.
4186 @item -femit-struct-debug-baseonly
4187 Emit debug information for struct-like types
4188 only when the base name of the compilation source file
4189 matches the base name of file in which the struct was defined.
4191 This option substantially reduces the size of debugging information,
4192 but at significant potential loss in type information to the debugger.
4193 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4194 See @option{-femit-struct-debug-detailed} for more detailed control.
4196 This option works only with DWARF 2.
4198 @item -femit-struct-debug-reduced
4199 Emit debug information for struct-like types
4200 only when the base name of the compilation source file
4201 matches the base name of file in which the type was defined,
4202 unless the struct is a template or defined in a system header.
4204 This option significantly reduces the size of debugging information,
4205 with some potential loss in type information to the debugger.
4206 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4207 See @option{-femit-struct-debug-detailed} for more detailed control.
4209 This option works only with DWARF 2.
4211 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4212 Specify the struct-like types
4213 for which the compiler will generate debug information.
4214 The intent is to reduce duplicate struct debug information
4215 between different object files within the same program.
4217 This option is a detailed version of
4218 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4219 which will serve for most needs.
4221 A specification has the syntax
4222 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4224 The optional first word limits the specification to
4225 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4226 A struct type is used directly when it is the type of a variable, member.
4227 Indirect uses arise through pointers to structs.
4228 That is, when use of an incomplete struct would be legal, the use is indirect.
4230 @samp{struct one direct; struct two * indirect;}.
4232 The optional second word limits the specification to
4233 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4234 Generic structs are a bit complicated to explain.
4235 For C++, these are non-explicit specializations of template classes,
4236 or non-template classes within the above.
4237 Other programming languages have generics,
4238 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4240 The third word specifies the source files for those
4241 structs for which the compiler will emit debug information.
4242 The values @samp{none} and @samp{any} have the normal meaning.
4243 The value @samp{base} means that
4244 the base of name of the file in which the type declaration appears
4245 must match the base of the name of the main compilation file.
4246 In practice, this means that
4247 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4248 but types declared in other header will not.
4249 The value @samp{sys} means those types satisfying @samp{base}
4250 or declared in system or compiler headers.
4252 You may need to experiment to determine the best settings for your application.
4254 The default is @samp{-femit-struct-debug-detailed=all}.
4256 This option works only with DWARF 2.
4258 @item -fno-merge-debug-strings
4259 @opindex fmerge-debug-strings
4260 @opindex fno-merge-debug-strings
4261 Direct the linker to merge together strings which are identical in
4262 different object files. This is not supported by all assemblers or
4263 linker. This decreases the size of the debug information in the
4264 output file at the cost of increasing link processing time. This is
4267 @item -fdebug-prefix-map=@var{old}=@var{new}
4268 @opindex fdebug-prefix-map
4269 When compiling files in directory @file{@var{old}}, record debugging
4270 information describing them as in @file{@var{new}} instead.
4272 @cindex @command{prof}
4275 Generate extra code to write profile information suitable for the
4276 analysis program @command{prof}. You must use this option when compiling
4277 the source files you want data about, and you must also use it when
4280 @cindex @command{gprof}
4283 Generate extra code to write profile information suitable for the
4284 analysis program @command{gprof}. You must use this option when compiling
4285 the source files you want data about, and you must also use it when
4290 Makes the compiler print out each function name as it is compiled, and
4291 print some statistics about each pass when it finishes.
4294 @opindex ftime-report
4295 Makes the compiler print some statistics about the time consumed by each
4296 pass when it finishes.
4299 @opindex fmem-report
4300 Makes the compiler print some statistics about permanent memory
4301 allocation when it finishes.
4303 @item -fpre-ipa-mem-report
4304 @opindex fpre-ipa-mem-report
4305 @item -fpost-ipa-mem-report
4306 @opindex fpost-ipa-mem-report
4307 Makes the compiler print some statistics about permanent memory
4308 allocation before or after interprocedural optimization.
4310 @item -fprofile-arcs
4311 @opindex fprofile-arcs
4312 Add code so that program flow @dfn{arcs} are instrumented. During
4313 execution the program records how many times each branch and call is
4314 executed and how many times it is taken or returns. When the compiled
4315 program exits it saves this data to a file called
4316 @file{@var{auxname}.gcda} for each source file. The data may be used for
4317 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4318 test coverage analysis (@option{-ftest-coverage}). Each object file's
4319 @var{auxname} is generated from the name of the output file, if
4320 explicitly specified and it is not the final executable, otherwise it is
4321 the basename of the source file. In both cases any suffix is removed
4322 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4323 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4324 @xref{Cross-profiling}.
4326 @cindex @command{gcov}
4330 This option is used to compile and link code instrumented for coverage
4331 analysis. The option is a synonym for @option{-fprofile-arcs}
4332 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4333 linking). See the documentation for those options for more details.
4338 Compile the source files with @option{-fprofile-arcs} plus optimization
4339 and code generation options. For test coverage analysis, use the
4340 additional @option{-ftest-coverage} option. You do not need to profile
4341 every source file in a program.
4344 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4345 (the latter implies the former).
4348 Run the program on a representative workload to generate the arc profile
4349 information. This may be repeated any number of times. You can run
4350 concurrent instances of your program, and provided that the file system
4351 supports locking, the data files will be correctly updated. Also
4352 @code{fork} calls are detected and correctly handled (double counting
4356 For profile-directed optimizations, compile the source files again with
4357 the same optimization and code generation options plus
4358 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4359 Control Optimization}).
4362 For test coverage analysis, use @command{gcov} to produce human readable
4363 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4364 @command{gcov} documentation for further information.
4368 With @option{-fprofile-arcs}, for each function of your program GCC
4369 creates a program flow graph, then finds a spanning tree for the graph.
4370 Only arcs that are not on the spanning tree have to be instrumented: the
4371 compiler adds code to count the number of times that these arcs are
4372 executed. When an arc is the only exit or only entrance to a block, the
4373 instrumentation code can be added to the block; otherwise, a new basic
4374 block must be created to hold the instrumentation code.
4377 @item -ftest-coverage
4378 @opindex ftest-coverage
4379 Produce a notes file that the @command{gcov} code-coverage utility
4380 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4381 show program coverage. Each source file's note file is called
4382 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4383 above for a description of @var{auxname} and instructions on how to
4384 generate test coverage data. Coverage data will match the source files
4385 more closely, if you do not optimize.
4387 @item -fdbg-cnt-list
4388 @opindex fdbg-cnt-list
4389 Print the name and the counter upperbound for all debug counters.
4391 @item -fdbg-cnt=@var{counter-value-list}
4393 Set the internal debug counter upperbound. @var{counter-value-list}
4394 is a comma-separated list of @var{name}:@var{value} pairs
4395 which sets the upperbound of each debug counter @var{name} to @var{value}.
4396 All debug counters have the initial upperbound of @var{UINT_MAX},
4397 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4398 e.g. With -fdbg-cnt=dce:10,tail_call:0
4399 dbg_cnt(dce) will return true only for first 10 invocations
4400 and dbg_cnt(tail_call) will return false always.
4402 @item -d@var{letters}
4403 @item -fdump-rtl-@var{pass}
4405 Says to make debugging dumps during compilation at times specified by
4406 @var{letters}. This is used for debugging the RTL-based passes of the
4407 compiler. The file names for most of the dumps are made by appending a
4408 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4409 from the name of the output file, if explicitly specified and it is not
4410 an executable, otherwise it is the basename of the source file. These
4411 switches may have different effects when @option{-E} is used for
4414 Most debug dumps can be enabled either passing a letter to the @option{-d}
4415 option, or with a long @option{-fdump-rtl} switch; here are the possible
4416 letters for use in @var{letters} and @var{pass}, and their meanings:
4421 Annotate the assembler output with miscellaneous debugging information.
4424 @itemx -fdump-rtl-bbro
4426 @opindex fdump-rtl-bbro
4427 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4430 @itemx -fdump-rtl-combine
4432 @opindex fdump-rtl-combine
4433 Dump after the RTL instruction combination pass, to the file
4434 @file{@var{file}.129r.combine}.
4437 @itemx -fdump-rtl-ce1
4438 @itemx -fdump-rtl-ce2
4440 @opindex fdump-rtl-ce1
4441 @opindex fdump-rtl-ce2
4442 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4443 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4444 and @option{-fdump-rtl-ce2} enable dumping after the second if
4445 conversion, to the file @file{@var{file}.130r.ce2}.
4448 @itemx -fdump-rtl-btl
4449 @itemx -fdump-rtl-dbr
4451 @opindex fdump-rtl-btl
4452 @opindex fdump-rtl-dbr
4453 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4454 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4455 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4456 scheduling, to @file{@var{file}.36.dbr}.
4460 Dump all macro definitions, at the end of preprocessing, in addition to
4464 @itemx -fdump-rtl-ce3
4466 @opindex fdump-rtl-ce3
4467 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4470 @itemx -fdump-rtl-cfg
4471 @itemx -fdump-rtl-life
4473 @opindex fdump-rtl-cfg
4474 @opindex fdump-rtl-life
4475 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4476 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4477 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4478 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4481 @itemx -fdump-rtl-greg
4483 @opindex fdump-rtl-greg
4484 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4487 @itemx -fdump-rtl-gcse
4488 @itemx -fdump-rtl-bypass
4490 @opindex fdump-rtl-gcse
4491 @opindex fdump-rtl-bypass
4492 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4493 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4494 enable dumping after jump bypassing and control flow optimizations, to
4495 @file{@var{file}.115r.bypass}.
4498 @itemx -fdump-rtl-eh
4500 @opindex fdump-rtl-eh
4501 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4504 @itemx -fdump-rtl-sibling
4506 @opindex fdump-rtl-sibling
4507 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4510 @itemx -fdump-rtl-jump
4512 @opindex fdump-rtl-jump
4513 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4516 @itemx -fdump-rtl-stack
4518 @opindex fdump-rtl-stack
4519 Dump after conversion from GCC's "flat register file" registers to the
4520 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4523 @itemx -fdump-rtl-lreg
4525 @opindex fdump-rtl-lreg
4526 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4529 @itemx -fdump-rtl-loop2
4531 @opindex fdump-rtl-loop2
4532 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4533 loop optimization pass, to @file{@var{file}.119r.loop2},
4534 @file{@var{file}.120r.loop2_init},
4535 @file{@var{file}.121r.loop2_invariant}, and
4536 @file{@var{file}.125r.loop2_done}.
4539 @itemx -fdump-rtl-sms
4541 @opindex fdump-rtl-sms
4542 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4545 @itemx -fdump-rtl-mach
4547 @opindex fdump-rtl-mach
4548 Dump after performing the machine dependent reorganization pass, to
4549 @file{@var{file}.155r.mach} if that pass exists.
4552 @itemx -fdump-rtl-rnreg
4554 @opindex fdump-rtl-rnreg
4555 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4558 @itemx -fdump-rtl-regmove
4560 @opindex fdump-rtl-regmove
4561 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4564 @itemx -fdump-rtl-postreload
4566 @opindex fdump-rtl-postreload
4567 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4570 @itemx -fdump-rtl-expand
4572 @opindex fdump-rtl-expand
4573 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4576 @itemx -fdump-rtl-sched2
4578 @opindex fdump-rtl-sched2
4579 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4582 @itemx -fdump-rtl-cse
4584 @opindex fdump-rtl-cse
4585 Dump after CSE (including the jump optimization that sometimes follows
4586 CSE), to @file{@var{file}.113r.cse}.
4589 @itemx -fdump-rtl-sched1
4591 @opindex fdump-rtl-sched1
4592 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4595 @itemx -fdump-rtl-cse2
4597 @opindex fdump-rtl-cse2
4598 Dump after the second CSE pass (including the jump optimization that
4599 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4602 @itemx -fdump-rtl-tracer
4604 @opindex fdump-rtl-tracer
4605 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4608 @itemx -fdump-rtl-vpt
4609 @itemx -fdump-rtl-vartrack
4611 @opindex fdump-rtl-vpt
4612 @opindex fdump-rtl-vartrack
4613 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4614 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4615 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4616 to @file{@var{file}.154r.vartrack}.
4619 @itemx -fdump-rtl-flow2
4621 @opindex fdump-rtl-flow2
4622 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4625 @itemx -fdump-rtl-peephole2
4627 @opindex fdump-rtl-peephole2
4628 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4631 @itemx -fdump-rtl-web
4633 @opindex fdump-rtl-web
4634 Dump after live range splitting, to @file{@var{file}.126r.web}.
4637 @itemx -fdump-rtl-all
4639 @opindex fdump-rtl-all
4640 Produce all the dumps listed above.
4644 Produce a core dump whenever an error occurs.
4648 Print statistics on memory usage, at the end of the run, to
4653 Annotate the assembler output with a comment indicating which
4654 pattern and alternative was used. The length of each instruction is
4659 Dump the RTL in the assembler output as a comment before each instruction.
4660 Also turns on @option{-dp} annotation.
4664 For each of the other indicated dump files (either with @option{-d} or
4665 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4666 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4670 Just generate RTL for a function instead of compiling it. Usually used
4671 with @samp{r} (@option{-fdump-rtl-expand}).
4675 Dump debugging information during parsing, to standard error.
4679 @opindex fdump-noaddr
4680 When doing debugging dumps (see @option{-d} option above), suppress
4681 address output. This makes it more feasible to use diff on debugging
4682 dumps for compiler invocations with different compiler binaries and/or
4683 different text / bss / data / heap / stack / dso start locations.
4685 @item -fdump-unnumbered
4686 @opindex fdump-unnumbered
4687 When doing debugging dumps (see @option{-d} option above), suppress instruction
4688 numbers and address output. This makes it more feasible to
4689 use diff on debugging dumps for compiler invocations with different
4690 options, in particular with and without @option{-g}.
4692 @item -fdump-translation-unit @r{(C++ only)}
4693 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4694 @opindex fdump-translation-unit
4695 Dump a representation of the tree structure for the entire translation
4696 unit to a file. The file name is made by appending @file{.tu} to the
4697 source file name. If the @samp{-@var{options}} form is used, @var{options}
4698 controls the details of the dump as described for the
4699 @option{-fdump-tree} options.
4701 @item -fdump-class-hierarchy @r{(C++ only)}
4702 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4703 @opindex fdump-class-hierarchy
4704 Dump a representation of each class's hierarchy and virtual function
4705 table layout to a file. The file name is made by appending @file{.class}
4706 to the source file name. If the @samp{-@var{options}} form is used,
4707 @var{options} controls the details of the dump as described for the
4708 @option{-fdump-tree} options.
4710 @item -fdump-ipa-@var{switch}
4712 Control the dumping at various stages of inter-procedural analysis
4713 language tree to a file. The file name is generated by appending a switch
4714 specific suffix to the source file name. The following dumps are possible:
4718 Enables all inter-procedural analysis dumps.
4721 Dumps information about call-graph optimization, unused function removal,
4722 and inlining decisions.
4725 Dump after function inlining.
4729 @item -fdump-tree-@var{switch}
4730 @itemx -fdump-tree-@var{switch}-@var{options}
4732 Control the dumping at various stages of processing the intermediate
4733 language tree to a file. The file name is generated by appending a switch
4734 specific suffix to the source file name. If the @samp{-@var{options}}
4735 form is used, @var{options} is a list of @samp{-} separated options that
4736 control the details of the dump. Not all options are applicable to all
4737 dumps, those which are not meaningful will be ignored. The following
4738 options are available
4742 Print the address of each node. Usually this is not meaningful as it
4743 changes according to the environment and source file. Its primary use
4744 is for tying up a dump file with a debug environment.
4746 Inhibit dumping of members of a scope or body of a function merely
4747 because that scope has been reached. Only dump such items when they
4748 are directly reachable by some other path. When dumping pretty-printed
4749 trees, this option inhibits dumping the bodies of control structures.
4751 Print a raw representation of the tree. By default, trees are
4752 pretty-printed into a C-like representation.
4754 Enable more detailed dumps (not honored by every dump option).
4756 Enable dumping various statistics about the pass (not honored by every dump
4759 Enable showing basic block boundaries (disabled in raw dumps).
4761 Enable showing virtual operands for every statement.
4763 Enable showing line numbers for statements.
4765 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4767 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4770 The following tree dumps are possible:
4774 Dump before any tree based optimization, to @file{@var{file}.original}.
4777 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4780 @opindex fdump-tree-gimple
4781 Dump each function before and after the gimplification pass to a file. The
4782 file name is made by appending @file{.gimple} to the source file name.
4785 @opindex fdump-tree-cfg
4786 Dump the control flow graph of each function to a file. The file name is
4787 made by appending @file{.cfg} to the source file name.
4790 @opindex fdump-tree-vcg
4791 Dump the control flow graph of each function to a file in VCG format. The
4792 file name is made by appending @file{.vcg} to the source file name. Note
4793 that if the file contains more than one function, the generated file cannot
4794 be used directly by VCG@. You will need to cut and paste each function's
4795 graph into its own separate file first.
4798 @opindex fdump-tree-ch
4799 Dump each function after copying loop headers. The file name is made by
4800 appending @file{.ch} to the source file name.
4803 @opindex fdump-tree-ssa
4804 Dump SSA related information to a file. The file name is made by appending
4805 @file{.ssa} to the source file name.
4808 @opindex fdump-tree-salias
4809 Dump structure aliasing variable information to a file. This file name
4810 is made by appending @file{.salias} to the source file name.
4813 @opindex fdump-tree-alias
4814 Dump aliasing information for each function. The file name is made by
4815 appending @file{.alias} to the source file name.
4818 @opindex fdump-tree-ccp
4819 Dump each function after CCP@. The file name is made by appending
4820 @file{.ccp} to the source file name.
4823 @opindex fdump-tree-storeccp
4824 Dump each function after STORE-CCP. The file name is made by appending
4825 @file{.storeccp} to the source file name.
4828 @opindex fdump-tree-pre
4829 Dump trees after partial redundancy elimination. The file name is made
4830 by appending @file{.pre} to the source file name.
4833 @opindex fdump-tree-fre
4834 Dump trees after full redundancy elimination. The file name is made
4835 by appending @file{.fre} to the source file name.
4838 @opindex fdump-tree-copyprop
4839 Dump trees after copy propagation. The file name is made
4840 by appending @file{.copyprop} to the source file name.
4842 @item store_copyprop
4843 @opindex fdump-tree-store_copyprop
4844 Dump trees after store copy-propagation. The file name is made
4845 by appending @file{.store_copyprop} to the source file name.
4848 @opindex fdump-tree-dce
4849 Dump each function after dead code elimination. The file name is made by
4850 appending @file{.dce} to the source file name.
4853 @opindex fdump-tree-mudflap
4854 Dump each function after adding mudflap instrumentation. The file name is
4855 made by appending @file{.mudflap} to the source file name.
4858 @opindex fdump-tree-sra
4859 Dump each function after performing scalar replacement of aggregates. The
4860 file name is made by appending @file{.sra} to the source file name.
4863 @opindex fdump-tree-sink
4864 Dump each function after performing code sinking. The file name is made
4865 by appending @file{.sink} to the source file name.
4868 @opindex fdump-tree-dom
4869 Dump each function after applying dominator tree optimizations. The file
4870 name is made by appending @file{.dom} to the source file name.
4873 @opindex fdump-tree-dse
4874 Dump each function after applying dead store elimination. The file
4875 name is made by appending @file{.dse} to the source file name.
4878 @opindex fdump-tree-phiopt
4879 Dump each function after optimizing PHI nodes into straightline code. The file
4880 name is made by appending @file{.phiopt} to the source file name.
4883 @opindex fdump-tree-forwprop
4884 Dump each function after forward propagating single use variables. The file
4885 name is made by appending @file{.forwprop} to the source file name.
4888 @opindex fdump-tree-copyrename
4889 Dump each function after applying the copy rename optimization. The file
4890 name is made by appending @file{.copyrename} to the source file name.
4893 @opindex fdump-tree-nrv
4894 Dump each function after applying the named return value optimization on
4895 generic trees. The file name is made by appending @file{.nrv} to the source
4899 @opindex fdump-tree-vect
4900 Dump each function after applying vectorization of loops. The file name is
4901 made by appending @file{.vect} to the source file name.
4904 @opindex fdump-tree-vrp
4905 Dump each function after Value Range Propagation (VRP). The file name
4906 is made by appending @file{.vrp} to the source file name.
4909 @opindex fdump-tree-all
4910 Enable all the available tree dumps with the flags provided in this option.
4913 @item -ftree-vectorizer-verbose=@var{n}
4914 @opindex ftree-vectorizer-verbose
4915 This option controls the amount of debugging output the vectorizer prints.
4916 This information is written to standard error, unless
4917 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4918 in which case it is output to the usual dump listing file, @file{.vect}.
4919 For @var{n}=0 no diagnostic information is reported.
4920 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4921 and the total number of loops that got vectorized.
4922 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4923 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4924 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4925 level that @option{-fdump-tree-vect-stats} uses.
4926 Higher verbosity levels mean either more information dumped for each
4927 reported loop, or same amount of information reported for more loops:
4928 If @var{n}=3, alignment related information is added to the reports.
4929 If @var{n}=4, data-references related information (e.g. memory dependences,
4930 memory access-patterns) is added to the reports.
4931 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4932 that did not pass the first analysis phase (i.e. may not be countable, or
4933 may have complicated control-flow).
4934 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4935 For @var{n}=7, all the information the vectorizer generates during its
4936 analysis and transformation is reported. This is the same verbosity level
4937 that @option{-fdump-tree-vect-details} uses.
4939 @item -frandom-seed=@var{string}
4940 @opindex frandom-string
4941 This option provides a seed that GCC uses when it would otherwise use
4942 random numbers. It is used to generate certain symbol names
4943 that have to be different in every compiled file. It is also used to
4944 place unique stamps in coverage data files and the object files that
4945 produce them. You can use the @option{-frandom-seed} option to produce
4946 reproducibly identical object files.
4948 The @var{string} should be different for every file you compile.
4950 @item -fsched-verbose=@var{n}
4951 @opindex fsched-verbose
4952 On targets that use instruction scheduling, this option controls the
4953 amount of debugging output the scheduler prints. This information is
4954 written to standard error, unless @option{-dS} or @option{-dR} is
4955 specified, in which case it is output to the usual dump
4956 listing file, @file{.sched} or @file{.sched2} respectively. However
4957 for @var{n} greater than nine, the output is always printed to standard
4960 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4961 same information as @option{-dRS}. For @var{n} greater than one, it
4962 also output basic block probabilities, detailed ready list information
4963 and unit/insn info. For @var{n} greater than two, it includes RTL
4964 at abort point, control-flow and regions info. And for @var{n} over
4965 four, @option{-fsched-verbose} also includes dependence info.
4969 Store the usual ``temporary'' intermediate files permanently; place them
4970 in the current directory and name them based on the source file. Thus,
4971 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4972 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4973 preprocessed @file{foo.i} output file even though the compiler now
4974 normally uses an integrated preprocessor.
4976 When used in combination with the @option{-x} command line option,
4977 @option{-save-temps} is sensible enough to avoid over writing an
4978 input source file with the same extension as an intermediate file.
4979 The corresponding intermediate file may be obtained by renaming the
4980 source file before using @option{-save-temps}.
4984 Report the CPU time taken by each subprocess in the compilation
4985 sequence. For C source files, this is the compiler proper and assembler
4986 (plus the linker if linking is done). The output looks like this:
4993 The first number on each line is the ``user time'', that is time spent
4994 executing the program itself. The second number is ``system time'',
4995 time spent executing operating system routines on behalf of the program.
4996 Both numbers are in seconds.
4998 @item -fvar-tracking
4999 @opindex fvar-tracking
5000 Run variable tracking pass. It computes where variables are stored at each
5001 position in code. Better debugging information is then generated
5002 (if the debugging information format supports this information).
5004 It is enabled by default when compiling with optimization (@option{-Os},
5005 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
5006 the debug info format supports it.
5008 @item -print-file-name=@var{library}
5009 @opindex print-file-name
5010 Print the full absolute name of the library file @var{library} that
5011 would be used when linking---and don't do anything else. With this
5012 option, GCC does not compile or link anything; it just prints the
5015 @item -print-multi-directory
5016 @opindex print-multi-directory
5017 Print the directory name corresponding to the multilib selected by any
5018 other switches present in the command line. This directory is supposed
5019 to exist in @env{GCC_EXEC_PREFIX}.
5021 @item -print-multi-lib
5022 @opindex print-multi-lib
5023 Print the mapping from multilib directory names to compiler switches
5024 that enable them. The directory name is separated from the switches by
5025 @samp{;}, and each switch starts with an @samp{@@} instead of the
5026 @samp{-}, without spaces between multiple switches. This is supposed to
5027 ease shell-processing.
5029 @item -print-prog-name=@var{program}
5030 @opindex print-prog-name
5031 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5033 @item -print-libgcc-file-name
5034 @opindex print-libgcc-file-name
5035 Same as @option{-print-file-name=libgcc.a}.
5037 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5038 but you do want to link with @file{libgcc.a}. You can do
5041 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5044 @item -print-search-dirs
5045 @opindex print-search-dirs
5046 Print the name of the configured installation directory and a list of
5047 program and library directories @command{gcc} will search---and don't do anything else.
5049 This is useful when @command{gcc} prints the error message
5050 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5051 To resolve this you either need to put @file{cpp0} and the other compiler
5052 components where @command{gcc} expects to find them, or you can set the environment
5053 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5054 Don't forget the trailing @samp{/}.
5055 @xref{Environment Variables}.
5057 @item -print-sysroot-headers-suffix
5058 @opindex print-sysroot-headers-suffix
5059 Print the suffix added to the target sysroot when searching for
5060 headers, or give an error if the compiler is not configured with such
5061 a suffix---and don't do anything else.
5064 @opindex dumpmachine
5065 Print the compiler's target machine (for example,
5066 @samp{i686-pc-linux-gnu})---and don't do anything else.
5069 @opindex dumpversion
5070 Print the compiler version (for example, @samp{3.0})---and don't do
5075 Print the compiler's built-in specs---and don't do anything else. (This
5076 is used when GCC itself is being built.) @xref{Spec Files}.
5078 @item -feliminate-unused-debug-types
5079 @opindex feliminate-unused-debug-types
5080 Normally, when producing DWARF2 output, GCC will emit debugging
5081 information for all types declared in a compilation
5082 unit, regardless of whether or not they are actually used
5083 in that compilation unit. Sometimes this is useful, such as
5084 if, in the debugger, you want to cast a value to a type that is
5085 not actually used in your program (but is declared). More often,
5086 however, this results in a significant amount of wasted space.
5087 With this option, GCC will avoid producing debug symbol output
5088 for types that are nowhere used in the source file being compiled.
5091 @node Optimize Options
5092 @section Options That Control Optimization
5093 @cindex optimize options
5094 @cindex options, optimization
5096 These options control various sorts of optimizations.
5098 Without any optimization option, the compiler's goal is to reduce the
5099 cost of compilation and to make debugging produce the expected
5100 results. Statements are independent: if you stop the program with a
5101 breakpoint between statements, you can then assign a new value to any
5102 variable or change the program counter to any other statement in the
5103 function and get exactly the results you would expect from the source
5106 Turning on optimization flags makes the compiler attempt to improve
5107 the performance and/or code size at the expense of compilation time
5108 and possibly the ability to debug the program.
5110 The compiler performs optimization based on the knowledge it has of
5111 the program. Optimization levels @option{-O} and above, in
5112 particular, enable @emph{unit-at-a-time} mode, which allows the
5113 compiler to consider information gained from later functions in
5114 the file when compiling a function. Compiling multiple files at
5115 once to a single output file in @emph{unit-at-a-time} mode allows
5116 the compiler to use information gained from all of the files when
5117 compiling each of them.
5119 Not all optimizations are controlled directly by a flag. Only
5120 optimizations that have a flag are listed.
5127 Optimize. Optimizing compilation takes somewhat more time, and a lot
5128 more memory for a large function.
5130 With @option{-O}, the compiler tries to reduce code size and execution
5131 time, without performing any optimizations that take a great deal of
5134 @option{-O} turns on the following optimization flags:
5137 -fcprop-registers @gol
5140 -fdelayed-branch @gol
5142 -fguess-branch-probability @gol
5143 -fif-conversion2 @gol
5144 -fif-conversion @gol
5145 -finline-small-functions @gol
5146 -fipa-pure-const @gol
5147 -fipa-reference @gol
5149 -fsplit-wide-types @gol
5152 -ftree-copyrename @gol
5154 -ftree-dominator-opts @gol
5161 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5162 where doing so does not interfere with debugging.
5166 Optimize even more. GCC performs nearly all supported optimizations
5167 that do not involve a space-speed tradeoff. The compiler does not
5168 perform loop unrolling or function inlining when you specify @option{-O2}.
5169 As compared to @option{-O}, this option increases both compilation time
5170 and the performance of the generated code.
5172 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5173 also turns on the following optimization flags:
5174 @gccoptlist{-fthread-jumps @gol
5175 -falign-functions -falign-jumps @gol
5176 -falign-loops -falign-labels @gol
5179 -fcse-follow-jumps -fcse-skip-blocks @gol
5180 -fdelete-null-pointer-checks @gol
5181 -fexpensive-optimizations @gol
5182 -fgcse -fgcse-lm @gol
5183 -foptimize-sibling-calls @gol
5186 -freorder-blocks -freorder-functions @gol
5187 -frerun-cse-after-loop @gol
5188 -fsched-interblock -fsched-spec @gol
5189 -fschedule-insns -fschedule-insns2 @gol
5190 -fstrict-aliasing -fstrict-overflow @gol
5194 Please note the warning under @option{-fgcse} about
5195 invoking @option{-O2} on programs that use computed gotos.
5199 Optimize yet more. @option{-O3} turns on all optimizations specified by
5200 @option{-O2} and also turns on the @option{-finline-functions},
5201 @option{-funswitch-loops}, @option{-fpredictive-commoning} and
5202 @option{-fgcse-after-reload} options.
5206 Reduce compilation time and make debugging produce the expected
5207 results. This is the default.
5211 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5212 do not typically increase code size. It also performs further
5213 optimizations designed to reduce code size.
5215 @option{-Os} disables the following optimization flags:
5216 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5217 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5218 -fprefetch-loop-arrays -ftree-vect-loop-version}
5220 If you use multiple @option{-O} options, with or without level numbers,
5221 the last such option is the one that is effective.
5224 Options of the form @option{-f@var{flag}} specify machine-independent
5225 flags. Most flags have both positive and negative forms; the negative
5226 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5227 below, only one of the forms is listed---the one you typically will
5228 use. You can figure out the other form by either removing @samp{no-}
5231 The following options control specific optimizations. They are either
5232 activated by @option{-O} options or are related to ones that are. You
5233 can use the following flags in the rare cases when ``fine-tuning'' of
5234 optimizations to be performed is desired.
5237 @item -fno-default-inline
5238 @opindex fno-default-inline
5239 Do not make member functions inline by default merely because they are
5240 defined inside the class scope (C++ only). Otherwise, when you specify
5241 @w{@option{-O}}, member functions defined inside class scope are compiled
5242 inline by default; i.e., you don't need to add @samp{inline} in front of
5243 the member function name.
5245 @item -fno-defer-pop
5246 @opindex fno-defer-pop
5247 Always pop the arguments to each function call as soon as that function
5248 returns. For machines which must pop arguments after a function call,
5249 the compiler normally lets arguments accumulate on the stack for several
5250 function calls and pops them all at once.
5252 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5254 @item -fforward-propagate
5255 @opindex fforward-propagate
5256 Perform a forward propagation pass on RTL. The pass tries to combine two
5257 instructions and checks if the result can be simplified. If loop unrolling
5258 is active, two passes are performed and the second is scheduled after
5261 This option is enabled by default at optimization levels @option{-O2},
5262 @option{-O3}, @option{-Os}.
5264 @item -fomit-frame-pointer
5265 @opindex fomit-frame-pointer
5266 Don't keep the frame pointer in a register for functions that
5267 don't need one. This avoids the instructions to save, set up and
5268 restore frame pointers; it also makes an extra register available
5269 in many functions. @strong{It also makes debugging impossible on
5272 On some machines, such as the VAX, this flag has no effect, because
5273 the standard calling sequence automatically handles the frame pointer
5274 and nothing is saved by pretending it doesn't exist. The
5275 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5276 whether a target machine supports this flag. @xref{Registers,,Register
5277 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5279 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5281 @item -foptimize-sibling-calls
5282 @opindex foptimize-sibling-calls
5283 Optimize sibling and tail recursive calls.
5285 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5289 Don't pay attention to the @code{inline} keyword. Normally this option
5290 is used to keep the compiler from expanding any functions inline.
5291 Note that if you are not optimizing, no functions can be expanded inline.
5293 @item -finline-small-functions
5294 @opindex finline-small-functions
5295 Integrate functions into their callers when their body is smaller than expected
5296 function call code (so overall size of program gets smaller). The compiler
5297 heuristically decides which functions are simple enough to be worth integrating
5300 Enabled at level @option{-O2}.
5302 @item -finline-functions
5303 @opindex finline-functions
5304 Integrate all simple functions into their callers. The compiler
5305 heuristically decides which functions are simple enough to be worth
5306 integrating in this way.
5308 If all calls to a given function are integrated, and the function is
5309 declared @code{static}, then the function is normally not output as
5310 assembler code in its own right.
5312 Enabled at level @option{-O3}.
5314 @item -finline-functions-called-once
5315 @opindex finline-functions-called-once
5316 Consider all @code{static} functions called once for inlining into their
5317 caller even if they are not marked @code{inline}. If a call to a given
5318 function is integrated, then the function is not output as assembler code
5321 Enabled if @option{-funit-at-a-time} is enabled.
5323 @item -fearly-inlining
5324 @opindex fearly-inlining
5325 Inline functions marked by @code{always_inline} and functions whose body seems
5326 smaller than the function call overhead early before doing
5327 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5328 makes profiling significantly cheaper and usually inlining faster on programs
5329 having large chains of nested wrapper functions.
5333 @item -finline-limit=@var{n}
5334 @opindex finline-limit
5335 By default, GCC limits the size of functions that can be inlined. This flag
5336 allows coarse control of this limit. @var{n} is the size of functions that
5337 can be inlined in number of pseudo instructions.
5339 Inlining is actually controlled by a number of parameters, which may be
5340 specified individually by using @option{--param @var{name}=@var{value}}.
5341 The @option{-finline-limit=@var{n}} option sets some of these parameters
5345 @item max-inline-insns-single
5346 is set to @var{n}/2.
5347 @item max-inline-insns-auto
5348 is set to @var{n}/2.
5351 See below for a documentation of the individual
5352 parameters controlling inlining and for the defaults of these parameters.
5354 @emph{Note:} there may be no value to @option{-finline-limit} that results
5355 in default behavior.
5357 @emph{Note:} pseudo instruction represents, in this particular context, an
5358 abstract measurement of function's size. In no way does it represent a count
5359 of assembly instructions and as such its exact meaning might change from one
5360 release to an another.
5362 @item -fkeep-inline-functions
5363 @opindex fkeep-inline-functions
5364 In C, emit @code{static} functions that are declared @code{inline}
5365 into the object file, even if the function has been inlined into all
5366 of its callers. This switch does not affect functions using the
5367 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5368 inline functions into the object file.
5370 @item -fkeep-static-consts
5371 @opindex fkeep-static-consts
5372 Emit variables declared @code{static const} when optimization isn't turned
5373 on, even if the variables aren't referenced.
5375 GCC enables this option by default. If you want to force the compiler to
5376 check if the variable was referenced, regardless of whether or not
5377 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5379 @item -fmerge-constants
5380 @opindex fmerge-constants
5381 Attempt to merge identical constants (string constants and floating point
5382 constants) across compilation units.
5384 This option is the default for optimized compilation if the assembler and
5385 linker support it. Use @option{-fno-merge-constants} to inhibit this
5388 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5390 @item -fmerge-all-constants
5391 @opindex fmerge-all-constants
5392 Attempt to merge identical constants and identical variables.
5394 This option implies @option{-fmerge-constants}. In addition to
5395 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5396 arrays or initialized constant variables with integral or floating point
5397 types. Languages like C or C++ require each non-automatic variable to
5398 have distinct location, so using this option will result in non-conforming
5401 @item -fmodulo-sched
5402 @opindex fmodulo-sched
5403 Perform swing modulo scheduling immediately before the first scheduling
5404 pass. This pass looks at innermost loops and reorders their
5405 instructions by overlapping different iterations.
5407 @item -fmodulo-sched-allow-regmoves
5408 @opindex fmodulo-sched-allow-regmoves
5409 Perform more aggressive SMS based modulo scheduling with register moves
5410 allowed. By setting this flag certain anti-dependences edges will be
5411 deleted which will trigger the generation of reg-moves based on the
5412 life-range analysis. This option is effective only with
5413 @option{-fmodulo-sched} enabled.
5415 @item -fno-branch-count-reg
5416 @opindex fno-branch-count-reg
5417 Do not use ``decrement and branch'' instructions on a count register,
5418 but instead generate a sequence of instructions that decrement a
5419 register, compare it against zero, then branch based upon the result.
5420 This option is only meaningful on architectures that support such
5421 instructions, which include x86, PowerPC, IA-64 and S/390.
5423 The default is @option{-fbranch-count-reg}.
5425 @item -fno-function-cse
5426 @opindex fno-function-cse
5427 Do not put function addresses in registers; make each instruction that
5428 calls a constant function contain the function's address explicitly.
5430 This option results in less efficient code, but some strange hacks
5431 that alter the assembler output may be confused by the optimizations
5432 performed when this option is not used.
5434 The default is @option{-ffunction-cse}
5436 @item -fno-zero-initialized-in-bss
5437 @opindex fno-zero-initialized-in-bss
5438 If the target supports a BSS section, GCC by default puts variables that
5439 are initialized to zero into BSS@. This can save space in the resulting
5442 This option turns off this behavior because some programs explicitly
5443 rely on variables going to the data section. E.g., so that the
5444 resulting executable can find the beginning of that section and/or make
5445 assumptions based on that.
5447 The default is @option{-fzero-initialized-in-bss}.
5449 @item -fmudflap -fmudflapth -fmudflapir
5453 @cindex bounds checking
5455 For front-ends that support it (C and C++), instrument all risky
5456 pointer/array dereferencing operations, some standard library
5457 string/heap functions, and some other associated constructs with
5458 range/validity tests. Modules so instrumented should be immune to
5459 buffer overflows, invalid heap use, and some other classes of C/C++
5460 programming errors. The instrumentation relies on a separate runtime
5461 library (@file{libmudflap}), which will be linked into a program if
5462 @option{-fmudflap} is given at link time. Run-time behavior of the
5463 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5464 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5467 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5468 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5469 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5470 instrumentation should ignore pointer reads. This produces less
5471 instrumentation (and therefore faster execution) and still provides
5472 some protection against outright memory corrupting writes, but allows
5473 erroneously read data to propagate within a program.
5475 @item -fthread-jumps
5476 @opindex fthread-jumps
5477 Perform optimizations where we check to see if a jump branches to a
5478 location where another comparison subsumed by the first is found. If
5479 so, the first branch is redirected to either the destination of the
5480 second branch or a point immediately following it, depending on whether
5481 the condition is known to be true or false.
5483 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5485 @item -fsplit-wide-types
5486 @opindex fsplit-wide-types
5487 When using a type that occupies multiple registers, such as @code{long
5488 long} on a 32-bit system, split the registers apart and allocate them
5489 independently. This normally generates better code for those types,
5490 but may make debugging more difficult.
5492 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5495 @item -fcse-follow-jumps
5496 @opindex fcse-follow-jumps
5497 In common subexpression elimination (CSE), scan through jump instructions
5498 when the target of the jump is not reached by any other path. For
5499 example, when CSE encounters an @code{if} statement with an
5500 @code{else} clause, CSE will follow the jump when the condition
5503 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5505 @item -fcse-skip-blocks
5506 @opindex fcse-skip-blocks
5507 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5508 follow jumps which conditionally skip over blocks. When CSE
5509 encounters a simple @code{if} statement with no else clause,
5510 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5511 body of the @code{if}.
5513 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5515 @item -frerun-cse-after-loop
5516 @opindex frerun-cse-after-loop
5517 Re-run common subexpression elimination after loop optimizations has been
5520 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5524 Perform a global common subexpression elimination pass.
5525 This pass also performs global constant and copy propagation.
5527 @emph{Note:} When compiling a program using computed gotos, a GCC
5528 extension, you may get better runtime performance if you disable
5529 the global common subexpression elimination pass by adding
5530 @option{-fno-gcse} to the command line.
5532 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5536 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5537 attempt to move loads which are only killed by stores into themselves. This
5538 allows a loop containing a load/store sequence to be changed to a load outside
5539 the loop, and a copy/store within the loop.
5541 Enabled by default when gcse is enabled.
5545 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5546 global common subexpression elimination. This pass will attempt to move
5547 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5548 loops containing a load/store sequence can be changed to a load before
5549 the loop and a store after the loop.
5551 Not enabled at any optimization level.
5555 When @option{-fgcse-las} is enabled, the global common subexpression
5556 elimination pass eliminates redundant loads that come after stores to the
5557 same memory location (both partial and full redundancies).
5559 Not enabled at any optimization level.
5561 @item -fgcse-after-reload
5562 @opindex fgcse-after-reload
5563 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5564 pass is performed after reload. The purpose of this pass is to cleanup
5567 @item -funsafe-loop-optimizations
5568 @opindex funsafe-loop-optimizations
5569 If given, the loop optimizer will assume that loop indices do not
5570 overflow, and that the loops with nontrivial exit condition are not
5571 infinite. This enables a wider range of loop optimizations even if
5572 the loop optimizer itself cannot prove that these assumptions are valid.
5573 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5574 if it finds this kind of loop.
5576 @item -fcrossjumping
5577 @opindex fcrossjumping
5578 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5579 resulting code may or may not perform better than without cross-jumping.
5581 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5583 @item -fauto-inc-dec
5584 @opindex fauto-inc-dec
5585 Combine increments or decrements of addresses with memory accesses.
5586 This pass is always skipped on architectures that do not have
5587 instructions to support this. Enabled by default at @option{-O} and
5588 higher on architectures that support this.
5592 Perform dead code elimination (DCE) on RTL.
5593 Enabled by default at @option{-O} and higher.
5597 Perform dead store elimination (DSE) on RTL.
5598 Enabled by default at @option{-O} and higher.
5600 @item -fif-conversion
5601 @opindex fif-conversion
5602 Attempt to transform conditional jumps into branch-less equivalents. This
5603 include use of conditional moves, min, max, set flags and abs instructions, and
5604 some tricks doable by standard arithmetics. The use of conditional execution
5605 on chips where it is available is controlled by @code{if-conversion2}.
5607 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5609 @item -fif-conversion2
5610 @opindex fif-conversion2
5611 Use conditional execution (where available) to transform conditional jumps into
5612 branch-less equivalents.
5614 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5616 @item -fdelete-null-pointer-checks
5617 @opindex fdelete-null-pointer-checks
5618 Use global dataflow analysis to identify and eliminate useless checks
5619 for null pointers. The compiler assumes that dereferencing a null
5620 pointer would have halted the program. If a pointer is checked after
5621 it has already been dereferenced, it cannot be null.
5623 In some environments, this assumption is not true, and programs can
5624 safely dereference null pointers. Use
5625 @option{-fno-delete-null-pointer-checks} to disable this optimization
5626 for programs which depend on that behavior.
5628 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5630 @item -fexpensive-optimizations
5631 @opindex fexpensive-optimizations
5632 Perform a number of minor optimizations that are relatively expensive.
5634 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5636 @item -foptimize-register-move
5638 @opindex foptimize-register-move
5640 Attempt to reassign register numbers in move instructions and as
5641 operands of other simple instructions in order to maximize the amount of
5642 register tying. This is especially helpful on machines with two-operand
5645 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5648 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5650 @item -fdelayed-branch
5651 @opindex fdelayed-branch
5652 If supported for the target machine, attempt to reorder instructions
5653 to exploit instruction slots available after delayed branch
5656 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5658 @item -fschedule-insns
5659 @opindex fschedule-insns
5660 If supported for the target machine, attempt to reorder instructions to
5661 eliminate execution stalls due to required data being unavailable. This
5662 helps machines that have slow floating point or memory load instructions
5663 by allowing other instructions to be issued until the result of the load
5664 or floating point instruction is required.
5666 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5668 @item -fschedule-insns2
5669 @opindex fschedule-insns2
5670 Similar to @option{-fschedule-insns}, but requests an additional pass of
5671 instruction scheduling after register allocation has been done. This is
5672 especially useful on machines with a relatively small number of
5673 registers and where memory load instructions take more than one cycle.
5675 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5677 @item -fno-sched-interblock
5678 @opindex fno-sched-interblock
5679 Don't schedule instructions across basic blocks. This is normally
5680 enabled by default when scheduling before register allocation, i.e.@:
5681 with @option{-fschedule-insns} or at @option{-O2} or higher.
5683 @item -fno-sched-spec
5684 @opindex fno-sched-spec
5685 Don't allow speculative motion of non-load instructions. This is normally
5686 enabled by default when scheduling before register allocation, i.e.@:
5687 with @option{-fschedule-insns} or at @option{-O2} or higher.
5689 @item -fsched-spec-load
5690 @opindex fsched-spec-load
5691 Allow speculative motion of some load instructions. This only makes
5692 sense when scheduling before register allocation, i.e.@: with
5693 @option{-fschedule-insns} or at @option{-O2} or higher.
5695 @item -fsched-spec-load-dangerous
5696 @opindex fsched-spec-load-dangerous
5697 Allow speculative motion of more load instructions. This only makes
5698 sense when scheduling before register allocation, i.e.@: with
5699 @option{-fschedule-insns} or at @option{-O2} or higher.
5701 @item -fsched-stalled-insns
5702 @itemx -fsched-stalled-insns=@var{n}
5703 @opindex fsched-stalled-insns
5704 Define how many insns (if any) can be moved prematurely from the queue
5705 of stalled insns into the ready list, during the second scheduling pass.
5706 @option{-fno-sched-stalled-insns} means that no insns will be moved
5707 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5708 on how many queued insns can be moved prematurely.
5709 @option{-fsched-stalled-insns} without a value is equivalent to
5710 @option{-fsched-stalled-insns=1}.
5712 @item -fsched-stalled-insns-dep
5713 @itemx -fsched-stalled-insns-dep=@var{n}
5714 @opindex fsched-stalled-insns-dep
5715 Define how many insn groups (cycles) will be examined for a dependency
5716 on a stalled insn that is candidate for premature removal from the queue
5717 of stalled insns. This has an effect only during the second scheduling pass,
5718 and only if @option{-fsched-stalled-insns} is used.
5719 @option{-fno-sched-stalled-insns-dep} is equivalent to
5720 @option{-fsched-stalled-insns-dep=0}.
5721 @option{-fsched-stalled-insns-dep} without a value is equivalent to
5722 @option{-fsched-stalled-insns-dep=1}.
5724 @item -fsched2-use-superblocks
5725 @opindex fsched2-use-superblocks
5726 When scheduling after register allocation, do use superblock scheduling
5727 algorithm. Superblock scheduling allows motion across basic block boundaries
5728 resulting on faster schedules. This option is experimental, as not all machine
5729 descriptions used by GCC model the CPU closely enough to avoid unreliable
5730 results from the algorithm.
5732 This only makes sense when scheduling after register allocation, i.e.@: with
5733 @option{-fschedule-insns2} or at @option{-O2} or higher.
5735 @item -fsched2-use-traces
5736 @opindex fsched2-use-traces
5737 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5738 allocation and additionally perform code duplication in order to increase the
5739 size of superblocks using tracer pass. See @option{-ftracer} for details on
5742 This mode should produce faster but significantly longer programs. Also
5743 without @option{-fbranch-probabilities} the traces constructed may not
5744 match the reality and hurt the performance. This only makes
5745 sense when scheduling after register allocation, i.e.@: with
5746 @option{-fschedule-insns2} or at @option{-O2} or higher.
5750 Eliminate redundant sign extension instructions and move the non-redundant
5751 ones to optimal placement using lazy code motion (LCM).
5753 @item -freschedule-modulo-scheduled-loops
5754 @opindex freschedule-modulo-scheduled-loops
5755 The modulo scheduling comes before the traditional scheduling, if a loop
5756 was modulo scheduled we may want to prevent the later scheduling passes
5757 from changing its schedule, we use this option to control that.
5759 @item -fcaller-saves
5760 @opindex fcaller-saves
5761 Enable values to be allocated in registers that will be clobbered by
5762 function calls, by emitting extra instructions to save and restore the
5763 registers around such calls. Such allocation is done only when it
5764 seems to result in better code than would otherwise be produced.
5766 This option is always enabled by default on certain machines, usually
5767 those which have no call-preserved registers to use instead.
5769 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5771 @item -ftree-reassoc
5772 @opindex ftree-reassoc
5773 Perform reassociation on trees. This flag is enabled by default
5774 at @option{-O} and higher.
5778 Perform partial redundancy elimination (PRE) on trees. This flag is
5779 enabled by default at @option{-O2} and @option{-O3}.
5783 Perform full redundancy elimination (FRE) on trees. The difference
5784 between FRE and PRE is that FRE only considers expressions
5785 that are computed on all paths leading to the redundant computation.
5786 This analysis is faster than PRE, though it exposes fewer redundancies.
5787 This flag is enabled by default at @option{-O} and higher.
5789 @item -ftree-copy-prop
5790 @opindex ftree-copy-prop
5791 Perform copy propagation on trees. This pass eliminates unnecessary
5792 copy operations. This flag is enabled by default at @option{-O} and
5796 @opindex ftree-salias
5797 Perform structural alias analysis on trees. This flag
5798 is enabled by default at @option{-O} and higher.
5800 @item -fipa-pure-const
5801 @opindex fipa-pure-const
5802 Discover which functions are pure or constant.
5803 Enabled by default at @option{-O} and higher.
5805 @item -fipa-reference
5806 @opindex fipa-reference
5807 Discover which static variables do not escape cannot escape the
5809 Enabled by default at @option{-O} and higher.
5811 @item -fipa-struct-reorg
5812 @opindex fipa-struct-reorg
5813 Perform structure reorganization optimization, that change C-like structures
5814 layout in order to better utilize spatial locality. This transformation is
5815 affective for programs containing arrays of structures. Available in two
5816 compilation modes: profile-based (enabled with @option{-fprofile-generate})
5817 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
5818 to provide the safety of this transformation. It works only in whole program
5819 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
5820 enabled. Structures considered @samp{cold} by this transformation are not
5821 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
5823 With this flag, the program debug info reflects a new structure layout.
5827 Perform interprocedural pointer analysis.
5831 Perform interprocedural constant propagation.
5832 This optimization analyzes the program to determine when values passed
5833 to functions are constants and then optimizes accordingly.
5834 This optimization can substantially increase performance
5835 if the application has constants passed to functions, but
5836 because this optimization can create multiple copies of functions,
5837 it may significantly increase code size.
5839 @item -fipa-matrix-reorg
5840 @opindex fipa-matrix-reorg
5841 Perform matrix flattening and transposing.
5842 Matrix flattening tries to replace a m-dimensional matrix
5843 with its equivalent n-dimensional matrix, where n < m.
5844 This reduces the level of indirection needed for accessing the elements
5845 of the matrix. The second optimization is matrix transposing that
5846 attemps to change the order of the matrix's dimensions in order to
5847 improve cache locality.
5848 Both optimizations need fwhole-program flag.
5849 Transposing is enabled only if profiling information is avaliable.
5854 Perform forward store motion on trees. This flag is
5855 enabled by default at @option{-O} and higher.
5859 Perform sparse conditional constant propagation (CCP) on trees. This
5860 pass only operates on local scalar variables and is enabled by default
5861 at @option{-O} and higher.
5863 @item -ftree-store-ccp
5864 @opindex ftree-store-ccp
5865 Perform sparse conditional constant propagation (CCP) on trees. This
5866 pass operates on both local scalar variables and memory stores and
5867 loads (global variables, structures, arrays, etc). This flag is
5868 enabled by default at @option{-O2} and higher.
5872 Perform dead code elimination (DCE) on trees. This flag is enabled by
5873 default at @option{-O} and higher.
5875 @item -ftree-dominator-opts
5876 @opindex ftree-dominator-opts
5877 Perform a variety of simple scalar cleanups (constant/copy
5878 propagation, redundancy elimination, range propagation and expression
5879 simplification) based on a dominator tree traversal. This also
5880 performs jump threading (to reduce jumps to jumps). This flag is
5881 enabled by default at @option{-O} and higher.
5885 Perform dead store elimination (DSE) on trees. A dead store is a store into
5886 a memory location which will later be overwritten by another store without
5887 any intervening loads. In this case the earlier store can be deleted. This
5888 flag is enabled by default at @option{-O} and higher.
5892 Perform loop header copying on trees. This is beneficial since it increases
5893 effectiveness of code motion optimizations. It also saves one jump. This flag
5894 is enabled by default at @option{-O} and higher. It is not enabled
5895 for @option{-Os}, since it usually increases code size.
5897 @item -ftree-loop-optimize
5898 @opindex ftree-loop-optimize
5899 Perform loop optimizations on trees. This flag is enabled by default
5900 at @option{-O} and higher.
5902 @item -ftree-loop-linear
5903 @opindex ftree-loop-linear
5904 Perform linear loop transformations on tree. This flag can improve cache
5905 performance and allow further loop optimizations to take place.
5907 @item -fcheck-data-deps
5908 @opindex fcheck-data-deps
5909 Compare the results of several data dependence analyzers. This option
5910 is used for debugging the data dependence analyzers.
5912 @item -ftree-loop-im
5913 @opindex ftree-loop-im
5914 Perform loop invariant motion on trees. This pass moves only invariants that
5915 would be hard to handle at RTL level (function calls, operations that expand to
5916 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5917 operands of conditions that are invariant out of the loop, so that we can use
5918 just trivial invariantness analysis in loop unswitching. The pass also includes
5921 @item -ftree-loop-ivcanon
5922 @opindex ftree-loop-ivcanon
5923 Create a canonical counter for number of iterations in the loop for that
5924 determining number of iterations requires complicated analysis. Later
5925 optimizations then may determine the number easily. Useful especially
5926 in connection with unrolling.
5930 Perform induction variable optimizations (strength reduction, induction
5931 variable merging and induction variable elimination) on trees.
5933 @item -ftree-parallelize-loops=n
5934 @opindex ftree-parallelize-loops
5935 Parallelize loops, i.e., split their iteration space to run in n threads.
5936 This is only possible for loops whose iterations are independent
5937 and can be arbitrarily reordered. The optimization is only
5938 profitable on multiprocessor machines, for loops that are CPU-intensive,
5939 rather than constrained e.g. by memory bandwidth. This option
5940 implies @option{-pthread}, and thus is only supported on targets
5941 that have support for @option{-pthread}.
5945 Perform scalar replacement of aggregates. This pass replaces structure
5946 references with scalars to prevent committing structures to memory too
5947 early. This flag is enabled by default at @option{-O} and higher.
5949 @item -ftree-copyrename
5950 @opindex ftree-copyrename
5951 Perform copy renaming on trees. This pass attempts to rename compiler
5952 temporaries to other variables at copy locations, usually resulting in
5953 variable names which more closely resemble the original variables. This flag
5954 is enabled by default at @option{-O} and higher.
5958 Perform temporary expression replacement during the SSA->normal phase. Single
5959 use/single def temporaries are replaced at their use location with their
5960 defining expression. This results in non-GIMPLE code, but gives the expanders
5961 much more complex trees to work on resulting in better RTL generation. This is
5962 enabled by default at @option{-O} and higher.
5964 @item -ftree-vectorize
5965 @opindex ftree-vectorize
5966 Perform loop vectorization on trees.
5968 @item -ftree-vect-loop-version
5969 @opindex ftree-vect-loop-version
5970 Perform loop versioning when doing loop vectorization on trees. When a loop
5971 appears to be vectorizable except that data alignment or data dependence cannot
5972 be determined at compile time then vectorized and non-vectorized versions of
5973 the loop are generated along with runtime checks for alignment or dependence
5974 to control which version is executed. This option is enabled by default
5975 except at level @option{-Os} where it is disabled.
5977 @item -fvect-cost-model
5978 @opindex fvect-cost-model
5979 Enable cost model for vectorization.
5983 Perform Value Range Propagation on trees. This is similar to the
5984 constant propagation pass, but instead of values, ranges of values are
5985 propagated. This allows the optimizers to remove unnecessary range
5986 checks like array bound checks and null pointer checks. This is
5987 enabled by default at @option{-O2} and higher. Null pointer check
5988 elimination is only done if @option{-fdelete-null-pointer-checks} is
5993 Perform tail duplication to enlarge superblock size. This transformation
5994 simplifies the control flow of the function allowing other optimizations to do
5997 @item -funroll-loops
5998 @opindex funroll-loops
5999 Unroll loops whose number of iterations can be determined at compile
6000 time or upon entry to the loop. @option{-funroll-loops} implies
6001 @option{-frerun-cse-after-loop}. This option makes code larger,
6002 and may or may not make it run faster.
6004 @item -funroll-all-loops
6005 @opindex funroll-all-loops
6006 Unroll all loops, even if their number of iterations is uncertain when
6007 the loop is entered. This usually makes programs run more slowly.
6008 @option{-funroll-all-loops} implies the same options as
6009 @option{-funroll-loops},
6011 @item -fsplit-ivs-in-unroller
6012 @opindex fsplit-ivs-in-unroller
6013 Enables expressing of values of induction variables in later iterations
6014 of the unrolled loop using the value in the first iteration. This breaks
6015 long dependency chains, thus improving efficiency of the scheduling passes.
6017 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6018 same effect. However in cases the loop body is more complicated than
6019 a single basic block, this is not reliable. It also does not work at all
6020 on some of the architectures due to restrictions in the CSE pass.
6022 This optimization is enabled by default.
6024 @item -fvariable-expansion-in-unroller
6025 @opindex fvariable-expansion-in-unroller
6026 With this option, the compiler will create multiple copies of some
6027 local variables when unrolling a loop which can result in superior code.
6029 @item -fpredictive-commoning
6030 @opindex fpredictive-commoning
6031 Perform predictive commoning optimization, i.e., reusing computations
6032 (especially memory loads and stores) performed in previous
6033 iterations of loops.
6035 This option is enabled at level @option{-O3}.
6037 @item -fprefetch-loop-arrays
6038 @opindex fprefetch-loop-arrays
6039 If supported by the target machine, generate instructions to prefetch
6040 memory to improve the performance of loops that access large arrays.
6042 This option may generate better or worse code; results are highly
6043 dependent on the structure of loops within the source code.
6045 Disabled at level @option{-Os}.
6048 @itemx -fno-peephole2
6049 @opindex fno-peephole
6050 @opindex fno-peephole2
6051 Disable any machine-specific peephole optimizations. The difference
6052 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6053 are implemented in the compiler; some targets use one, some use the
6054 other, a few use both.
6056 @option{-fpeephole} is enabled by default.
6057 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6059 @item -fno-guess-branch-probability
6060 @opindex fno-guess-branch-probability
6061 Do not guess branch probabilities using heuristics.
6063 GCC will use heuristics to guess branch probabilities if they are
6064 not provided by profiling feedback (@option{-fprofile-arcs}). These
6065 heuristics are based on the control flow graph. If some branch probabilities
6066 are specified by @samp{__builtin_expect}, then the heuristics will be
6067 used to guess branch probabilities for the rest of the control flow graph,
6068 taking the @samp{__builtin_expect} info into account. The interactions
6069 between the heuristics and @samp{__builtin_expect} can be complex, and in
6070 some cases, it may be useful to disable the heuristics so that the effects
6071 of @samp{__builtin_expect} are easier to understand.
6073 The default is @option{-fguess-branch-probability} at levels
6074 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6076 @item -freorder-blocks
6077 @opindex freorder-blocks
6078 Reorder basic blocks in the compiled function in order to reduce number of
6079 taken branches and improve code locality.
6081 Enabled at levels @option{-O2}, @option{-O3}.
6083 @item -freorder-blocks-and-partition
6084 @opindex freorder-blocks-and-partition
6085 In addition to reordering basic blocks in the compiled function, in order
6086 to reduce number of taken branches, partitions hot and cold basic blocks
6087 into separate sections of the assembly and .o files, to improve
6088 paging and cache locality performance.
6090 This optimization is automatically turned off in the presence of
6091 exception handling, for linkonce sections, for functions with a user-defined
6092 section attribute and on any architecture that does not support named
6095 @item -freorder-functions
6096 @opindex freorder-functions
6097 Reorder functions in the object file in order to
6098 improve code locality. This is implemented by using special
6099 subsections @code{.text.hot} for most frequently executed functions and
6100 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6101 the linker so object file format must support named sections and linker must
6102 place them in a reasonable way.
6104 Also profile feedback must be available in to make this option effective. See
6105 @option{-fprofile-arcs} for details.
6107 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6109 @item -fstrict-aliasing
6110 @opindex fstrict-aliasing
6111 Allows the compiler to assume the strictest aliasing rules applicable to
6112 the language being compiled. For C (and C++), this activates
6113 optimizations based on the type of expressions. In particular, an
6114 object of one type is assumed never to reside at the same address as an
6115 object of a different type, unless the types are almost the same. For
6116 example, an @code{unsigned int} can alias an @code{int}, but not a
6117 @code{void*} or a @code{double}. A character type may alias any other
6120 Pay special attention to code like this:
6133 The practice of reading from a different union member than the one most
6134 recently written to (called ``type-punning'') is common. Even with
6135 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6136 is accessed through the union type. So, the code above will work as
6137 expected. However, this code might not:
6148 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6150 @item -fstrict-overflow
6151 @opindex fstrict-overflow
6152 Allow the compiler to assume strict signed overflow rules, depending
6153 on the language being compiled. For C (and C++) this means that
6154 overflow when doing arithmetic with signed numbers is undefined, which
6155 means that the compiler may assume that it will not happen. This
6156 permits various optimizations. For example, the compiler will assume
6157 that an expression like @code{i + 10 > i} will always be true for
6158 signed @code{i}. This assumption is only valid if signed overflow is
6159 undefined, as the expression is false if @code{i + 10} overflows when
6160 using twos complement arithmetic. When this option is in effect any
6161 attempt to determine whether an operation on signed numbers will
6162 overflow must be written carefully to not actually involve overflow.
6164 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6165 that signed overflow is fully defined: it wraps. When
6166 @option{-fwrapv} is used, there is no difference between
6167 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
6168 @option{-fwrapv} certain types of overflow are permitted. For
6169 example, if the compiler gets an overflow when doing arithmetic on
6170 constants, the overflowed value can still be used with
6171 @option{-fwrapv}, but not otherwise.
6173 The @option{-fstrict-overflow} option is enabled at levels
6174 @option{-O2}, @option{-O3}, @option{-Os}.
6176 @item -falign-functions
6177 @itemx -falign-functions=@var{n}
6178 @opindex falign-functions
6179 Align the start of functions to the next power-of-two greater than
6180 @var{n}, skipping up to @var{n} bytes. For instance,
6181 @option{-falign-functions=32} aligns functions to the next 32-byte
6182 boundary, but @option{-falign-functions=24} would align to the next
6183 32-byte boundary only if this can be done by skipping 23 bytes or less.
6185 @option{-fno-align-functions} and @option{-falign-functions=1} are
6186 equivalent and mean that functions will not be aligned.
6188 Some assemblers only support this flag when @var{n} is a power of two;
6189 in that case, it is rounded up.
6191 If @var{n} is not specified or is zero, use a machine-dependent default.
6193 Enabled at levels @option{-O2}, @option{-O3}.
6195 @item -falign-labels
6196 @itemx -falign-labels=@var{n}
6197 @opindex falign-labels
6198 Align all branch targets to a power-of-two boundary, skipping up to
6199 @var{n} bytes like @option{-falign-functions}. This option can easily
6200 make code slower, because it must insert dummy operations for when the
6201 branch target is reached in the usual flow of the code.
6203 @option{-fno-align-labels} and @option{-falign-labels=1} are
6204 equivalent and mean that labels will not be aligned.
6206 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6207 are greater than this value, then their values are used instead.
6209 If @var{n} is not specified or is zero, use a machine-dependent default
6210 which is very likely to be @samp{1}, meaning no alignment.
6212 Enabled at levels @option{-O2}, @option{-O3}.
6215 @itemx -falign-loops=@var{n}
6216 @opindex falign-loops
6217 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6218 like @option{-falign-functions}. The hope is that the loop will be
6219 executed many times, which will make up for any execution of the dummy
6222 @option{-fno-align-loops} and @option{-falign-loops=1} are
6223 equivalent and mean that loops will not be aligned.
6225 If @var{n} is not specified or is zero, use a machine-dependent default.
6227 Enabled at levels @option{-O2}, @option{-O3}.
6230 @itemx -falign-jumps=@var{n}
6231 @opindex falign-jumps
6232 Align branch targets to a power-of-two boundary, for branch targets
6233 where the targets can only be reached by jumping, skipping up to @var{n}
6234 bytes like @option{-falign-functions}. In this case, no dummy operations
6237 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6238 equivalent and mean that loops will not be aligned.
6240 If @var{n} is not specified or is zero, use a machine-dependent default.
6242 Enabled at levels @option{-O2}, @option{-O3}.
6244 @item -funit-at-a-time
6245 @opindex funit-at-a-time
6246 Parse the whole compilation unit before starting to produce code.
6247 This allows some extra optimizations to take place but consumes
6248 more memory (in general). There are some compatibility issues
6249 with @emph{unit-at-a-time} mode:
6252 enabling @emph{unit-at-a-time} mode may change the order
6253 in which functions, variables, and top-level @code{asm} statements
6254 are emitted, and will likely break code relying on some particular
6255 ordering. The majority of such top-level @code{asm} statements,
6256 though, can be replaced by @code{section} attributes. The
6257 @option{fno-toplevel-reorder} option may be used to keep the ordering
6258 used in the input file, at the cost of some optimizations.
6261 @emph{unit-at-a-time} mode removes unreferenced static variables
6262 and functions. This may result in undefined references
6263 when an @code{asm} statement refers directly to variables or functions
6264 that are otherwise unused. In that case either the variable/function
6265 shall be listed as an operand of the @code{asm} statement operand or,
6266 in the case of top-level @code{asm} statements the attribute @code{used}
6267 shall be used on the declaration.
6270 Static functions now can use non-standard passing conventions that
6271 may break @code{asm} statements calling functions directly. Again,
6272 attribute @code{used} will prevent this behavior.
6275 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
6276 but this scheme may not be supported by future releases of GCC@.
6278 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6280 @item -fno-toplevel-reorder
6281 @opindex fno-toplevel-reorder
6282 Do not reorder top-level functions, variables, and @code{asm}
6283 statements. Output them in the same order that they appear in the
6284 input file. When this option is used, unreferenced static variables
6285 will not be removed. This option is intended to support existing code
6286 which relies on a particular ordering. For new code, it is better to
6291 Constructs webs as commonly used for register allocation purposes and assign
6292 each web individual pseudo register. This allows the register allocation pass
6293 to operate on pseudos directly, but also strengthens several other optimization
6294 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6295 however, make debugging impossible, since variables will no longer stay in a
6298 Enabled by default with @option{-funroll-loops}.
6300 @item -fwhole-program
6301 @opindex fwhole-program
6302 Assume that the current compilation unit represents whole program being
6303 compiled. All public functions and variables with the exception of @code{main}
6304 and those merged by attribute @code{externally_visible} become static functions
6305 and in a affect gets more aggressively optimized by interprocedural optimizers.
6306 While this option is equivalent to proper use of @code{static} keyword for
6307 programs consisting of single file, in combination with option
6308 @option{--combine} this flag can be used to compile most of smaller scale C
6309 programs since the functions and variables become local for the whole combined
6310 compilation unit, not for the single source file itself.
6312 This option is not supported for Fortran programs.
6314 @item -fcprop-registers
6315 @opindex fcprop-registers
6316 After register allocation and post-register allocation instruction splitting,
6317 we perform a copy-propagation pass to try to reduce scheduling dependencies
6318 and occasionally eliminate the copy.
6320 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6322 @item -fprofile-generate
6323 @opindex fprofile-generate
6325 Enable options usually used for instrumenting application to produce
6326 profile useful for later recompilation with profile feedback based
6327 optimization. You must use @option{-fprofile-generate} both when
6328 compiling and when linking your program.
6330 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6333 @opindex fprofile-use
6334 Enable profile feedback directed optimizations, and optimizations
6335 generally profitable only with profile feedback available.
6337 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6338 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6340 By default, GCC emits an error message if the feedback profiles do not
6341 match the source code. This error can be turned into a warning by using
6342 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6346 The following options control compiler behavior regarding floating
6347 point arithmetic. These options trade off between speed and
6348 correctness. All must be specifically enabled.
6352 @opindex ffloat-store
6353 Do not store floating point variables in registers, and inhibit other
6354 options that might change whether a floating point value is taken from a
6357 @cindex floating point precision
6358 This option prevents undesirable excess precision on machines such as
6359 the 68000 where the floating registers (of the 68881) keep more
6360 precision than a @code{double} is supposed to have. Similarly for the
6361 x86 architecture. For most programs, the excess precision does only
6362 good, but a few programs rely on the precise definition of IEEE floating
6363 point. Use @option{-ffloat-store} for such programs, after modifying
6364 them to store all pertinent intermediate computations into variables.
6368 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6369 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6370 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6372 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6374 This option is not turned on by any @option{-O} option since
6375 it can result in incorrect output for programs which depend on
6376 an exact implementation of IEEE or ISO rules/specifications for
6377 math functions. It may, however, yield faster code for programs
6378 that do not require the guarantees of these specifications.
6380 @item -fno-math-errno
6381 @opindex fno-math-errno
6382 Do not set ERRNO after calling math functions that are executed
6383 with a single instruction, e.g., sqrt. A program that relies on
6384 IEEE exceptions for math error handling may want to use this flag
6385 for speed while maintaining IEEE arithmetic compatibility.
6387 This option is not turned on by any @option{-O} option since
6388 it can result in incorrect output for programs which depend on
6389 an exact implementation of IEEE or ISO rules/specifications for
6390 math functions. It may, however, yield faster code for programs
6391 that do not require the guarantees of these specifications.
6393 The default is @option{-fmath-errno}.
6395 On Darwin systems, the math library never sets @code{errno}. There is
6396 therefore no reason for the compiler to consider the possibility that
6397 it might, and @option{-fno-math-errno} is the default.
6399 @item -funsafe-math-optimizations
6400 @opindex funsafe-math-optimizations
6402 Allow optimizations for floating-point arithmetic that (a) assume
6403 that arguments and results are valid and (b) may violate IEEE or
6404 ANSI standards. When used at link-time, it may include libraries
6405 or startup files that change the default FPU control word or other
6406 similar optimizations.
6408 This option is not turned on by any @option{-O} option since
6409 it can result in incorrect output for programs which depend on
6410 an exact implementation of IEEE or ISO rules/specifications for
6411 math functions. It may, however, yield faster code for programs
6412 that do not require the guarantees of these specifications.
6413 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6414 @option{-fassociative-math} and @option{-freciprocal-math}.
6416 The default is @option{-fno-unsafe-math-optimizations}.
6418 @item -fassociative-math
6419 @opindex fassociative-math
6421 Allow re-association of operands in series of floating-point operations.
6422 This violates the ISO C and C++ language standard by possibly changing
6423 computation result. NOTE: re-ordering may change the sign of zero as
6424 well as ignore NaNs and inhibit or create underflow or overflow (and
6425 thus cannot be used on a code which relies on rounding behavior like
6426 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6427 and thus may not be used when ordered comparisons are required.
6428 This option requires that both @option{-fno-signed-zeros} and
6429 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6430 much sense with @option{-frounding-math}.
6432 The default is @option{-fno-associative-math}.
6434 @item -freciprocal-math
6435 @opindex freciprocal-math
6437 Allow the reciprocal of a value to be used instead of dividing by
6438 the value if this enables optimizations. For example @code{x / y}
6439 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6440 is subject to common subexpression elimination. Note that this loses
6441 precision and increases the number of flops operating on the value.
6443 The default is @option{-fno-reciprocal-math}.
6445 @item -ffinite-math-only
6446 @opindex ffinite-math-only
6447 Allow optimizations for floating-point arithmetic that assume
6448 that arguments and results are not NaNs or +-Infs.
6450 This option is not turned on by any @option{-O} option since
6451 it can result in incorrect output for programs which depend on
6452 an exact implementation of IEEE or ISO rules/specifications for
6453 math functions. It may, however, yield faster code for programs
6454 that do not require the guarantees of these specifications.
6456 The default is @option{-fno-finite-math-only}.
6458 @item -fno-signed-zeros
6459 @opindex fno-signed-zeros
6460 Allow optimizations for floating point arithmetic that ignore the
6461 signedness of zero. IEEE arithmetic specifies the behavior of
6462 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6463 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6464 This option implies that the sign of a zero result isn't significant.
6466 The default is @option{-fsigned-zeros}.
6468 @item -fno-trapping-math
6469 @opindex fno-trapping-math
6470 Compile code assuming that floating-point operations cannot generate
6471 user-visible traps. These traps include division by zero, overflow,
6472 underflow, inexact result and invalid operation. This option requires
6473 that @option{-fno-signaling-nans} be in effect. Setting this option may
6474 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6476 This option should never be turned on by any @option{-O} option since
6477 it can result in incorrect output for programs which depend on
6478 an exact implementation of IEEE or ISO rules/specifications for
6481 The default is @option{-ftrapping-math}.
6483 @item -frounding-math
6484 @opindex frounding-math
6485 Disable transformations and optimizations that assume default floating
6486 point rounding behavior. This is round-to-zero for all floating point
6487 to integer conversions, and round-to-nearest for all other arithmetic
6488 truncations. This option should be specified for programs that change
6489 the FP rounding mode dynamically, or that may be executed with a
6490 non-default rounding mode. This option disables constant folding of
6491 floating point expressions at compile-time (which may be affected by
6492 rounding mode) and arithmetic transformations that are unsafe in the
6493 presence of sign-dependent rounding modes.
6495 The default is @option{-fno-rounding-math}.
6497 This option is experimental and does not currently guarantee to
6498 disable all GCC optimizations that are affected by rounding mode.
6499 Future versions of GCC may provide finer control of this setting
6500 using C99's @code{FENV_ACCESS} pragma. This command line option
6501 will be used to specify the default state for @code{FENV_ACCESS}.
6503 @item -frtl-abstract-sequences
6504 @opindex frtl-abstract-sequences
6505 It is a size optimization method. This option is to find identical
6506 sequences of code, which can be turned into pseudo-procedures and
6507 then replace all occurrences with calls to the newly created
6508 subroutine. It is kind of an opposite of @option{-finline-functions}.
6509 This optimization runs at RTL level.
6511 @item -fsignaling-nans
6512 @opindex fsignaling-nans
6513 Compile code assuming that IEEE signaling NaNs may generate user-visible
6514 traps during floating-point operations. Setting this option disables
6515 optimizations that may change the number of exceptions visible with
6516 signaling NaNs. This option implies @option{-ftrapping-math}.
6518 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6521 The default is @option{-fno-signaling-nans}.
6523 This option is experimental and does not currently guarantee to
6524 disable all GCC optimizations that affect signaling NaN behavior.
6526 @item -fsingle-precision-constant
6527 @opindex fsingle-precision-constant
6528 Treat floating point constant as single precision constant instead of
6529 implicitly converting it to double precision constant.
6531 @item -fcx-limited-range
6532 @opindex fcx-limited-range
6533 When enabled, this option states that a range reduction step is not
6534 needed when performing complex division. The default is
6535 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
6537 This option controls the default setting of the ISO C99
6538 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6543 The following options control optimizations that may improve
6544 performance, but are not enabled by any @option{-O} options. This
6545 section includes experimental options that may produce broken code.
6548 @item -fbranch-probabilities
6549 @opindex fbranch-probabilities
6550 After running a program compiled with @option{-fprofile-arcs}
6551 (@pxref{Debugging Options,, Options for Debugging Your Program or
6552 @command{gcc}}), you can compile it a second time using
6553 @option{-fbranch-probabilities}, to improve optimizations based on
6554 the number of times each branch was taken. When the program
6555 compiled with @option{-fprofile-arcs} exits it saves arc execution
6556 counts to a file called @file{@var{sourcename}.gcda} for each source
6557 file. The information in this data file is very dependent on the
6558 structure of the generated code, so you must use the same source code
6559 and the same optimization options for both compilations.
6561 With @option{-fbranch-probabilities}, GCC puts a
6562 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6563 These can be used to improve optimization. Currently, they are only
6564 used in one place: in @file{reorg.c}, instead of guessing which path a
6565 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6566 exactly determine which path is taken more often.
6568 @item -fprofile-values
6569 @opindex fprofile-values
6570 If combined with @option{-fprofile-arcs}, it adds code so that some
6571 data about values of expressions in the program is gathered.
6573 With @option{-fbranch-probabilities}, it reads back the data gathered
6574 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6575 notes to instructions for their later usage in optimizations.
6577 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6581 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6582 a code to gather information about values of expressions.
6584 With @option{-fbranch-probabilities}, it reads back the data gathered
6585 and actually performs the optimizations based on them.
6586 Currently the optimizations include specialization of division operation
6587 using the knowledge about the value of the denominator.
6589 @item -frename-registers
6590 @opindex frename-registers
6591 Attempt to avoid false dependencies in scheduled code by making use
6592 of registers left over after register allocation. This optimization
6593 will most benefit processors with lots of registers. Depending on the
6594 debug information format adopted by the target, however, it can
6595 make debugging impossible, since variables will no longer stay in
6596 a ``home register''.
6598 Enabled by default with @option{-funroll-loops}.
6602 Perform tail duplication to enlarge superblock size. This transformation
6603 simplifies the control flow of the function allowing other optimizations to do
6606 Enabled with @option{-fprofile-use}.
6608 @item -funroll-loops
6609 @opindex funroll-loops
6610 Unroll loops whose number of iterations can be determined at compile time or
6611 upon entry to the loop. @option{-funroll-loops} implies
6612 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6613 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6614 small constant number of iterations). This option makes code larger, and may
6615 or may not make it run faster.
6617 Enabled with @option{-fprofile-use}.
6619 @item -funroll-all-loops
6620 @opindex funroll-all-loops
6621 Unroll all loops, even if their number of iterations is uncertain when
6622 the loop is entered. This usually makes programs run more slowly.
6623 @option{-funroll-all-loops} implies the same options as
6624 @option{-funroll-loops}.
6627 @opindex fpeel-loops
6628 Peels the loops for that there is enough information that they do not
6629 roll much (from profile feedback). It also turns on complete loop peeling
6630 (i.e.@: complete removal of loops with small constant number of iterations).
6632 Enabled with @option{-fprofile-use}.
6634 @item -fmove-loop-invariants
6635 @opindex fmove-loop-invariants
6636 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6637 at level @option{-O1}
6639 @item -funswitch-loops
6640 @opindex funswitch-loops
6641 Move branches with loop invariant conditions out of the loop, with duplicates
6642 of the loop on both branches (modified according to result of the condition).
6644 @item -ffunction-sections
6645 @itemx -fdata-sections
6646 @opindex ffunction-sections
6647 @opindex fdata-sections
6648 Place each function or data item into its own section in the output
6649 file if the target supports arbitrary sections. The name of the
6650 function or the name of the data item determines the section's name
6653 Use these options on systems where the linker can perform optimizations
6654 to improve locality of reference in the instruction space. Most systems
6655 using the ELF object format and SPARC processors running Solaris 2 have
6656 linkers with such optimizations. AIX may have these optimizations in
6659 Only use these options when there are significant benefits from doing
6660 so. When you specify these options, the assembler and linker will
6661 create larger object and executable files and will also be slower.
6662 You will not be able to use @code{gprof} on all systems if you
6663 specify this option and you may have problems with debugging if
6664 you specify both this option and @option{-g}.
6666 @item -fbranch-target-load-optimize
6667 @opindex fbranch-target-load-optimize
6668 Perform branch target register load optimization before prologue / epilogue
6670 The use of target registers can typically be exposed only during reload,
6671 thus hoisting loads out of loops and doing inter-block scheduling needs
6672 a separate optimization pass.
6674 @item -fbranch-target-load-optimize2
6675 @opindex fbranch-target-load-optimize2
6676 Perform branch target register load optimization after prologue / epilogue
6679 @item -fbtr-bb-exclusive
6680 @opindex fbtr-bb-exclusive
6681 When performing branch target register load optimization, don't reuse
6682 branch target registers in within any basic block.
6684 @item -fstack-protector
6685 @opindex fstack-protector
6686 Emit extra code to check for buffer overflows, such as stack smashing
6687 attacks. This is done by adding a guard variable to functions with
6688 vulnerable objects. This includes functions that call alloca, and
6689 functions with buffers larger than 8 bytes. The guards are initialized
6690 when a function is entered and then checked when the function exits.
6691 If a guard check fails, an error message is printed and the program exits.
6693 @item -fstack-protector-all
6694 @opindex fstack-protector-all
6695 Like @option{-fstack-protector} except that all functions are protected.
6697 @item -fsection-anchors
6698 @opindex fsection-anchors
6699 Try to reduce the number of symbolic address calculations by using
6700 shared ``anchor'' symbols to address nearby objects. This transformation
6701 can help to reduce the number of GOT entries and GOT accesses on some
6704 For example, the implementation of the following function @code{foo}:
6708 int foo (void) @{ return a + b + c; @}
6711 would usually calculate the addresses of all three variables, but if you
6712 compile it with @option{-fsection-anchors}, it will access the variables
6713 from a common anchor point instead. The effect is similar to the
6714 following pseudocode (which isn't valid C):
6719 register int *xr = &x;
6720 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6724 Not all targets support this option.
6726 @item --param @var{name}=@var{value}
6728 In some places, GCC uses various constants to control the amount of
6729 optimization that is done. For example, GCC will not inline functions
6730 that contain more that a certain number of instructions. You can
6731 control some of these constants on the command-line using the
6732 @option{--param} option.
6734 The names of specific parameters, and the meaning of the values, are
6735 tied to the internals of the compiler, and are subject to change
6736 without notice in future releases.
6738 In each case, the @var{value} is an integer. The allowable choices for
6739 @var{name} are given in the following table:
6742 @item salias-max-implicit-fields
6743 The maximum number of fields in a variable without direct
6744 structure accesses for which structure aliasing will consider trying
6745 to track each field. The default is 5
6747 @item salias-max-array-elements
6748 The maximum number of elements an array can have and its elements
6749 still be tracked individually by structure aliasing. The default is 4
6751 @item sra-max-structure-size
6752 The maximum structure size, in bytes, at which the scalar replacement
6753 of aggregates (SRA) optimization will perform block copies. The
6754 default value, 0, implies that GCC will select the most appropriate
6757 @item sra-field-structure-ratio
6758 The threshold ratio (as a percentage) between instantiated fields and
6759 the complete structure size. We say that if the ratio of the number
6760 of bytes in instantiated fields to the number of bytes in the complete
6761 structure exceeds this parameter, then block copies are not used. The
6764 @item struct-reorg-cold-struct-ratio
6765 The threshold ratio (as a percentage) between a structure frequency
6766 and the frequency of the hottest structure in the program. This parameter
6767 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
6768 We say that if the ratio of a structure frequency, calculated by profiling,
6769 to the hottest structure frequency in the program is less than this
6770 parameter, then structure reorganization is not applied to this structure.
6773 @item max-crossjump-edges
6774 The maximum number of incoming edges to consider for crossjumping.
6775 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6776 the number of edges incoming to each block. Increasing values mean
6777 more aggressive optimization, making the compile time increase with
6778 probably small improvement in executable size.
6780 @item min-crossjump-insns
6781 The minimum number of instructions which must be matched at the end
6782 of two blocks before crossjumping will be performed on them. This
6783 value is ignored in the case where all instructions in the block being
6784 crossjumped from are matched. The default value is 5.
6786 @item max-grow-copy-bb-insns
6787 The maximum code size expansion factor when copying basic blocks
6788 instead of jumping. The expansion is relative to a jump instruction.
6789 The default value is 8.
6791 @item max-goto-duplication-insns
6792 The maximum number of instructions to duplicate to a block that jumps
6793 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6794 passes, GCC factors computed gotos early in the compilation process,
6795 and unfactors them as late as possible. Only computed jumps at the
6796 end of a basic blocks with no more than max-goto-duplication-insns are
6797 unfactored. The default value is 8.
6799 @item max-delay-slot-insn-search
6800 The maximum number of instructions to consider when looking for an
6801 instruction to fill a delay slot. If more than this arbitrary number of
6802 instructions is searched, the time savings from filling the delay slot
6803 will be minimal so stop searching. Increasing values mean more
6804 aggressive optimization, making the compile time increase with probably
6805 small improvement in executable run time.
6807 @item max-delay-slot-live-search
6808 When trying to fill delay slots, the maximum number of instructions to
6809 consider when searching for a block with valid live register
6810 information. Increasing this arbitrarily chosen value means more
6811 aggressive optimization, increasing the compile time. This parameter
6812 should be removed when the delay slot code is rewritten to maintain the
6815 @item max-gcse-memory
6816 The approximate maximum amount of memory that will be allocated in
6817 order to perform the global common subexpression elimination
6818 optimization. If more memory than specified is required, the
6819 optimization will not be done.
6821 @item max-gcse-passes
6822 The maximum number of passes of GCSE to run. The default is 1.
6824 @item max-pending-list-length
6825 The maximum number of pending dependencies scheduling will allow
6826 before flushing the current state and starting over. Large functions
6827 with few branches or calls can create excessively large lists which
6828 needlessly consume memory and resources.
6830 @item max-inline-insns-single
6831 Several parameters control the tree inliner used in gcc.
6832 This number sets the maximum number of instructions (counted in GCC's
6833 internal representation) in a single function that the tree inliner
6834 will consider for inlining. This only affects functions declared
6835 inline and methods implemented in a class declaration (C++).
6836 The default value is 450.
6838 @item max-inline-insns-auto
6839 When you use @option{-finline-functions} (included in @option{-O3}),
6840 a lot of functions that would otherwise not be considered for inlining
6841 by the compiler will be investigated. To those functions, a different
6842 (more restrictive) limit compared to functions declared inline can
6844 The default value is 90.
6846 @item large-function-insns
6847 The limit specifying really large functions. For functions larger than this
6848 limit after inlining inlining is constrained by
6849 @option{--param large-function-growth}. This parameter is useful primarily
6850 to avoid extreme compilation time caused by non-linear algorithms used by the
6852 This parameter is ignored when @option{-funit-at-a-time} is not used.
6853 The default value is 2700.
6855 @item large-function-growth
6856 Specifies maximal growth of large function caused by inlining in percents.
6857 This parameter is ignored when @option{-funit-at-a-time} is not used.
6858 The default value is 100 which limits large function growth to 2.0 times
6861 @item large-unit-insns
6862 The limit specifying large translation unit. Growth caused by inlining of
6863 units larger than this limit is limited by @option{--param inline-unit-growth}.
6864 For small units this might be too tight (consider unit consisting of function A
6865 that is inline and B that just calls A three time. If B is small relative to
6866 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6867 large units consisting of small inlineable functions however the overall unit
6868 growth limit is needed to avoid exponential explosion of code size. Thus for
6869 smaller units, the size is increased to @option{--param large-unit-insns}
6870 before applying @option{--param inline-unit-growth}. The default is 10000
6872 @item inline-unit-growth
6873 Specifies maximal overall growth of the compilation unit caused by inlining.
6874 This parameter is ignored when @option{-funit-at-a-time} is not used.
6875 The default value is 30 which limits unit growth to 1.3 times the original
6878 @item large-stack-frame
6879 The limit specifying large stack frames. While inlining the algorithm is trying
6880 to not grow past this limit too much. Default value is 256 bytes.
6882 @item large-stack-frame-growth
6883 Specifies maximal growth of large stack frames caused by inlining in percents.
6884 The default value is 1000 which limits large stack frame growth to 11 times
6887 @item max-inline-insns-recursive
6888 @itemx max-inline-insns-recursive-auto
6889 Specifies maximum number of instructions out-of-line copy of self recursive inline
6890 function can grow into by performing recursive inlining.
6892 For functions declared inline @option{--param max-inline-insns-recursive} is
6893 taken into account. For function not declared inline, recursive inlining
6894 happens only when @option{-finline-functions} (included in @option{-O3}) is
6895 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6896 default value is 450.
6898 @item max-inline-recursive-depth
6899 @itemx max-inline-recursive-depth-auto
6900 Specifies maximum recursion depth used by the recursive inlining.
6902 For functions declared inline @option{--param max-inline-recursive-depth} is
6903 taken into account. For function not declared inline, recursive inlining
6904 happens only when @option{-finline-functions} (included in @option{-O3}) is
6905 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6908 @item min-inline-recursive-probability
6909 Recursive inlining is profitable only for function having deep recursion
6910 in average and can hurt for function having little recursion depth by
6911 increasing the prologue size or complexity of function body to other
6914 When profile feedback is available (see @option{-fprofile-generate}) the actual
6915 recursion depth can be guessed from probability that function will recurse via
6916 given call expression. This parameter limits inlining only to call expression
6917 whose probability exceeds given threshold (in percents). The default value is
6920 @item inline-call-cost
6921 Specify cost of call instruction relative to simple arithmetics operations
6922 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6923 functions and at the same time increases size of leaf function that is believed to
6924 reduce function size by being inlined. In effect it increases amount of
6925 inlining for code having large abstraction penalty (many functions that just
6926 pass the arguments to other functions) and decrease inlining for code with low
6927 abstraction penalty. The default value is 12.
6929 @item min-vect-loop-bound
6930 The minimum number of iterations under which a loop will not get vectorized
6931 when @option{-ftree-vectorize} is used. The number of iterations after
6932 vectorization needs to be greater than the value specified by this option
6933 to allow vectorization. The default value is 0.
6935 @item max-unrolled-insns
6936 The maximum number of instructions that a loop should have if that loop
6937 is unrolled, and if the loop is unrolled, it determines how many times
6938 the loop code is unrolled.
6940 @item max-average-unrolled-insns
6941 The maximum number of instructions biased by probabilities of their execution
6942 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6943 it determines how many times the loop code is unrolled.
6945 @item max-unroll-times
6946 The maximum number of unrollings of a single loop.
6948 @item max-peeled-insns
6949 The maximum number of instructions that a loop should have if that loop
6950 is peeled, and if the loop is peeled, it determines how many times
6951 the loop code is peeled.
6953 @item max-peel-times
6954 The maximum number of peelings of a single loop.
6956 @item max-completely-peeled-insns
6957 The maximum number of insns of a completely peeled loop.
6959 @item max-completely-peel-times
6960 The maximum number of iterations of a loop to be suitable for complete peeling.
6962 @item max-unswitch-insns
6963 The maximum number of insns of an unswitched loop.
6965 @item max-unswitch-level
6966 The maximum number of branches unswitched in a single loop.
6969 The minimum cost of an expensive expression in the loop invariant motion.
6971 @item iv-consider-all-candidates-bound
6972 Bound on number of candidates for induction variables below that
6973 all candidates are considered for each use in induction variable
6974 optimizations. Only the most relevant candidates are considered
6975 if there are more candidates, to avoid quadratic time complexity.
6977 @item iv-max-considered-uses
6978 The induction variable optimizations give up on loops that contain more
6979 induction variable uses.
6981 @item iv-always-prune-cand-set-bound
6982 If number of candidates in the set is smaller than this value,
6983 we always try to remove unnecessary ivs from the set during its
6984 optimization when a new iv is added to the set.
6986 @item scev-max-expr-size
6987 Bound on size of expressions used in the scalar evolutions analyzer.
6988 Large expressions slow the analyzer.
6990 @item omega-max-vars
6991 The maximum number of variables in an Omega constraint system.
6992 The default value is 128.
6994 @item omega-max-geqs
6995 The maximum number of inequalities in an Omega constraint system.
6996 The default value is 256.
6999 The maximum number of equalities in an Omega constraint system.
7000 The default value is 128.
7002 @item omega-max-wild-cards
7003 The maximum number of wildcard variables that the Omega solver will
7004 be able to insert. The default value is 18.
7006 @item omega-hash-table-size
7007 The size of the hash table in the Omega solver. The default value is
7010 @item omega-max-keys
7011 The maximal number of keys used by the Omega solver. The default
7014 @item omega-eliminate-redundant-constraints
7015 When set to 1, use expensive methods to eliminate all redundant
7016 constraints. The default value is 0.
7018 @item vect-max-version-for-alignment-checks
7019 The maximum number of runtime checks that can be performed when
7020 doing loop versioning for alignment in the vectorizer. See option
7021 ftree-vect-loop-version for more information.
7023 @item vect-max-version-for-alias-checks
7024 The maximum number of runtime checks that can be performed when
7025 doing loop versioning for alias in the vectorizer. See option
7026 ftree-vect-loop-version for more information.
7028 @item max-iterations-to-track
7030 The maximum number of iterations of a loop the brute force algorithm
7031 for analysis of # of iterations of the loop tries to evaluate.
7033 @item hot-bb-count-fraction
7034 Select fraction of the maximal count of repetitions of basic block in program
7035 given basic block needs to have to be considered hot.
7037 @item hot-bb-frequency-fraction
7038 Select fraction of the maximal frequency of executions of basic block in
7039 function given basic block needs to have to be considered hot
7041 @item max-predicted-iterations
7042 The maximum number of loop iterations we predict statically. This is useful
7043 in cases where function contain single loop with known bound and other loop
7044 with unknown. We predict the known number of iterations correctly, while
7045 the unknown number of iterations average to roughly 10. This means that the
7046 loop without bounds would appear artificially cold relative to the other one.
7048 @item align-threshold
7050 Select fraction of the maximal frequency of executions of basic block in
7051 function given basic block will get aligned.
7053 @item align-loop-iterations
7055 A loop expected to iterate at lest the selected number of iterations will get
7058 @item tracer-dynamic-coverage
7059 @itemx tracer-dynamic-coverage-feedback
7061 This value is used to limit superblock formation once the given percentage of
7062 executed instructions is covered. This limits unnecessary code size
7065 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7066 feedback is available. The real profiles (as opposed to statically estimated
7067 ones) are much less balanced allowing the threshold to be larger value.
7069 @item tracer-max-code-growth
7070 Stop tail duplication once code growth has reached given percentage. This is
7071 rather hokey argument, as most of the duplicates will be eliminated later in
7072 cross jumping, so it may be set to much higher values than is the desired code
7075 @item tracer-min-branch-ratio
7077 Stop reverse growth when the reverse probability of best edge is less than this
7078 threshold (in percent).
7080 @item tracer-min-branch-ratio
7081 @itemx tracer-min-branch-ratio-feedback
7083 Stop forward growth if the best edge do have probability lower than this
7086 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7087 compilation for profile feedback and one for compilation without. The value
7088 for compilation with profile feedback needs to be more conservative (higher) in
7089 order to make tracer effective.
7091 @item max-cse-path-length
7093 Maximum number of basic blocks on path that cse considers. The default is 10.
7096 The maximum instructions CSE process before flushing. The default is 1000.
7098 @item max-aliased-vops
7100 Maximum number of virtual operands per function allowed to represent
7101 aliases before triggering the alias partitioning heuristic. Alias
7102 partitioning reduces compile times and memory consumption needed for
7103 aliasing at the expense of precision loss in alias information. The
7104 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7107 Notice that if a function contains more memory statements than the
7108 value of this parameter, it is not really possible to achieve this
7109 reduction. In this case, the compiler will use the number of memory
7110 statements as the value for @option{max-aliased-vops}.
7112 @item avg-aliased-vops
7114 Average number of virtual operands per statement allowed to represent
7115 aliases before triggering the alias partitioning heuristic. This
7116 works in conjunction with @option{max-aliased-vops}. If a function
7117 contains more than @option{max-aliased-vops} virtual operators, then
7118 memory symbols will be grouped into memory partitions until either the
7119 total number of virtual operators is below @option{max-aliased-vops}
7120 or the average number of virtual operators per memory statement is
7121 below @option{avg-aliased-vops}. The default value for this parameter
7122 is 1 for -O1 and -O2, and 3 for -O3.
7124 @item ggc-min-expand
7126 GCC uses a garbage collector to manage its own memory allocation. This
7127 parameter specifies the minimum percentage by which the garbage
7128 collector's heap should be allowed to expand between collections.
7129 Tuning this may improve compilation speed; it has no effect on code
7132 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7133 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7134 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7135 GCC is not able to calculate RAM on a particular platform, the lower
7136 bound of 30% is used. Setting this parameter and
7137 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7138 every opportunity. This is extremely slow, but can be useful for
7141 @item ggc-min-heapsize
7143 Minimum size of the garbage collector's heap before it begins bothering
7144 to collect garbage. The first collection occurs after the heap expands
7145 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7146 tuning this may improve compilation speed, and has no effect on code
7149 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7150 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7151 with a lower bound of 4096 (four megabytes) and an upper bound of
7152 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7153 particular platform, the lower bound is used. Setting this parameter
7154 very large effectively disables garbage collection. Setting this
7155 parameter and @option{ggc-min-expand} to zero causes a full collection
7156 to occur at every opportunity.
7158 @item max-reload-search-insns
7159 The maximum number of instruction reload should look backward for equivalent
7160 register. Increasing values mean more aggressive optimization, making the
7161 compile time increase with probably slightly better performance. The default
7164 @item max-cselib-memory-locations
7165 The maximum number of memory locations cselib should take into account.
7166 Increasing values mean more aggressive optimization, making the compile time
7167 increase with probably slightly better performance. The default value is 500.
7169 @item max-flow-memory-locations
7170 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
7171 The default value is 100.
7173 @item reorder-blocks-duplicate
7174 @itemx reorder-blocks-duplicate-feedback
7176 Used by basic block reordering pass to decide whether to use unconditional
7177 branch or duplicate the code on its destination. Code is duplicated when its
7178 estimated size is smaller than this value multiplied by the estimated size of
7179 unconditional jump in the hot spots of the program.
7181 The @option{reorder-block-duplicate-feedback} is used only when profile
7182 feedback is available and may be set to higher values than
7183 @option{reorder-block-duplicate} since information about the hot spots is more
7186 @item max-sched-ready-insns
7187 The maximum number of instructions ready to be issued the scheduler should
7188 consider at any given time during the first scheduling pass. Increasing
7189 values mean more thorough searches, making the compilation time increase
7190 with probably little benefit. The default value is 100.
7192 @item max-sched-region-blocks
7193 The maximum number of blocks in a region to be considered for
7194 interblock scheduling. The default value is 10.
7196 @item max-sched-region-insns
7197 The maximum number of insns in a region to be considered for
7198 interblock scheduling. The default value is 100.
7201 The minimum probability (in percents) of reaching a source block
7202 for interblock speculative scheduling. The default value is 40.
7204 @item max-sched-extend-regions-iters
7205 The maximum number of iterations through CFG to extend regions.
7206 0 - disable region extension,
7207 N - do at most N iterations.
7208 The default value is 0.
7210 @item max-sched-insn-conflict-delay
7211 The maximum conflict delay for an insn to be considered for speculative motion.
7212 The default value is 3.
7214 @item sched-spec-prob-cutoff
7215 The minimal probability of speculation success (in percents), so that
7216 speculative insn will be scheduled.
7217 The default value is 40.
7219 @item max-last-value-rtl
7221 The maximum size measured as number of RTLs that can be recorded in an expression
7222 in combiner for a pseudo register as last known value of that register. The default
7225 @item integer-share-limit
7226 Small integer constants can use a shared data structure, reducing the
7227 compiler's memory usage and increasing its speed. This sets the maximum
7228 value of a shared integer constant. The default value is 256.
7230 @item min-virtual-mappings
7231 Specifies the minimum number of virtual mappings in the incremental
7232 SSA updater that should be registered to trigger the virtual mappings
7233 heuristic defined by virtual-mappings-ratio. The default value is
7236 @item virtual-mappings-ratio
7237 If the number of virtual mappings is virtual-mappings-ratio bigger
7238 than the number of virtual symbols to be updated, then the incremental
7239 SSA updater switches to a full update for those symbols. The default
7242 @item ssp-buffer-size
7243 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
7244 protection when @option{-fstack-protection} is used.
7246 @item max-jump-thread-duplication-stmts
7247 Maximum number of statements allowed in a block that needs to be
7248 duplicated when threading jumps.
7250 @item max-fields-for-field-sensitive
7251 Maximum number of fields in a structure we will treat in
7252 a field sensitive manner during pointer analysis.
7254 @item prefetch-latency
7255 Estimate on average number of instructions that are executed before
7256 prefetch finishes. The distance we prefetch ahead is proportional
7257 to this constant. Increasing this number may also lead to less
7258 streams being prefetched (see @option{simultaneous-prefetches}).
7260 @item simultaneous-prefetches
7261 Maximum number of prefetches that can run at the same time.
7263 @item l1-cache-line-size
7264 The size of cache line in L1 cache, in bytes.
7267 The size of L1 cache, in kilobytes.
7270 The size of L2 cache, in kilobytes.
7272 @item use-canonical-types
7273 Whether the compiler should use the ``canonical'' type system. By
7274 default, this should always be 1, which uses a more efficient internal
7275 mechanism for comparing types in C++ and Objective-C++. However, if
7276 bugs in the canonical type system are causing compilation failures,
7277 set this value to 0 to disable canonical types.
7279 @item max-partial-antic-length
7280 Maximum length of the partial antic set computed during the tree
7281 partial redundancy elimination optimization (@option{-ftree-pre}) when
7282 optimizing at @option{-O3} and above. For some sorts of source code
7283 the enhanced partial redundancy elimination optimization can run away,
7284 consuming all of the memory available on the host machine. This
7285 parameter sets a limit on the length of the sets that are computed,
7286 which prevents the runaway behaviour. Setting a value of 0 for
7287 this paramter will allow an unlimited set length.
7289 @item sccvn-max-scc-size
7290 Maximum size of a strongly connected component (SCC) during SCCVN
7291 processing. If this limit is hit, SCCVN processing for the whole
7292 function will not be done and optimizations depending on it will
7293 be disabled. The default maximum SCC size is 10000.
7298 @node Preprocessor Options
7299 @section Options Controlling the Preprocessor
7300 @cindex preprocessor options
7301 @cindex options, preprocessor
7303 These options control the C preprocessor, which is run on each C source
7304 file before actual compilation.
7306 If you use the @option{-E} option, nothing is done except preprocessing.
7307 Some of these options make sense only together with @option{-E} because
7308 they cause the preprocessor output to be unsuitable for actual
7313 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7314 and pass @var{option} directly through to the preprocessor. If
7315 @var{option} contains commas, it is split into multiple options at the
7316 commas. However, many options are modified, translated or interpreted
7317 by the compiler driver before being passed to the preprocessor, and
7318 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7319 interface is undocumented and subject to change, so whenever possible
7320 you should avoid using @option{-Wp} and let the driver handle the
7323 @item -Xpreprocessor @var{option}
7324 @opindex preprocessor
7325 Pass @var{option} as an option to the preprocessor. You can use this to
7326 supply system-specific preprocessor options which GCC does not know how to
7329 If you want to pass an option that takes an argument, you must use
7330 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7333 @include cppopts.texi
7335 @node Assembler Options
7336 @section Passing Options to the Assembler
7338 @c prevent bad page break with this line
7339 You can pass options to the assembler.
7342 @item -Wa,@var{option}
7344 Pass @var{option} as an option to the assembler. If @var{option}
7345 contains commas, it is split into multiple options at the commas.
7347 @item -Xassembler @var{option}
7349 Pass @var{option} as an option to the assembler. You can use this to
7350 supply system-specific assembler options which GCC does not know how to
7353 If you want to pass an option that takes an argument, you must use
7354 @option{-Xassembler} twice, once for the option and once for the argument.
7359 @section Options for Linking
7360 @cindex link options
7361 @cindex options, linking
7363 These options come into play when the compiler links object files into
7364 an executable output file. They are meaningless if the compiler is
7365 not doing a link step.
7369 @item @var{object-file-name}
7370 A file name that does not end in a special recognized suffix is
7371 considered to name an object file or library. (Object files are
7372 distinguished from libraries by the linker according to the file
7373 contents.) If linking is done, these object files are used as input
7382 If any of these options is used, then the linker is not run, and
7383 object file names should not be used as arguments. @xref{Overall
7387 @item -l@var{library}
7388 @itemx -l @var{library}
7390 Search the library named @var{library} when linking. (The second
7391 alternative with the library as a separate argument is only for
7392 POSIX compliance and is not recommended.)
7394 It makes a difference where in the command you write this option; the
7395 linker searches and processes libraries and object files in the order they
7396 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7397 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7398 to functions in @samp{z}, those functions may not be loaded.
7400 The linker searches a standard list of directories for the library,
7401 which is actually a file named @file{lib@var{library}.a}. The linker
7402 then uses this file as if it had been specified precisely by name.
7404 The directories searched include several standard system directories
7405 plus any that you specify with @option{-L}.
7407 Normally the files found this way are library files---archive files
7408 whose members are object files. The linker handles an archive file by
7409 scanning through it for members which define symbols that have so far
7410 been referenced but not defined. But if the file that is found is an
7411 ordinary object file, it is linked in the usual fashion. The only
7412 difference between using an @option{-l} option and specifying a file name
7413 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7414 and searches several directories.
7418 You need this special case of the @option{-l} option in order to
7419 link an Objective-C or Objective-C++ program.
7422 @opindex nostartfiles
7423 Do not use the standard system startup files when linking.
7424 The standard system libraries are used normally, unless @option{-nostdlib}
7425 or @option{-nodefaultlibs} is used.
7427 @item -nodefaultlibs
7428 @opindex nodefaultlibs
7429 Do not use the standard system libraries when linking.
7430 Only the libraries you specify will be passed to the linker.
7431 The standard startup files are used normally, unless @option{-nostartfiles}
7432 is used. The compiler may generate calls to @code{memcmp},
7433 @code{memset}, @code{memcpy} and @code{memmove}.
7434 These entries are usually resolved by entries in
7435 libc. These entry points should be supplied through some other
7436 mechanism when this option is specified.
7440 Do not use the standard system startup files or libraries when linking.
7441 No startup files and only the libraries you specify will be passed to
7442 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7443 @code{memcpy} and @code{memmove}.
7444 These entries are usually resolved by entries in
7445 libc. These entry points should be supplied through some other
7446 mechanism when this option is specified.
7448 @cindex @option{-lgcc}, use with @option{-nostdlib}
7449 @cindex @option{-nostdlib} and unresolved references
7450 @cindex unresolved references and @option{-nostdlib}
7451 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7452 @cindex @option{-nodefaultlibs} and unresolved references
7453 @cindex unresolved references and @option{-nodefaultlibs}
7454 One of the standard libraries bypassed by @option{-nostdlib} and
7455 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7456 that GCC uses to overcome shortcomings of particular machines, or special
7457 needs for some languages.
7458 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7459 Collection (GCC) Internals},
7460 for more discussion of @file{libgcc.a}.)
7461 In most cases, you need @file{libgcc.a} even when you want to avoid
7462 other standard libraries. In other words, when you specify @option{-nostdlib}
7463 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7464 This ensures that you have no unresolved references to internal GCC
7465 library subroutines. (For example, @samp{__main}, used to ensure C++
7466 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7467 GNU Compiler Collection (GCC) Internals}.)
7471 Produce a position independent executable on targets which support it.
7472 For predictable results, you must also specify the same set of options
7473 that were used to generate code (@option{-fpie}, @option{-fPIE},
7474 or model suboptions) when you specify this option.
7478 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7479 that support it. This instructs the linker to add all symbols, not
7480 only used ones, to the dynamic symbol table. This option is needed
7481 for some uses of @code{dlopen} or to allow obtaining backtraces
7482 from within a program.
7486 Remove all symbol table and relocation information from the executable.
7490 On systems that support dynamic linking, this prevents linking with the shared
7491 libraries. On other systems, this option has no effect.
7495 Produce a shared object which can then be linked with other objects to
7496 form an executable. Not all systems support this option. For predictable
7497 results, you must also specify the same set of options that were used to
7498 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7499 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7500 needs to build supplementary stub code for constructors to work. On
7501 multi-libbed systems, @samp{gcc -shared} must select the correct support
7502 libraries to link against. Failing to supply the correct flags may lead
7503 to subtle defects. Supplying them in cases where they are not necessary
7506 @item -shared-libgcc
7507 @itemx -static-libgcc
7508 @opindex shared-libgcc
7509 @opindex static-libgcc
7510 On systems that provide @file{libgcc} as a shared library, these options
7511 force the use of either the shared or static version respectively.
7512 If no shared version of @file{libgcc} was built when the compiler was
7513 configured, these options have no effect.
7515 There are several situations in which an application should use the
7516 shared @file{libgcc} instead of the static version. The most common
7517 of these is when the application wishes to throw and catch exceptions
7518 across different shared libraries. In that case, each of the libraries
7519 as well as the application itself should use the shared @file{libgcc}.
7521 Therefore, the G++ and GCJ drivers automatically add
7522 @option{-shared-libgcc} whenever you build a shared library or a main
7523 executable, because C++ and Java programs typically use exceptions, so
7524 this is the right thing to do.
7526 If, instead, you use the GCC driver to create shared libraries, you may
7527 find that they will not always be linked with the shared @file{libgcc}.
7528 If GCC finds, at its configuration time, that you have a non-GNU linker
7529 or a GNU linker that does not support option @option{--eh-frame-hdr},
7530 it will link the shared version of @file{libgcc} into shared libraries
7531 by default. Otherwise, it will take advantage of the linker and optimize
7532 away the linking with the shared version of @file{libgcc}, linking with
7533 the static version of libgcc by default. This allows exceptions to
7534 propagate through such shared libraries, without incurring relocation
7535 costs at library load time.
7537 However, if a library or main executable is supposed to throw or catch
7538 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7539 for the languages used in the program, or using the option
7540 @option{-shared-libgcc}, such that it is linked with the shared
7545 Bind references to global symbols when building a shared object. Warn
7546 about any unresolved references (unless overridden by the link editor
7547 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7550 @item -Xlinker @var{option}
7552 Pass @var{option} as an option to the linker. You can use this to
7553 supply system-specific linker options which GCC does not know how to
7556 If you want to pass an option that takes an argument, you must use
7557 @option{-Xlinker} twice, once for the option and once for the argument.
7558 For example, to pass @option{-assert definitions}, you must write
7559 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7560 @option{-Xlinker "-assert definitions"}, because this passes the entire
7561 string as a single argument, which is not what the linker expects.
7563 @item -Wl,@var{option}
7565 Pass @var{option} as an option to the linker. If @var{option} contains
7566 commas, it is split into multiple options at the commas.
7568 @item -u @var{symbol}
7570 Pretend the symbol @var{symbol} is undefined, to force linking of
7571 library modules to define it. You can use @option{-u} multiple times with
7572 different symbols to force loading of additional library modules.
7575 @node Directory Options
7576 @section Options for Directory Search
7577 @cindex directory options
7578 @cindex options, directory search
7581 These options specify directories to search for header files, for
7582 libraries and for parts of the compiler:
7587 Add the directory @var{dir} to the head of the list of directories to be
7588 searched for header files. This can be used to override a system header
7589 file, substituting your own version, since these directories are
7590 searched before the system header file directories. However, you should
7591 not use this option to add directories that contain vendor-supplied
7592 system header files (use @option{-isystem} for that). If you use more than
7593 one @option{-I} option, the directories are scanned in left-to-right
7594 order; the standard system directories come after.
7596 If a standard system include directory, or a directory specified with
7597 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7598 option will be ignored. The directory will still be searched but as a
7599 system directory at its normal position in the system include chain.
7600 This is to ensure that GCC's procedure to fix buggy system headers and
7601 the ordering for the include_next directive are not inadvertently changed.
7602 If you really need to change the search order for system directories,
7603 use the @option{-nostdinc} and/or @option{-isystem} options.
7605 @item -iquote@var{dir}
7607 Add the directory @var{dir} to the head of the list of directories to
7608 be searched for header files only for the case of @samp{#include
7609 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7610 otherwise just like @option{-I}.
7614 Add directory @var{dir} to the list of directories to be searched
7617 @item -B@var{prefix}
7619 This option specifies where to find the executables, libraries,
7620 include files, and data files of the compiler itself.
7622 The compiler driver program runs one or more of the subprograms
7623 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7624 @var{prefix} as a prefix for each program it tries to run, both with and
7625 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7627 For each subprogram to be run, the compiler driver first tries the
7628 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7629 was not specified, the driver tries two standard prefixes, which are
7630 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7631 those results in a file name that is found, the unmodified program
7632 name is searched for using the directories specified in your
7633 @env{PATH} environment variable.
7635 The compiler will check to see if the path provided by the @option{-B}
7636 refers to a directory, and if necessary it will add a directory
7637 separator character at the end of the path.
7639 @option{-B} prefixes that effectively specify directory names also apply
7640 to libraries in the linker, because the compiler translates these
7641 options into @option{-L} options for the linker. They also apply to
7642 includes files in the preprocessor, because the compiler translates these
7643 options into @option{-isystem} options for the preprocessor. In this case,
7644 the compiler appends @samp{include} to the prefix.
7646 The run-time support file @file{libgcc.a} can also be searched for using
7647 the @option{-B} prefix, if needed. If it is not found there, the two
7648 standard prefixes above are tried, and that is all. The file is left
7649 out of the link if it is not found by those means.
7651 Another way to specify a prefix much like the @option{-B} prefix is to use
7652 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7655 As a special kludge, if the path provided by @option{-B} is
7656 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7657 9, then it will be replaced by @file{[dir/]include}. This is to help
7658 with boot-strapping the compiler.
7660 @item -specs=@var{file}
7662 Process @var{file} after the compiler reads in the standard @file{specs}
7663 file, in order to override the defaults that the @file{gcc} driver
7664 program uses when determining what switches to pass to @file{cc1},
7665 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7666 @option{-specs=@var{file}} can be specified on the command line, and they
7667 are processed in order, from left to right.
7669 @item --sysroot=@var{dir}
7671 Use @var{dir} as the logical root directory for headers and libraries.
7672 For example, if the compiler would normally search for headers in
7673 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7674 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7676 If you use both this option and the @option{-isysroot} option, then
7677 the @option{--sysroot} option will apply to libraries, but the
7678 @option{-isysroot} option will apply to header files.
7680 The GNU linker (beginning with version 2.16) has the necessary support
7681 for this option. If your linker does not support this option, the
7682 header file aspect of @option{--sysroot} will still work, but the
7683 library aspect will not.
7687 This option has been deprecated. Please use @option{-iquote} instead for
7688 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7689 Any directories you specify with @option{-I} options before the @option{-I-}
7690 option are searched only for the case of @samp{#include "@var{file}"};
7691 they are not searched for @samp{#include <@var{file}>}.
7693 If additional directories are specified with @option{-I} options after
7694 the @option{-I-}, these directories are searched for all @samp{#include}
7695 directives. (Ordinarily @emph{all} @option{-I} directories are used
7698 In addition, the @option{-I-} option inhibits the use of the current
7699 directory (where the current input file came from) as the first search
7700 directory for @samp{#include "@var{file}"}. There is no way to
7701 override this effect of @option{-I-}. With @option{-I.} you can specify
7702 searching the directory which was current when the compiler was
7703 invoked. That is not exactly the same as what the preprocessor does
7704 by default, but it is often satisfactory.
7706 @option{-I-} does not inhibit the use of the standard system directories
7707 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7714 @section Specifying subprocesses and the switches to pass to them
7717 @command{gcc} is a driver program. It performs its job by invoking a
7718 sequence of other programs to do the work of compiling, assembling and
7719 linking. GCC interprets its command-line parameters and uses these to
7720 deduce which programs it should invoke, and which command-line options
7721 it ought to place on their command lines. This behavior is controlled
7722 by @dfn{spec strings}. In most cases there is one spec string for each
7723 program that GCC can invoke, but a few programs have multiple spec
7724 strings to control their behavior. The spec strings built into GCC can
7725 be overridden by using the @option{-specs=} command-line switch to specify
7728 @dfn{Spec files} are plaintext files that are used to construct spec
7729 strings. They consist of a sequence of directives separated by blank
7730 lines. The type of directive is determined by the first non-whitespace
7731 character on the line and it can be one of the following:
7734 @item %@var{command}
7735 Issues a @var{command} to the spec file processor. The commands that can
7739 @item %include <@var{file}>
7741 Search for @var{file} and insert its text at the current point in the
7744 @item %include_noerr <@var{file}>
7745 @cindex %include_noerr
7746 Just like @samp{%include}, but do not generate an error message if the include
7747 file cannot be found.
7749 @item %rename @var{old_name} @var{new_name}
7751 Rename the spec string @var{old_name} to @var{new_name}.
7755 @item *[@var{spec_name}]:
7756 This tells the compiler to create, override or delete the named spec
7757 string. All lines after this directive up to the next directive or
7758 blank line are considered to be the text for the spec string. If this
7759 results in an empty string then the spec will be deleted. (Or, if the
7760 spec did not exist, then nothing will happened.) Otherwise, if the spec
7761 does not currently exist a new spec will be created. If the spec does
7762 exist then its contents will be overridden by the text of this
7763 directive, unless the first character of that text is the @samp{+}
7764 character, in which case the text will be appended to the spec.
7766 @item [@var{suffix}]:
7767 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7768 and up to the next directive or blank line are considered to make up the
7769 spec string for the indicated suffix. When the compiler encounters an
7770 input file with the named suffix, it will processes the spec string in
7771 order to work out how to compile that file. For example:
7778 This says that any input file whose name ends in @samp{.ZZ} should be
7779 passed to the program @samp{z-compile}, which should be invoked with the
7780 command-line switch @option{-input} and with the result of performing the
7781 @samp{%i} substitution. (See below.)
7783 As an alternative to providing a spec string, the text that follows a
7784 suffix directive can be one of the following:
7787 @item @@@var{language}
7788 This says that the suffix is an alias for a known @var{language}. This is
7789 similar to using the @option{-x} command-line switch to GCC to specify a
7790 language explicitly. For example:
7797 Says that .ZZ files are, in fact, C++ source files.
7800 This causes an error messages saying:
7803 @var{name} compiler not installed on this system.
7807 GCC already has an extensive list of suffixes built into it.
7808 This directive will add an entry to the end of the list of suffixes, but
7809 since the list is searched from the end backwards, it is effectively
7810 possible to override earlier entries using this technique.
7814 GCC has the following spec strings built into it. Spec files can
7815 override these strings or create their own. Note that individual
7816 targets can also add their own spec strings to this list.
7819 asm Options to pass to the assembler
7820 asm_final Options to pass to the assembler post-processor
7821 cpp Options to pass to the C preprocessor
7822 cc1 Options to pass to the C compiler
7823 cc1plus Options to pass to the C++ compiler
7824 endfile Object files to include at the end of the link
7825 link Options to pass to the linker
7826 lib Libraries to include on the command line to the linker
7827 libgcc Decides which GCC support library to pass to the linker
7828 linker Sets the name of the linker
7829 predefines Defines to be passed to the C preprocessor
7830 signed_char Defines to pass to CPP to say whether @code{char} is signed
7832 startfile Object files to include at the start of the link
7835 Here is a small example of a spec file:
7841 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7844 This example renames the spec called @samp{lib} to @samp{old_lib} and
7845 then overrides the previous definition of @samp{lib} with a new one.
7846 The new definition adds in some extra command-line options before
7847 including the text of the old definition.
7849 @dfn{Spec strings} are a list of command-line options to be passed to their
7850 corresponding program. In addition, the spec strings can contain
7851 @samp{%}-prefixed sequences to substitute variable text or to
7852 conditionally insert text into the command line. Using these constructs
7853 it is possible to generate quite complex command lines.
7855 Here is a table of all defined @samp{%}-sequences for spec
7856 strings. Note that spaces are not generated automatically around the
7857 results of expanding these sequences. Therefore you can concatenate them
7858 together or combine them with constant text in a single argument.
7862 Substitute one @samp{%} into the program name or argument.
7865 Substitute the name of the input file being processed.
7868 Substitute the basename of the input file being processed.
7869 This is the substring up to (and not including) the last period
7870 and not including the directory.
7873 This is the same as @samp{%b}, but include the file suffix (text after
7877 Marks the argument containing or following the @samp{%d} as a
7878 temporary file name, so that that file will be deleted if GCC exits
7879 successfully. Unlike @samp{%g}, this contributes no text to the
7882 @item %g@var{suffix}
7883 Substitute a file name that has suffix @var{suffix} and is chosen
7884 once per compilation, and mark the argument in the same way as
7885 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7886 name is now chosen in a way that is hard to predict even when previously
7887 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7888 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7889 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7890 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7891 was simply substituted with a file name chosen once per compilation,
7892 without regard to any appended suffix (which was therefore treated
7893 just like ordinary text), making such attacks more likely to succeed.
7895 @item %u@var{suffix}
7896 Like @samp{%g}, but generates a new temporary file name even if
7897 @samp{%u@var{suffix}} was already seen.
7899 @item %U@var{suffix}
7900 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7901 new one if there is no such last file name. In the absence of any
7902 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7903 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7904 would involve the generation of two distinct file names, one
7905 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7906 simply substituted with a file name chosen for the previous @samp{%u},
7907 without regard to any appended suffix.
7909 @item %j@var{suffix}
7910 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7911 writable, and if save-temps is off; otherwise, substitute the name
7912 of a temporary file, just like @samp{%u}. This temporary file is not
7913 meant for communication between processes, but rather as a junk
7916 @item %|@var{suffix}
7917 @itemx %m@var{suffix}
7918 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7919 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7920 all. These are the two most common ways to instruct a program that it
7921 should read from standard input or write to standard output. If you
7922 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7923 construct: see for example @file{f/lang-specs.h}.
7925 @item %.@var{SUFFIX}
7926 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7927 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7928 terminated by the next space or %.
7931 Marks the argument containing or following the @samp{%w} as the
7932 designated output file of this compilation. This puts the argument
7933 into the sequence of arguments that @samp{%o} will substitute later.
7936 Substitutes the names of all the output files, with spaces
7937 automatically placed around them. You should write spaces
7938 around the @samp{%o} as well or the results are undefined.
7939 @samp{%o} is for use in the specs for running the linker.
7940 Input files whose names have no recognized suffix are not compiled
7941 at all, but they are included among the output files, so they will
7945 Substitutes the suffix for object files. Note that this is
7946 handled specially when it immediately follows @samp{%g, %u, or %U},
7947 because of the need for those to form complete file names. The
7948 handling is such that @samp{%O} is treated exactly as if it had already
7949 been substituted, except that @samp{%g, %u, and %U} do not currently
7950 support additional @var{suffix} characters following @samp{%O} as they would
7951 following, for example, @samp{.o}.
7954 Substitutes the standard macro predefinitions for the
7955 current target machine. Use this when running @code{cpp}.
7958 Like @samp{%p}, but puts @samp{__} before and after the name of each
7959 predefined macro, except for macros that start with @samp{__} or with
7960 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7964 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7965 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7966 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7967 and @option{-imultilib} as necessary.
7970 Current argument is the name of a library or startup file of some sort.
7971 Search for that file in a standard list of directories and substitute
7972 the full name found.
7975 Print @var{str} as an error message. @var{str} is terminated by a newline.
7976 Use this when inconsistent options are detected.
7979 Substitute the contents of spec string @var{name} at this point.
7982 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7984 @item %x@{@var{option}@}
7985 Accumulate an option for @samp{%X}.
7988 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7992 Output the accumulated assembler options specified by @option{-Wa}.
7995 Output the accumulated preprocessor options specified by @option{-Wp}.
7998 Process the @code{asm} spec. This is used to compute the
7999 switches to be passed to the assembler.
8002 Process the @code{asm_final} spec. This is a spec string for
8003 passing switches to an assembler post-processor, if such a program is
8007 Process the @code{link} spec. This is the spec for computing the
8008 command line passed to the linker. Typically it will make use of the
8009 @samp{%L %G %S %D and %E} sequences.
8012 Dump out a @option{-L} option for each directory that GCC believes might
8013 contain startup files. If the target supports multilibs then the
8014 current multilib directory will be prepended to each of these paths.
8017 Process the @code{lib} spec. This is a spec string for deciding which
8018 libraries should be included on the command line to the linker.
8021 Process the @code{libgcc} spec. This is a spec string for deciding
8022 which GCC support library should be included on the command line to the linker.
8025 Process the @code{startfile} spec. This is a spec for deciding which
8026 object files should be the first ones passed to the linker. Typically
8027 this might be a file named @file{crt0.o}.
8030 Process the @code{endfile} spec. This is a spec string that specifies
8031 the last object files that will be passed to the linker.
8034 Process the @code{cpp} spec. This is used to construct the arguments
8035 to be passed to the C preprocessor.
8038 Process the @code{cc1} spec. This is used to construct the options to be
8039 passed to the actual C compiler (@samp{cc1}).
8042 Process the @code{cc1plus} spec. This is used to construct the options to be
8043 passed to the actual C++ compiler (@samp{cc1plus}).
8046 Substitute the variable part of a matched option. See below.
8047 Note that each comma in the substituted string is replaced by
8051 Remove all occurrences of @code{-S} from the command line. Note---this
8052 command is position dependent. @samp{%} commands in the spec string
8053 before this one will see @code{-S}, @samp{%} commands in the spec string
8054 after this one will not.
8056 @item %:@var{function}(@var{args})
8057 Call the named function @var{function}, passing it @var{args}.
8058 @var{args} is first processed as a nested spec string, then split
8059 into an argument vector in the usual fashion. The function returns
8060 a string which is processed as if it had appeared literally as part
8061 of the current spec.
8063 The following built-in spec functions are provided:
8067 The @code{getenv} spec function takes two arguments: an environment
8068 variable name and a string. If the environment variable is not
8069 defined, a fatal error is issued. Otherwise, the return value is the
8070 value of the environment variable concatenated with the string. For
8071 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8074 %:getenv(TOPDIR /include)
8077 expands to @file{/path/to/top/include}.
8079 @item @code{if-exists}
8080 The @code{if-exists} spec function takes one argument, an absolute
8081 pathname to a file. If the file exists, @code{if-exists} returns the
8082 pathname. Here is a small example of its usage:
8086 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
8089 @item @code{if-exists-else}
8090 The @code{if-exists-else} spec function is similar to the @code{if-exists}
8091 spec function, except that it takes two arguments. The first argument is
8092 an absolute pathname to a file. If the file exists, @code{if-exists-else}
8093 returns the pathname. If it does not exist, it returns the second argument.
8094 This way, @code{if-exists-else} can be used to select one file or another,
8095 based on the existence of the first. Here is a small example of its usage:
8099 crt0%O%s %:if-exists(crti%O%s) \
8100 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8103 @item @code{replace-outfile}
8104 The @code{replace-outfile} spec function takes two arguments. It looks for the
8105 first argument in the outfiles array and replaces it with the second argument. Here
8106 is a small example of its usage:
8109 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8112 @item @code{print-asm-header}
8113 The @code{print-asm-header} function takes no arguments and simply
8114 prints a banner like:
8120 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8123 It is used to separate compiler options from assembler options
8124 in the @option{--target-help} output.
8128 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8129 If that switch was not specified, this substitutes nothing. Note that
8130 the leading dash is omitted when specifying this option, and it is
8131 automatically inserted if the substitution is performed. Thus the spec
8132 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8133 and would output the command line option @option{-foo}.
8135 @item %W@{@code{S}@}
8136 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8139 @item %@{@code{S}*@}
8140 Substitutes all the switches specified to GCC whose names start
8141 with @code{-S}, but which also take an argument. This is used for
8142 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8143 GCC considers @option{-o foo} as being
8144 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
8145 text, including the space. Thus two arguments would be generated.
8147 @item %@{@code{S}*&@code{T}*@}
8148 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8149 (the order of @code{S} and @code{T} in the spec is not significant).
8150 There can be any number of ampersand-separated variables; for each the
8151 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8153 @item %@{@code{S}:@code{X}@}
8154 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8156 @item %@{!@code{S}:@code{X}@}
8157 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8159 @item %@{@code{S}*:@code{X}@}
8160 Substitutes @code{X} if one or more switches whose names start with
8161 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8162 once, no matter how many such switches appeared. However, if @code{%*}
8163 appears somewhere in @code{X}, then @code{X} will be substituted once
8164 for each matching switch, with the @code{%*} replaced by the part of
8165 that switch that matched the @code{*}.
8167 @item %@{.@code{S}:@code{X}@}
8168 Substitutes @code{X}, if processing a file with suffix @code{S}.
8170 @item %@{!.@code{S}:@code{X}@}
8171 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8173 @item %@{,@code{S}:@code{X}@}
8174 Substitutes @code{X}, if processing a file for language @code{S}.
8176 @item %@{!,@code{S}:@code{X}@}
8177 Substitutes @code{X}, if not processing a file for language @code{S}.
8179 @item %@{@code{S}|@code{P}:@code{X}@}
8180 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8181 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8182 @code{*} sequences as well, although they have a stronger binding than
8183 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8184 alternatives must be starred, and only the first matching alternative
8187 For example, a spec string like this:
8190 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8193 will output the following command-line options from the following input
8194 command-line options:
8199 -d fred.c -foo -baz -boggle
8200 -d jim.d -bar -baz -boggle
8203 @item %@{S:X; T:Y; :D@}
8205 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8206 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8207 be as many clauses as you need. This may be combined with @code{.},
8208 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8213 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8214 construct may contain other nested @samp{%} constructs or spaces, or
8215 even newlines. They are processed as usual, as described above.
8216 Trailing white space in @code{X} is ignored. White space may also
8217 appear anywhere on the left side of the colon in these constructs,
8218 except between @code{.} or @code{*} and the corresponding word.
8220 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8221 handled specifically in these constructs. If another value of
8222 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8223 @option{-W} switch is found later in the command line, the earlier
8224 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8225 just one letter, which passes all matching options.
8227 The character @samp{|} at the beginning of the predicate text is used to
8228 indicate that a command should be piped to the following command, but
8229 only if @option{-pipe} is specified.
8231 It is built into GCC which switches take arguments and which do not.
8232 (You might think it would be useful to generalize this to allow each
8233 compiler's spec to say which switches take arguments. But this cannot
8234 be done in a consistent fashion. GCC cannot even decide which input
8235 files have been specified without knowing which switches take arguments,
8236 and it must know which input files to compile in order to tell which
8239 GCC also knows implicitly that arguments starting in @option{-l} are to be
8240 treated as compiler output files, and passed to the linker in their
8241 proper position among the other output files.
8243 @c man begin OPTIONS
8245 @node Target Options
8246 @section Specifying Target Machine and Compiler Version
8247 @cindex target options
8248 @cindex cross compiling
8249 @cindex specifying machine version
8250 @cindex specifying compiler version and target machine
8251 @cindex compiler version, specifying
8252 @cindex target machine, specifying
8254 The usual way to run GCC is to run the executable called @file{gcc}, or
8255 @file{<machine>-gcc} when cross-compiling, or
8256 @file{<machine>-gcc-<version>} to run a version other than the one that
8257 was installed last. Sometimes this is inconvenient, so GCC provides
8258 options that will switch to another cross-compiler or version.
8261 @item -b @var{machine}
8263 The argument @var{machine} specifies the target machine for compilation.
8265 The value to use for @var{machine} is the same as was specified as the
8266 machine type when configuring GCC as a cross-compiler. For
8267 example, if a cross-compiler was configured with @samp{configure
8268 arm-elf}, meaning to compile for an arm processor with elf binaries,
8269 then you would specify @option{-b arm-elf} to run that cross compiler.
8270 Because there are other options beginning with @option{-b}, the
8271 configuration must contain a hyphen.
8273 @item -V @var{version}
8275 The argument @var{version} specifies which version of GCC to run.
8276 This is useful when multiple versions are installed. For example,
8277 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8280 The @option{-V} and @option{-b} options work by running the
8281 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8282 use them if you can just run that directly.
8284 @node Submodel Options
8285 @section Hardware Models and Configurations
8286 @cindex submodel options
8287 @cindex specifying hardware config
8288 @cindex hardware models and configurations, specifying
8289 @cindex machine dependent options
8291 Earlier we discussed the standard option @option{-b} which chooses among
8292 different installed compilers for completely different target
8293 machines, such as VAX vs.@: 68000 vs.@: 80386.
8295 In addition, each of these target machine types can have its own
8296 special options, starting with @samp{-m}, to choose among various
8297 hardware models or configurations---for example, 68010 vs 68020,
8298 floating coprocessor or none. A single installed version of the
8299 compiler can compile for any model or configuration, according to the
8302 Some configurations of the compiler also support additional special
8303 options, usually for compatibility with other compilers on the same
8306 @c This list is ordered alphanumerically by subsection name.
8307 @c It should be the same order and spelling as these options are listed
8308 @c in Machine Dependent Options
8314 * Blackfin Options::
8318 * DEC Alpha Options::
8319 * DEC Alpha/VMS Options::
8321 * GNU/Linux Options::
8324 * i386 and x86-64 Options::
8337 * RS/6000 and PowerPC Options::
8338 * S/390 and zSeries Options::
8343 * System V Options::
8348 * Xstormy16 Options::
8354 @subsection ARC Options
8357 These options are defined for ARC implementations:
8362 Compile code for little endian mode. This is the default.
8366 Compile code for big endian mode.
8369 @opindex mmangle-cpu
8370 Prepend the name of the cpu to all public symbol names.
8371 In multiple-processor systems, there are many ARC variants with different
8372 instruction and register set characteristics. This flag prevents code
8373 compiled for one cpu to be linked with code compiled for another.
8374 No facility exists for handling variants that are ``almost identical''.
8375 This is an all or nothing option.
8377 @item -mcpu=@var{cpu}
8379 Compile code for ARC variant @var{cpu}.
8380 Which variants are supported depend on the configuration.
8381 All variants support @option{-mcpu=base}, this is the default.
8383 @item -mtext=@var{text-section}
8384 @itemx -mdata=@var{data-section}
8385 @itemx -mrodata=@var{readonly-data-section}
8389 Put functions, data, and readonly data in @var{text-section},
8390 @var{data-section}, and @var{readonly-data-section} respectively
8391 by default. This can be overridden with the @code{section} attribute.
8392 @xref{Variable Attributes}.
8397 @subsection ARM Options
8400 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8404 @item -mabi=@var{name}
8406 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8407 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8410 @opindex mapcs-frame
8411 Generate a stack frame that is compliant with the ARM Procedure Call
8412 Standard for all functions, even if this is not strictly necessary for
8413 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8414 with this option will cause the stack frames not to be generated for
8415 leaf functions. The default is @option{-mno-apcs-frame}.
8419 This is a synonym for @option{-mapcs-frame}.
8422 @c not currently implemented
8423 @item -mapcs-stack-check
8424 @opindex mapcs-stack-check
8425 Generate code to check the amount of stack space available upon entry to
8426 every function (that actually uses some stack space). If there is
8427 insufficient space available then either the function
8428 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8429 called, depending upon the amount of stack space required. The run time
8430 system is required to provide these functions. The default is
8431 @option{-mno-apcs-stack-check}, since this produces smaller code.
8433 @c not currently implemented
8435 @opindex mapcs-float
8436 Pass floating point arguments using the float point registers. This is
8437 one of the variants of the APCS@. This option is recommended if the
8438 target hardware has a floating point unit or if a lot of floating point
8439 arithmetic is going to be performed by the code. The default is
8440 @option{-mno-apcs-float}, since integer only code is slightly increased in
8441 size if @option{-mapcs-float} is used.
8443 @c not currently implemented
8444 @item -mapcs-reentrant
8445 @opindex mapcs-reentrant
8446 Generate reentrant, position independent code. The default is
8447 @option{-mno-apcs-reentrant}.
8450 @item -mthumb-interwork
8451 @opindex mthumb-interwork
8452 Generate code which supports calling between the ARM and Thumb
8453 instruction sets. Without this option the two instruction sets cannot
8454 be reliably used inside one program. The default is
8455 @option{-mno-thumb-interwork}, since slightly larger code is generated
8456 when @option{-mthumb-interwork} is specified.
8458 @item -mno-sched-prolog
8459 @opindex mno-sched-prolog
8460 Prevent the reordering of instructions in the function prolog, or the
8461 merging of those instruction with the instructions in the function's
8462 body. This means that all functions will start with a recognizable set
8463 of instructions (or in fact one of a choice from a small set of
8464 different function prologues), and this information can be used to
8465 locate the start if functions inside an executable piece of code. The
8466 default is @option{-msched-prolog}.
8469 @opindex mhard-float
8470 Generate output containing floating point instructions. This is the
8474 @opindex msoft-float
8475 Generate output containing library calls for floating point.
8476 @strong{Warning:} the requisite libraries are not available for all ARM
8477 targets. Normally the facilities of the machine's usual C compiler are
8478 used, but this cannot be done directly in cross-compilation. You must make
8479 your own arrangements to provide suitable library functions for
8482 @option{-msoft-float} changes the calling convention in the output file;
8483 therefore, it is only useful if you compile @emph{all} of a program with
8484 this option. In particular, you need to compile @file{libgcc.a}, the
8485 library that comes with GCC, with @option{-msoft-float} in order for
8488 @item -mfloat-abi=@var{name}
8490 Specifies which ABI to use for floating point values. Permissible values
8491 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8493 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8494 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8495 of floating point instructions, but still uses the soft-float calling
8498 @item -mlittle-endian
8499 @opindex mlittle-endian
8500 Generate code for a processor running in little-endian mode. This is
8501 the default for all standard configurations.
8504 @opindex mbig-endian
8505 Generate code for a processor running in big-endian mode; the default is
8506 to compile code for a little-endian processor.
8508 @item -mwords-little-endian
8509 @opindex mwords-little-endian
8510 This option only applies when generating code for big-endian processors.
8511 Generate code for a little-endian word order but a big-endian byte
8512 order. That is, a byte order of the form @samp{32107654}. Note: this
8513 option should only be used if you require compatibility with code for
8514 big-endian ARM processors generated by versions of the compiler prior to
8517 @item -mcpu=@var{name}
8519 This specifies the name of the target ARM processor. GCC uses this name
8520 to determine what kind of instructions it can emit when generating
8521 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8522 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8523 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8524 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8525 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8526 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8527 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8528 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8529 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8530 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8531 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8532 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8533 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8534 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8535 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
8536 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8538 @itemx -mtune=@var{name}
8540 This option is very similar to the @option{-mcpu=} option, except that
8541 instead of specifying the actual target processor type, and hence
8542 restricting which instructions can be used, it specifies that GCC should
8543 tune the performance of the code as if the target were of the type
8544 specified in this option, but still choosing the instructions that it
8545 will generate based on the cpu specified by a @option{-mcpu=} option.
8546 For some ARM implementations better performance can be obtained by using
8549 @item -march=@var{name}
8551 This specifies the name of the target ARM architecture. GCC uses this
8552 name to determine what kind of instructions it can emit when generating
8553 assembly code. This option can be used in conjunction with or instead
8554 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8555 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8556 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8557 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
8558 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
8560 @item -mfpu=@var{name}
8561 @itemx -mfpe=@var{number}
8562 @itemx -mfp=@var{number}
8566 This specifies what floating point hardware (or hardware emulation) is
8567 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8568 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
8569 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8570 with older versions of GCC@.
8572 If @option{-msoft-float} is specified this specifies the format of
8573 floating point values.
8575 @item -mstructure-size-boundary=@var{n}
8576 @opindex mstructure-size-boundary
8577 The size of all structures and unions will be rounded up to a multiple
8578 of the number of bits set by this option. Permissible values are 8, 32
8579 and 64. The default value varies for different toolchains. For the COFF
8580 targeted toolchain the default value is 8. A value of 64 is only allowed
8581 if the underlying ABI supports it.
8583 Specifying the larger number can produce faster, more efficient code, but
8584 can also increase the size of the program. Different values are potentially
8585 incompatible. Code compiled with one value cannot necessarily expect to
8586 work with code or libraries compiled with another value, if they exchange
8587 information using structures or unions.
8589 @item -mabort-on-noreturn
8590 @opindex mabort-on-noreturn
8591 Generate a call to the function @code{abort} at the end of a
8592 @code{noreturn} function. It will be executed if the function tries to
8596 @itemx -mno-long-calls
8597 @opindex mlong-calls
8598 @opindex mno-long-calls
8599 Tells the compiler to perform function calls by first loading the
8600 address of the function into a register and then performing a subroutine
8601 call on this register. This switch is needed if the target function
8602 will lie outside of the 64 megabyte addressing range of the offset based
8603 version of subroutine call instruction.
8605 Even if this switch is enabled, not all function calls will be turned
8606 into long calls. The heuristic is that static functions, functions
8607 which have the @samp{short-call} attribute, functions that are inside
8608 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8609 definitions have already been compiled within the current compilation
8610 unit, will not be turned into long calls. The exception to this rule is
8611 that weak function definitions, functions with the @samp{long-call}
8612 attribute or the @samp{section} attribute, and functions that are within
8613 the scope of a @samp{#pragma long_calls} directive, will always be
8614 turned into long calls.
8616 This feature is not enabled by default. Specifying
8617 @option{-mno-long-calls} will restore the default behavior, as will
8618 placing the function calls within the scope of a @samp{#pragma
8619 long_calls_off} directive. Note these switches have no effect on how
8620 the compiler generates code to handle function calls via function
8623 @item -mnop-fun-dllimport
8624 @opindex mnop-fun-dllimport
8625 Disable support for the @code{dllimport} attribute.
8627 @item -msingle-pic-base
8628 @opindex msingle-pic-base
8629 Treat the register used for PIC addressing as read-only, rather than
8630 loading it in the prologue for each function. The run-time system is
8631 responsible for initializing this register with an appropriate value
8632 before execution begins.
8634 @item -mpic-register=@var{reg}
8635 @opindex mpic-register
8636 Specify the register to be used for PIC addressing. The default is R10
8637 unless stack-checking is enabled, when R9 is used.
8639 @item -mcirrus-fix-invalid-insns
8640 @opindex mcirrus-fix-invalid-insns
8641 @opindex mno-cirrus-fix-invalid-insns
8642 Insert NOPs into the instruction stream to in order to work around
8643 problems with invalid Maverick instruction combinations. This option
8644 is only valid if the @option{-mcpu=ep9312} option has been used to
8645 enable generation of instructions for the Cirrus Maverick floating
8646 point co-processor. This option is not enabled by default, since the
8647 problem is only present in older Maverick implementations. The default
8648 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8651 @item -mpoke-function-name
8652 @opindex mpoke-function-name
8653 Write the name of each function into the text section, directly
8654 preceding the function prologue. The generated code is similar to this:
8658 .ascii "arm_poke_function_name", 0
8661 .word 0xff000000 + (t1 - t0)
8662 arm_poke_function_name
8664 stmfd sp!, @{fp, ip, lr, pc@}
8668 When performing a stack backtrace, code can inspect the value of
8669 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8670 location @code{pc - 12} and the top 8 bits are set, then we know that
8671 there is a function name embedded immediately preceding this location
8672 and has length @code{((pc[-3]) & 0xff000000)}.
8676 Generate code for the Thumb instruction set. The default is to
8677 use the 32-bit ARM instruction set.
8678 This option automatically enables either 16-bit Thumb-1 or
8679 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8680 and @option{-march=@var{name}} options.
8683 @opindex mtpcs-frame
8684 Generate a stack frame that is compliant with the Thumb Procedure Call
8685 Standard for all non-leaf functions. (A leaf function is one that does
8686 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8688 @item -mtpcs-leaf-frame
8689 @opindex mtpcs-leaf-frame
8690 Generate a stack frame that is compliant with the Thumb Procedure Call
8691 Standard for all leaf functions. (A leaf function is one that does
8692 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8694 @item -mcallee-super-interworking
8695 @opindex mcallee-super-interworking
8696 Gives all externally visible functions in the file being compiled an ARM
8697 instruction set header which switches to Thumb mode before executing the
8698 rest of the function. This allows these functions to be called from
8699 non-interworking code.
8701 @item -mcaller-super-interworking
8702 @opindex mcaller-super-interworking
8703 Allows calls via function pointers (including virtual functions) to
8704 execute correctly regardless of whether the target code has been
8705 compiled for interworking or not. There is a small overhead in the cost
8706 of executing a function pointer if this option is enabled.
8708 @item -mtp=@var{name}
8710 Specify the access model for the thread local storage pointer. The valid
8711 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8712 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8713 (supported in the arm6k architecture), and @option{auto}, which uses the
8714 best available method for the selected processor. The default setting is
8720 @subsection AVR Options
8723 These options are defined for AVR implementations:
8726 @item -mmcu=@var{mcu}
8728 Specify ATMEL AVR instruction set or MCU type.
8730 Instruction set avr1 is for the minimal AVR core, not supported by the C
8731 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8732 attiny11, attiny12, attiny15, attiny28).
8734 Instruction set avr2 (default) is for the classic AVR core with up to
8735 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8736 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8737 at90c8534, at90s8535).
8739 Instruction set avr3 is for the classic AVR core with up to 128K program
8740 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8742 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8743 memory space (MCU types: atmega8, atmega83, atmega85).
8745 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8746 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8747 atmega64, atmega128, at43usb355, at94k).
8751 Output instruction sizes to the asm file.
8753 @item -minit-stack=@var{N}
8754 @opindex minit-stack
8755 Specify the initial stack address, which may be a symbol or numeric value,
8756 @samp{__stack} is the default.
8758 @item -mno-interrupts
8759 @opindex mno-interrupts
8760 Generated code is not compatible with hardware interrupts.
8761 Code size will be smaller.
8763 @item -mcall-prologues
8764 @opindex mcall-prologues
8765 Functions prologues/epilogues expanded as call to appropriate
8766 subroutines. Code size will be smaller.
8768 @item -mno-tablejump
8769 @opindex mno-tablejump
8770 Do not generate tablejump insns which sometimes increase code size.
8773 @opindex mtiny-stack
8774 Change only the low 8 bits of the stack pointer.
8778 Assume int to be 8 bit integer. This affects the sizes of all types: A
8779 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8780 and long long will be 4 bytes. Please note that this option does not
8781 comply to the C standards, but it will provide you with smaller code
8785 @node Blackfin Options
8786 @subsection Blackfin Options
8787 @cindex Blackfin Options
8790 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
8792 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8793 can be one of @samp{bf522}, @samp{bf523}, @samp{bf524},
8794 @samp{bf525}, @samp{bf526}, @samp{bf527},
8795 @samp{bf531}, @samp{bf532}, @samp{bf533}, @samp{bf534},
8796 @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
8797 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
8799 The optional @var{sirevision} specifies the silicon revision of the target
8800 Blackfin processor. Any workarounds available for the targeted silicon revision
8801 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
8802 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
8803 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
8804 hexadecimal digits representing the major and minor numbers in the silicon
8805 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
8806 is not defined. If @var{sirevision} is @samp{any}, the
8807 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
8808 If this optional @var{sirevision} is not used, GCC assumes the latest known
8809 silicon revision of the targeted Blackfin processor.
8811 Support for @samp{bf561} is incomplete. For @samp{bf561},
8812 Only the processor macro is defined.
8813 Without this option, @samp{bf532} is used as the processor by default.
8814 The corresponding predefined processor macros for @var{cpu} is to
8815 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
8816 provided by libgloss to be linked in if @option{-msim} is not given.
8820 Specifies that the program will be run on the simulator. This causes
8821 the simulator BSP provided by libgloss to be linked in. This option
8822 has effect only for @samp{bfin-elf} toolchain.
8823 Certain other options, such as @option{-mid-shared-library} and
8824 @option{-mfdpic}, imply @option{-msim}.
8826 @item -momit-leaf-frame-pointer
8827 @opindex momit-leaf-frame-pointer
8828 Don't keep the frame pointer in a register for leaf functions. This
8829 avoids the instructions to save, set up and restore frame pointers and
8830 makes an extra register available in leaf functions. The option
8831 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8832 which might make debugging harder.
8834 @item -mspecld-anomaly
8835 @opindex mspecld-anomaly
8836 When enabled, the compiler will ensure that the generated code does not
8837 contain speculative loads after jump instructions. If this option is used,
8838 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
8840 @item -mno-specld-anomaly
8841 @opindex mno-specld-anomaly
8842 Don't generate extra code to prevent speculative loads from occurring.
8844 @item -mcsync-anomaly
8845 @opindex mcsync-anomaly
8846 When enabled, the compiler will ensure that the generated code does not
8847 contain CSYNC or SSYNC instructions too soon after conditional branches.
8848 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
8850 @item -mno-csync-anomaly
8851 @opindex mno-csync-anomaly
8852 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8853 occurring too soon after a conditional branch.
8857 When enabled, the compiler is free to take advantage of the knowledge that
8858 the entire program fits into the low 64k of memory.
8861 @opindex mno-low-64k
8862 Assume that the program is arbitrarily large. This is the default.
8864 @item -mstack-check-l1
8865 @opindex mstack-check-l1
8866 Do stack checking using information placed into L1 scratchpad memory by the
8869 @item -mid-shared-library
8870 @opindex mid-shared-library
8871 Generate code that supports shared libraries via the library ID method.
8872 This allows for execute in place and shared libraries in an environment
8873 without virtual memory management. This option implies @option{-fPIC}.
8874 With a @samp{bfin-elf} target, this option implies @option{-msim}.
8876 @item -mno-id-shared-library
8877 @opindex mno-id-shared-library
8878 Generate code that doesn't assume ID based shared libraries are being used.
8879 This is the default.
8881 @item -mleaf-id-shared-library
8882 @opindex mleaf-id-shared-library
8883 Generate code that supports shared libraries via the library ID method,
8884 but assumes that this library or executable won't link against any other
8885 ID shared libraries. That allows the compiler to use faster code for jumps
8888 @item -mno-leaf-id-shared-library
8889 @opindex mno-leaf-id-shared-library
8890 Do not assume that the code being compiled won't link against any ID shared
8891 libraries. Slower code will be generated for jump and call insns.
8893 @item -mshared-library-id=n
8894 @opindex mshared-library-id
8895 Specified the identification number of the ID based shared library being
8896 compiled. Specifying a value of 0 will generate more compact code, specifying
8897 other values will force the allocation of that number to the current
8898 library but is no more space or time efficient than omitting this option.
8902 Generate code that allows the data segment to be located in a different
8903 area of memory from the text segment. This allows for execute in place in
8904 an environment without virtual memory management by eliminating relocations
8905 against the text section.
8908 @opindex mno-sep-data
8909 Generate code that assumes that the data segment follows the text segment.
8910 This is the default.
8913 @itemx -mno-long-calls
8914 @opindex mlong-calls
8915 @opindex mno-long-calls
8916 Tells the compiler to perform function calls by first loading the
8917 address of the function into a register and then performing a subroutine
8918 call on this register. This switch is needed if the target function
8919 will lie outside of the 24 bit addressing range of the offset based
8920 version of subroutine call instruction.
8922 This feature is not enabled by default. Specifying
8923 @option{-mno-long-calls} will restore the default behavior. Note these
8924 switches have no effect on how the compiler generates code to handle
8925 function calls via function pointers.
8929 Link with the fast floating-point library. This library relaxes some of
8930 the IEEE floating-point standard's rules for checking inputs against
8931 Not-a-Number (NAN), in the interest of performance.
8934 @opindex minline-plt
8935 Enable inlining of PLT entries in function calls to functions that are
8936 not known to bind locally. It has no effect without @option{-mfdpic}.
8940 @subsection CRIS Options
8941 @cindex CRIS Options
8943 These options are defined specifically for the CRIS ports.
8946 @item -march=@var{architecture-type}
8947 @itemx -mcpu=@var{architecture-type}
8950 Generate code for the specified architecture. The choices for
8951 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8952 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8953 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8956 @item -mtune=@var{architecture-type}
8958 Tune to @var{architecture-type} everything applicable about the generated
8959 code, except for the ABI and the set of available instructions. The
8960 choices for @var{architecture-type} are the same as for
8961 @option{-march=@var{architecture-type}}.
8963 @item -mmax-stack-frame=@var{n}
8964 @opindex mmax-stack-frame
8965 Warn when the stack frame of a function exceeds @var{n} bytes.
8967 @item -melinux-stacksize=@var{n}
8968 @opindex melinux-stacksize
8969 Only available with the @samp{cris-axis-aout} target. Arranges for
8970 indications in the program to the kernel loader that the stack of the
8971 program should be set to @var{n} bytes.
8977 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8978 @option{-march=v3} and @option{-march=v8} respectively.
8980 @item -mmul-bug-workaround
8981 @itemx -mno-mul-bug-workaround
8982 @opindex mmul-bug-workaround
8983 @opindex mno-mul-bug-workaround
8984 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8985 models where it applies. This option is active by default.
8989 Enable CRIS-specific verbose debug-related information in the assembly
8990 code. This option also has the effect to turn off the @samp{#NO_APP}
8991 formatted-code indicator to the assembler at the beginning of the
8996 Do not use condition-code results from previous instruction; always emit
8997 compare and test instructions before use of condition codes.
8999 @item -mno-side-effects
9000 @opindex mno-side-effects
9001 Do not emit instructions with side-effects in addressing modes other than
9005 @itemx -mno-stack-align
9007 @itemx -mno-data-align
9008 @itemx -mconst-align
9009 @itemx -mno-const-align
9010 @opindex mstack-align
9011 @opindex mno-stack-align
9012 @opindex mdata-align
9013 @opindex mno-data-align
9014 @opindex mconst-align
9015 @opindex mno-const-align
9016 These options (no-options) arranges (eliminate arrangements) for the
9017 stack-frame, individual data and constants to be aligned for the maximum
9018 single data access size for the chosen CPU model. The default is to
9019 arrange for 32-bit alignment. ABI details such as structure layout are
9020 not affected by these options.
9028 Similar to the stack- data- and const-align options above, these options
9029 arrange for stack-frame, writable data and constants to all be 32-bit,
9030 16-bit or 8-bit aligned. The default is 32-bit alignment.
9032 @item -mno-prologue-epilogue
9033 @itemx -mprologue-epilogue
9034 @opindex mno-prologue-epilogue
9035 @opindex mprologue-epilogue
9036 With @option{-mno-prologue-epilogue}, the normal function prologue and
9037 epilogue that sets up the stack-frame are omitted and no return
9038 instructions or return sequences are generated in the code. Use this
9039 option only together with visual inspection of the compiled code: no
9040 warnings or errors are generated when call-saved registers must be saved,
9041 or storage for local variable needs to be allocated.
9047 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9048 instruction sequences that load addresses for functions from the PLT part
9049 of the GOT rather than (traditional on other architectures) calls to the
9050 PLT@. The default is @option{-mgotplt}.
9054 Legacy no-op option only recognized with the cris-axis-aout target.
9058 Legacy no-op option only recognized with the cris-axis-elf and
9059 cris-axis-linux-gnu targets.
9063 Only recognized with the cris-axis-aout target, where it selects a
9064 GNU/linux-like multilib, include files and instruction set for
9069 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9073 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
9074 to link with input-output functions from a simulator library. Code,
9075 initialized data and zero-initialized data are allocated consecutively.
9079 Like @option{-sim}, but pass linker options to locate initialized data at
9080 0x40000000 and zero-initialized data at 0x80000000.
9084 @subsection CRX Options
9087 These options are defined specifically for the CRX ports.
9093 Enable the use of multiply-accumulate instructions. Disabled by default.
9097 Push instructions will be used to pass outgoing arguments when functions
9098 are called. Enabled by default.
9101 @node Darwin Options
9102 @subsection Darwin Options
9103 @cindex Darwin options
9105 These options are defined for all architectures running the Darwin operating
9108 FSF GCC on Darwin does not create ``fat'' object files; it will create
9109 an object file for the single architecture that it was built to
9110 target. Apple's GCC on Darwin does create ``fat'' files if multiple
9111 @option{-arch} options are used; it does so by running the compiler or
9112 linker multiple times and joining the results together with
9115 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9116 @samp{i686}) is determined by the flags that specify the ISA
9117 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
9118 @option{-force_cpusubtype_ALL} option can be used to override this.
9120 The Darwin tools vary in their behavior when presented with an ISA
9121 mismatch. The assembler, @file{as}, will only permit instructions to
9122 be used that are valid for the subtype of the file it is generating,
9123 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9124 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9125 and print an error if asked to create a shared library with a less
9126 restrictive subtype than its input files (for instance, trying to put
9127 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
9128 for executables, @file{ld}, will quietly give the executable the most
9129 restrictive subtype of any of its input files.
9134 Add the framework directory @var{dir} to the head of the list of
9135 directories to be searched for header files. These directories are
9136 interleaved with those specified by @option{-I} options and are
9137 scanned in a left-to-right order.
9139 A framework directory is a directory with frameworks in it. A
9140 framework is a directory with a @samp{"Headers"} and/or
9141 @samp{"PrivateHeaders"} directory contained directly in it that ends
9142 in @samp{".framework"}. The name of a framework is the name of this
9143 directory excluding the @samp{".framework"}. Headers associated with
9144 the framework are found in one of those two directories, with
9145 @samp{"Headers"} being searched first. A subframework is a framework
9146 directory that is in a framework's @samp{"Frameworks"} directory.
9147 Includes of subframework headers can only appear in a header of a
9148 framework that contains the subframework, or in a sibling subframework
9149 header. Two subframeworks are siblings if they occur in the same
9150 framework. A subframework should not have the same name as a
9151 framework, a warning will be issued if this is violated. Currently a
9152 subframework cannot have subframeworks, in the future, the mechanism
9153 may be extended to support this. The standard frameworks can be found
9154 in @samp{"/System/Library/Frameworks"} and
9155 @samp{"/Library/Frameworks"}. An example include looks like
9156 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9157 the name of the framework and header.h is found in the
9158 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9160 @item -iframework@var{dir}
9162 Like @option{-F} except the directory is a treated as a system
9163 directory. The main difference between this @option{-iframework} and
9164 @option{-F} is that with @option{-iframework} the compiler does not
9165 warn about constructs contained within header files found via
9166 @var{dir}. This option is valid only for the C family of languages.
9170 Emit debugging information for symbols that are used. For STABS
9171 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9172 This is by default ON@.
9176 Emit debugging information for all symbols and types.
9178 @item -mmacosx-version-min=@var{version}
9179 The earliest version of MacOS X that this executable will run on
9180 is @var{version}. Typical values of @var{version} include @code{10.1},
9181 @code{10.2}, and @code{10.3.9}.
9183 If the compiler was built to use the system's headers by default,
9184 then the default for this option is the system version on which the
9185 compiler is running, otherwise the default is to make choices which
9186 are compatible with as many systems and code bases as possible.
9190 Enable kernel development mode. The @option{-mkernel} option sets
9191 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9192 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9193 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9194 applicable. This mode also sets @option{-mno-altivec},
9195 @option{-msoft-float}, @option{-fno-builtin} and
9196 @option{-mlong-branch} for PowerPC targets.
9198 @item -mone-byte-bool
9199 @opindex mone-byte-bool
9200 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9201 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9202 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9203 option has no effect on x86.
9205 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9206 to generate code that is not binary compatible with code generated
9207 without that switch. Using this switch may require recompiling all
9208 other modules in a program, including system libraries. Use this
9209 switch to conform to a non-default data model.
9211 @item -mfix-and-continue
9212 @itemx -ffix-and-continue
9213 @itemx -findirect-data
9214 @opindex mfix-and-continue
9215 @opindex ffix-and-continue
9216 @opindex findirect-data
9217 Generate code suitable for fast turn around development. Needed to
9218 enable gdb to dynamically load @code{.o} files into already running
9219 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9220 are provided for backwards compatibility.
9224 Loads all members of static archive libraries.
9225 See man ld(1) for more information.
9227 @item -arch_errors_fatal
9228 @opindex arch_errors_fatal
9229 Cause the errors having to do with files that have the wrong architecture
9233 @opindex bind_at_load
9234 Causes the output file to be marked such that the dynamic linker will
9235 bind all undefined references when the file is loaded or launched.
9239 Produce a Mach-o bundle format file.
9240 See man ld(1) for more information.
9242 @item -bundle_loader @var{executable}
9243 @opindex bundle_loader
9244 This option specifies the @var{executable} that will be loading the build
9245 output file being linked. See man ld(1) for more information.
9249 When passed this option, GCC will produce a dynamic library instead of
9250 an executable when linking, using the Darwin @file{libtool} command.
9252 @item -force_cpusubtype_ALL
9253 @opindex force_cpusubtype_ALL
9254 This causes GCC's output file to have the @var{ALL} subtype, instead of
9255 one controlled by the @option{-mcpu} or @option{-march} option.
9257 @item -allowable_client @var{client_name}
9259 @itemx -compatibility_version
9260 @itemx -current_version
9262 @itemx -dependency-file
9264 @itemx -dylinker_install_name
9266 @itemx -exported_symbols_list
9268 @itemx -flat_namespace
9269 @itemx -force_flat_namespace
9270 @itemx -headerpad_max_install_names
9273 @itemx -install_name
9274 @itemx -keep_private_externs
9275 @itemx -multi_module
9276 @itemx -multiply_defined
9277 @itemx -multiply_defined_unused
9279 @itemx -no_dead_strip_inits_and_terms
9280 @itemx -nofixprebinding
9283 @itemx -noseglinkedit
9284 @itemx -pagezero_size
9286 @itemx -prebind_all_twolevel_modules
9287 @itemx -private_bundle
9288 @itemx -read_only_relocs
9290 @itemx -sectobjectsymbols
9294 @itemx -sectobjectsymbols
9297 @itemx -segs_read_only_addr
9298 @itemx -segs_read_write_addr
9299 @itemx -seg_addr_table
9300 @itemx -seg_addr_table_filename
9303 @itemx -segs_read_only_addr
9304 @itemx -segs_read_write_addr
9305 @itemx -single_module
9308 @itemx -sub_umbrella
9309 @itemx -twolevel_namespace
9312 @itemx -unexported_symbols_list
9313 @itemx -weak_reference_mismatches
9316 @opindex allowable_client
9317 @opindex client_name
9318 @opindex compatibility_version
9319 @opindex current_version
9321 @opindex dependency-file
9323 @opindex dylinker_install_name
9325 @opindex exported_symbols_list
9327 @opindex flat_namespace
9328 @opindex force_flat_namespace
9329 @opindex headerpad_max_install_names
9332 @opindex install_name
9333 @opindex keep_private_externs
9334 @opindex multi_module
9335 @opindex multiply_defined
9336 @opindex multiply_defined_unused
9338 @opindex no_dead_strip_inits_and_terms
9339 @opindex nofixprebinding
9340 @opindex nomultidefs
9342 @opindex noseglinkedit
9343 @opindex pagezero_size
9345 @opindex prebind_all_twolevel_modules
9346 @opindex private_bundle
9347 @opindex read_only_relocs
9349 @opindex sectobjectsymbols
9353 @opindex sectobjectsymbols
9356 @opindex segs_read_only_addr
9357 @opindex segs_read_write_addr
9358 @opindex seg_addr_table
9359 @opindex seg_addr_table_filename
9360 @opindex seglinkedit
9362 @opindex segs_read_only_addr
9363 @opindex segs_read_write_addr
9364 @opindex single_module
9366 @opindex sub_library
9367 @opindex sub_umbrella
9368 @opindex twolevel_namespace
9371 @opindex unexported_symbols_list
9372 @opindex weak_reference_mismatches
9373 @opindex whatsloaded
9375 These options are passed to the Darwin linker. The Darwin linker man page
9376 describes them in detail.
9379 @node DEC Alpha Options
9380 @subsection DEC Alpha Options
9382 These @samp{-m} options are defined for the DEC Alpha implementations:
9385 @item -mno-soft-float
9387 @opindex mno-soft-float
9388 @opindex msoft-float
9389 Use (do not use) the hardware floating-point instructions for
9390 floating-point operations. When @option{-msoft-float} is specified,
9391 functions in @file{libgcc.a} will be used to perform floating-point
9392 operations. Unless they are replaced by routines that emulate the
9393 floating-point operations, or compiled in such a way as to call such
9394 emulations routines, these routines will issue floating-point
9395 operations. If you are compiling for an Alpha without floating-point
9396 operations, you must ensure that the library is built so as not to call
9399 Note that Alpha implementations without floating-point operations are
9400 required to have floating-point registers.
9405 @opindex mno-fp-regs
9406 Generate code that uses (does not use) the floating-point register set.
9407 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9408 register set is not used, floating point operands are passed in integer
9409 registers as if they were integers and floating-point results are passed
9410 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9411 so any function with a floating-point argument or return value called by code
9412 compiled with @option{-mno-fp-regs} must also be compiled with that
9415 A typical use of this option is building a kernel that does not use,
9416 and hence need not save and restore, any floating-point registers.
9420 The Alpha architecture implements floating-point hardware optimized for
9421 maximum performance. It is mostly compliant with the IEEE floating
9422 point standard. However, for full compliance, software assistance is
9423 required. This option generates code fully IEEE compliant code
9424 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9425 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9426 defined during compilation. The resulting code is less efficient but is
9427 able to correctly support denormalized numbers and exceptional IEEE
9428 values such as not-a-number and plus/minus infinity. Other Alpha
9429 compilers call this option @option{-ieee_with_no_inexact}.
9431 @item -mieee-with-inexact
9432 @opindex mieee-with-inexact
9433 This is like @option{-mieee} except the generated code also maintains
9434 the IEEE @var{inexact-flag}. Turning on this option causes the
9435 generated code to implement fully-compliant IEEE math. In addition to
9436 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9437 macro. On some Alpha implementations the resulting code may execute
9438 significantly slower than the code generated by default. Since there is
9439 very little code that depends on the @var{inexact-flag}, you should
9440 normally not specify this option. Other Alpha compilers call this
9441 option @option{-ieee_with_inexact}.
9443 @item -mfp-trap-mode=@var{trap-mode}
9444 @opindex mfp-trap-mode
9445 This option controls what floating-point related traps are enabled.
9446 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9447 The trap mode can be set to one of four values:
9451 This is the default (normal) setting. The only traps that are enabled
9452 are the ones that cannot be disabled in software (e.g., division by zero
9456 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9460 Like @samp{u}, but the instructions are marked to be safe for software
9461 completion (see Alpha architecture manual for details).
9464 Like @samp{su}, but inexact traps are enabled as well.
9467 @item -mfp-rounding-mode=@var{rounding-mode}
9468 @opindex mfp-rounding-mode
9469 Selects the IEEE rounding mode. Other Alpha compilers call this option
9470 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9475 Normal IEEE rounding mode. Floating point numbers are rounded towards
9476 the nearest machine number or towards the even machine number in case
9480 Round towards minus infinity.
9483 Chopped rounding mode. Floating point numbers are rounded towards zero.
9486 Dynamic rounding mode. A field in the floating point control register
9487 (@var{fpcr}, see Alpha architecture reference manual) controls the
9488 rounding mode in effect. The C library initializes this register for
9489 rounding towards plus infinity. Thus, unless your program modifies the
9490 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9493 @item -mtrap-precision=@var{trap-precision}
9494 @opindex mtrap-precision
9495 In the Alpha architecture, floating point traps are imprecise. This
9496 means without software assistance it is impossible to recover from a
9497 floating trap and program execution normally needs to be terminated.
9498 GCC can generate code that can assist operating system trap handlers
9499 in determining the exact location that caused a floating point trap.
9500 Depending on the requirements of an application, different levels of
9501 precisions can be selected:
9505 Program precision. This option is the default and means a trap handler
9506 can only identify which program caused a floating point exception.
9509 Function precision. The trap handler can determine the function that
9510 caused a floating point exception.
9513 Instruction precision. The trap handler can determine the exact
9514 instruction that caused a floating point exception.
9517 Other Alpha compilers provide the equivalent options called
9518 @option{-scope_safe} and @option{-resumption_safe}.
9520 @item -mieee-conformant
9521 @opindex mieee-conformant
9522 This option marks the generated code as IEEE conformant. You must not
9523 use this option unless you also specify @option{-mtrap-precision=i} and either
9524 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9525 is to emit the line @samp{.eflag 48} in the function prologue of the
9526 generated assembly file. Under DEC Unix, this has the effect that
9527 IEEE-conformant math library routines will be linked in.
9529 @item -mbuild-constants
9530 @opindex mbuild-constants
9531 Normally GCC examines a 32- or 64-bit integer constant to
9532 see if it can construct it from smaller constants in two or three
9533 instructions. If it cannot, it will output the constant as a literal and
9534 generate code to load it from the data segment at runtime.
9536 Use this option to require GCC to construct @emph{all} integer constants
9537 using code, even if it takes more instructions (the maximum is six).
9539 You would typically use this option to build a shared library dynamic
9540 loader. Itself a shared library, it must relocate itself in memory
9541 before it can find the variables and constants in its own data segment.
9547 Select whether to generate code to be assembled by the vendor-supplied
9548 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9566 Indicate whether GCC should generate code to use the optional BWX,
9567 CIX, FIX and MAX instruction sets. The default is to use the instruction
9568 sets supported by the CPU type specified via @option{-mcpu=} option or that
9569 of the CPU on which GCC was built if none was specified.
9574 @opindex mfloat-ieee
9575 Generate code that uses (does not use) VAX F and G floating point
9576 arithmetic instead of IEEE single and double precision.
9578 @item -mexplicit-relocs
9579 @itemx -mno-explicit-relocs
9580 @opindex mexplicit-relocs
9581 @opindex mno-explicit-relocs
9582 Older Alpha assemblers provided no way to generate symbol relocations
9583 except via assembler macros. Use of these macros does not allow
9584 optimal instruction scheduling. GNU binutils as of version 2.12
9585 supports a new syntax that allows the compiler to explicitly mark
9586 which relocations should apply to which instructions. This option
9587 is mostly useful for debugging, as GCC detects the capabilities of
9588 the assembler when it is built and sets the default accordingly.
9592 @opindex msmall-data
9593 @opindex mlarge-data
9594 When @option{-mexplicit-relocs} is in effect, static data is
9595 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9596 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9597 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9598 16-bit relocations off of the @code{$gp} register. This limits the
9599 size of the small data area to 64KB, but allows the variables to be
9600 directly accessed via a single instruction.
9602 The default is @option{-mlarge-data}. With this option the data area
9603 is limited to just below 2GB@. Programs that require more than 2GB of
9604 data must use @code{malloc} or @code{mmap} to allocate the data in the
9605 heap instead of in the program's data segment.
9607 When generating code for shared libraries, @option{-fpic} implies
9608 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9612 @opindex msmall-text
9613 @opindex mlarge-text
9614 When @option{-msmall-text} is used, the compiler assumes that the
9615 code of the entire program (or shared library) fits in 4MB, and is
9616 thus reachable with a branch instruction. When @option{-msmall-data}
9617 is used, the compiler can assume that all local symbols share the
9618 same @code{$gp} value, and thus reduce the number of instructions
9619 required for a function call from 4 to 1.
9621 The default is @option{-mlarge-text}.
9623 @item -mcpu=@var{cpu_type}
9625 Set the instruction set and instruction scheduling parameters for
9626 machine type @var{cpu_type}. You can specify either the @samp{EV}
9627 style name or the corresponding chip number. GCC supports scheduling
9628 parameters for the EV4, EV5 and EV6 family of processors and will
9629 choose the default values for the instruction set from the processor
9630 you specify. If you do not specify a processor type, GCC will default
9631 to the processor on which the compiler was built.
9633 Supported values for @var{cpu_type} are
9639 Schedules as an EV4 and has no instruction set extensions.
9643 Schedules as an EV5 and has no instruction set extensions.
9647 Schedules as an EV5 and supports the BWX extension.
9652 Schedules as an EV5 and supports the BWX and MAX extensions.
9656 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9660 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9663 @item -mtune=@var{cpu_type}
9665 Set only the instruction scheduling parameters for machine type
9666 @var{cpu_type}. The instruction set is not changed.
9668 @item -mmemory-latency=@var{time}
9669 @opindex mmemory-latency
9670 Sets the latency the scheduler should assume for typical memory
9671 references as seen by the application. This number is highly
9672 dependent on the memory access patterns used by the application
9673 and the size of the external cache on the machine.
9675 Valid options for @var{time} are
9679 A decimal number representing clock cycles.
9685 The compiler contains estimates of the number of clock cycles for
9686 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9687 (also called Dcache, Scache, and Bcache), as well as to main memory.
9688 Note that L3 is only valid for EV5.
9693 @node DEC Alpha/VMS Options
9694 @subsection DEC Alpha/VMS Options
9696 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9699 @item -mvms-return-codes
9700 @opindex mvms-return-codes
9701 Return VMS condition codes from main. The default is to return POSIX
9702 style condition (e.g.@: error) codes.
9706 @subsection FRV Options
9713 Only use the first 32 general purpose registers.
9718 Use all 64 general purpose registers.
9723 Use only the first 32 floating point registers.
9728 Use all 64 floating point registers
9731 @opindex mhard-float
9733 Use hardware instructions for floating point operations.
9736 @opindex msoft-float
9738 Use library routines for floating point operations.
9743 Dynamically allocate condition code registers.
9748 Do not try to dynamically allocate condition code registers, only
9749 use @code{icc0} and @code{fcc0}.
9754 Change ABI to use double word insns.
9759 Do not use double word instructions.
9764 Use floating point double instructions.
9769 Do not use floating point double instructions.
9774 Use media instructions.
9779 Do not use media instructions.
9784 Use multiply and add/subtract instructions.
9789 Do not use multiply and add/subtract instructions.
9794 Select the FDPIC ABI, that uses function descriptors to represent
9795 pointers to functions. Without any PIC/PIE-related options, it
9796 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9797 assumes GOT entries and small data are within a 12-bit range from the
9798 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9799 are computed with 32 bits.
9800 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9803 @opindex minline-plt
9805 Enable inlining of PLT entries in function calls to functions that are
9806 not known to bind locally. It has no effect without @option{-mfdpic}.
9807 It's enabled by default if optimizing for speed and compiling for
9808 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9809 optimization option such as @option{-O3} or above is present in the
9815 Assume a large TLS segment when generating thread-local code.
9820 Do not assume a large TLS segment when generating thread-local code.
9825 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9826 that is known to be in read-only sections. It's enabled by default,
9827 except for @option{-fpic} or @option{-fpie}: even though it may help
9828 make the global offset table smaller, it trades 1 instruction for 4.
9829 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9830 one of which may be shared by multiple symbols, and it avoids the need
9831 for a GOT entry for the referenced symbol, so it's more likely to be a
9832 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9834 @item -multilib-library-pic
9835 @opindex multilib-library-pic
9837 Link with the (library, not FD) pic libraries. It's implied by
9838 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9839 @option{-fpic} without @option{-mfdpic}. You should never have to use
9845 Follow the EABI requirement of always creating a frame pointer whenever
9846 a stack frame is allocated. This option is enabled by default and can
9847 be disabled with @option{-mno-linked-fp}.
9850 @opindex mlong-calls
9852 Use indirect addressing to call functions outside the current
9853 compilation unit. This allows the functions to be placed anywhere
9854 within the 32-bit address space.
9856 @item -malign-labels
9857 @opindex malign-labels
9859 Try to align labels to an 8-byte boundary by inserting nops into the
9860 previous packet. This option only has an effect when VLIW packing
9861 is enabled. It doesn't create new packets; it merely adds nops to
9865 @opindex mlibrary-pic
9867 Generate position-independent EABI code.
9872 Use only the first four media accumulator registers.
9877 Use all eight media accumulator registers.
9882 Pack VLIW instructions.
9887 Do not pack VLIW instructions.
9892 Do not mark ABI switches in e_flags.
9897 Enable the use of conditional-move instructions (default).
9899 This switch is mainly for debugging the compiler and will likely be removed
9900 in a future version.
9902 @item -mno-cond-move
9903 @opindex mno-cond-move
9905 Disable the use of conditional-move instructions.
9907 This switch is mainly for debugging the compiler and will likely be removed
9908 in a future version.
9913 Enable the use of conditional set instructions (default).
9915 This switch is mainly for debugging the compiler and will likely be removed
9916 in a future version.
9921 Disable the use of conditional set instructions.
9923 This switch is mainly for debugging the compiler and will likely be removed
9924 in a future version.
9929 Enable the use of conditional execution (default).
9931 This switch is mainly for debugging the compiler and will likely be removed
9932 in a future version.
9934 @item -mno-cond-exec
9935 @opindex mno-cond-exec
9937 Disable the use of conditional execution.
9939 This switch is mainly for debugging the compiler and will likely be removed
9940 in a future version.
9943 @opindex mvliw-branch
9945 Run a pass to pack branches into VLIW instructions (default).
9947 This switch is mainly for debugging the compiler and will likely be removed
9948 in a future version.
9950 @item -mno-vliw-branch
9951 @opindex mno-vliw-branch
9953 Do not run a pass to pack branches into VLIW instructions.
9955 This switch is mainly for debugging the compiler and will likely be removed
9956 in a future version.
9958 @item -mmulti-cond-exec
9959 @opindex mmulti-cond-exec
9961 Enable optimization of @code{&&} and @code{||} in conditional execution
9964 This switch is mainly for debugging the compiler and will likely be removed
9965 in a future version.
9967 @item -mno-multi-cond-exec
9968 @opindex mno-multi-cond-exec
9970 Disable optimization of @code{&&} and @code{||} in conditional execution.
9972 This switch is mainly for debugging the compiler and will likely be removed
9973 in a future version.
9975 @item -mnested-cond-exec
9976 @opindex mnested-cond-exec
9978 Enable nested conditional execution optimizations (default).
9980 This switch is mainly for debugging the compiler and will likely be removed
9981 in a future version.
9983 @item -mno-nested-cond-exec
9984 @opindex mno-nested-cond-exec
9986 Disable nested conditional execution optimizations.
9988 This switch is mainly for debugging the compiler and will likely be removed
9989 in a future version.
9991 @item -moptimize-membar
9992 @opindex moptimize-membar
9994 This switch removes redundant @code{membar} instructions from the
9995 compiler generated code. It is enabled by default.
9997 @item -mno-optimize-membar
9998 @opindex mno-optimize-membar
10000 This switch disables the automatic removal of redundant @code{membar}
10001 instructions from the generated code.
10003 @item -mtomcat-stats
10004 @opindex mtomcat-stats
10006 Cause gas to print out tomcat statistics.
10008 @item -mcpu=@var{cpu}
10011 Select the processor type for which to generate code. Possible values are
10012 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
10013 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
10017 @node GNU/Linux Options
10018 @subsection GNU/Linux Options
10020 These @samp{-m} options are defined for GNU/Linux targets:
10025 Use the GNU C library instead of uClibc. This is the default except
10026 on @samp{*-*-linux-*uclibc*} targets.
10030 Use uClibc instead of the GNU C library. This is the default on
10031 @samp{*-*-linux-*uclibc*} targets.
10034 @node H8/300 Options
10035 @subsection H8/300 Options
10037 These @samp{-m} options are defined for the H8/300 implementations:
10042 Shorten some address references at link time, when possible; uses the
10043 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
10044 ld, Using ld}, for a fuller description.
10048 Generate code for the H8/300H@.
10052 Generate code for the H8S@.
10056 Generate code for the H8S and H8/300H in the normal mode. This switch
10057 must be used either with @option{-mh} or @option{-ms}.
10061 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
10065 Make @code{int} data 32 bits by default.
10068 @opindex malign-300
10069 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
10070 The default for the H8/300H and H8S is to align longs and floats on 4
10072 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
10073 This option has no effect on the H8/300.
10077 @subsection HPPA Options
10078 @cindex HPPA Options
10080 These @samp{-m} options are defined for the HPPA family of computers:
10083 @item -march=@var{architecture-type}
10085 Generate code for the specified architecture. The choices for
10086 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
10087 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
10088 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
10089 architecture option for your machine. Code compiled for lower numbered
10090 architectures will run on higher numbered architectures, but not the
10093 @item -mpa-risc-1-0
10094 @itemx -mpa-risc-1-1
10095 @itemx -mpa-risc-2-0
10096 @opindex mpa-risc-1-0
10097 @opindex mpa-risc-1-1
10098 @opindex mpa-risc-2-0
10099 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10102 @opindex mbig-switch
10103 Generate code suitable for big switch tables. Use this option only if
10104 the assembler/linker complain about out of range branches within a switch
10107 @item -mjump-in-delay
10108 @opindex mjump-in-delay
10109 Fill delay slots of function calls with unconditional jump instructions
10110 by modifying the return pointer for the function call to be the target
10111 of the conditional jump.
10113 @item -mdisable-fpregs
10114 @opindex mdisable-fpregs
10115 Prevent floating point registers from being used in any manner. This is
10116 necessary for compiling kernels which perform lazy context switching of
10117 floating point registers. If you use this option and attempt to perform
10118 floating point operations, the compiler will abort.
10120 @item -mdisable-indexing
10121 @opindex mdisable-indexing
10122 Prevent the compiler from using indexing address modes. This avoids some
10123 rather obscure problems when compiling MIG generated code under MACH@.
10125 @item -mno-space-regs
10126 @opindex mno-space-regs
10127 Generate code that assumes the target has no space registers. This allows
10128 GCC to generate faster indirect calls and use unscaled index address modes.
10130 Such code is suitable for level 0 PA systems and kernels.
10132 @item -mfast-indirect-calls
10133 @opindex mfast-indirect-calls
10134 Generate code that assumes calls never cross space boundaries. This
10135 allows GCC to emit code which performs faster indirect calls.
10137 This option will not work in the presence of shared libraries or nested
10140 @item -mfixed-range=@var{register-range}
10141 @opindex mfixed-range
10142 Generate code treating the given register range as fixed registers.
10143 A fixed register is one that the register allocator can not use. This is
10144 useful when compiling kernel code. A register range is specified as
10145 two registers separated by a dash. Multiple register ranges can be
10146 specified separated by a comma.
10148 @item -mlong-load-store
10149 @opindex mlong-load-store
10150 Generate 3-instruction load and store sequences as sometimes required by
10151 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
10154 @item -mportable-runtime
10155 @opindex mportable-runtime
10156 Use the portable calling conventions proposed by HP for ELF systems.
10160 Enable the use of assembler directives only GAS understands.
10162 @item -mschedule=@var{cpu-type}
10164 Schedule code according to the constraints for the machine type
10165 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10166 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10167 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10168 proper scheduling option for your machine. The default scheduling is
10172 @opindex mlinker-opt
10173 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10174 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10175 linkers in which they give bogus error messages when linking some programs.
10178 @opindex msoft-float
10179 Generate output containing library calls for floating point.
10180 @strong{Warning:} the requisite libraries are not available for all HPPA
10181 targets. Normally the facilities of the machine's usual C compiler are
10182 used, but this cannot be done directly in cross-compilation. You must make
10183 your own arrangements to provide suitable library functions for
10184 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
10185 does provide software floating point support.
10187 @option{-msoft-float} changes the calling convention in the output file;
10188 therefore, it is only useful if you compile @emph{all} of a program with
10189 this option. In particular, you need to compile @file{libgcc.a}, the
10190 library that comes with GCC, with @option{-msoft-float} in order for
10195 Generate the predefine, @code{_SIO}, for server IO@. The default is
10196 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10197 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10198 options are available under HP-UX and HI-UX@.
10202 Use GNU ld specific options. This passes @option{-shared} to ld when
10203 building a shared library. It is the default when GCC is configured,
10204 explicitly or implicitly, with the GNU linker. This option does not
10205 have any affect on which ld is called, it only changes what parameters
10206 are passed to that ld. The ld that is called is determined by the
10207 @option{--with-ld} configure option, GCC's program search path, and
10208 finally by the user's @env{PATH}. The linker used by GCC can be printed
10209 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10210 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10214 Use HP ld specific options. This passes @option{-b} to ld when building
10215 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10216 links. It is the default when GCC is configured, explicitly or
10217 implicitly, with the HP linker. This option does not have any affect on
10218 which ld is called, it only changes what parameters are passed to that
10219 ld. The ld that is called is determined by the @option{--with-ld}
10220 configure option, GCC's program search path, and finally by the user's
10221 @env{PATH}. The linker used by GCC can be printed using @samp{which
10222 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10223 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10226 @opindex mno-long-calls
10227 Generate code that uses long call sequences. This ensures that a call
10228 is always able to reach linker generated stubs. The default is to generate
10229 long calls only when the distance from the call site to the beginning
10230 of the function or translation unit, as the case may be, exceeds a
10231 predefined limit set by the branch type being used. The limits for
10232 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10233 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10236 Distances are measured from the beginning of functions when using the
10237 @option{-ffunction-sections} option, or when using the @option{-mgas}
10238 and @option{-mno-portable-runtime} options together under HP-UX with
10241 It is normally not desirable to use this option as it will degrade
10242 performance. However, it may be useful in large applications,
10243 particularly when partial linking is used to build the application.
10245 The types of long calls used depends on the capabilities of the
10246 assembler and linker, and the type of code being generated. The
10247 impact on systems that support long absolute calls, and long pic
10248 symbol-difference or pc-relative calls should be relatively small.
10249 However, an indirect call is used on 32-bit ELF systems in pic code
10250 and it is quite long.
10252 @item -munix=@var{unix-std}
10254 Generate compiler predefines and select a startfile for the specified
10255 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10256 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10257 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10258 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10259 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10262 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10263 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10264 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10265 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10266 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10267 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10269 It is @emph{important} to note that this option changes the interfaces
10270 for various library routines. It also affects the operational behavior
10271 of the C library. Thus, @emph{extreme} care is needed in using this
10274 Library code that is intended to operate with more than one UNIX
10275 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10276 as appropriate. Most GNU software doesn't provide this capability.
10280 Suppress the generation of link options to search libdld.sl when the
10281 @option{-static} option is specified on HP-UX 10 and later.
10285 The HP-UX implementation of setlocale in libc has a dependency on
10286 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10287 when the @option{-static} option is specified, special link options
10288 are needed to resolve this dependency.
10290 On HP-UX 10 and later, the GCC driver adds the necessary options to
10291 link with libdld.sl when the @option{-static} option is specified.
10292 This causes the resulting binary to be dynamic. On the 64-bit port,
10293 the linkers generate dynamic binaries by default in any case. The
10294 @option{-nolibdld} option can be used to prevent the GCC driver from
10295 adding these link options.
10299 Add support for multithreading with the @dfn{dce thread} library
10300 under HP-UX@. This option sets flags for both the preprocessor and
10304 @node i386 and x86-64 Options
10305 @subsection Intel 386 and AMD x86-64 Options
10306 @cindex i386 Options
10307 @cindex x86-64 Options
10308 @cindex Intel 386 Options
10309 @cindex AMD x86-64 Options
10311 These @samp{-m} options are defined for the i386 and x86-64 family of
10315 @item -mtune=@var{cpu-type}
10317 Tune to @var{cpu-type} everything applicable about the generated code, except
10318 for the ABI and the set of available instructions. The choices for
10319 @var{cpu-type} are:
10322 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10323 If you know the CPU on which your code will run, then you should use
10324 the corresponding @option{-mtune} option instead of
10325 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10326 of your application will have, then you should use this option.
10328 As new processors are deployed in the marketplace, the behavior of this
10329 option will change. Therefore, if you upgrade to a newer version of
10330 GCC, the code generated option will change to reflect the processors
10331 that were most common when that version of GCC was released.
10333 There is no @option{-march=generic} option because @option{-march}
10334 indicates the instruction set the compiler can use, and there is no
10335 generic instruction set applicable to all processors. In contrast,
10336 @option{-mtune} indicates the processor (or, in this case, collection of
10337 processors) for which the code is optimized.
10339 This selects the CPU to tune for at compilation time by determining
10340 the processor type of the compiling machine. Using @option{-mtune=native}
10341 will produce code optimized for the local machine under the constraints
10342 of the selected instruction set. Using @option{-march=native} will
10343 enable all instruction subsets supported by the local machine (hence
10344 the result might not run on different machines).
10346 Original Intel's i386 CPU@.
10348 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10349 @item i586, pentium
10350 Intel Pentium CPU with no MMX support.
10352 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10354 Intel PentiumPro CPU@.
10356 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10357 instruction set will be used, so the code will run on all i686 family chips.
10359 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10360 @item pentium3, pentium3m
10361 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10364 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10365 support. Used by Centrino notebooks.
10366 @item pentium4, pentium4m
10367 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10369 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10372 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10373 SSE2 and SSE3 instruction set support.
10375 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10376 instruction set support.
10378 AMD K6 CPU with MMX instruction set support.
10380 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
10381 @item athlon, athlon-tbird
10382 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
10384 @item athlon-4, athlon-xp, athlon-mp
10385 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
10386 instruction set support.
10387 @item k8, opteron, athlon64, athlon-fx
10388 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10389 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
10390 @item k8-sse3, opteron-sse3, athlon64-sse3
10391 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10392 @item amdfam10, barcelona
10393 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10394 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10395 instruction set extensions.)
10397 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10400 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
10401 instruction set support.
10403 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
10404 implemented for this chip.)
10406 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10407 implemented for this chip.)
10409 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10412 While picking a specific @var{cpu-type} will schedule things appropriately
10413 for that particular chip, the compiler will not generate any code that
10414 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10417 @item -march=@var{cpu-type}
10419 Generate instructions for the machine type @var{cpu-type}. The choices
10420 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10421 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10423 @item -mcpu=@var{cpu-type}
10425 A deprecated synonym for @option{-mtune}.
10427 @item -mfpmath=@var{unit}
10429 Generate floating point arithmetics for selected unit @var{unit}. The choices
10430 for @var{unit} are:
10434 Use the standard 387 floating point coprocessor present majority of chips and
10435 emulated otherwise. Code compiled with this option will run almost everywhere.
10436 The temporary results are computed in 80bit precision instead of precision
10437 specified by the type resulting in slightly different results compared to most
10438 of other chips. See @option{-ffloat-store} for more detailed description.
10440 This is the default choice for i386 compiler.
10443 Use scalar floating point instructions present in the SSE instruction set.
10444 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10445 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10446 instruction set supports only single precision arithmetics, thus the double and
10447 extended precision arithmetics is still done using 387. Later version, present
10448 only in Pentium4 and the future AMD x86-64 chips supports double precision
10451 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10452 or @option{-msse2} switches to enable SSE extensions and make this option
10453 effective. For the x86-64 compiler, these extensions are enabled by default.
10455 The resulting code should be considerably faster in the majority of cases and avoid
10456 the numerical instability problems of 387 code, but may break some existing
10457 code that expects temporaries to be 80bit.
10459 This is the default choice for the x86-64 compiler.
10462 Attempt to utilize both instruction sets at once. This effectively double the
10463 amount of available registers and on chips with separate execution units for
10464 387 and SSE the execution resources too. Use this option with care, as it is
10465 still experimental, because the GCC register allocator does not model separate
10466 functional units well resulting in instable performance.
10469 @item -masm=@var{dialect}
10470 @opindex masm=@var{dialect}
10471 Output asm instructions using selected @var{dialect}. Supported
10472 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10473 not support @samp{intel}.
10476 @itemx -mno-ieee-fp
10478 @opindex mno-ieee-fp
10479 Control whether or not the compiler uses IEEE floating point
10480 comparisons. These handle correctly the case where the result of a
10481 comparison is unordered.
10484 @opindex msoft-float
10485 Generate output containing library calls for floating point.
10486 @strong{Warning:} the requisite libraries are not part of GCC@.
10487 Normally the facilities of the machine's usual C compiler are used, but
10488 this can't be done directly in cross-compilation. You must make your
10489 own arrangements to provide suitable library functions for
10492 On machines where a function returns floating point results in the 80387
10493 register stack, some floating point opcodes may be emitted even if
10494 @option{-msoft-float} is used.
10496 @item -mno-fp-ret-in-387
10497 @opindex mno-fp-ret-in-387
10498 Do not use the FPU registers for return values of functions.
10500 The usual calling convention has functions return values of types
10501 @code{float} and @code{double} in an FPU register, even if there
10502 is no FPU@. The idea is that the operating system should emulate
10505 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10506 in ordinary CPU registers instead.
10508 @item -mno-fancy-math-387
10509 @opindex mno-fancy-math-387
10510 Some 387 emulators do not support the @code{sin}, @code{cos} and
10511 @code{sqrt} instructions for the 387. Specify this option to avoid
10512 generating those instructions. This option is the default on FreeBSD,
10513 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10514 indicates that the target cpu will always have an FPU and so the
10515 instruction will not need emulation. As of revision 2.6.1, these
10516 instructions are not generated unless you also use the
10517 @option{-funsafe-math-optimizations} switch.
10519 @item -malign-double
10520 @itemx -mno-align-double
10521 @opindex malign-double
10522 @opindex mno-align-double
10523 Control whether GCC aligns @code{double}, @code{long double}, and
10524 @code{long long} variables on a two word boundary or a one word
10525 boundary. Aligning @code{double} variables on a two word boundary will
10526 produce code that runs somewhat faster on a @samp{Pentium} at the
10527 expense of more memory.
10529 On x86-64, @option{-malign-double} is enabled by default.
10531 @strong{Warning:} if you use the @option{-malign-double} switch,
10532 structures containing the above types will be aligned differently than
10533 the published application binary interface specifications for the 386
10534 and will not be binary compatible with structures in code compiled
10535 without that switch.
10537 @item -m96bit-long-double
10538 @itemx -m128bit-long-double
10539 @opindex m96bit-long-double
10540 @opindex m128bit-long-double
10541 These switches control the size of @code{long double} type. The i386
10542 application binary interface specifies the size to be 96 bits,
10543 so @option{-m96bit-long-double} is the default in 32 bit mode.
10545 Modern architectures (Pentium and newer) would prefer @code{long double}
10546 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10547 conforming to the ABI, this would not be possible. So specifying a
10548 @option{-m128bit-long-double} will align @code{long double}
10549 to a 16 byte boundary by padding the @code{long double} with an additional
10552 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10553 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10555 Notice that neither of these options enable any extra precision over the x87
10556 standard of 80 bits for a @code{long double}.
10558 @strong{Warning:} if you override the default value for your target ABI, the
10559 structures and arrays containing @code{long double} variables will change
10560 their size as well as function calling convention for function taking
10561 @code{long double} will be modified. Hence they will not be binary
10562 compatible with arrays or structures in code compiled without that switch.
10564 @item -mmlarge-data-threshold=@var{number}
10565 @opindex mlarge-data-threshold=@var{number}
10566 When @option{-mcmodel=medium} is specified, the data greater than
10567 @var{threshold} are placed in large data section. This value must be the
10568 same across all object linked into the binary and defaults to 65535.
10572 Use a different function-calling convention, in which functions that
10573 take a fixed number of arguments return with the @code{ret} @var{num}
10574 instruction, which pops their arguments while returning. This saves one
10575 instruction in the caller since there is no need to pop the arguments
10578 You can specify that an individual function is called with this calling
10579 sequence with the function attribute @samp{stdcall}. You can also
10580 override the @option{-mrtd} option by using the function attribute
10581 @samp{cdecl}. @xref{Function Attributes}.
10583 @strong{Warning:} this calling convention is incompatible with the one
10584 normally used on Unix, so you cannot use it if you need to call
10585 libraries compiled with the Unix compiler.
10587 Also, you must provide function prototypes for all functions that
10588 take variable numbers of arguments (including @code{printf});
10589 otherwise incorrect code will be generated for calls to those
10592 In addition, seriously incorrect code will result if you call a
10593 function with too many arguments. (Normally, extra arguments are
10594 harmlessly ignored.)
10596 @item -mregparm=@var{num}
10598 Control how many registers are used to pass integer arguments. By
10599 default, no registers are used to pass arguments, and at most 3
10600 registers can be used. You can control this behavior for a specific
10601 function by using the function attribute @samp{regparm}.
10602 @xref{Function Attributes}.
10604 @strong{Warning:} if you use this switch, and
10605 @var{num} is nonzero, then you must build all modules with the same
10606 value, including any libraries. This includes the system libraries and
10610 @opindex msseregparm
10611 Use SSE register passing conventions for float and double arguments
10612 and return values. You can control this behavior for a specific
10613 function by using the function attribute @samp{sseregparm}.
10614 @xref{Function Attributes}.
10616 @strong{Warning:} if you use this switch then you must build all
10617 modules with the same value, including any libraries. This includes
10618 the system libraries and startup modules.
10627 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10628 is specified, the significands of results of floating-point operations are
10629 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10630 significands of results of floating-point operations to 53 bits (double
10631 precision) and @option{-mpc80} rounds the significands of results of
10632 floating-point operations to 64 bits (extended double precision), which is
10633 the default. When this option is used, floating-point operations in higher
10634 precisions are not available to the programmer without setting the FPU
10635 control word explicitly.
10637 Setting the rounding of floating-point operations to less than the default
10638 80 bits can speed some programs by 2% or more. Note that some mathematical
10639 libraries assume that extended precision (80 bit) floating-point operations
10640 are enabled by default; routines in such libraries could suffer significant
10641 loss of accuracy, typically through so-called "catastrophic cancellation",
10642 when this option is used to set the precision to less than extended precision.
10644 @item -mstackrealign
10645 @opindex mstackrealign
10646 Realign the stack at entry. On the Intel x86, the
10647 @option{-mstackrealign} option will generate an alternate prologue and
10648 epilogue that realigns the runtime stack. This supports mixing legacy
10649 codes that keep a 4-byte aligned stack with modern codes that keep a
10650 16-byte stack for SSE compatibility. The alternate prologue and
10651 epilogue are slower and bigger than the regular ones, and the
10652 alternate prologue requires an extra scratch register; this lowers the
10653 number of registers available if used in conjunction with the
10654 @code{regparm} attribute. The @option{-mstackrealign} option is
10655 incompatible with the nested function prologue; this is considered a
10656 hard error. See also the attribute @code{force_align_arg_pointer},
10657 applicable to individual functions.
10659 @item -mpreferred-stack-boundary=@var{num}
10660 @opindex mpreferred-stack-boundary
10661 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10662 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10663 the default is 4 (16 bytes or 128 bits).
10665 On Pentium and PentiumPro, @code{double} and @code{long double} values
10666 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10667 suffer significant run time performance penalties. On Pentium III, the
10668 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10669 properly if it is not 16 byte aligned.
10671 To ensure proper alignment of this values on the stack, the stack boundary
10672 must be as aligned as that required by any value stored on the stack.
10673 Further, every function must be generated such that it keeps the stack
10674 aligned. Thus calling a function compiled with a higher preferred
10675 stack boundary from a function compiled with a lower preferred stack
10676 boundary will most likely misalign the stack. It is recommended that
10677 libraries that use callbacks always use the default setting.
10679 This extra alignment does consume extra stack space, and generally
10680 increases code size. Code that is sensitive to stack space usage, such
10681 as embedded systems and operating system kernels, may want to reduce the
10682 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10716 These switches enable or disable the use of instructions in the MMX,
10717 SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4A, SSE5, ABM or 3DNow!@: extended
10719 These extensions are also available as built-in functions: see
10720 @ref{X86 Built-in Functions}, for details of the functions enabled and
10721 disabled by these switches.
10723 To have SSE/SSE2 instructions generated automatically from floating-point
10724 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10726 These options will enable GCC to use these extended instructions in
10727 generated code, even without @option{-mfpmath=sse}. Applications which
10728 perform runtime CPU detection must compile separate files for each
10729 supported architecture, using the appropriate flags. In particular,
10730 the file containing the CPU detection code should be compiled without
10735 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10736 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10737 data types. This is useful for high resolution counters that could be updated
10738 by multiple processors (or cores). This instruction is generated as part of
10739 atomic built-in functions: see @ref{Atomic Builtins} for details.
10743 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10744 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10745 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10746 SAHF are load and store instructions, respectively, for certain status flags.
10747 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10748 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10752 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10753 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Rhapson step
10754 to increase precision instead of DIVSS and SQRTSS (and their vectorized
10755 variants) for single precision floating point arguments. These instructions
10756 are generated only when @option{-funsafe-math-optimizations} is enabled
10757 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
10758 Note that while the throughput of the sequence is higher than the throughput
10759 of the non-reciprocal instruction, the precision of the sequence can be
10760 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
10762 @item -mveclibabi=@var{type}
10763 @opindex mveclibabi
10764 Specifies the ABI type to use for vectorizing intrinsics using an
10765 external library. Supported types are @code{acml} for the AMD
10766 math core library style of interfacing. GCC will currently emit
10767 calls to @code{__vrd2_sin}, @code{__vrd2_cos}, @code{__vrd2_exp},
10768 @code{__vrd2_log}, @code{__vrd2_log2}, @code{__vrd2_log10},
10769 @code{__vrs4_sinf}, @code{__vrs4_cosf}, @code{__vrs4_expf},
10770 @code{__vrs4_logf}, @code{__vrs4_log2f}, @code{__vrs4_log10f}
10771 and @code{__vrs4_powf} when using this type and @option{-ftree-vectorize}
10772 is enabled. A ACML ABI compatible library will have to be specified
10776 @itemx -mno-push-args
10777 @opindex mpush-args
10778 @opindex mno-push-args
10779 Use PUSH operations to store outgoing parameters. This method is shorter
10780 and usually equally fast as method using SUB/MOV operations and is enabled
10781 by default. In some cases disabling it may improve performance because of
10782 improved scheduling and reduced dependencies.
10784 @item -maccumulate-outgoing-args
10785 @opindex maccumulate-outgoing-args
10786 If enabled, the maximum amount of space required for outgoing arguments will be
10787 computed in the function prologue. This is faster on most modern CPUs
10788 because of reduced dependencies, improved scheduling and reduced stack usage
10789 when preferred stack boundary is not equal to 2. The drawback is a notable
10790 increase in code size. This switch implies @option{-mno-push-args}.
10794 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10795 on thread-safe exception handling must compile and link all code with the
10796 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10797 @option{-D_MT}; when linking, it links in a special thread helper library
10798 @option{-lmingwthrd} which cleans up per thread exception handling data.
10800 @item -mno-align-stringops
10801 @opindex mno-align-stringops
10802 Do not align destination of inlined string operations. This switch reduces
10803 code size and improves performance in case the destination is already aligned,
10804 but GCC doesn't know about it.
10806 @item -minline-all-stringops
10807 @opindex minline-all-stringops
10808 By default GCC inlines string operations only when destination is known to be
10809 aligned at least to 4 byte boundary. This enables more inlining, increase code
10810 size, but may improve performance of code that depends on fast memcpy, strlen
10811 and memset for short lengths.
10813 @item -minline-stringops-dynamically
10814 @opindex minline-stringops-dynamically
10815 For string operation of unknown size, inline runtime checks so for small
10816 blocks inline code is used, while for large blocks library call is used.
10818 @item -mstringop-strategy=@var{alg}
10819 @opindex mstringop-strategy=@var{alg}
10820 Overwrite internal decision heuristic about particular algorithm to inline
10821 string operation with. The allowed values are @code{rep_byte},
10822 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10823 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10824 expanding inline loop, @code{libcall} for always expanding library call.
10826 @item -momit-leaf-frame-pointer
10827 @opindex momit-leaf-frame-pointer
10828 Don't keep the frame pointer in a register for leaf functions. This
10829 avoids the instructions to save, set up and restore frame pointers and
10830 makes an extra register available in leaf functions. The option
10831 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10832 which might make debugging harder.
10834 @item -mtls-direct-seg-refs
10835 @itemx -mno-tls-direct-seg-refs
10836 @opindex mtls-direct-seg-refs
10837 Controls whether TLS variables may be accessed with offsets from the
10838 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10839 or whether the thread base pointer must be added. Whether or not this
10840 is legal depends on the operating system, and whether it maps the
10841 segment to cover the entire TLS area.
10843 For systems that use GNU libc, the default is on.
10846 @itemx -mno-fused-madd
10847 @opindex mfused-madd
10848 Enable automatic generation of fused floating point multiply-add instructions
10849 if the ISA supports such instructions. The -mfused-madd option is on by
10850 default. The fused multiply-add instructions have a different
10851 rounding behavior compared to executing a multiply followed by an add.
10854 These @samp{-m} switches are supported in addition to the above
10855 on AMD x86-64 processors in 64-bit environments.
10862 Generate code for a 32-bit or 64-bit environment.
10863 The 32-bit environment sets int, long and pointer to 32 bits and
10864 generates code that runs on any i386 system.
10865 The 64-bit environment sets int to 32 bits and long and pointer
10866 to 64 bits and generates code for AMD's x86-64 architecture. For
10867 darwin only the -m64 option turns off the @option{-fno-pic} and
10868 @option{-mdynamic-no-pic} options.
10870 @item -mno-red-zone
10871 @opindex no-red-zone
10872 Do not use a so called red zone for x86-64 code. The red zone is mandated
10873 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10874 stack pointer that will not be modified by signal or interrupt handlers
10875 and therefore can be used for temporary data without adjusting the stack
10876 pointer. The flag @option{-mno-red-zone} disables this red zone.
10878 @item -mcmodel=small
10879 @opindex mcmodel=small
10880 Generate code for the small code model: the program and its symbols must
10881 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10882 Programs can be statically or dynamically linked. This is the default
10885 @item -mcmodel=kernel
10886 @opindex mcmodel=kernel
10887 Generate code for the kernel code model. The kernel runs in the
10888 negative 2 GB of the address space.
10889 This model has to be used for Linux kernel code.
10891 @item -mcmodel=medium
10892 @opindex mcmodel=medium
10893 Generate code for the medium model: The program is linked in the lower 2
10894 GB of the address space but symbols can be located anywhere in the
10895 address space. Programs can be statically or dynamically linked, but
10896 building of shared libraries are not supported with the medium model.
10898 @item -mcmodel=large
10899 @opindex mcmodel=large
10900 Generate code for the large model: This model makes no assumptions
10901 about addresses and sizes of sections.
10904 @node IA-64 Options
10905 @subsection IA-64 Options
10906 @cindex IA-64 Options
10908 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10912 @opindex mbig-endian
10913 Generate code for a big endian target. This is the default for HP-UX@.
10915 @item -mlittle-endian
10916 @opindex mlittle-endian
10917 Generate code for a little endian target. This is the default for AIX5
10923 @opindex mno-gnu-as
10924 Generate (or don't) code for the GNU assembler. This is the default.
10925 @c Also, this is the default if the configure option @option{--with-gnu-as}
10931 @opindex mno-gnu-ld
10932 Generate (or don't) code for the GNU linker. This is the default.
10933 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10938 Generate code that does not use a global pointer register. The result
10939 is not position independent code, and violates the IA-64 ABI@.
10941 @item -mvolatile-asm-stop
10942 @itemx -mno-volatile-asm-stop
10943 @opindex mvolatile-asm-stop
10944 @opindex mno-volatile-asm-stop
10945 Generate (or don't) a stop bit immediately before and after volatile asm
10948 @item -mregister-names
10949 @itemx -mno-register-names
10950 @opindex mregister-names
10951 @opindex mno-register-names
10952 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10953 the stacked registers. This may make assembler output more readable.
10959 Disable (or enable) optimizations that use the small data section. This may
10960 be useful for working around optimizer bugs.
10962 @item -mconstant-gp
10963 @opindex mconstant-gp
10964 Generate code that uses a single constant global pointer value. This is
10965 useful when compiling kernel code.
10969 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10970 This is useful when compiling firmware code.
10972 @item -minline-float-divide-min-latency
10973 @opindex minline-float-divide-min-latency
10974 Generate code for inline divides of floating point values
10975 using the minimum latency algorithm.
10977 @item -minline-float-divide-max-throughput
10978 @opindex minline-float-divide-max-throughput
10979 Generate code for inline divides of floating point values
10980 using the maximum throughput algorithm.
10982 @item -minline-int-divide-min-latency
10983 @opindex minline-int-divide-min-latency
10984 Generate code for inline divides of integer values
10985 using the minimum latency algorithm.
10987 @item -minline-int-divide-max-throughput
10988 @opindex minline-int-divide-max-throughput
10989 Generate code for inline divides of integer values
10990 using the maximum throughput algorithm.
10992 @item -minline-sqrt-min-latency
10993 @opindex minline-sqrt-min-latency
10994 Generate code for inline square roots
10995 using the minimum latency algorithm.
10997 @item -minline-sqrt-max-throughput
10998 @opindex minline-sqrt-max-throughput
10999 Generate code for inline square roots
11000 using the maximum throughput algorithm.
11002 @item -mno-dwarf2-asm
11003 @itemx -mdwarf2-asm
11004 @opindex mno-dwarf2-asm
11005 @opindex mdwarf2-asm
11006 Don't (or do) generate assembler code for the DWARF2 line number debugging
11007 info. This may be useful when not using the GNU assembler.
11009 @item -mearly-stop-bits
11010 @itemx -mno-early-stop-bits
11011 @opindex mearly-stop-bits
11012 @opindex mno-early-stop-bits
11013 Allow stop bits to be placed earlier than immediately preceding the
11014 instruction that triggered the stop bit. This can improve instruction
11015 scheduling, but does not always do so.
11017 @item -mfixed-range=@var{register-range}
11018 @opindex mfixed-range
11019 Generate code treating the given register range as fixed registers.
11020 A fixed register is one that the register allocator can not use. This is
11021 useful when compiling kernel code. A register range is specified as
11022 two registers separated by a dash. Multiple register ranges can be
11023 specified separated by a comma.
11025 @item -mtls-size=@var{tls-size}
11027 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
11030 @item -mtune=@var{cpu-type}
11032 Tune the instruction scheduling for a particular CPU, Valid values are
11033 itanium, itanium1, merced, itanium2, and mckinley.
11039 Add support for multithreading using the POSIX threads library. This
11040 option sets flags for both the preprocessor and linker. It does
11041 not affect the thread safety of object code produced by the compiler or
11042 that of libraries supplied with it. These are HP-UX specific flags.
11048 Generate code for a 32-bit or 64-bit environment.
11049 The 32-bit environment sets int, long and pointer to 32 bits.
11050 The 64-bit environment sets int to 32 bits and long and pointer
11051 to 64 bits. These are HP-UX specific flags.
11053 @item -mno-sched-br-data-spec
11054 @itemx -msched-br-data-spec
11055 @opindex mno-sched-br-data-spec
11056 @opindex msched-br-data-spec
11057 (Dis/En)able data speculative scheduling before reload.
11058 This will result in generation of the ld.a instructions and
11059 the corresponding check instructions (ld.c / chk.a).
11060 The default is 'disable'.
11062 @item -msched-ar-data-spec
11063 @itemx -mno-sched-ar-data-spec
11064 @opindex msched-ar-data-spec
11065 @opindex mno-sched-ar-data-spec
11066 (En/Dis)able data speculative scheduling after reload.
11067 This will result in generation of the ld.a instructions and
11068 the corresponding check instructions (ld.c / chk.a).
11069 The default is 'enable'.
11071 @item -mno-sched-control-spec
11072 @itemx -msched-control-spec
11073 @opindex mno-sched-control-spec
11074 @opindex msched-control-spec
11075 (Dis/En)able control speculative scheduling. This feature is
11076 available only during region scheduling (i.e.@: before reload).
11077 This will result in generation of the ld.s instructions and
11078 the corresponding check instructions chk.s .
11079 The default is 'disable'.
11081 @item -msched-br-in-data-spec
11082 @itemx -mno-sched-br-in-data-spec
11083 @opindex msched-br-in-data-spec
11084 @opindex mno-sched-br-in-data-spec
11085 (En/Dis)able speculative scheduling of the instructions that
11086 are dependent on the data speculative loads before reload.
11087 This is effective only with @option{-msched-br-data-spec} enabled.
11088 The default is 'enable'.
11090 @item -msched-ar-in-data-spec
11091 @itemx -mno-sched-ar-in-data-spec
11092 @opindex msched-ar-in-data-spec
11093 @opindex mno-sched-ar-in-data-spec
11094 (En/Dis)able speculative scheduling of the instructions that
11095 are dependent on the data speculative loads after reload.
11096 This is effective only with @option{-msched-ar-data-spec} enabled.
11097 The default is 'enable'.
11099 @item -msched-in-control-spec
11100 @itemx -mno-sched-in-control-spec
11101 @opindex msched-in-control-spec
11102 @opindex mno-sched-in-control-spec
11103 (En/Dis)able speculative scheduling of the instructions that
11104 are dependent on the control speculative loads.
11105 This is effective only with @option{-msched-control-spec} enabled.
11106 The default is 'enable'.
11109 @itemx -mno-sched-ldc
11110 @opindex msched-ldc
11111 @opindex mno-sched-ldc
11112 (En/Dis)able use of simple data speculation checks ld.c .
11113 If disabled, only chk.a instructions will be emitted to check
11114 data speculative loads.
11115 The default is 'enable'.
11117 @item -mno-sched-control-ldc
11118 @itemx -msched-control-ldc
11119 @opindex mno-sched-control-ldc
11120 @opindex msched-control-ldc
11121 (Dis/En)able use of ld.c instructions to check control speculative loads.
11122 If enabled, in case of control speculative load with no speculatively
11123 scheduled dependent instructions this load will be emitted as ld.sa and
11124 ld.c will be used to check it.
11125 The default is 'disable'.
11127 @item -mno-sched-spec-verbose
11128 @itemx -msched-spec-verbose
11129 @opindex mno-sched-spec-verbose
11130 @opindex msched-spec-verbose
11131 (Dis/En)able printing of the information about speculative motions.
11133 @item -mno-sched-prefer-non-data-spec-insns
11134 @itemx -msched-prefer-non-data-spec-insns
11135 @opindex mno-sched-prefer-non-data-spec-insns
11136 @opindex msched-prefer-non-data-spec-insns
11137 If enabled, data speculative instructions will be chosen for schedule
11138 only if there are no other choices at the moment. This will make
11139 the use of the data speculation much more conservative.
11140 The default is 'disable'.
11142 @item -mno-sched-prefer-non-control-spec-insns
11143 @itemx -msched-prefer-non-control-spec-insns
11144 @opindex mno-sched-prefer-non-control-spec-insns
11145 @opindex msched-prefer-non-control-spec-insns
11146 If enabled, control speculative instructions will be chosen for schedule
11147 only if there are no other choices at the moment. This will make
11148 the use of the control speculation much more conservative.
11149 The default is 'disable'.
11151 @item -mno-sched-count-spec-in-critical-path
11152 @itemx -msched-count-spec-in-critical-path
11153 @opindex mno-sched-count-spec-in-critical-path
11154 @opindex msched-count-spec-in-critical-path
11155 If enabled, speculative dependencies will be considered during
11156 computation of the instructions priorities. This will make the use of the
11157 speculation a bit more conservative.
11158 The default is 'disable'.
11163 @subsection M32C Options
11164 @cindex M32C options
11167 @item -mcpu=@var{name}
11169 Select the CPU for which code is generated. @var{name} may be one of
11170 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11171 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11172 the M32C/80 series.
11176 Specifies that the program will be run on the simulator. This causes
11177 an alternate runtime library to be linked in which supports, for
11178 example, file I/O. You must not use this option when generating
11179 programs that will run on real hardware; you must provide your own
11180 runtime library for whatever I/O functions are needed.
11182 @item -memregs=@var{number}
11184 Specifies the number of memory-based pseudo-registers GCC will use
11185 during code generation. These pseudo-registers will be used like real
11186 registers, so there is a tradeoff between GCC's ability to fit the
11187 code into available registers, and the performance penalty of using
11188 memory instead of registers. Note that all modules in a program must
11189 be compiled with the same value for this option. Because of that, you
11190 must not use this option with the default runtime libraries gcc
11195 @node M32R/D Options
11196 @subsection M32R/D Options
11197 @cindex M32R/D options
11199 These @option{-m} options are defined for Renesas M32R/D architectures:
11204 Generate code for the M32R/2@.
11208 Generate code for the M32R/X@.
11212 Generate code for the M32R@. This is the default.
11214 @item -mmodel=small
11215 @opindex mmodel=small
11216 Assume all objects live in the lower 16MB of memory (so that their addresses
11217 can be loaded with the @code{ld24} instruction), and assume all subroutines
11218 are reachable with the @code{bl} instruction.
11219 This is the default.
11221 The addressability of a particular object can be set with the
11222 @code{model} attribute.
11224 @item -mmodel=medium
11225 @opindex mmodel=medium
11226 Assume objects may be anywhere in the 32-bit address space (the compiler
11227 will generate @code{seth/add3} instructions to load their addresses), and
11228 assume all subroutines are reachable with the @code{bl} instruction.
11230 @item -mmodel=large
11231 @opindex mmodel=large
11232 Assume objects may be anywhere in the 32-bit address space (the compiler
11233 will generate @code{seth/add3} instructions to load their addresses), and
11234 assume subroutines may not be reachable with the @code{bl} instruction
11235 (the compiler will generate the much slower @code{seth/add3/jl}
11236 instruction sequence).
11239 @opindex msdata=none
11240 Disable use of the small data area. Variables will be put into
11241 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11242 @code{section} attribute has been specified).
11243 This is the default.
11245 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11246 Objects may be explicitly put in the small data area with the
11247 @code{section} attribute using one of these sections.
11249 @item -msdata=sdata
11250 @opindex msdata=sdata
11251 Put small global and static data in the small data area, but do not
11252 generate special code to reference them.
11255 @opindex msdata=use
11256 Put small global and static data in the small data area, and generate
11257 special instructions to reference them.
11261 @cindex smaller data references
11262 Put global and static objects less than or equal to @var{num} bytes
11263 into the small data or bss sections instead of the normal data or bss
11264 sections. The default value of @var{num} is 8.
11265 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11266 for this option to have any effect.
11268 All modules should be compiled with the same @option{-G @var{num}} value.
11269 Compiling with different values of @var{num} may or may not work; if it
11270 doesn't the linker will give an error message---incorrect code will not be
11275 Makes the M32R specific code in the compiler display some statistics
11276 that might help in debugging programs.
11278 @item -malign-loops
11279 @opindex malign-loops
11280 Align all loops to a 32-byte boundary.
11282 @item -mno-align-loops
11283 @opindex mno-align-loops
11284 Do not enforce a 32-byte alignment for loops. This is the default.
11286 @item -missue-rate=@var{number}
11287 @opindex missue-rate=@var{number}
11288 Issue @var{number} instructions per cycle. @var{number} can only be 1
11291 @item -mbranch-cost=@var{number}
11292 @opindex mbranch-cost=@var{number}
11293 @var{number} can only be 1 or 2. If it is 1 then branches will be
11294 preferred over conditional code, if it is 2, then the opposite will
11297 @item -mflush-trap=@var{number}
11298 @opindex mflush-trap=@var{number}
11299 Specifies the trap number to use to flush the cache. The default is
11300 12. Valid numbers are between 0 and 15 inclusive.
11302 @item -mno-flush-trap
11303 @opindex mno-flush-trap
11304 Specifies that the cache cannot be flushed by using a trap.
11306 @item -mflush-func=@var{name}
11307 @opindex mflush-func=@var{name}
11308 Specifies the name of the operating system function to call to flush
11309 the cache. The default is @emph{_flush_cache}, but a function call
11310 will only be used if a trap is not available.
11312 @item -mno-flush-func
11313 @opindex mno-flush-func
11314 Indicates that there is no OS function for flushing the cache.
11318 @node M680x0 Options
11319 @subsection M680x0 Options
11320 @cindex M680x0 options
11322 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11323 The default settings depend on which architecture was selected when
11324 the compiler was configured; the defaults for the most common choices
11328 @item -march=@var{arch}
11330 Generate code for a specific M680x0 or ColdFire instruction set
11331 architecture. Permissible values of @var{arch} for M680x0
11332 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11333 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11334 architectures are selected according to Freescale's ISA classification
11335 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11336 @samp{isab} and @samp{isac}.
11338 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11339 code for a ColdFire target. The @var{arch} in this macro is one of the
11340 @option{-march} arguments given above.
11342 When used together, @option{-march} and @option{-mtune} select code
11343 that runs on a family of similar processors but that is optimized
11344 for a particular microarchitecture.
11346 @item -mcpu=@var{cpu}
11348 Generate code for a specific M680x0 or ColdFire processor.
11349 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11350 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11351 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11352 below, which also classifies the CPUs into families:
11354 @multitable @columnfractions 0.20 0.80
11355 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11356 @item @samp{51qe} @tab @samp{51qe}
11357 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11358 @item @samp{5206e} @tab @samp{5206e}
11359 @item @samp{5208} @tab @samp{5207} @samp{5208}
11360 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11361 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11362 @item @samp{5216} @tab @samp{5214} @samp{5216}
11363 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11364 @item @samp{5225} @tab @samp{5224} @samp{5225}
11365 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11366 @item @samp{5249} @tab @samp{5249}
11367 @item @samp{5250} @tab @samp{5250}
11368 @item @samp{5271} @tab @samp{5270} @samp{5271}
11369 @item @samp{5272} @tab @samp{5272}
11370 @item @samp{5275} @tab @samp{5274} @samp{5275}
11371 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11372 @item @samp{5307} @tab @samp{5307}
11373 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11374 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11375 @item @samp{5407} @tab @samp{5407}
11376 @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}
11379 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11380 @var{arch} is compatible with @var{cpu}. Other combinations of
11381 @option{-mcpu} and @option{-march} are rejected.
11383 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11384 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11385 where the value of @var{family} is given by the table above.
11387 @item -mtune=@var{tune}
11389 Tune the code for a particular microarchitecture, within the
11390 constraints set by @option{-march} and @option{-mcpu}.
11391 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11392 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11393 and @samp{cpu32}. The ColdFire microarchitectures
11394 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11396 You can also use @option{-mtune=68020-40} for code that needs
11397 to run relatively well on 68020, 68030 and 68040 targets.
11398 @option{-mtune=68020-60} is similar but includes 68060 targets
11399 as well. These two options select the same tuning decisions as
11400 @option{-m68020-40} and @option{-m68020-60} respectively.
11402 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11403 when tuning for 680x0 architecture @var{arch}. It also defines
11404 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11405 option is used. If gcc is tuning for a range of architectures,
11406 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11407 it defines the macros for every architecture in the range.
11409 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11410 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11411 of the arguments given above.
11417 Generate output for a 68000. This is the default
11418 when the compiler is configured for 68000-based systems.
11419 It is equivalent to @option{-march=68000}.
11421 Use this option for microcontrollers with a 68000 or EC000 core,
11422 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11426 Generate output for a 68010. This is the default
11427 when the compiler is configured for 68010-based systems.
11428 It is equivalent to @option{-march=68010}.
11434 Generate output for a 68020. This is the default
11435 when the compiler is configured for 68020-based systems.
11436 It is equivalent to @option{-march=68020}.
11440 Generate output for a 68030. This is the default when the compiler is
11441 configured for 68030-based systems. It is equivalent to
11442 @option{-march=68030}.
11446 Generate output for a 68040. This is the default when the compiler is
11447 configured for 68040-based systems. It is equivalent to
11448 @option{-march=68040}.
11450 This option inhibits the use of 68881/68882 instructions that have to be
11451 emulated by software on the 68040. Use this option if your 68040 does not
11452 have code to emulate those instructions.
11456 Generate output for a 68060. This is the default when the compiler is
11457 configured for 68060-based systems. It is equivalent to
11458 @option{-march=68060}.
11460 This option inhibits the use of 68020 and 68881/68882 instructions that
11461 have to be emulated by software on the 68060. Use this option if your 68060
11462 does not have code to emulate those instructions.
11466 Generate output for a CPU32. This is the default
11467 when the compiler is configured for CPU32-based systems.
11468 It is equivalent to @option{-march=cpu32}.
11470 Use this option for microcontrollers with a
11471 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11472 68336, 68340, 68341, 68349 and 68360.
11476 Generate output for a 520X ColdFire CPU. This is the default
11477 when the compiler is configured for 520X-based systems.
11478 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11479 in favor of that option.
11481 Use this option for microcontroller with a 5200 core, including
11482 the MCF5202, MCF5203, MCF5204 and MCF5206.
11486 Generate output for a 5206e ColdFire CPU. The option is now
11487 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11491 Generate output for a member of the ColdFire 528X family.
11492 The option is now deprecated in favor of the equivalent
11493 @option{-mcpu=528x}.
11497 Generate output for a ColdFire 5307 CPU. The option is now deprecated
11498 in favor of the equivalent @option{-mcpu=5307}.
11502 Generate output for a ColdFire 5407 CPU. The option is now deprecated
11503 in favor of the equivalent @option{-mcpu=5407}.
11507 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11508 This includes use of hardware floating point instructions.
11509 The option is equivalent to @option{-mcpu=547x}, and is now
11510 deprecated in favor of that option.
11514 Generate output for a 68040, without using any of the new instructions.
11515 This results in code which can run relatively efficiently on either a
11516 68020/68881 or a 68030 or a 68040. The generated code does use the
11517 68881 instructions that are emulated on the 68040.
11519 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11523 Generate output for a 68060, without using any of the new instructions.
11524 This results in code which can run relatively efficiently on either a
11525 68020/68881 or a 68030 or a 68040. The generated code does use the
11526 68881 instructions that are emulated on the 68060.
11528 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11532 @opindex mhard-float
11534 Generate floating-point instructions. This is the default for 68020
11535 and above, and for ColdFire devices that have an FPU. It defines the
11536 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11537 on ColdFire targets.
11540 @opindex msoft-float
11541 Do not generate floating-point instructions; use library calls instead.
11542 This is the default for 68000, 68010, and 68832 targets. It is also
11543 the default for ColdFire devices that have no FPU.
11549 Generate (do not generate) ColdFire hardware divide and remainder
11550 instructions. If @option{-march} is used without @option{-mcpu},
11551 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11552 architectures. Otherwise, the default is taken from the target CPU
11553 (either the default CPU, or the one specified by @option{-mcpu}). For
11554 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11555 @option{-mcpu=5206e}.
11557 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11561 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11562 Additionally, parameters passed on the stack are also aligned to a
11563 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11567 Do not consider type @code{int} to be 16 bits wide. This is the default.
11570 @itemx -mno-bitfield
11571 @opindex mnobitfield
11572 @opindex mno-bitfield
11573 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11574 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11578 Do use the bit-field instructions. The @option{-m68020} option implies
11579 @option{-mbitfield}. This is the default if you use a configuration
11580 designed for a 68020.
11584 Use a different function-calling convention, in which functions
11585 that take a fixed number of arguments return with the @code{rtd}
11586 instruction, which pops their arguments while returning. This
11587 saves one instruction in the caller since there is no need to pop
11588 the arguments there.
11590 This calling convention is incompatible with the one normally
11591 used on Unix, so you cannot use it if you need to call libraries
11592 compiled with the Unix compiler.
11594 Also, you must provide function prototypes for all functions that
11595 take variable numbers of arguments (including @code{printf});
11596 otherwise incorrect code will be generated for calls to those
11599 In addition, seriously incorrect code will result if you call a
11600 function with too many arguments. (Normally, extra arguments are
11601 harmlessly ignored.)
11603 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11604 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11608 Do not use the calling conventions selected by @option{-mrtd}.
11609 This is the default.
11612 @itemx -mno-align-int
11613 @opindex malign-int
11614 @opindex mno-align-int
11615 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11616 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11617 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11618 Aligning variables on 32-bit boundaries produces code that runs somewhat
11619 faster on processors with 32-bit busses at the expense of more memory.
11621 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11622 align structures containing the above types differently than
11623 most published application binary interface specifications for the m68k.
11627 Use the pc-relative addressing mode of the 68000 directly, instead of
11628 using a global offset table. At present, this option implies @option{-fpic},
11629 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11630 not presently supported with @option{-mpcrel}, though this could be supported for
11631 68020 and higher processors.
11633 @item -mno-strict-align
11634 @itemx -mstrict-align
11635 @opindex mno-strict-align
11636 @opindex mstrict-align
11637 Do not (do) assume that unaligned memory references will be handled by
11641 Generate code that allows the data segment to be located in a different
11642 area of memory from the text segment. This allows for execute in place in
11643 an environment without virtual memory management. This option implies
11646 @item -mno-sep-data
11647 Generate code that assumes that the data segment follows the text segment.
11648 This is the default.
11650 @item -mid-shared-library
11651 Generate code that supports shared libraries via the library ID method.
11652 This allows for execute in place and shared libraries in an environment
11653 without virtual memory management. This option implies @option{-fPIC}.
11655 @item -mno-id-shared-library
11656 Generate code that doesn't assume ID based shared libraries are being used.
11657 This is the default.
11659 @item -mshared-library-id=n
11660 Specified the identification number of the ID based shared library being
11661 compiled. Specifying a value of 0 will generate more compact code, specifying
11662 other values will force the allocation of that number to the current
11663 library but is no more space or time efficient than omitting this option.
11667 @node M68hc1x Options
11668 @subsection M68hc1x Options
11669 @cindex M68hc1x options
11671 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11672 microcontrollers. The default values for these options depends on
11673 which style of microcontroller was selected when the compiler was configured;
11674 the defaults for the most common choices are given below.
11681 Generate output for a 68HC11. This is the default
11682 when the compiler is configured for 68HC11-based systems.
11688 Generate output for a 68HC12. This is the default
11689 when the compiler is configured for 68HC12-based systems.
11695 Generate output for a 68HCS12.
11697 @item -mauto-incdec
11698 @opindex mauto-incdec
11699 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11706 Enable the use of 68HC12 min and max instructions.
11709 @itemx -mno-long-calls
11710 @opindex mlong-calls
11711 @opindex mno-long-calls
11712 Treat all calls as being far away (near). If calls are assumed to be
11713 far away, the compiler will use the @code{call} instruction to
11714 call a function and the @code{rtc} instruction for returning.
11718 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11720 @item -msoft-reg-count=@var{count}
11721 @opindex msoft-reg-count
11722 Specify the number of pseudo-soft registers which are used for the
11723 code generation. The maximum number is 32. Using more pseudo-soft
11724 register may or may not result in better code depending on the program.
11725 The default is 4 for 68HC11 and 2 for 68HC12.
11729 @node MCore Options
11730 @subsection MCore Options
11731 @cindex MCore options
11733 These are the @samp{-m} options defined for the Motorola M*Core
11739 @itemx -mno-hardlit
11741 @opindex mno-hardlit
11742 Inline constants into the code stream if it can be done in two
11743 instructions or less.
11749 Use the divide instruction. (Enabled by default).
11751 @item -mrelax-immediate
11752 @itemx -mno-relax-immediate
11753 @opindex mrelax-immediate
11754 @opindex mno-relax-immediate
11755 Allow arbitrary sized immediates in bit operations.
11757 @item -mwide-bitfields
11758 @itemx -mno-wide-bitfields
11759 @opindex mwide-bitfields
11760 @opindex mno-wide-bitfields
11761 Always treat bit-fields as int-sized.
11763 @item -m4byte-functions
11764 @itemx -mno-4byte-functions
11765 @opindex m4byte-functions
11766 @opindex mno-4byte-functions
11767 Force all functions to be aligned to a four byte boundary.
11769 @item -mcallgraph-data
11770 @itemx -mno-callgraph-data
11771 @opindex mcallgraph-data
11772 @opindex mno-callgraph-data
11773 Emit callgraph information.
11776 @itemx -mno-slow-bytes
11777 @opindex mslow-bytes
11778 @opindex mno-slow-bytes
11779 Prefer word access when reading byte quantities.
11781 @item -mlittle-endian
11782 @itemx -mbig-endian
11783 @opindex mlittle-endian
11784 @opindex mbig-endian
11785 Generate code for a little endian target.
11791 Generate code for the 210 processor.
11795 @subsection MIPS Options
11796 @cindex MIPS options
11802 Generate big-endian code.
11806 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11809 @item -march=@var{arch}
11811 Generate code that will run on @var{arch}, which can be the name of a
11812 generic MIPS ISA, or the name of a particular processor.
11814 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11815 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11816 The processor names are:
11817 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11818 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11819 @samp{5kc}, @samp{5kf},
11821 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11822 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11823 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11824 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11827 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11828 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11829 @samp{rm7000}, @samp{rm9000},
11832 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11833 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11834 The special value @samp{from-abi} selects the
11835 most compatible architecture for the selected ABI (that is,
11836 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11838 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11839 (for example, @samp{-march=r2k}). Prefixes are optional, and
11840 @samp{vr} may be written @samp{r}.
11842 Names of the form @samp{@var{n}f2_1} refer to processors with
11843 FPUs clocked at half the rate of the core, names of the form
11844 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
11845 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
11846 processors with FPUs clocked a ratio of 3:2 with respect to the core.
11847 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
11848 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
11849 accepted as synonyms for @samp{@var{n}f1_1}.
11851 GCC defines two macros based on the value of this option. The first
11852 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11853 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11854 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11855 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11856 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11858 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11859 above. In other words, it will have the full prefix and will not
11860 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11861 the macro names the resolved architecture (either @samp{"mips1"} or
11862 @samp{"mips3"}). It names the default architecture when no
11863 @option{-march} option is given.
11865 @item -mtune=@var{arch}
11867 Optimize for @var{arch}. Among other things, this option controls
11868 the way instructions are scheduled, and the perceived cost of arithmetic
11869 operations. The list of @var{arch} values is the same as for
11872 When this option is not used, GCC will optimize for the processor
11873 specified by @option{-march}. By using @option{-march} and
11874 @option{-mtune} together, it is possible to generate code that will
11875 run on a family of processors, but optimize the code for one
11876 particular member of that family.
11878 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11879 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11880 @samp{-march} ones described above.
11884 Equivalent to @samp{-march=mips1}.
11888 Equivalent to @samp{-march=mips2}.
11892 Equivalent to @samp{-march=mips3}.
11896 Equivalent to @samp{-march=mips4}.
11900 Equivalent to @samp{-march=mips32}.
11904 Equivalent to @samp{-march=mips32r2}.
11908 Equivalent to @samp{-march=mips64}.
11913 @opindex mno-mips16
11914 Generate (do not generate) MIPS16 code. If GCC is targetting a
11915 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11917 MIPS16 code generation can also be controlled on a per-function basis
11918 by means of @code{mips16} and @code{nomips16} attributes.
11919 @xref{Function Attributes}, for more information.
11921 @item -mflip-mips16
11922 @opindex mflip-mips16
11923 Generate MIPS16 code on alternating functions. This option is provided
11924 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
11925 not intended for ordinary use in compiling user code.
11927 @item -minterlink-mips16
11928 @itemx -mno-interlink-mips16
11929 @opindex minterlink-mips16
11930 @opindex mno-interlink-mips16
11931 Require (do not require) that non-MIPS16 code be link-compatible with
11934 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
11935 it must either use a call or an indirect jump. @option{-minterlink-mips16}
11936 therefore disables direct jumps unless GCC knows that the target of the
11937 jump is not MIPS16.
11949 Generate code for the given ABI@.
11951 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11952 generates 64-bit code when you select a 64-bit architecture, but you
11953 can use @option{-mgp32} to get 32-bit code instead.
11955 For information about the O64 ABI, see
11956 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11958 GCC supports a variant of the o32 ABI in which floating-point registers
11959 are 64 rather than 32 bits wide. You can select this combination with
11960 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11961 and @samp{mfhc1} instructions and is therefore only supported for
11962 MIPS32R2 processors.
11964 The register assignments for arguments and return values remain the
11965 same, but each scalar value is passed in a single 64-bit register
11966 rather than a pair of 32-bit registers. For example, scalar
11967 floating-point values are returned in @samp{$f0} only, not a
11968 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
11969 remains the same, but all 64 bits are saved.
11972 @itemx -mno-abicalls
11974 @opindex mno-abicalls
11975 Generate (do not generate) code that is suitable for SVR4-style
11976 dynamic objects. @option{-mabicalls} is the default for SVR4-based
11981 Generate (do not generate) code that is fully position-independent,
11982 and that can therefore be linked into shared libraries. This option
11983 only affects @option{-mabicalls}.
11985 All @option{-mabicalls} code has traditionally been position-independent,
11986 regardless of options like @option{-fPIC} and @option{-fpic}. However,
11987 as an extension, the GNU toolchain allows executables to use absolute
11988 accesses for locally-binding symbols. It can also use shorter GP
11989 initialization sequences and generate direct calls to locally-defined
11990 functions. This mode is selected by @option{-mno-shared}.
11992 @option{-mno-shared} depends on binutils 2.16 or higher and generates
11993 objects that can only be linked by the GNU linker. However, the option
11994 does not affect the ABI of the final executable; it only affects the ABI
11995 of relocatable objects. Using @option{-mno-shared} will generally make
11996 executables both smaller and quicker.
11998 @option{-mshared} is the default.
12004 Lift (do not lift) the usual restrictions on the size of the global
12007 GCC normally uses a single instruction to load values from the GOT@.
12008 While this is relatively efficient, it will only work if the GOT
12009 is smaller than about 64k. Anything larger will cause the linker
12010 to report an error such as:
12012 @cindex relocation truncated to fit (MIPS)
12014 relocation truncated to fit: R_MIPS_GOT16 foobar
12017 If this happens, you should recompile your code with @option{-mxgot}.
12018 It should then work with very large GOTs, although it will also be
12019 less efficient, since it will take three instructions to fetch the
12020 value of a global symbol.
12022 Note that some linkers can create multiple GOTs. If you have such a
12023 linker, you should only need to use @option{-mxgot} when a single object
12024 file accesses more than 64k's worth of GOT entries. Very few do.
12026 These options have no effect unless GCC is generating position
12031 Assume that general-purpose registers are 32 bits wide.
12035 Assume that general-purpose registers are 64 bits wide.
12039 Assume that floating-point registers are 32 bits wide.
12043 Assume that floating-point registers are 64 bits wide.
12046 @opindex mhard-float
12047 Use floating-point coprocessor instructions.
12050 @opindex msoft-float
12051 Do not use floating-point coprocessor instructions. Implement
12052 floating-point calculations using library calls instead.
12054 @item -msingle-float
12055 @opindex msingle-float
12056 Assume that the floating-point coprocessor only supports single-precision
12059 @item -mdouble-float
12060 @opindex mdouble-float
12061 Assume that the floating-point coprocessor supports double-precision
12062 operations. This is the default.
12068 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
12069 implement atomic memory built-in functions. When neither option is
12070 specified, GCC will use the instructions if the target architecture
12073 @option{-mllsc} is useful if the runtime environment can emulate the
12074 instructions and @option{-mno-llsc} can be useful when compiling for
12075 nonstandard ISAs. You can make either option the default by
12076 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
12077 respectively. @option{--with-llsc} is the default for some
12078 configurations; see the installation documentation for details.
12084 Use (do not use) revision 1 of the MIPS DSP ASE.
12085 @xref{MIPS DSP Built-in Functions}. This option defines the
12086 preprocessor macro @samp{__mips_dsp}. It also defines
12087 @samp{__mips_dsp_rev} to 1.
12093 Use (do not use) revision 2 of the MIPS DSP ASE.
12094 @xref{MIPS DSP Built-in Functions}. This option defines the
12095 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
12096 It also defines @samp{__mips_dsp_rev} to 2.
12099 @itemx -mno-smartmips
12100 @opindex msmartmips
12101 @opindex mno-smartmips
12102 Use (do not use) the MIPS SmartMIPS ASE.
12104 @item -mpaired-single
12105 @itemx -mno-paired-single
12106 @opindex mpaired-single
12107 @opindex mno-paired-single
12108 Use (do not use) paired-single floating-point instructions.
12109 @xref{MIPS Paired-Single Support}. This option requires
12110 hardware floating-point support to be enabled.
12116 Use (do not use) MIPS Digital Media Extension instructions.
12117 This option can only be used when generating 64-bit code and requires
12118 hardware floating-point support to be enabled.
12123 @opindex mno-mips3d
12124 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
12125 The option @option{-mips3d} implies @option{-mpaired-single}.
12131 Use (do not use) MT Multithreading instructions.
12135 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
12136 an explanation of the default and the way that the pointer size is
12141 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12143 The default size of @code{int}s, @code{long}s and pointers depends on
12144 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
12145 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12146 32-bit @code{long}s. Pointers are the same size as @code{long}s,
12147 or the same size as integer registers, whichever is smaller.
12153 Assume (do not assume) that all symbols have 32-bit values, regardless
12154 of the selected ABI@. This option is useful in combination with
12155 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12156 to generate shorter and faster references to symbolic addresses.
12160 Put definitions of externally-visible data in a small data section
12161 if that data is no bigger than @var{num} bytes. GCC can then access
12162 the data more efficiently; see @option{-mgpopt} for details.
12164 The default @option{-G} option depends on the configuration.
12166 @item -mlocal-sdata
12167 @itemx -mno-local-sdata
12168 @opindex mlocal-sdata
12169 @opindex mno-local-sdata
12170 Extend (do not extend) the @option{-G} behavior to local data too,
12171 such as to static variables in C. @option{-mlocal-sdata} is the
12172 default for all configurations.
12174 If the linker complains that an application is using too much small data,
12175 you might want to try rebuilding the less performance-critical parts with
12176 @option{-mno-local-sdata}. You might also want to build large
12177 libraries with @option{-mno-local-sdata}, so that the libraries leave
12178 more room for the main program.
12180 @item -mextern-sdata
12181 @itemx -mno-extern-sdata
12182 @opindex mextern-sdata
12183 @opindex mno-extern-sdata
12184 Assume (do not assume) that externally-defined data will be in
12185 a small data section if that data is within the @option{-G} limit.
12186 @option{-mextern-sdata} is the default for all configurations.
12188 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12189 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12190 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12191 is placed in a small data section. If @var{Var} is defined by another
12192 module, you must either compile that module with a high-enough
12193 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12194 definition. If @var{Var} is common, you must link the application
12195 with a high-enough @option{-G} setting.
12197 The easiest way of satisfying these restrictions is to compile
12198 and link every module with the same @option{-G} option. However,
12199 you may wish to build a library that supports several different
12200 small data limits. You can do this by compiling the library with
12201 the highest supported @option{-G} setting and additionally using
12202 @option{-mno-extern-sdata} to stop the library from making assumptions
12203 about externally-defined data.
12209 Use (do not use) GP-relative accesses for symbols that are known to be
12210 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12211 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12214 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12215 might not hold the value of @code{_gp}. For example, if the code is
12216 part of a library that might be used in a boot monitor, programs that
12217 call boot monitor routines will pass an unknown value in @code{$gp}.
12218 (In such situations, the boot monitor itself would usually be compiled
12219 with @option{-G0}.)
12221 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12222 @option{-mno-extern-sdata}.
12224 @item -membedded-data
12225 @itemx -mno-embedded-data
12226 @opindex membedded-data
12227 @opindex mno-embedded-data
12228 Allocate variables to the read-only data section first if possible, then
12229 next in the small data section if possible, otherwise in data. This gives
12230 slightly slower code than the default, but reduces the amount of RAM required
12231 when executing, and thus may be preferred for some embedded systems.
12233 @item -muninit-const-in-rodata
12234 @itemx -mno-uninit-const-in-rodata
12235 @opindex muninit-const-in-rodata
12236 @opindex mno-uninit-const-in-rodata
12237 Put uninitialized @code{const} variables in the read-only data section.
12238 This option is only meaningful in conjunction with @option{-membedded-data}.
12240 @item -mcode-readable=@var{setting}
12241 @opindex mcode-readable
12242 Specify whether GCC may generate code that reads from executable sections.
12243 There are three possible settings:
12246 @item -mcode-readable=yes
12247 Instructions may freely access executable sections. This is the
12250 @item -mcode-readable=pcrel
12251 MIPS16 PC-relative load instructions can access executable sections,
12252 but other instructions must not do so. This option is useful on 4KSc
12253 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12254 It is also useful on processors that can be configured to have a dual
12255 instruction/data SRAM interface and that, like the M4K, automatically
12256 redirect PC-relative loads to the instruction RAM.
12258 @item -mcode-readable=no
12259 Instructions must not access executable sections. This option can be
12260 useful on targets that are configured to have a dual instruction/data
12261 SRAM interface but that (unlike the M4K) do not automatically redirect
12262 PC-relative loads to the instruction RAM.
12265 @item -msplit-addresses
12266 @itemx -mno-split-addresses
12267 @opindex msplit-addresses
12268 @opindex mno-split-addresses
12269 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12270 relocation operators. This option has been superseded by
12271 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12273 @item -mexplicit-relocs
12274 @itemx -mno-explicit-relocs
12275 @opindex mexplicit-relocs
12276 @opindex mno-explicit-relocs
12277 Use (do not use) assembler relocation operators when dealing with symbolic
12278 addresses. The alternative, selected by @option{-mno-explicit-relocs},
12279 is to use assembler macros instead.
12281 @option{-mexplicit-relocs} is the default if GCC was configured
12282 to use an assembler that supports relocation operators.
12284 @item -mcheck-zero-division
12285 @itemx -mno-check-zero-division
12286 @opindex mcheck-zero-division
12287 @opindex mno-check-zero-division
12288 Trap (do not trap) on integer division by zero.
12290 The default is @option{-mcheck-zero-division}.
12292 @item -mdivide-traps
12293 @itemx -mdivide-breaks
12294 @opindex mdivide-traps
12295 @opindex mdivide-breaks
12296 MIPS systems check for division by zero by generating either a
12297 conditional trap or a break instruction. Using traps results in
12298 smaller code, but is only supported on MIPS II and later. Also, some
12299 versions of the Linux kernel have a bug that prevents trap from
12300 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
12301 allow conditional traps on architectures that support them and
12302 @option{-mdivide-breaks} to force the use of breaks.
12304 The default is usually @option{-mdivide-traps}, but this can be
12305 overridden at configure time using @option{--with-divide=breaks}.
12306 Divide-by-zero checks can be completely disabled using
12307 @option{-mno-check-zero-division}.
12312 @opindex mno-memcpy
12313 Force (do not force) the use of @code{memcpy()} for non-trivial block
12314 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
12315 most constant-sized copies.
12318 @itemx -mno-long-calls
12319 @opindex mlong-calls
12320 @opindex mno-long-calls
12321 Disable (do not disable) use of the @code{jal} instruction. Calling
12322 functions using @code{jal} is more efficient but requires the caller
12323 and callee to be in the same 256 megabyte segment.
12325 This option has no effect on abicalls code. The default is
12326 @option{-mno-long-calls}.
12332 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
12333 instructions, as provided by the R4650 ISA@.
12336 @itemx -mno-fused-madd
12337 @opindex mfused-madd
12338 @opindex mno-fused-madd
12339 Enable (disable) use of the floating point multiply-accumulate
12340 instructions, when they are available. The default is
12341 @option{-mfused-madd}.
12343 When multiply-accumulate instructions are used, the intermediate
12344 product is calculated to infinite precision and is not subject to
12345 the FCSR Flush to Zero bit. This may be undesirable in some
12350 Tell the MIPS assembler to not run its preprocessor over user
12351 assembler files (with a @samp{.s} suffix) when assembling them.
12354 @itemx -mno-fix-r4000
12355 @opindex mfix-r4000
12356 @opindex mno-fix-r4000
12357 Work around certain R4000 CPU errata:
12360 A double-word or a variable shift may give an incorrect result if executed
12361 immediately after starting an integer division.
12363 A double-word or a variable shift may give an incorrect result if executed
12364 while an integer multiplication is in progress.
12366 An integer division may give an incorrect result if started in a delay slot
12367 of a taken branch or a jump.
12371 @itemx -mno-fix-r4400
12372 @opindex mfix-r4400
12373 @opindex mno-fix-r4400
12374 Work around certain R4400 CPU errata:
12377 A double-word or a variable shift may give an incorrect result if executed
12378 immediately after starting an integer division.
12382 @itemx -mno-fix-vr4120
12383 @opindex mfix-vr4120
12384 Work around certain VR4120 errata:
12387 @code{dmultu} does not always produce the correct result.
12389 @code{div} and @code{ddiv} do not always produce the correct result if one
12390 of the operands is negative.
12392 The workarounds for the division errata rely on special functions in
12393 @file{libgcc.a}. At present, these functions are only provided by
12394 the @code{mips64vr*-elf} configurations.
12396 Other VR4120 errata require a nop to be inserted between certain pairs of
12397 instructions. These errata are handled by the assembler, not by GCC itself.
12400 @opindex mfix-vr4130
12401 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
12402 workarounds are implemented by the assembler rather than by GCC,
12403 although GCC will avoid using @code{mflo} and @code{mfhi} if the
12404 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
12405 instructions are available instead.
12408 @itemx -mno-fix-sb1
12410 Work around certain SB-1 CPU core errata.
12411 (This flag currently works around the SB-1 revision 2
12412 ``F1'' and ``F2'' floating point errata.)
12414 @item -mflush-func=@var{func}
12415 @itemx -mno-flush-func
12416 @opindex mflush-func
12417 Specifies the function to call to flush the I and D caches, or to not
12418 call any such function. If called, the function must take the same
12419 arguments as the common @code{_flush_func()}, that is, the address of the
12420 memory range for which the cache is being flushed, the size of the
12421 memory range, and the number 3 (to flush both caches). The default
12422 depends on the target GCC was configured for, but commonly is either
12423 @samp{_flush_func} or @samp{__cpu_flush}.
12425 @item mbranch-cost=@var{num}
12426 @opindex mbranch-cost
12427 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12428 This cost is only a heuristic and is not guaranteed to produce
12429 consistent results across releases. A zero cost redundantly selects
12430 the default, which is based on the @option{-mtune} setting.
12432 @item -mbranch-likely
12433 @itemx -mno-branch-likely
12434 @opindex mbranch-likely
12435 @opindex mno-branch-likely
12436 Enable or disable use of Branch Likely instructions, regardless of the
12437 default for the selected architecture. By default, Branch Likely
12438 instructions may be generated if they are supported by the selected
12439 architecture. An exception is for the MIPS32 and MIPS64 architectures
12440 and processors which implement those architectures; for those, Branch
12441 Likely instructions will not be generated by default because the MIPS32
12442 and MIPS64 architectures specifically deprecate their use.
12444 @item -mfp-exceptions
12445 @itemx -mno-fp-exceptions
12446 @opindex mfp-exceptions
12447 Specifies whether FP exceptions are enabled. This affects how we schedule
12448 FP instructions for some processors. The default is that FP exceptions are
12451 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12452 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12455 @item -mvr4130-align
12456 @itemx -mno-vr4130-align
12457 @opindex mvr4130-align
12458 The VR4130 pipeline is two-way superscalar, but can only issue two
12459 instructions together if the first one is 8-byte aligned. When this
12460 option is enabled, GCC will align pairs of instructions that it
12461 thinks should execute in parallel.
12463 This option only has an effect when optimizing for the VR4130.
12464 It normally makes code faster, but at the expense of making it bigger.
12465 It is enabled by default at optimization level @option{-O3}.
12469 @subsection MMIX Options
12470 @cindex MMIX Options
12472 These options are defined for the MMIX:
12476 @itemx -mno-libfuncs
12478 @opindex mno-libfuncs
12479 Specify that intrinsic library functions are being compiled, passing all
12480 values in registers, no matter the size.
12483 @itemx -mno-epsilon
12485 @opindex mno-epsilon
12486 Generate floating-point comparison instructions that compare with respect
12487 to the @code{rE} epsilon register.
12489 @item -mabi=mmixware
12491 @opindex mabi-mmixware
12493 Generate code that passes function parameters and return values that (in
12494 the called function) are seen as registers @code{$0} and up, as opposed to
12495 the GNU ABI which uses global registers @code{$231} and up.
12497 @item -mzero-extend
12498 @itemx -mno-zero-extend
12499 @opindex mzero-extend
12500 @opindex mno-zero-extend
12501 When reading data from memory in sizes shorter than 64 bits, use (do not
12502 use) zero-extending load instructions by default, rather than
12503 sign-extending ones.
12506 @itemx -mno-knuthdiv
12508 @opindex mno-knuthdiv
12509 Make the result of a division yielding a remainder have the same sign as
12510 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12511 remainder follows the sign of the dividend. Both methods are
12512 arithmetically valid, the latter being almost exclusively used.
12514 @item -mtoplevel-symbols
12515 @itemx -mno-toplevel-symbols
12516 @opindex mtoplevel-symbols
12517 @opindex mno-toplevel-symbols
12518 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12519 code can be used with the @code{PREFIX} assembly directive.
12523 Generate an executable in the ELF format, rather than the default
12524 @samp{mmo} format used by the @command{mmix} simulator.
12526 @item -mbranch-predict
12527 @itemx -mno-branch-predict
12528 @opindex mbranch-predict
12529 @opindex mno-branch-predict
12530 Use (do not use) the probable-branch instructions, when static branch
12531 prediction indicates a probable branch.
12533 @item -mbase-addresses
12534 @itemx -mno-base-addresses
12535 @opindex mbase-addresses
12536 @opindex mno-base-addresses
12537 Generate (do not generate) code that uses @emph{base addresses}. Using a
12538 base address automatically generates a request (handled by the assembler
12539 and the linker) for a constant to be set up in a global register. The
12540 register is used for one or more base address requests within the range 0
12541 to 255 from the value held in the register. The generally leads to short
12542 and fast code, but the number of different data items that can be
12543 addressed is limited. This means that a program that uses lots of static
12544 data may require @option{-mno-base-addresses}.
12546 @item -msingle-exit
12547 @itemx -mno-single-exit
12548 @opindex msingle-exit
12549 @opindex mno-single-exit
12550 Force (do not force) generated code to have a single exit point in each
12554 @node MN10300 Options
12555 @subsection MN10300 Options
12556 @cindex MN10300 options
12558 These @option{-m} options are defined for Matsushita MN10300 architectures:
12563 Generate code to avoid bugs in the multiply instructions for the MN10300
12564 processors. This is the default.
12566 @item -mno-mult-bug
12567 @opindex mno-mult-bug
12568 Do not generate code to avoid bugs in the multiply instructions for the
12569 MN10300 processors.
12573 Generate code which uses features specific to the AM33 processor.
12577 Do not generate code which uses features specific to the AM33 processor. This
12580 @item -mreturn-pointer-on-d0
12581 @opindex mreturn-pointer-on-d0
12582 When generating a function which returns a pointer, return the pointer
12583 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12584 only in a0, and attempts to call such functions without a prototype
12585 would result in errors. Note that this option is on by default; use
12586 @option{-mno-return-pointer-on-d0} to disable it.
12590 Do not link in the C run-time initialization object file.
12594 Indicate to the linker that it should perform a relaxation optimization pass
12595 to shorten branches, calls and absolute memory addresses. This option only
12596 has an effect when used on the command line for the final link step.
12598 This option makes symbolic debugging impossible.
12602 @subsection MT Options
12605 These @option{-m} options are defined for Morpho MT architectures:
12609 @item -march=@var{cpu-type}
12611 Generate code that will run on @var{cpu-type}, which is the name of a system
12612 representing a certain processor type. Possible values for
12613 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
12614 @samp{ms1-16-003} and @samp{ms2}.
12616 When this option is not used, the default is @option{-march=ms1-16-002}.
12620 Use byte loads and stores when generating code.
12624 Do not use byte loads and stores when generating code.
12628 Use simulator runtime
12632 Do not link in the C run-time initialization object file
12633 @file{crti.o}. Other run-time initialization and termination files
12634 such as @file{startup.o} and @file{exit.o} are still included on the
12635 linker command line.
12639 @node PDP-11 Options
12640 @subsection PDP-11 Options
12641 @cindex PDP-11 Options
12643 These options are defined for the PDP-11:
12648 Use hardware FPP floating point. This is the default. (FIS floating
12649 point on the PDP-11/40 is not supported.)
12652 @opindex msoft-float
12653 Do not use hardware floating point.
12657 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12661 Return floating-point results in memory. This is the default.
12665 Generate code for a PDP-11/40.
12669 Generate code for a PDP-11/45. This is the default.
12673 Generate code for a PDP-11/10.
12675 @item -mbcopy-builtin
12676 @opindex bcopy-builtin
12677 Use inline @code{movmemhi} patterns for copying memory. This is the
12682 Do not use inline @code{movmemhi} patterns for copying memory.
12688 Use 16-bit @code{int}. This is the default.
12694 Use 32-bit @code{int}.
12697 @itemx -mno-float32
12699 @opindex mno-float32
12700 Use 64-bit @code{float}. This is the default.
12703 @itemx -mno-float64
12705 @opindex mno-float64
12706 Use 32-bit @code{float}.
12710 Use @code{abshi2} pattern. This is the default.
12714 Do not use @code{abshi2} pattern.
12716 @item -mbranch-expensive
12717 @opindex mbranch-expensive
12718 Pretend that branches are expensive. This is for experimenting with
12719 code generation only.
12721 @item -mbranch-cheap
12722 @opindex mbranch-cheap
12723 Do not pretend that branches are expensive. This is the default.
12727 Generate code for a system with split I&D@.
12731 Generate code for a system without split I&D@. This is the default.
12735 Use Unix assembler syntax. This is the default when configured for
12736 @samp{pdp11-*-bsd}.
12740 Use DEC assembler syntax. This is the default when configured for any
12741 PDP-11 target other than @samp{pdp11-*-bsd}.
12744 @node PowerPC Options
12745 @subsection PowerPC Options
12746 @cindex PowerPC options
12748 These are listed under @xref{RS/6000 and PowerPC Options}.
12750 @node RS/6000 and PowerPC Options
12751 @subsection IBM RS/6000 and PowerPC Options
12752 @cindex RS/6000 and PowerPC Options
12753 @cindex IBM RS/6000 and PowerPC Options
12755 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12762 @itemx -mno-powerpc
12763 @itemx -mpowerpc-gpopt
12764 @itemx -mno-powerpc-gpopt
12765 @itemx -mpowerpc-gfxopt
12766 @itemx -mno-powerpc-gfxopt
12768 @itemx -mno-powerpc64
12772 @itemx -mno-popcntb
12780 @itemx -mno-hard-dfp
12784 @opindex mno-power2
12786 @opindex mno-powerpc
12787 @opindex mpowerpc-gpopt
12788 @opindex mno-powerpc-gpopt
12789 @opindex mpowerpc-gfxopt
12790 @opindex mno-powerpc-gfxopt
12791 @opindex mpowerpc64
12792 @opindex mno-powerpc64
12796 @opindex mno-popcntb
12802 @opindex mno-mfpgpr
12804 @opindex mno-hard-dfp
12805 GCC supports two related instruction set architectures for the
12806 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12807 instructions supported by the @samp{rios} chip set used in the original
12808 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12809 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12810 the IBM 4xx, 6xx, and follow-on microprocessors.
12812 Neither architecture is a subset of the other. However there is a
12813 large common subset of instructions supported by both. An MQ
12814 register is included in processors supporting the POWER architecture.
12816 You use these options to specify which instructions are available on the
12817 processor you are using. The default value of these options is
12818 determined when configuring GCC@. Specifying the
12819 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12820 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12821 rather than the options listed above.
12823 The @option{-mpower} option allows GCC to generate instructions that
12824 are found only in the POWER architecture and to use the MQ register.
12825 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12826 to generate instructions that are present in the POWER2 architecture but
12827 not the original POWER architecture.
12829 The @option{-mpowerpc} option allows GCC to generate instructions that
12830 are found only in the 32-bit subset of the PowerPC architecture.
12831 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12832 GCC to use the optional PowerPC architecture instructions in the
12833 General Purpose group, including floating-point square root. Specifying
12834 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12835 use the optional PowerPC architecture instructions in the Graphics
12836 group, including floating-point select.
12838 The @option{-mmfcrf} option allows GCC to generate the move from
12839 condition register field instruction implemented on the POWER4
12840 processor and other processors that support the PowerPC V2.01
12842 The @option{-mpopcntb} option allows GCC to generate the popcount and
12843 double precision FP reciprocal estimate instruction implemented on the
12844 POWER5 processor and other processors that support the PowerPC V2.02
12846 The @option{-mfprnd} option allows GCC to generate the FP round to
12847 integer instructions implemented on the POWER5+ processor and other
12848 processors that support the PowerPC V2.03 architecture.
12849 The @option{-mcmpb} option allows GCC to generate the compare bytes
12850 instruction implemented on the POWER6 processor and other processors
12851 that support the PowerPC V2.05 architecture.
12852 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12853 general purpose register instructions implemented on the POWER6X
12854 processor and other processors that support the extended PowerPC V2.05
12856 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
12857 point instructions implemented on some POWER processors.
12859 The @option{-mpowerpc64} option allows GCC to generate the additional
12860 64-bit instructions that are found in the full PowerPC64 architecture
12861 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12862 @option{-mno-powerpc64}.
12864 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12865 will use only the instructions in the common subset of both
12866 architectures plus some special AIX common-mode calls, and will not use
12867 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12868 permits GCC to use any instruction from either architecture and to
12869 allow use of the MQ register; specify this for the Motorola MPC601.
12871 @item -mnew-mnemonics
12872 @itemx -mold-mnemonics
12873 @opindex mnew-mnemonics
12874 @opindex mold-mnemonics
12875 Select which mnemonics to use in the generated assembler code. With
12876 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12877 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12878 assembler mnemonics defined for the POWER architecture. Instructions
12879 defined in only one architecture have only one mnemonic; GCC uses that
12880 mnemonic irrespective of which of these options is specified.
12882 GCC defaults to the mnemonics appropriate for the architecture in
12883 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
12884 value of these option. Unless you are building a cross-compiler, you
12885 should normally not specify either @option{-mnew-mnemonics} or
12886 @option{-mold-mnemonics}, but should instead accept the default.
12888 @item -mcpu=@var{cpu_type}
12890 Set architecture type, register usage, choice of mnemonics, and
12891 instruction scheduling parameters for machine type @var{cpu_type}.
12892 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
12893 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
12894 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
12895 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
12896 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
12897 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
12898 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
12899 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
12900 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
12901 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
12903 @option{-mcpu=common} selects a completely generic processor. Code
12904 generated under this option will run on any POWER or PowerPC processor.
12905 GCC will use only the instructions in the common subset of both
12906 architectures, and will not use the MQ register. GCC assumes a generic
12907 processor model for scheduling purposes.
12909 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
12910 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
12911 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
12912 types, with an appropriate, generic processor model assumed for
12913 scheduling purposes.
12915 The other options specify a specific processor. Code generated under
12916 those options will run best on that processor, and may not run at all on
12919 The @option{-mcpu} options automatically enable or disable the
12922 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
12923 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
12924 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
12926 The particular options set for any particular CPU will vary between
12927 compiler versions, depending on what setting seems to produce optimal
12928 code for that CPU; it doesn't necessarily reflect the actual hardware's
12929 capabilities. If you wish to set an individual option to a particular
12930 value, you may specify it after the @option{-mcpu} option, like
12931 @samp{-mcpu=970 -mno-altivec}.
12933 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
12934 not enabled or disabled by the @option{-mcpu} option at present because
12935 AIX does not have full support for these options. You may still
12936 enable or disable them individually if you're sure it'll work in your
12939 @item -mtune=@var{cpu_type}
12941 Set the instruction scheduling parameters for machine type
12942 @var{cpu_type}, but do not set the architecture type, register usage, or
12943 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
12944 values for @var{cpu_type} are used for @option{-mtune} as for
12945 @option{-mcpu}. If both are specified, the code generated will use the
12946 architecture, registers, and mnemonics set by @option{-mcpu}, but the
12947 scheduling parameters set by @option{-mtune}.
12953 Generate code to compute division as reciprocal estimate and iterative
12954 refinement, creating opportunities for increased throughput. This
12955 feature requires: optional PowerPC Graphics instruction set for single
12956 precision and FRE instruction for double precision, assuming divides
12957 cannot generate user-visible traps, and the domain values not include
12958 Infinities, denormals or zero denominator.
12961 @itemx -mno-altivec
12963 @opindex mno-altivec
12964 Generate code that uses (does not use) AltiVec instructions, and also
12965 enable the use of built-in functions that allow more direct access to
12966 the AltiVec instruction set. You may also need to set
12967 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
12973 @opindex mno-vrsave
12974 Generate VRSAVE instructions when generating AltiVec code.
12977 @opindex msecure-plt
12978 Generate code that allows ld and ld.so to build executables and shared
12979 libraries with non-exec .plt and .got sections. This is a PowerPC
12980 32-bit SYSV ABI option.
12984 Generate code that uses a BSS .plt section that ld.so fills in, and
12985 requires .plt and .got sections that are both writable and executable.
12986 This is a PowerPC 32-bit SYSV ABI option.
12992 This switch enables or disables the generation of ISEL instructions.
12994 @item -misel=@var{yes/no}
12995 This switch has been deprecated. Use @option{-misel} and
12996 @option{-mno-isel} instead.
13002 This switch enables or disables the generation of SPE simd
13008 @opindex mno-paired
13009 This switch enables or disables the generation of PAIRED simd
13012 @item -mspe=@var{yes/no}
13013 This option has been deprecated. Use @option{-mspe} and
13014 @option{-mno-spe} instead.
13016 @item -mfloat-gprs=@var{yes/single/double/no}
13017 @itemx -mfloat-gprs
13018 @opindex mfloat-gprs
13019 This switch enables or disables the generation of floating point
13020 operations on the general purpose registers for architectures that
13023 The argument @var{yes} or @var{single} enables the use of
13024 single-precision floating point operations.
13026 The argument @var{double} enables the use of single and
13027 double-precision floating point operations.
13029 The argument @var{no} disables floating point operations on the
13030 general purpose registers.
13032 This option is currently only available on the MPC854x.
13038 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
13039 targets (including GNU/Linux). The 32-bit environment sets int, long
13040 and pointer to 32 bits and generates code that runs on any PowerPC
13041 variant. The 64-bit environment sets int to 32 bits and long and
13042 pointer to 64 bits, and generates code for PowerPC64, as for
13043 @option{-mpowerpc64}.
13046 @itemx -mno-fp-in-toc
13047 @itemx -mno-sum-in-toc
13048 @itemx -mminimal-toc
13050 @opindex mno-fp-in-toc
13051 @opindex mno-sum-in-toc
13052 @opindex mminimal-toc
13053 Modify generation of the TOC (Table Of Contents), which is created for
13054 every executable file. The @option{-mfull-toc} option is selected by
13055 default. In that case, GCC will allocate at least one TOC entry for
13056 each unique non-automatic variable reference in your program. GCC
13057 will also place floating-point constants in the TOC@. However, only
13058 16,384 entries are available in the TOC@.
13060 If you receive a linker error message that saying you have overflowed
13061 the available TOC space, you can reduce the amount of TOC space used
13062 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
13063 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
13064 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
13065 generate code to calculate the sum of an address and a constant at
13066 run-time instead of putting that sum into the TOC@. You may specify one
13067 or both of these options. Each causes GCC to produce very slightly
13068 slower and larger code at the expense of conserving TOC space.
13070 If you still run out of space in the TOC even when you specify both of
13071 these options, specify @option{-mminimal-toc} instead. This option causes
13072 GCC to make only one TOC entry for every file. When you specify this
13073 option, GCC will produce code that is slower and larger but which
13074 uses extremely little TOC space. You may wish to use this option
13075 only on files that contain less frequently executed code.
13081 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
13082 @code{long} type, and the infrastructure needed to support them.
13083 Specifying @option{-maix64} implies @option{-mpowerpc64} and
13084 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
13085 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
13088 @itemx -mno-xl-compat
13089 @opindex mxl-compat
13090 @opindex mno-xl-compat
13091 Produce code that conforms more closely to IBM XL compiler semantics
13092 when using AIX-compatible ABI. Pass floating-point arguments to
13093 prototyped functions beyond the register save area (RSA) on the stack
13094 in addition to argument FPRs. Do not assume that most significant
13095 double in 128-bit long double value is properly rounded when comparing
13096 values and converting to double. Use XL symbol names for long double
13099 The AIX calling convention was extended but not initially documented to
13100 handle an obscure K&R C case of calling a function that takes the
13101 address of its arguments with fewer arguments than declared. IBM XL
13102 compilers access floating point arguments which do not fit in the
13103 RSA from the stack when a subroutine is compiled without
13104 optimization. Because always storing floating-point arguments on the
13105 stack is inefficient and rarely needed, this option is not enabled by
13106 default and only is necessary when calling subroutines compiled by IBM
13107 XL compilers without optimization.
13111 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
13112 application written to use message passing with special startup code to
13113 enable the application to run. The system must have PE installed in the
13114 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13115 must be overridden with the @option{-specs=} option to specify the
13116 appropriate directory location. The Parallel Environment does not
13117 support threads, so the @option{-mpe} option and the @option{-pthread}
13118 option are incompatible.
13120 @item -malign-natural
13121 @itemx -malign-power
13122 @opindex malign-natural
13123 @opindex malign-power
13124 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13125 @option{-malign-natural} overrides the ABI-defined alignment of larger
13126 types, such as floating-point doubles, on their natural size-based boundary.
13127 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13128 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
13130 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13134 @itemx -mhard-float
13135 @opindex msoft-float
13136 @opindex mhard-float
13137 Generate code that does not use (uses) the floating-point register set.
13138 Software floating point emulation is provided if you use the
13139 @option{-msoft-float} option, and pass the option to GCC when linking.
13142 @itemx -mno-multiple
13144 @opindex mno-multiple
13145 Generate code that uses (does not use) the load multiple word
13146 instructions and the store multiple word instructions. These
13147 instructions are generated by default on POWER systems, and not
13148 generated on PowerPC systems. Do not use @option{-mmultiple} on little
13149 endian PowerPC systems, since those instructions do not work when the
13150 processor is in little endian mode. The exceptions are PPC740 and
13151 PPC750 which permit the instructions usage in little endian mode.
13156 @opindex mno-string
13157 Generate code that uses (does not use) the load string instructions
13158 and the store string word instructions to save multiple registers and
13159 do small block moves. These instructions are generated by default on
13160 POWER systems, and not generated on PowerPC systems. Do not use
13161 @option{-mstring} on little endian PowerPC systems, since those
13162 instructions do not work when the processor is in little endian mode.
13163 The exceptions are PPC740 and PPC750 which permit the instructions
13164 usage in little endian mode.
13169 @opindex mno-update
13170 Generate code that uses (does not use) the load or store instructions
13171 that update the base register to the address of the calculated memory
13172 location. These instructions are generated by default. If you use
13173 @option{-mno-update}, there is a small window between the time that the
13174 stack pointer is updated and the address of the previous frame is
13175 stored, which means code that walks the stack frame across interrupts or
13176 signals may get corrupted data.
13179 @itemx -mno-fused-madd
13180 @opindex mfused-madd
13181 @opindex mno-fused-madd
13182 Generate code that uses (does not use) the floating point multiply and
13183 accumulate instructions. These instructions are generated by default if
13184 hardware floating is used.
13190 Generate code that uses (does not use) the half-word multiply and
13191 multiply-accumulate instructions on the IBM 405 and 440 processors.
13192 These instructions are generated by default when targetting those
13199 Generate code that uses (does not use) the string-search @samp{dlmzb}
13200 instruction on the IBM 405 and 440 processors. This instruction is
13201 generated by default when targetting those processors.
13203 @item -mno-bit-align
13205 @opindex mno-bit-align
13206 @opindex mbit-align
13207 On System V.4 and embedded PowerPC systems do not (do) force structures
13208 and unions that contain bit-fields to be aligned to the base type of the
13211 For example, by default a structure containing nothing but 8
13212 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
13213 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
13214 the structure would be aligned to a 1 byte boundary and be one byte in
13217 @item -mno-strict-align
13218 @itemx -mstrict-align
13219 @opindex mno-strict-align
13220 @opindex mstrict-align
13221 On System V.4 and embedded PowerPC systems do not (do) assume that
13222 unaligned memory references will be handled by the system.
13224 @item -mrelocatable
13225 @itemx -mno-relocatable
13226 @opindex mrelocatable
13227 @opindex mno-relocatable
13228 On embedded PowerPC systems generate code that allows (does not allow)
13229 the program to be relocated to a different address at runtime. If you
13230 use @option{-mrelocatable} on any module, all objects linked together must
13231 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
13233 @item -mrelocatable-lib
13234 @itemx -mno-relocatable-lib
13235 @opindex mrelocatable-lib
13236 @opindex mno-relocatable-lib
13237 On embedded PowerPC systems generate code that allows (does not allow)
13238 the program to be relocated to a different address at runtime. Modules
13239 compiled with @option{-mrelocatable-lib} can be linked with either modules
13240 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
13241 with modules compiled with the @option{-mrelocatable} options.
13247 On System V.4 and embedded PowerPC systems do not (do) assume that
13248 register 2 contains a pointer to a global area pointing to the addresses
13249 used in the program.
13252 @itemx -mlittle-endian
13254 @opindex mlittle-endian
13255 On System V.4 and embedded PowerPC systems compile code for the
13256 processor in little endian mode. The @option{-mlittle-endian} option is
13257 the same as @option{-mlittle}.
13260 @itemx -mbig-endian
13262 @opindex mbig-endian
13263 On System V.4 and embedded PowerPC systems compile code for the
13264 processor in big endian mode. The @option{-mbig-endian} option is
13265 the same as @option{-mbig}.
13267 @item -mdynamic-no-pic
13268 @opindex mdynamic-no-pic
13269 On Darwin and Mac OS X systems, compile code so that it is not
13270 relocatable, but that its external references are relocatable. The
13271 resulting code is suitable for applications, but not shared
13274 @item -mprioritize-restricted-insns=@var{priority}
13275 @opindex mprioritize-restricted-insns
13276 This option controls the priority that is assigned to
13277 dispatch-slot restricted instructions during the second scheduling
13278 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
13279 @var{no/highest/second-highest} priority to dispatch slot restricted
13282 @item -msched-costly-dep=@var{dependence_type}
13283 @opindex msched-costly-dep
13284 This option controls which dependences are considered costly
13285 by the target during instruction scheduling. The argument
13286 @var{dependence_type} takes one of the following values:
13287 @var{no}: no dependence is costly,
13288 @var{all}: all dependences are costly,
13289 @var{true_store_to_load}: a true dependence from store to load is costly,
13290 @var{store_to_load}: any dependence from store to load is costly,
13291 @var{number}: any dependence which latency >= @var{number} is costly.
13293 @item -minsert-sched-nops=@var{scheme}
13294 @opindex minsert-sched-nops
13295 This option controls which nop insertion scheme will be used during
13296 the second scheduling pass. The argument @var{scheme} takes one of the
13298 @var{no}: Don't insert nops.
13299 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
13300 according to the scheduler's grouping.
13301 @var{regroup_exact}: Insert nops to force costly dependent insns into
13302 separate groups. Insert exactly as many nops as needed to force an insn
13303 to a new group, according to the estimated processor grouping.
13304 @var{number}: Insert nops to force costly dependent insns into
13305 separate groups. Insert @var{number} nops to force an insn to a new group.
13308 @opindex mcall-sysv
13309 On System V.4 and embedded PowerPC systems compile code using calling
13310 conventions that adheres to the March 1995 draft of the System V
13311 Application Binary Interface, PowerPC processor supplement. This is the
13312 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
13314 @item -mcall-sysv-eabi
13315 @opindex mcall-sysv-eabi
13316 Specify both @option{-mcall-sysv} and @option{-meabi} options.
13318 @item -mcall-sysv-noeabi
13319 @opindex mcall-sysv-noeabi
13320 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
13322 @item -mcall-solaris
13323 @opindex mcall-solaris
13324 On System V.4 and embedded PowerPC systems compile code for the Solaris
13328 @opindex mcall-linux
13329 On System V.4 and embedded PowerPC systems compile code for the
13330 Linux-based GNU system.
13334 On System V.4 and embedded PowerPC systems compile code for the
13335 Hurd-based GNU system.
13337 @item -mcall-netbsd
13338 @opindex mcall-netbsd
13339 On System V.4 and embedded PowerPC systems compile code for the
13340 NetBSD operating system.
13342 @item -maix-struct-return
13343 @opindex maix-struct-return
13344 Return all structures in memory (as specified by the AIX ABI)@.
13346 @item -msvr4-struct-return
13347 @opindex msvr4-struct-return
13348 Return structures smaller than 8 bytes in registers (as specified by the
13351 @item -mabi=@var{abi-type}
13353 Extend the current ABI with a particular extension, or remove such extension.
13354 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
13355 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
13359 Extend the current ABI with SPE ABI extensions. This does not change
13360 the default ABI, instead it adds the SPE ABI extensions to the current
13364 @opindex mabi=no-spe
13365 Disable Booke SPE ABI extensions for the current ABI@.
13367 @item -mabi=ibmlongdouble
13368 @opindex mabi=ibmlongdouble
13369 Change the current ABI to use IBM extended precision long double.
13370 This is a PowerPC 32-bit SYSV ABI option.
13372 @item -mabi=ieeelongdouble
13373 @opindex mabi=ieeelongdouble
13374 Change the current ABI to use IEEE extended precision long double.
13375 This is a PowerPC 32-bit Linux ABI option.
13378 @itemx -mno-prototype
13379 @opindex mprototype
13380 @opindex mno-prototype
13381 On System V.4 and embedded PowerPC systems assume that all calls to
13382 variable argument functions are properly prototyped. Otherwise, the
13383 compiler must insert an instruction before every non prototyped call to
13384 set or clear bit 6 of the condition code register (@var{CR}) to
13385 indicate whether floating point values were passed in the floating point
13386 registers in case the function takes a variable arguments. With
13387 @option{-mprototype}, only calls to prototyped variable argument functions
13388 will set or clear the bit.
13392 On embedded PowerPC systems, assume that the startup module is called
13393 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
13394 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
13399 On embedded PowerPC systems, assume that the startup module is called
13400 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
13405 On embedded PowerPC systems, assume that the startup module is called
13406 @file{crt0.o} and the standard C libraries are @file{libads.a} and
13409 @item -myellowknife
13410 @opindex myellowknife
13411 On embedded PowerPC systems, assume that the startup module is called
13412 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
13417 On System V.4 and embedded PowerPC systems, specify that you are
13418 compiling for a VxWorks system.
13422 Specify that you are compiling for the WindISS simulation environment.
13426 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
13427 header to indicate that @samp{eabi} extended relocations are used.
13433 On System V.4 and embedded PowerPC systems do (do not) adhere to the
13434 Embedded Applications Binary Interface (eabi) which is a set of
13435 modifications to the System V.4 specifications. Selecting @option{-meabi}
13436 means that the stack is aligned to an 8 byte boundary, a function
13437 @code{__eabi} is called to from @code{main} to set up the eabi
13438 environment, and the @option{-msdata} option can use both @code{r2} and
13439 @code{r13} to point to two separate small data areas. Selecting
13440 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
13441 do not call an initialization function from @code{main}, and the
13442 @option{-msdata} option will only use @code{r13} to point to a single
13443 small data area. The @option{-meabi} option is on by default if you
13444 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
13447 @opindex msdata=eabi
13448 On System V.4 and embedded PowerPC systems, put small initialized
13449 @code{const} global and static data in the @samp{.sdata2} section, which
13450 is pointed to by register @code{r2}. Put small initialized
13451 non-@code{const} global and static data in the @samp{.sdata} section,
13452 which is pointed to by register @code{r13}. Put small uninitialized
13453 global and static data in the @samp{.sbss} section, which is adjacent to
13454 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13455 incompatible with the @option{-mrelocatable} option. The
13456 @option{-msdata=eabi} option also sets the @option{-memb} option.
13459 @opindex msdata=sysv
13460 On System V.4 and embedded PowerPC systems, put small global and static
13461 data in the @samp{.sdata} section, which is pointed to by register
13462 @code{r13}. Put small uninitialized global and static data in the
13463 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13464 The @option{-msdata=sysv} option is incompatible with the
13465 @option{-mrelocatable} option.
13467 @item -msdata=default
13469 @opindex msdata=default
13471 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13472 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13473 same as @option{-msdata=sysv}.
13476 @opindex msdata-data
13477 On System V.4 and embedded PowerPC systems, put small global
13478 data in the @samp{.sdata} section. Put small uninitialized global
13479 data in the @samp{.sbss} section. Do not use register @code{r13}
13480 to address small data however. This is the default behavior unless
13481 other @option{-msdata} options are used.
13485 @opindex msdata=none
13487 On embedded PowerPC systems, put all initialized global and static data
13488 in the @samp{.data} section, and all uninitialized data in the
13489 @samp{.bss} section.
13493 @cindex smaller data references (PowerPC)
13494 @cindex .sdata/.sdata2 references (PowerPC)
13495 On embedded PowerPC systems, put global and static items less than or
13496 equal to @var{num} bytes into the small data or bss sections instead of
13497 the normal data or bss section. By default, @var{num} is 8. The
13498 @option{-G @var{num}} switch is also passed to the linker.
13499 All modules should be compiled with the same @option{-G @var{num}} value.
13502 @itemx -mno-regnames
13504 @opindex mno-regnames
13505 On System V.4 and embedded PowerPC systems do (do not) emit register
13506 names in the assembly language output using symbolic forms.
13509 @itemx -mno-longcall
13511 @opindex mno-longcall
13512 By default assume that all calls are far away so that a longer more
13513 expensive calling sequence is required. This is required for calls
13514 further than 32 megabytes (33,554,432 bytes) from the current location.
13515 A short call will be generated if the compiler knows
13516 the call cannot be that far away. This setting can be overridden by
13517 the @code{shortcall} function attribute, or by @code{#pragma
13520 Some linkers are capable of detecting out-of-range calls and generating
13521 glue code on the fly. On these systems, long calls are unnecessary and
13522 generate slower code. As of this writing, the AIX linker can do this,
13523 as can the GNU linker for PowerPC/64. It is planned to add this feature
13524 to the GNU linker for 32-bit PowerPC systems as well.
13526 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13527 callee, L42'', plus a ``branch island'' (glue code). The two target
13528 addresses represent the callee and the ``branch island''. The
13529 Darwin/PPC linker will prefer the first address and generate a ``bl
13530 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13531 otherwise, the linker will generate ``bl L42'' to call the ``branch
13532 island''. The ``branch island'' is appended to the body of the
13533 calling function; it computes the full 32-bit address of the callee
13536 On Mach-O (Darwin) systems, this option directs the compiler emit to
13537 the glue for every direct call, and the Darwin linker decides whether
13538 to use or discard it.
13540 In the future, we may cause GCC to ignore all longcall specifications
13541 when the linker is known to generate glue.
13545 Adds support for multithreading with the @dfn{pthreads} library.
13546 This option sets flags for both the preprocessor and linker.
13550 @node S/390 and zSeries Options
13551 @subsection S/390 and zSeries Options
13552 @cindex S/390 and zSeries Options
13554 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13558 @itemx -msoft-float
13559 @opindex mhard-float
13560 @opindex msoft-float
13561 Use (do not use) the hardware floating-point instructions and registers
13562 for floating-point operations. When @option{-msoft-float} is specified,
13563 functions in @file{libgcc.a} will be used to perform floating-point
13564 operations. When @option{-mhard-float} is specified, the compiler
13565 generates IEEE floating-point instructions. This is the default.
13567 @item -mlong-double-64
13568 @itemx -mlong-double-128
13569 @opindex mlong-double-64
13570 @opindex mlong-double-128
13571 These switches control the size of @code{long double} type. A size
13572 of 64bit makes the @code{long double} type equivalent to the @code{double}
13573 type. This is the default.
13576 @itemx -mno-backchain
13577 @opindex mbackchain
13578 @opindex mno-backchain
13579 Store (do not store) the address of the caller's frame as backchain pointer
13580 into the callee's stack frame.
13581 A backchain may be needed to allow debugging using tools that do not understand
13582 DWARF-2 call frame information.
13583 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13584 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13585 the backchain is placed into the topmost word of the 96/160 byte register
13588 In general, code compiled with @option{-mbackchain} is call-compatible with
13589 code compiled with @option{-mmo-backchain}; however, use of the backchain
13590 for debugging purposes usually requires that the whole binary is built with
13591 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13592 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13593 to build a linux kernel use @option{-msoft-float}.
13595 The default is to not maintain the backchain.
13597 @item -mpacked-stack
13598 @item -mno-packed-stack
13599 @opindex mpacked-stack
13600 @opindex mno-packed-stack
13601 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13602 specified, the compiler uses the all fields of the 96/160 byte register save
13603 area only for their default purpose; unused fields still take up stack space.
13604 When @option{-mpacked-stack} is specified, register save slots are densely
13605 packed at the top of the register save area; unused space is reused for other
13606 purposes, allowing for more efficient use of the available stack space.
13607 However, when @option{-mbackchain} is also in effect, the topmost word of
13608 the save area is always used to store the backchain, and the return address
13609 register is always saved two words below the backchain.
13611 As long as the stack frame backchain is not used, code generated with
13612 @option{-mpacked-stack} is call-compatible with code generated with
13613 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13614 S/390 or zSeries generated code that uses the stack frame backchain at run
13615 time, not just for debugging purposes. Such code is not call-compatible
13616 with code compiled with @option{-mpacked-stack}. Also, note that the
13617 combination of @option{-mbackchain},
13618 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13619 to build a linux kernel use @option{-msoft-float}.
13621 The default is to not use the packed stack layout.
13624 @itemx -mno-small-exec
13625 @opindex msmall-exec
13626 @opindex mno-small-exec
13627 Generate (or do not generate) code using the @code{bras} instruction
13628 to do subroutine calls.
13629 This only works reliably if the total executable size does not
13630 exceed 64k. The default is to use the @code{basr} instruction instead,
13631 which does not have this limitation.
13637 When @option{-m31} is specified, generate code compliant to the
13638 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13639 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13640 particular to generate 64-bit instructions. For the @samp{s390}
13641 targets, the default is @option{-m31}, while the @samp{s390x}
13642 targets default to @option{-m64}.
13648 When @option{-mzarch} is specified, generate code using the
13649 instructions available on z/Architecture.
13650 When @option{-mesa} is specified, generate code using the
13651 instructions available on ESA/390. Note that @option{-mesa} is
13652 not possible with @option{-m64}.
13653 When generating code compliant to the GNU/Linux for S/390 ABI,
13654 the default is @option{-mesa}. When generating code compliant
13655 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13661 Generate (or do not generate) code using the @code{mvcle} instruction
13662 to perform block moves. When @option{-mno-mvcle} is specified,
13663 use a @code{mvc} loop instead. This is the default unless optimizing for
13670 Print (or do not print) additional debug information when compiling.
13671 The default is to not print debug information.
13673 @item -march=@var{cpu-type}
13675 Generate code that will run on @var{cpu-type}, which is the name of a system
13676 representing a certain processor type. Possible values for
13677 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13678 When generating code using the instructions available on z/Architecture,
13679 the default is @option{-march=z900}. Otherwise, the default is
13680 @option{-march=g5}.
13682 @item -mtune=@var{cpu-type}
13684 Tune to @var{cpu-type} everything applicable about the generated code,
13685 except for the ABI and the set of available instructions.
13686 The list of @var{cpu-type} values is the same as for @option{-march}.
13687 The default is the value used for @option{-march}.
13690 @itemx -mno-tpf-trace
13691 @opindex mtpf-trace
13692 @opindex mno-tpf-trace
13693 Generate code that adds (does not add) in TPF OS specific branches to trace
13694 routines in the operating system. This option is off by default, even
13695 when compiling for the TPF OS@.
13698 @itemx -mno-fused-madd
13699 @opindex mfused-madd
13700 @opindex mno-fused-madd
13701 Generate code that uses (does not use) the floating point multiply and
13702 accumulate instructions. These instructions are generated by default if
13703 hardware floating point is used.
13705 @item -mwarn-framesize=@var{framesize}
13706 @opindex mwarn-framesize
13707 Emit a warning if the current function exceeds the given frame size. Because
13708 this is a compile time check it doesn't need to be a real problem when the program
13709 runs. It is intended to identify functions which most probably cause
13710 a stack overflow. It is useful to be used in an environment with limited stack
13711 size e.g.@: the linux kernel.
13713 @item -mwarn-dynamicstack
13714 @opindex mwarn-dynamicstack
13715 Emit a warning if the function calls alloca or uses dynamically
13716 sized arrays. This is generally a bad idea with a limited stack size.
13718 @item -mstack-guard=@var{stack-guard}
13719 @item -mstack-size=@var{stack-size}
13720 @opindex mstack-guard
13721 @opindex mstack-size
13722 If these options are provided the s390 back end emits additional instructions in
13723 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13724 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13725 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13726 the frame size of the compiled function is chosen.
13727 These options are intended to be used to help debugging stack overflow problems.
13728 The additionally emitted code causes only little overhead and hence can also be
13729 used in production like systems without greater performance degradation. The given
13730 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13731 @var{stack-guard} without exceeding 64k.
13732 In order to be efficient the extra code makes the assumption that the stack starts
13733 at an address aligned to the value given by @var{stack-size}.
13734 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13737 @node Score Options
13738 @subsection Score Options
13739 @cindex Score Options
13741 These options are defined for Score implementations:
13746 Compile code for big endian mode. This is the default.
13750 Compile code for little endian mode.
13754 Disable generate bcnz instruction.
13758 Enable generate unaligned load and store instruction.
13762 Enable the use of multiply-accumulate instructions. Disabled by default.
13766 Specify the SCORE5 as the target architecture.
13770 Specify the SCORE5U of the target architecture.
13774 Specify the SCORE7 as the target architecture. This is the default.
13778 Specify the SCORE7D as the target architecture.
13782 @subsection SH Options
13784 These @samp{-m} options are defined for the SH implementations:
13789 Generate code for the SH1.
13793 Generate code for the SH2.
13796 Generate code for the SH2e.
13800 Generate code for the SH3.
13804 Generate code for the SH3e.
13808 Generate code for the SH4 without a floating-point unit.
13810 @item -m4-single-only
13811 @opindex m4-single-only
13812 Generate code for the SH4 with a floating-point unit that only
13813 supports single-precision arithmetic.
13817 Generate code for the SH4 assuming the floating-point unit is in
13818 single-precision mode by default.
13822 Generate code for the SH4.
13826 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13827 floating-point unit is not used.
13829 @item -m4a-single-only
13830 @opindex m4a-single-only
13831 Generate code for the SH4a, in such a way that no double-precision
13832 floating point operations are used.
13835 @opindex m4a-single
13836 Generate code for the SH4a assuming the floating-point unit is in
13837 single-precision mode by default.
13841 Generate code for the SH4a.
13845 Same as @option{-m4a-nofpu}, except that it implicitly passes
13846 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13847 instructions at the moment.
13851 Compile code for the processor in big endian mode.
13855 Compile code for the processor in little endian mode.
13859 Align doubles at 64-bit boundaries. Note that this changes the calling
13860 conventions, and thus some functions from the standard C library will
13861 not work unless you recompile it first with @option{-mdalign}.
13865 Shorten some address references at link time, when possible; uses the
13866 linker option @option{-relax}.
13870 Use 32-bit offsets in @code{switch} tables. The default is to use
13875 Enable the use of the instruction @code{fmovd}.
13879 Comply with the calling conventions defined by Renesas.
13883 Comply with the calling conventions defined by Renesas.
13887 Comply with the calling conventions defined for GCC before the Renesas
13888 conventions were available. This option is the default for all
13889 targets of the SH toolchain except for @samp{sh-symbianelf}.
13892 @opindex mnomacsave
13893 Mark the @code{MAC} register as call-clobbered, even if
13894 @option{-mhitachi} is given.
13898 Increase IEEE-compliance of floating-point code.
13899 At the moment, this is equivalent to @option{-fno-finite-math-only}.
13900 When generating 16 bit SH opcodes, getting IEEE-conforming results for
13901 comparisons of NANs / infinities incurs extra overhead in every
13902 floating point comparison, therefore the default is set to
13903 @option{-ffinite-math-only}.
13905 @item -minline-ic_invalidate
13906 @opindex minline-ic_invalidate
13907 Inline code to invalidate instruction cache entries after setting up
13908 nested function trampolines.
13909 This option has no effect if -musermode is in effect and the selected
13910 code generation option (e.g. -m4) does not allow the use of the icbi
13912 If the selected code generation option does not allow the use of the icbi
13913 instruction, and -musermode is not in effect, the inlined code will
13914 manipulate the instruction cache address array directly with an associative
13915 write. This not only requires privileged mode, but it will also
13916 fail if the cache line had been mapped via the TLB and has become unmapped.
13920 Dump instruction size and location in the assembly code.
13923 @opindex mpadstruct
13924 This option is deprecated. It pads structures to multiple of 4 bytes,
13925 which is incompatible with the SH ABI@.
13929 Optimize for space instead of speed. Implied by @option{-Os}.
13932 @opindex mprefergot
13933 When generating position-independent code, emit function calls using
13934 the Global Offset Table instead of the Procedure Linkage Table.
13938 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
13939 if the inlined code would not work in user mode.
13940 This is the default when the target is @code{sh-*-linux*}.
13942 @item -multcost=@var{number}
13943 @opindex multcost=@var{number}
13944 Set the cost to assume for a multiply insn.
13946 @item -mdiv=@var{strategy}
13947 @opindex mdiv=@var{strategy}
13948 Set the division strategy to use for SHmedia code. @var{strategy} must be
13949 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
13950 inv:call2, inv:fp .
13951 "fp" performs the operation in floating point. This has a very high latency,
13952 but needs only a few instructions, so it might be a good choice if
13953 your code has enough easily exploitable ILP to allow the compiler to
13954 schedule the floating point instructions together with other instructions.
13955 Division by zero causes a floating point exception.
13956 "inv" uses integer operations to calculate the inverse of the divisor,
13957 and then multiplies the dividend with the inverse. This strategy allows
13958 cse and hoisting of the inverse calculation. Division by zero calculates
13959 an unspecified result, but does not trap.
13960 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
13961 have been found, or if the entire operation has been hoisted to the same
13962 place, the last stages of the inverse calculation are intertwined with the
13963 final multiply to reduce the overall latency, at the expense of using a few
13964 more instructions, and thus offering fewer scheduling opportunities with
13966 "call" calls a library function that usually implements the inv:minlat
13968 This gives high code density for m5-*media-nofpu compilations.
13969 "call2" uses a different entry point of the same library function, where it
13970 assumes that a pointer to a lookup table has already been set up, which
13971 exposes the pointer load to cse / code hoisting optimizations.
13972 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
13973 code generation, but if the code stays unoptimized, revert to the "call",
13974 "call2", or "fp" strategies, respectively. Note that the
13975 potentially-trapping side effect of division by zero is carried by a
13976 separate instruction, so it is possible that all the integer instructions
13977 are hoisted out, but the marker for the side effect stays where it is.
13978 A recombination to fp operations or a call is not possible in that case.
13979 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
13980 that the inverse calculation was nor separated from the multiply, they speed
13981 up division where the dividend fits into 20 bits (plus sign where applicable),
13982 by inserting a test to skip a number of operations in this case; this test
13983 slows down the case of larger dividends. inv20u assumes the case of a such
13984 a small dividend to be unlikely, and inv20l assumes it to be likely.
13986 @item -mdivsi3_libfunc=@var{name}
13987 @opindex mdivsi3_libfunc=@var{name}
13988 Set the name of the library function used for 32 bit signed division to
13989 @var{name}. This only affect the name used in the call and inv:call
13990 division strategies, and the compiler will still expect the same
13991 sets of input/output/clobbered registers as if this option was not present.
13993 @item -madjust-unroll
13994 @opindex madjust-unroll
13995 Throttle unrolling to avoid thrashing target registers.
13996 This option only has an effect if the gcc code base supports the
13997 TARGET_ADJUST_UNROLL_MAX target hook.
13999 @item -mindexed-addressing
14000 @opindex mindexed-addressing
14001 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
14002 This is only safe if the hardware and/or OS implement 32 bit wrap-around
14003 semantics for the indexed addressing mode. The architecture allows the
14004 implementation of processors with 64 bit MMU, which the OS could use to
14005 get 32 bit addressing, but since no current hardware implementation supports
14006 this or any other way to make the indexed addressing mode safe to use in
14007 the 32 bit ABI, the default is -mno-indexed-addressing.
14009 @item -mgettrcost=@var{number}
14010 @opindex mgettrcost=@var{number}
14011 Set the cost assumed for the gettr instruction to @var{number}.
14012 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
14016 Assume pt* instructions won't trap. This will generally generate better
14017 scheduled code, but is unsafe on current hardware. The current architecture
14018 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
14019 This has the unintentional effect of making it unsafe to schedule ptabs /
14020 ptrel before a branch, or hoist it out of a loop. For example,
14021 __do_global_ctors, a part of libgcc that runs constructors at program
14022 startup, calls functions in a list which is delimited by @minus{}1. With the
14023 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
14024 That means that all the constructors will be run a bit quicker, but when
14025 the loop comes to the end of the list, the program crashes because ptabs
14026 loads @minus{}1 into a target register. Since this option is unsafe for any
14027 hardware implementing the current architecture specification, the default
14028 is -mno-pt-fixed. Unless the user specifies a specific cost with
14029 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
14030 this deters register allocation using target registers for storing
14033 @item -minvalid-symbols
14034 @opindex minvalid-symbols
14035 Assume symbols might be invalid. Ordinary function symbols generated by
14036 the compiler will always be valid to load with movi/shori/ptabs or
14037 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
14038 to generate symbols that will cause ptabs / ptrel to trap.
14039 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
14040 It will then prevent cross-basic-block cse, hoisting and most scheduling
14041 of symbol loads. The default is @option{-mno-invalid-symbols}.
14044 @node SPARC Options
14045 @subsection SPARC Options
14046 @cindex SPARC options
14048 These @samp{-m} options are supported on the SPARC:
14051 @item -mno-app-regs
14053 @opindex mno-app-regs
14055 Specify @option{-mapp-regs} to generate output using the global registers
14056 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
14059 To be fully SVR4 ABI compliant at the cost of some performance loss,
14060 specify @option{-mno-app-regs}. You should compile libraries and system
14061 software with this option.
14064 @itemx -mhard-float
14066 @opindex mhard-float
14067 Generate output containing floating point instructions. This is the
14071 @itemx -msoft-float
14073 @opindex msoft-float
14074 Generate output containing library calls for floating point.
14075 @strong{Warning:} the requisite libraries are not available for all SPARC
14076 targets. Normally the facilities of the machine's usual C compiler are
14077 used, but this cannot be done directly in cross-compilation. You must make
14078 your own arrangements to provide suitable library functions for
14079 cross-compilation. The embedded targets @samp{sparc-*-aout} and
14080 @samp{sparclite-*-*} do provide software floating point support.
14082 @option{-msoft-float} changes the calling convention in the output file;
14083 therefore, it is only useful if you compile @emph{all} of a program with
14084 this option. In particular, you need to compile @file{libgcc.a}, the
14085 library that comes with GCC, with @option{-msoft-float} in order for
14088 @item -mhard-quad-float
14089 @opindex mhard-quad-float
14090 Generate output containing quad-word (long double) floating point
14093 @item -msoft-quad-float
14094 @opindex msoft-quad-float
14095 Generate output containing library calls for quad-word (long double)
14096 floating point instructions. The functions called are those specified
14097 in the SPARC ABI@. This is the default.
14099 As of this writing, there are no SPARC implementations that have hardware
14100 support for the quad-word floating point instructions. They all invoke
14101 a trap handler for one of these instructions, and then the trap handler
14102 emulates the effect of the instruction. Because of the trap handler overhead,
14103 this is much slower than calling the ABI library routines. Thus the
14104 @option{-msoft-quad-float} option is the default.
14106 @item -mno-unaligned-doubles
14107 @itemx -munaligned-doubles
14108 @opindex mno-unaligned-doubles
14109 @opindex munaligned-doubles
14110 Assume that doubles have 8 byte alignment. This is the default.
14112 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
14113 alignment only if they are contained in another type, or if they have an
14114 absolute address. Otherwise, it assumes they have 4 byte alignment.
14115 Specifying this option avoids some rare compatibility problems with code
14116 generated by other compilers. It is not the default because it results
14117 in a performance loss, especially for floating point code.
14119 @item -mno-faster-structs
14120 @itemx -mfaster-structs
14121 @opindex mno-faster-structs
14122 @opindex mfaster-structs
14123 With @option{-mfaster-structs}, the compiler assumes that structures
14124 should have 8 byte alignment. This enables the use of pairs of
14125 @code{ldd} and @code{std} instructions for copies in structure
14126 assignment, in place of twice as many @code{ld} and @code{st} pairs.
14127 However, the use of this changed alignment directly violates the SPARC
14128 ABI@. Thus, it's intended only for use on targets where the developer
14129 acknowledges that their resulting code will not be directly in line with
14130 the rules of the ABI@.
14132 @item -mimpure-text
14133 @opindex mimpure-text
14134 @option{-mimpure-text}, used in addition to @option{-shared}, tells
14135 the compiler to not pass @option{-z text} to the linker when linking a
14136 shared object. Using this option, you can link position-dependent
14137 code into a shared object.
14139 @option{-mimpure-text} suppresses the ``relocations remain against
14140 allocatable but non-writable sections'' linker error message.
14141 However, the necessary relocations will trigger copy-on-write, and the
14142 shared object is not actually shared across processes. Instead of
14143 using @option{-mimpure-text}, you should compile all source code with
14144 @option{-fpic} or @option{-fPIC}.
14146 This option is only available on SunOS and Solaris.
14148 @item -mcpu=@var{cpu_type}
14150 Set the instruction set, register set, and instruction scheduling parameters
14151 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
14152 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
14153 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
14154 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
14155 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
14157 Default instruction scheduling parameters are used for values that select
14158 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
14159 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
14161 Here is a list of each supported architecture and their supported
14166 v8: supersparc, hypersparc
14167 sparclite: f930, f934, sparclite86x
14169 v9: ultrasparc, ultrasparc3, niagara, niagara2
14172 By default (unless configured otherwise), GCC generates code for the V7
14173 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
14174 additionally optimizes it for the Cypress CY7C602 chip, as used in the
14175 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
14176 SPARCStation 1, 2, IPX etc.
14178 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
14179 architecture. The only difference from V7 code is that the compiler emits
14180 the integer multiply and integer divide instructions which exist in SPARC-V8
14181 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
14182 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
14185 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
14186 the SPARC architecture. This adds the integer multiply, integer divide step
14187 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
14188 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
14189 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
14190 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
14191 MB86934 chip, which is the more recent SPARClite with FPU@.
14193 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
14194 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
14195 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
14196 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
14197 optimizes it for the TEMIC SPARClet chip.
14199 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
14200 architecture. This adds 64-bit integer and floating-point move instructions,
14201 3 additional floating-point condition code registers and conditional move
14202 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
14203 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
14204 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
14205 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
14206 @option{-mcpu=niagara}, the compiler additionally optimizes it for
14207 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
14208 additionally optimizes it for Sun UltraSPARC T2 chips.
14210 @item -mtune=@var{cpu_type}
14212 Set the instruction scheduling parameters for machine type
14213 @var{cpu_type}, but do not set the instruction set or register set that the
14214 option @option{-mcpu=@var{cpu_type}} would.
14216 The same values for @option{-mcpu=@var{cpu_type}} can be used for
14217 @option{-mtune=@var{cpu_type}}, but the only useful values are those
14218 that select a particular cpu implementation. Those are @samp{cypress},
14219 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
14220 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
14221 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
14226 @opindex mno-v8plus
14227 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
14228 difference from the V8 ABI is that the global and out registers are
14229 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
14230 mode for all SPARC-V9 processors.
14236 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
14237 Visual Instruction Set extensions. The default is @option{-mno-vis}.
14240 These @samp{-m} options are supported in addition to the above
14241 on SPARC-V9 processors in 64-bit environments:
14244 @item -mlittle-endian
14245 @opindex mlittle-endian
14246 Generate code for a processor running in little-endian mode. It is only
14247 available for a few configurations and most notably not on Solaris and Linux.
14253 Generate code for a 32-bit or 64-bit environment.
14254 The 32-bit environment sets int, long and pointer to 32 bits.
14255 The 64-bit environment sets int to 32 bits and long and pointer
14258 @item -mcmodel=medlow
14259 @opindex mcmodel=medlow
14260 Generate code for the Medium/Low code model: 64-bit addresses, programs
14261 must be linked in the low 32 bits of memory. Programs can be statically
14262 or dynamically linked.
14264 @item -mcmodel=medmid
14265 @opindex mcmodel=medmid
14266 Generate code for the Medium/Middle code model: 64-bit addresses, programs
14267 must be linked in the low 44 bits of memory, the text and data segments must
14268 be less than 2GB in size and the data segment must be located within 2GB of
14271 @item -mcmodel=medany
14272 @opindex mcmodel=medany
14273 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
14274 may be linked anywhere in memory, the text and data segments must be less
14275 than 2GB in size and the data segment must be located within 2GB of the
14278 @item -mcmodel=embmedany
14279 @opindex mcmodel=embmedany
14280 Generate code for the Medium/Anywhere code model for embedded systems:
14281 64-bit addresses, the text and data segments must be less than 2GB in
14282 size, both starting anywhere in memory (determined at link time). The
14283 global register %g4 points to the base of the data segment. Programs
14284 are statically linked and PIC is not supported.
14287 @itemx -mno-stack-bias
14288 @opindex mstack-bias
14289 @opindex mno-stack-bias
14290 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
14291 frame pointer if present, are offset by @minus{}2047 which must be added back
14292 when making stack frame references. This is the default in 64-bit mode.
14293 Otherwise, assume no such offset is present.
14296 These switches are supported in addition to the above on Solaris:
14301 Add support for multithreading using the Solaris threads library. This
14302 option sets flags for both the preprocessor and linker. This option does
14303 not affect the thread safety of object code produced by the compiler or
14304 that of libraries supplied with it.
14308 Add support for multithreading using the POSIX threads library. This
14309 option sets flags for both the preprocessor and linker. This option does
14310 not affect the thread safety of object code produced by the compiler or
14311 that of libraries supplied with it.
14315 This is a synonym for @option{-pthreads}.
14319 @subsection SPU Options
14320 @cindex SPU options
14322 These @samp{-m} options are supported on the SPU:
14326 @itemx -merror-reloc
14327 @opindex mwarn-reloc
14328 @opindex merror-reloc
14330 The loader for SPU does not handle dynamic relocations. By default, GCC
14331 will give an error when it generates code that requires a dynamic
14332 relocation. @option{-mno-error-reloc} disables the error,
14333 @option{-mwarn-reloc} will generate a warning instead.
14336 @itemx -munsafe-dma
14338 @opindex munsafe-dma
14340 Instructions which initiate or test completion of DMA must not be
14341 reordered with respect to loads and stores of the memory which is being
14342 accessed. Users typically address this problem using the volatile
14343 keyword, but that can lead to inefficient code in places where the
14344 memory is known to not change. Rather than mark the memory as volatile
14345 we treat the DMA instructions as potentially effecting all memory. With
14346 @option{-munsafe-dma} users must use the volatile keyword to protect
14349 @item -mbranch-hints
14350 @opindex mbranch-hints
14352 By default, GCC will generate a branch hint instruction to avoid
14353 pipeline stalls for always taken or probably taken branches. A hint
14354 will not be generated closer than 8 instructions away from its branch.
14355 There is little reason to disable them, except for debugging purposes,
14356 or to make an object a little bit smaller.
14360 @opindex msmall-mem
14361 @opindex mlarge-mem
14363 By default, GCC generates code assuming that addresses are never larger
14364 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
14365 a full 32 bit address.
14370 By default, GCC links against startup code that assumes the SPU-style
14371 main function interface (which has an unconventional parameter list).
14372 With @option{-mstdmain}, GCC will link your program against startup
14373 code that assumes a C99-style interface to @code{main}, including a
14374 local copy of @code{argv} strings.
14376 @item -mfixed-range=@var{register-range}
14377 @opindex mfixed-range
14378 Generate code treating the given register range as fixed registers.
14379 A fixed register is one that the register allocator can not use. This is
14380 useful when compiling kernel code. A register range is specified as
14381 two registers separated by a dash. Multiple register ranges can be
14382 specified separated by a comma.
14386 @node System V Options
14387 @subsection Options for System V
14389 These additional options are available on System V Release 4 for
14390 compatibility with other compilers on those systems:
14395 Create a shared object.
14396 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
14400 Identify the versions of each tool used by the compiler, in a
14401 @code{.ident} assembler directive in the output.
14405 Refrain from adding @code{.ident} directives to the output file (this is
14408 @item -YP,@var{dirs}
14410 Search the directories @var{dirs}, and no others, for libraries
14411 specified with @option{-l}.
14413 @item -Ym,@var{dir}
14415 Look in the directory @var{dir} to find the M4 preprocessor.
14416 The assembler uses this option.
14417 @c This is supposed to go with a -Yd for predefined M4 macro files, but
14418 @c the generic assembler that comes with Solaris takes just -Ym.
14422 @subsection V850 Options
14423 @cindex V850 Options
14425 These @samp{-m} options are defined for V850 implementations:
14429 @itemx -mno-long-calls
14430 @opindex mlong-calls
14431 @opindex mno-long-calls
14432 Treat all calls as being far away (near). If calls are assumed to be
14433 far away, the compiler will always load the functions address up into a
14434 register, and call indirect through the pointer.
14440 Do not optimize (do optimize) basic blocks that use the same index
14441 pointer 4 or more times to copy pointer into the @code{ep} register, and
14442 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14443 option is on by default if you optimize.
14445 @item -mno-prolog-function
14446 @itemx -mprolog-function
14447 @opindex mno-prolog-function
14448 @opindex mprolog-function
14449 Do not use (do use) external functions to save and restore registers
14450 at the prologue and epilogue of a function. The external functions
14451 are slower, but use less code space if more than one function saves
14452 the same number of registers. The @option{-mprolog-function} option
14453 is on by default if you optimize.
14457 Try to make the code as small as possible. At present, this just turns
14458 on the @option{-mep} and @option{-mprolog-function} options.
14460 @item -mtda=@var{n}
14462 Put static or global variables whose size is @var{n} bytes or less into
14463 the tiny data area that register @code{ep} points to. The tiny data
14464 area can hold up to 256 bytes in total (128 bytes for byte references).
14466 @item -msda=@var{n}
14468 Put static or global variables whose size is @var{n} bytes or less into
14469 the small data area that register @code{gp} points to. The small data
14470 area can hold up to 64 kilobytes.
14472 @item -mzda=@var{n}
14474 Put static or global variables whose size is @var{n} bytes or less into
14475 the first 32 kilobytes of memory.
14479 Specify that the target processor is the V850.
14482 @opindex mbig-switch
14483 Generate code suitable for big switch tables. Use this option only if
14484 the assembler/linker complain about out of range branches within a switch
14489 This option will cause r2 and r5 to be used in the code generated by
14490 the compiler. This setting is the default.
14492 @item -mno-app-regs
14493 @opindex mno-app-regs
14494 This option will cause r2 and r5 to be treated as fixed registers.
14498 Specify that the target processor is the V850E1. The preprocessor
14499 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14500 this option is used.
14504 Specify that the target processor is the V850E@. The preprocessor
14505 constant @samp{__v850e__} will be defined if this option is used.
14507 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14508 are defined then a default target processor will be chosen and the
14509 relevant @samp{__v850*__} preprocessor constant will be defined.
14511 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14512 defined, regardless of which processor variant is the target.
14514 @item -mdisable-callt
14515 @opindex mdisable-callt
14516 This option will suppress generation of the CALLT instruction for the
14517 v850e and v850e1 flavors of the v850 architecture. The default is
14518 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14523 @subsection VAX Options
14524 @cindex VAX options
14526 These @samp{-m} options are defined for the VAX:
14531 Do not output certain jump instructions (@code{aobleq} and so on)
14532 that the Unix assembler for the VAX cannot handle across long
14537 Do output those jump instructions, on the assumption that you
14538 will assemble with the GNU assembler.
14542 Output code for g-format floating point numbers instead of d-format.
14545 @node VxWorks Options
14546 @subsection VxWorks Options
14547 @cindex VxWorks Options
14549 The options in this section are defined for all VxWorks targets.
14550 Options specific to the target hardware are listed with the other
14551 options for that target.
14556 GCC can generate code for both VxWorks kernels and real time processes
14557 (RTPs). This option switches from the former to the latter. It also
14558 defines the preprocessor macro @code{__RTP__}.
14561 @opindex non-static
14562 Link an RTP executable against shared libraries rather than static
14563 libraries. The options @option{-static} and @option{-shared} can
14564 also be used for RTPs (@pxref{Link Options}); @option{-static}
14571 These options are passed down to the linker. They are defined for
14572 compatibility with Diab.
14575 @opindex Xbind-lazy
14576 Enable lazy binding of function calls. This option is equivalent to
14577 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14581 Disable lazy binding of function calls. This option is the default and
14582 is defined for compatibility with Diab.
14585 @node x86-64 Options
14586 @subsection x86-64 Options
14587 @cindex x86-64 options
14589 These are listed under @xref{i386 and x86-64 Options}.
14591 @node Xstormy16 Options
14592 @subsection Xstormy16 Options
14593 @cindex Xstormy16 Options
14595 These options are defined for Xstormy16:
14600 Choose startup files and linker script suitable for the simulator.
14603 @node Xtensa Options
14604 @subsection Xtensa Options
14605 @cindex Xtensa Options
14607 These options are supported for Xtensa targets:
14611 @itemx -mno-const16
14613 @opindex mno-const16
14614 Enable or disable use of @code{CONST16} instructions for loading
14615 constant values. The @code{CONST16} instruction is currently not a
14616 standard option from Tensilica. When enabled, @code{CONST16}
14617 instructions are always used in place of the standard @code{L32R}
14618 instructions. The use of @code{CONST16} is enabled by default only if
14619 the @code{L32R} instruction is not available.
14622 @itemx -mno-fused-madd
14623 @opindex mfused-madd
14624 @opindex mno-fused-madd
14625 Enable or disable use of fused multiply/add and multiply/subtract
14626 instructions in the floating-point option. This has no effect if the
14627 floating-point option is not also enabled. Disabling fused multiply/add
14628 and multiply/subtract instructions forces the compiler to use separate
14629 instructions for the multiply and add/subtract operations. This may be
14630 desirable in some cases where strict IEEE 754-compliant results are
14631 required: the fused multiply add/subtract instructions do not round the
14632 intermediate result, thereby producing results with @emph{more} bits of
14633 precision than specified by the IEEE standard. Disabling fused multiply
14634 add/subtract instructions also ensures that the program output is not
14635 sensitive to the compiler's ability to combine multiply and add/subtract
14638 @item -mtext-section-literals
14639 @itemx -mno-text-section-literals
14640 @opindex mtext-section-literals
14641 @opindex mno-text-section-literals
14642 Control the treatment of literal pools. The default is
14643 @option{-mno-text-section-literals}, which places literals in a separate
14644 section in the output file. This allows the literal pool to be placed
14645 in a data RAM/ROM, and it also allows the linker to combine literal
14646 pools from separate object files to remove redundant literals and
14647 improve code size. With @option{-mtext-section-literals}, the literals
14648 are interspersed in the text section in order to keep them as close as
14649 possible to their references. This may be necessary for large assembly
14652 @item -mtarget-align
14653 @itemx -mno-target-align
14654 @opindex mtarget-align
14655 @opindex mno-target-align
14656 When this option is enabled, GCC instructs the assembler to
14657 automatically align instructions to reduce branch penalties at the
14658 expense of some code density. The assembler attempts to widen density
14659 instructions to align branch targets and the instructions following call
14660 instructions. If there are not enough preceding safe density
14661 instructions to align a target, no widening will be performed. The
14662 default is @option{-mtarget-align}. These options do not affect the
14663 treatment of auto-aligned instructions like @code{LOOP}, which the
14664 assembler will always align, either by widening density instructions or
14665 by inserting no-op instructions.
14668 @itemx -mno-longcalls
14669 @opindex mlongcalls
14670 @opindex mno-longcalls
14671 When this option is enabled, GCC instructs the assembler to translate
14672 direct calls to indirect calls unless it can determine that the target
14673 of a direct call is in the range allowed by the call instruction. This
14674 translation typically occurs for calls to functions in other source
14675 files. Specifically, the assembler translates a direct @code{CALL}
14676 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14677 The default is @option{-mno-longcalls}. This option should be used in
14678 programs where the call target can potentially be out of range. This
14679 option is implemented in the assembler, not the compiler, so the
14680 assembly code generated by GCC will still show direct call
14681 instructions---look at the disassembled object code to see the actual
14682 instructions. Note that the assembler will use an indirect call for
14683 every cross-file call, not just those that really will be out of range.
14686 @node zSeries Options
14687 @subsection zSeries Options
14688 @cindex zSeries options
14690 These are listed under @xref{S/390 and zSeries Options}.
14692 @node Code Gen Options
14693 @section Options for Code Generation Conventions
14694 @cindex code generation conventions
14695 @cindex options, code generation
14696 @cindex run-time options
14698 These machine-independent options control the interface conventions
14699 used in code generation.
14701 Most of them have both positive and negative forms; the negative form
14702 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14703 one of the forms is listed---the one which is not the default. You
14704 can figure out the other form by either removing @samp{no-} or adding
14708 @item -fbounds-check
14709 @opindex fbounds-check
14710 For front-ends that support it, generate additional code to check that
14711 indices used to access arrays are within the declared range. This is
14712 currently only supported by the Java and Fortran front-ends, where
14713 this option defaults to true and false respectively.
14717 This option generates traps for signed overflow on addition, subtraction,
14718 multiplication operations.
14722 This option instructs the compiler to assume that signed arithmetic
14723 overflow of addition, subtraction and multiplication wraps around
14724 using twos-complement representation. This flag enables some optimizations
14725 and disables others. This option is enabled by default for the Java
14726 front-end, as required by the Java language specification.
14729 @opindex fexceptions
14730 Enable exception handling. Generates extra code needed to propagate
14731 exceptions. For some targets, this implies GCC will generate frame
14732 unwind information for all functions, which can produce significant data
14733 size overhead, although it does not affect execution. If you do not
14734 specify this option, GCC will enable it by default for languages like
14735 C++ which normally require exception handling, and disable it for
14736 languages like C that do not normally require it. However, you may need
14737 to enable this option when compiling C code that needs to interoperate
14738 properly with exception handlers written in C++. You may also wish to
14739 disable this option if you are compiling older C++ programs that don't
14740 use exception handling.
14742 @item -fnon-call-exceptions
14743 @opindex fnon-call-exceptions
14744 Generate code that allows trapping instructions to throw exceptions.
14745 Note that this requires platform-specific runtime support that does
14746 not exist everywhere. Moreover, it only allows @emph{trapping}
14747 instructions to throw exceptions, i.e.@: memory references or floating
14748 point instructions. It does not allow exceptions to be thrown from
14749 arbitrary signal handlers such as @code{SIGALRM}.
14751 @item -funwind-tables
14752 @opindex funwind-tables
14753 Similar to @option{-fexceptions}, except that it will just generate any needed
14754 static data, but will not affect the generated code in any other way.
14755 You will normally not enable this option; instead, a language processor
14756 that needs this handling would enable it on your behalf.
14758 @item -fasynchronous-unwind-tables
14759 @opindex fasynchronous-unwind-tables
14760 Generate unwind table in dwarf2 format, if supported by target machine. The
14761 table is exact at each instruction boundary, so it can be used for stack
14762 unwinding from asynchronous events (such as debugger or garbage collector).
14764 @item -fpcc-struct-return
14765 @opindex fpcc-struct-return
14766 Return ``short'' @code{struct} and @code{union} values in memory like
14767 longer ones, rather than in registers. This convention is less
14768 efficient, but it has the advantage of allowing intercallability between
14769 GCC-compiled files and files compiled with other compilers, particularly
14770 the Portable C Compiler (pcc).
14772 The precise convention for returning structures in memory depends
14773 on the target configuration macros.
14775 Short structures and unions are those whose size and alignment match
14776 that of some integer type.
14778 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14779 switch is not binary compatible with code compiled with the
14780 @option{-freg-struct-return} switch.
14781 Use it to conform to a non-default application binary interface.
14783 @item -freg-struct-return
14784 @opindex freg-struct-return
14785 Return @code{struct} and @code{union} values in registers when possible.
14786 This is more efficient for small structures than
14787 @option{-fpcc-struct-return}.
14789 If you specify neither @option{-fpcc-struct-return} nor
14790 @option{-freg-struct-return}, GCC defaults to whichever convention is
14791 standard for the target. If there is no standard convention, GCC
14792 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14793 the principal compiler. In those cases, we can choose the standard, and
14794 we chose the more efficient register return alternative.
14796 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14797 switch is not binary compatible with code compiled with the
14798 @option{-fpcc-struct-return} switch.
14799 Use it to conform to a non-default application binary interface.
14801 @item -fshort-enums
14802 @opindex fshort-enums
14803 Allocate to an @code{enum} type only as many bytes as it needs for the
14804 declared range of possible values. Specifically, the @code{enum} type
14805 will be equivalent to the smallest integer type which has enough room.
14807 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14808 code that is not binary compatible with code generated without that switch.
14809 Use it to conform to a non-default application binary interface.
14811 @item -fshort-double
14812 @opindex fshort-double
14813 Use the same size for @code{double} as for @code{float}.
14815 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14816 code that is not binary compatible with code generated without that switch.
14817 Use it to conform to a non-default application binary interface.
14819 @item -fshort-wchar
14820 @opindex fshort-wchar
14821 Override the underlying type for @samp{wchar_t} to be @samp{short
14822 unsigned int} instead of the default for the target. This option is
14823 useful for building programs to run under WINE@.
14825 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14826 code that is not binary compatible with code generated without that switch.
14827 Use it to conform to a non-default application binary interface.
14830 @opindex fno-common
14831 In C, allocate even uninitialized global variables in the data section of the
14832 object file, rather than generating them as common blocks. This has the
14833 effect that if the same variable is declared (without @code{extern}) in
14834 two different compilations, you will get an error when you link them.
14835 The only reason this might be useful is if you wish to verify that the
14836 program will work on other systems which always work this way.
14840 Ignore the @samp{#ident} directive.
14842 @item -finhibit-size-directive
14843 @opindex finhibit-size-directive
14844 Don't output a @code{.size} assembler directive, or anything else that
14845 would cause trouble if the function is split in the middle, and the
14846 two halves are placed at locations far apart in memory. This option is
14847 used when compiling @file{crtstuff.c}; you should not need to use it
14850 @item -fverbose-asm
14851 @opindex fverbose-asm
14852 Put extra commentary information in the generated assembly code to
14853 make it more readable. This option is generally only of use to those
14854 who actually need to read the generated assembly code (perhaps while
14855 debugging the compiler itself).
14857 @option{-fno-verbose-asm}, the default, causes the
14858 extra information to be omitted and is useful when comparing two assembler
14861 @item -frecord-gcc-switches
14862 @opindex frecord-gcc-switches
14863 This switch causes the command line that was used to invoke the
14864 compiler to be recorded into the object file that is being created.
14865 This switch is only implemented on some targets and the exact format
14866 of the recording is target and binary file format dependent, but it
14867 usually takes the form of a section containing ASCII text. This
14868 switch is related to the @option{-fverbose-asm} switch, but that
14869 switch only records information in the assembler output file as
14870 comments, so it never reaches the object file.
14874 @cindex global offset table
14876 Generate position-independent code (PIC) suitable for use in a shared
14877 library, if supported for the target machine. Such code accesses all
14878 constant addresses through a global offset table (GOT)@. The dynamic
14879 loader resolves the GOT entries when the program starts (the dynamic
14880 loader is not part of GCC; it is part of the operating system). If
14881 the GOT size for the linked executable exceeds a machine-specific
14882 maximum size, you get an error message from the linker indicating that
14883 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14884 instead. (These maximums are 8k on the SPARC and 32k
14885 on the m68k and RS/6000. The 386 has no such limit.)
14887 Position-independent code requires special support, and therefore works
14888 only on certain machines. For the 386, GCC supports PIC for System V
14889 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14890 position-independent.
14892 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14897 If supported for the target machine, emit position-independent code,
14898 suitable for dynamic linking and avoiding any limit on the size of the
14899 global offset table. This option makes a difference on the m68k,
14900 PowerPC and SPARC@.
14902 Position-independent code requires special support, and therefore works
14903 only on certain machines.
14905 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14912 These options are similar to @option{-fpic} and @option{-fPIC}, but
14913 generated position independent code can be only linked into executables.
14914 Usually these options are used when @option{-pie} GCC option will be
14915 used during linking.
14917 @option{-fpie} and @option{-fPIE} both define the macros
14918 @code{__pie__} and @code{__PIE__}. The macros have the value 1
14919 for @option{-fpie} and 2 for @option{-fPIE}.
14921 @item -fno-jump-tables
14922 @opindex fno-jump-tables
14923 Do not use jump tables for switch statements even where it would be
14924 more efficient than other code generation strategies. This option is
14925 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14926 building code which forms part of a dynamic linker and cannot
14927 reference the address of a jump table. On some targets, jump tables
14928 do not require a GOT and this option is not needed.
14930 @item -ffixed-@var{reg}
14932 Treat the register named @var{reg} as a fixed register; generated code
14933 should never refer to it (except perhaps as a stack pointer, frame
14934 pointer or in some other fixed role).
14936 @var{reg} must be the name of a register. The register names accepted
14937 are machine-specific and are defined in the @code{REGISTER_NAMES}
14938 macro in the machine description macro file.
14940 This flag does not have a negative form, because it specifies a
14943 @item -fcall-used-@var{reg}
14944 @opindex fcall-used
14945 Treat the register named @var{reg} as an allocable register that is
14946 clobbered by function calls. It may be allocated for temporaries or
14947 variables that do not live across a call. Functions compiled this way
14948 will not save and restore the register @var{reg}.
14950 It is an error to used this flag with the frame pointer or stack pointer.
14951 Use of this flag for other registers that have fixed pervasive roles in
14952 the machine's execution model will produce disastrous results.
14954 This flag does not have a negative form, because it specifies a
14957 @item -fcall-saved-@var{reg}
14958 @opindex fcall-saved
14959 Treat the register named @var{reg} as an allocable register saved by
14960 functions. It may be allocated even for temporaries or variables that
14961 live across a call. Functions compiled this way will save and restore
14962 the register @var{reg} if they use it.
14964 It is an error to used this flag with the frame pointer or stack pointer.
14965 Use of this flag for other registers that have fixed pervasive roles in
14966 the machine's execution model will produce disastrous results.
14968 A different sort of disaster will result from the use of this flag for
14969 a register in which function values may be returned.
14971 This flag does not have a negative form, because it specifies a
14974 @item -fpack-struct[=@var{n}]
14975 @opindex fpack-struct
14976 Without a value specified, pack all structure members together without
14977 holes. When a value is specified (which must be a small power of two), pack
14978 structure members according to this value, representing the maximum
14979 alignment (that is, objects with default alignment requirements larger than
14980 this will be output potentially unaligned at the next fitting location.
14982 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14983 code that is not binary compatible with code generated without that switch.
14984 Additionally, it makes the code suboptimal.
14985 Use it to conform to a non-default application binary interface.
14987 @item -finstrument-functions
14988 @opindex finstrument-functions
14989 Generate instrumentation calls for entry and exit to functions. Just
14990 after function entry and just before function exit, the following
14991 profiling functions will be called with the address of the current
14992 function and its call site. (On some platforms,
14993 @code{__builtin_return_address} does not work beyond the current
14994 function, so the call site information may not be available to the
14995 profiling functions otherwise.)
14998 void __cyg_profile_func_enter (void *this_fn,
15000 void __cyg_profile_func_exit (void *this_fn,
15004 The first argument is the address of the start of the current function,
15005 which may be looked up exactly in the symbol table.
15007 This instrumentation is also done for functions expanded inline in other
15008 functions. The profiling calls will indicate where, conceptually, the
15009 inline function is entered and exited. This means that addressable
15010 versions of such functions must be available. If all your uses of a
15011 function are expanded inline, this may mean an additional expansion of
15012 code size. If you use @samp{extern inline} in your C code, an
15013 addressable version of such functions must be provided. (This is
15014 normally the case anyways, but if you get lucky and the optimizer always
15015 expands the functions inline, you might have gotten away without
15016 providing static copies.)
15018 A function may be given the attribute @code{no_instrument_function}, in
15019 which case this instrumentation will not be done. This can be used, for
15020 example, for the profiling functions listed above, high-priority
15021 interrupt routines, and any functions from which the profiling functions
15022 cannot safely be called (perhaps signal handlers, if the profiling
15023 routines generate output or allocate memory).
15025 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
15026 @opindex finstrument-functions-exclude-file-list
15028 Set the list of functions that are excluded from instrumentation (see
15029 the description of @code{-finstrument-functions}). If the file that
15030 contains a function definition matches with one of @var{file}, then
15031 that function is not instrumented. The match is done on substrings:
15032 if the @var{file} parameter is a substring of the file name, it is
15033 considered to be a match.
15036 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
15037 will exclude any inline function defined in files whose pathnames
15038 contain @code{/bits/stl} or @code{include/sys}.
15040 If, for some reason, you want to include letter @code{','} in one of
15041 @var{sym}, write @code{'\,'}. For example,
15042 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
15043 (note the single quote surrounding the option).
15045 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
15046 @opindex finstrument-functions-exclude-function-list
15048 This is similar to @code{-finstrument-functions-exclude-file-list},
15049 but this option sets the list of function names to be excluded from
15050 instrumentation. The function name to be matched is its user-visible
15051 name, such as @code{vector<int> blah(const vector<int> &)}, not the
15052 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
15053 match is done on substrings: if the @var{sym} parameter is a substring
15054 of the function name, it is considered to be a match.
15056 @item -fstack-check
15057 @opindex fstack-check
15058 Generate code to verify that you do not go beyond the boundary of the
15059 stack. You should specify this flag if you are running in an
15060 environment with multiple threads, but only rarely need to specify it in
15061 a single-threaded environment since stack overflow is automatically
15062 detected on nearly all systems if there is only one stack.
15064 Note that this switch does not actually cause checking to be done; the
15065 operating system must do that. The switch causes generation of code
15066 to ensure that the operating system sees the stack being extended.
15068 @item -fstack-limit-register=@var{reg}
15069 @itemx -fstack-limit-symbol=@var{sym}
15070 @itemx -fno-stack-limit
15071 @opindex fstack-limit-register
15072 @opindex fstack-limit-symbol
15073 @opindex fno-stack-limit
15074 Generate code to ensure that the stack does not grow beyond a certain value,
15075 either the value of a register or the address of a symbol. If the stack
15076 would grow beyond the value, a signal is raised. For most targets,
15077 the signal is raised before the stack overruns the boundary, so
15078 it is possible to catch the signal without taking special precautions.
15080 For instance, if the stack starts at absolute address @samp{0x80000000}
15081 and grows downwards, you can use the flags
15082 @option{-fstack-limit-symbol=__stack_limit} and
15083 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
15084 of 128KB@. Note that this may only work with the GNU linker.
15086 @cindex aliasing of parameters
15087 @cindex parameters, aliased
15088 @item -fargument-alias
15089 @itemx -fargument-noalias
15090 @itemx -fargument-noalias-global
15091 @itemx -fargument-noalias-anything
15092 @opindex fargument-alias
15093 @opindex fargument-noalias
15094 @opindex fargument-noalias-global
15095 @opindex fargument-noalias-anything
15096 Specify the possible relationships among parameters and between
15097 parameters and global data.
15099 @option{-fargument-alias} specifies that arguments (parameters) may
15100 alias each other and may alias global storage.@*
15101 @option{-fargument-noalias} specifies that arguments do not alias
15102 each other, but may alias global storage.@*
15103 @option{-fargument-noalias-global} specifies that arguments do not
15104 alias each other and do not alias global storage.
15105 @option{-fargument-noalias-anything} specifies that arguments do not
15106 alias any other storage.
15108 Each language will automatically use whatever option is required by
15109 the language standard. You should not need to use these options yourself.
15111 @item -fleading-underscore
15112 @opindex fleading-underscore
15113 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
15114 change the way C symbols are represented in the object file. One use
15115 is to help link with legacy assembly code.
15117 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
15118 generate code that is not binary compatible with code generated without that
15119 switch. Use it to conform to a non-default application binary interface.
15120 Not all targets provide complete support for this switch.
15122 @item -ftls-model=@var{model}
15123 @opindex ftls-model
15124 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
15125 The @var{model} argument should be one of @code{global-dynamic},
15126 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
15128 The default without @option{-fpic} is @code{initial-exec}; with
15129 @option{-fpic} the default is @code{global-dynamic}.
15131 @item -fvisibility=@var{default|internal|hidden|protected}
15132 @opindex fvisibility
15133 Set the default ELF image symbol visibility to the specified option---all
15134 symbols will be marked with this unless overridden within the code.
15135 Using this feature can very substantially improve linking and
15136 load times of shared object libraries, produce more optimized
15137 code, provide near-perfect API export and prevent symbol clashes.
15138 It is @strong{strongly} recommended that you use this in any shared objects
15141 Despite the nomenclature, @code{default} always means public ie;
15142 available to be linked against from outside the shared object.
15143 @code{protected} and @code{internal} are pretty useless in real-world
15144 usage so the only other commonly used option will be @code{hidden}.
15145 The default if @option{-fvisibility} isn't specified is
15146 @code{default}, i.e., make every
15147 symbol public---this causes the same behavior as previous versions of
15150 A good explanation of the benefits offered by ensuring ELF
15151 symbols have the correct visibility is given by ``How To Write
15152 Shared Libraries'' by Ulrich Drepper (which can be found at
15153 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
15154 solution made possible by this option to marking things hidden when
15155 the default is public is to make the default hidden and mark things
15156 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
15157 and @code{__attribute__ ((visibility("default")))} instead of
15158 @code{__declspec(dllexport)} you get almost identical semantics with
15159 identical syntax. This is a great boon to those working with
15160 cross-platform projects.
15162 For those adding visibility support to existing code, you may find
15163 @samp{#pragma GCC visibility} of use. This works by you enclosing
15164 the declarations you wish to set visibility for with (for example)
15165 @samp{#pragma GCC visibility push(hidden)} and
15166 @samp{#pragma GCC visibility pop}.
15167 Bear in mind that symbol visibility should be viewed @strong{as
15168 part of the API interface contract} and thus all new code should
15169 always specify visibility when it is not the default ie; declarations
15170 only for use within the local DSO should @strong{always} be marked explicitly
15171 as hidden as so to avoid PLT indirection overheads---making this
15172 abundantly clear also aids readability and self-documentation of the code.
15173 Note that due to ISO C++ specification requirements, operator new and
15174 operator delete must always be of default visibility.
15176 Be aware that headers from outside your project, in particular system
15177 headers and headers from any other library you use, may not be
15178 expecting to be compiled with visibility other than the default. You
15179 may need to explicitly say @samp{#pragma GCC visibility push(default)}
15180 before including any such headers.
15182 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
15183 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
15184 no modifications. However, this means that calls to @samp{extern}
15185 functions with no explicit visibility will use the PLT, so it is more
15186 effective to use @samp{__attribute ((visibility))} and/or
15187 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
15188 declarations should be treated as hidden.
15190 Note that @samp{-fvisibility} does affect C++ vague linkage
15191 entities. This means that, for instance, an exception class that will
15192 be thrown between DSOs must be explicitly marked with default
15193 visibility so that the @samp{type_info} nodes will be unified between
15196 An overview of these techniques, their benefits and how to use them
15197 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
15203 @node Environment Variables
15204 @section Environment Variables Affecting GCC
15205 @cindex environment variables
15207 @c man begin ENVIRONMENT
15208 This section describes several environment variables that affect how GCC
15209 operates. Some of them work by specifying directories or prefixes to use
15210 when searching for various kinds of files. Some are used to specify other
15211 aspects of the compilation environment.
15213 Note that you can also specify places to search using options such as
15214 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
15215 take precedence over places specified using environment variables, which
15216 in turn take precedence over those specified by the configuration of GCC@.
15217 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
15218 GNU Compiler Collection (GCC) Internals}.
15223 @c @itemx LC_COLLATE
15225 @c @itemx LC_MONETARY
15226 @c @itemx LC_NUMERIC
15231 @c @findex LC_COLLATE
15232 @findex LC_MESSAGES
15233 @c @findex LC_MONETARY
15234 @c @findex LC_NUMERIC
15238 These environment variables control the way that GCC uses
15239 localization information that allow GCC to work with different
15240 national conventions. GCC inspects the locale categories
15241 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
15242 so. These locale categories can be set to any value supported by your
15243 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
15244 Kingdom encoded in UTF-8.
15246 The @env{LC_CTYPE} environment variable specifies character
15247 classification. GCC uses it to determine the character boundaries in
15248 a string; this is needed for some multibyte encodings that contain quote
15249 and escape characters that would otherwise be interpreted as a string
15252 The @env{LC_MESSAGES} environment variable specifies the language to
15253 use in diagnostic messages.
15255 If the @env{LC_ALL} environment variable is set, it overrides the value
15256 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
15257 and @env{LC_MESSAGES} default to the value of the @env{LANG}
15258 environment variable. If none of these variables are set, GCC
15259 defaults to traditional C English behavior.
15263 If @env{TMPDIR} is set, it specifies the directory to use for temporary
15264 files. GCC uses temporary files to hold the output of one stage of
15265 compilation which is to be used as input to the next stage: for example,
15266 the output of the preprocessor, which is the input to the compiler
15269 @item GCC_EXEC_PREFIX
15270 @findex GCC_EXEC_PREFIX
15271 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
15272 names of the subprograms executed by the compiler. No slash is added
15273 when this prefix is combined with the name of a subprogram, but you can
15274 specify a prefix that ends with a slash if you wish.
15276 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
15277 an appropriate prefix to use based on the pathname it was invoked with.
15279 If GCC cannot find the subprogram using the specified prefix, it
15280 tries looking in the usual places for the subprogram.
15282 The default value of @env{GCC_EXEC_PREFIX} is
15283 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
15284 the installed compiler. In many cases @var{prefix} is the value
15285 of @code{prefix} when you ran the @file{configure} script.
15287 Other prefixes specified with @option{-B} take precedence over this prefix.
15289 This prefix is also used for finding files such as @file{crt0.o} that are
15292 In addition, the prefix is used in an unusual way in finding the
15293 directories to search for header files. For each of the standard
15294 directories whose name normally begins with @samp{/usr/local/lib/gcc}
15295 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
15296 replacing that beginning with the specified prefix to produce an
15297 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
15298 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
15299 These alternate directories are searched first; the standard directories
15300 come next. If a standard directory begins with the configured
15301 @var{prefix} then the value of @var{prefix} is replaced by
15302 @env{GCC_EXEC_PREFIX} when looking for header files.
15304 @item COMPILER_PATH
15305 @findex COMPILER_PATH
15306 The value of @env{COMPILER_PATH} is a colon-separated list of
15307 directories, much like @env{PATH}. GCC tries the directories thus
15308 specified when searching for subprograms, if it can't find the
15309 subprograms using @env{GCC_EXEC_PREFIX}.
15312 @findex LIBRARY_PATH
15313 The value of @env{LIBRARY_PATH} is a colon-separated list of
15314 directories, much like @env{PATH}. When configured as a native compiler,
15315 GCC tries the directories thus specified when searching for special
15316 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15317 using GCC also uses these directories when searching for ordinary
15318 libraries for the @option{-l} option (but directories specified with
15319 @option{-L} come first).
15323 @cindex locale definition
15324 This variable is used to pass locale information to the compiler. One way in
15325 which this information is used is to determine the character set to be used
15326 when character literals, string literals and comments are parsed in C and C++.
15327 When the compiler is configured to allow multibyte characters,
15328 the following values for @env{LANG} are recognized:
15332 Recognize JIS characters.
15334 Recognize SJIS characters.
15336 Recognize EUCJP characters.
15339 If @env{LANG} is not defined, or if it has some other value, then the
15340 compiler will use mblen and mbtowc as defined by the default locale to
15341 recognize and translate multibyte characters.
15345 Some additional environments variables affect the behavior of the
15348 @include cppenv.texi
15352 @node Precompiled Headers
15353 @section Using Precompiled Headers
15354 @cindex precompiled headers
15355 @cindex speed of compilation
15357 Often large projects have many header files that are included in every
15358 source file. The time the compiler takes to process these header files
15359 over and over again can account for nearly all of the time required to
15360 build the project. To make builds faster, GCC allows users to
15361 `precompile' a header file; then, if builds can use the precompiled
15362 header file they will be much faster.
15364 To create a precompiled header file, simply compile it as you would any
15365 other file, if necessary using the @option{-x} option to make the driver
15366 treat it as a C or C++ header file. You will probably want to use a
15367 tool like @command{make} to keep the precompiled header up-to-date when
15368 the headers it contains change.
15370 A precompiled header file will be searched for when @code{#include} is
15371 seen in the compilation. As it searches for the included file
15372 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15373 compiler looks for a precompiled header in each directory just before it
15374 looks for the include file in that directory. The name searched for is
15375 the name specified in the @code{#include} with @samp{.gch} appended. If
15376 the precompiled header file can't be used, it is ignored.
15378 For instance, if you have @code{#include "all.h"}, and you have
15379 @file{all.h.gch} in the same directory as @file{all.h}, then the
15380 precompiled header file will be used if possible, and the original
15381 header will be used otherwise.
15383 Alternatively, you might decide to put the precompiled header file in a
15384 directory and use @option{-I} to ensure that directory is searched
15385 before (or instead of) the directory containing the original header.
15386 Then, if you want to check that the precompiled header file is always
15387 used, you can put a file of the same name as the original header in this
15388 directory containing an @code{#error} command.
15390 This also works with @option{-include}. So yet another way to use
15391 precompiled headers, good for projects not designed with precompiled
15392 header files in mind, is to simply take most of the header files used by
15393 a project, include them from another header file, precompile that header
15394 file, and @option{-include} the precompiled header. If the header files
15395 have guards against multiple inclusion, they will be skipped because
15396 they've already been included (in the precompiled header).
15398 If you need to precompile the same header file for different
15399 languages, targets, or compiler options, you can instead make a
15400 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15401 header in the directory, perhaps using @option{-o}. It doesn't matter
15402 what you call the files in the directory, every precompiled header in
15403 the directory will be considered. The first precompiled header
15404 encountered in the directory that is valid for this compilation will
15405 be used; they're searched in no particular order.
15407 There are many other possibilities, limited only by your imagination,
15408 good sense, and the constraints of your build system.
15410 A precompiled header file can be used only when these conditions apply:
15414 Only one precompiled header can be used in a particular compilation.
15417 A precompiled header can't be used once the first C token is seen. You
15418 can have preprocessor directives before a precompiled header; you can
15419 even include a precompiled header from inside another header, so long as
15420 there are no C tokens before the @code{#include}.
15423 The precompiled header file must be produced for the same language as
15424 the current compilation. You can't use a C precompiled header for a C++
15428 The precompiled header file must have been produced by the same compiler
15429 binary as the current compilation is using.
15432 Any macros defined before the precompiled header is included must
15433 either be defined in the same way as when the precompiled header was
15434 generated, or must not affect the precompiled header, which usually
15435 means that they don't appear in the precompiled header at all.
15437 The @option{-D} option is one way to define a macro before a
15438 precompiled header is included; using a @code{#define} can also do it.
15439 There are also some options that define macros implicitly, like
15440 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15443 @item If debugging information is output when using the precompiled
15444 header, using @option{-g} or similar, the same kind of debugging information
15445 must have been output when building the precompiled header. However,
15446 a precompiled header built using @option{-g} can be used in a compilation
15447 when no debugging information is being output.
15449 @item The same @option{-m} options must generally be used when building
15450 and using the precompiled header. @xref{Submodel Options},
15451 for any cases where this rule is relaxed.
15453 @item Each of the following options must be the same when building and using
15454 the precompiled header:
15456 @gccoptlist{-fexceptions -funit-at-a-time}
15459 Some other command-line options starting with @option{-f},
15460 @option{-p}, or @option{-O} must be defined in the same way as when
15461 the precompiled header was generated. At present, it's not clear
15462 which options are safe to change and which are not; the safest choice
15463 is to use exactly the same options when generating and using the
15464 precompiled header. The following are known to be safe:
15466 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15467 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15468 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15473 For all of these except the last, the compiler will automatically
15474 ignore the precompiled header if the conditions aren't met. If you
15475 find an option combination that doesn't work and doesn't cause the
15476 precompiled header to be ignored, please consider filing a bug report,
15479 If you do use differing options when generating and using the
15480 precompiled header, the actual behavior will be a mixture of the
15481 behavior for the options. For instance, if you use @option{-g} to
15482 generate the precompiled header but not when using it, you may or may
15483 not get debugging information for routines in the precompiled header.
15485 @node Running Protoize
15486 @section Running Protoize
15488 The program @code{protoize} is an optional part of GCC@. You can use
15489 it to add prototypes to a program, thus converting the program to ISO
15490 C in one respect. The companion program @code{unprotoize} does the
15491 reverse: it removes argument types from any prototypes that are found.
15493 When you run these programs, you must specify a set of source files as
15494 command line arguments. The conversion programs start out by compiling
15495 these files to see what functions they define. The information gathered
15496 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15498 After scanning comes actual conversion. The specified files are all
15499 eligible to be converted; any files they include (whether sources or
15500 just headers) are eligible as well.
15502 But not all the eligible files are converted. By default,
15503 @code{protoize} and @code{unprotoize} convert only source and header
15504 files in the current directory. You can specify additional directories
15505 whose files should be converted with the @option{-d @var{directory}}
15506 option. You can also specify particular files to exclude with the
15507 @option{-x @var{file}} option. A file is converted if it is eligible, its
15508 directory name matches one of the specified directory names, and its
15509 name within the directory has not been excluded.
15511 Basic conversion with @code{protoize} consists of rewriting most
15512 function definitions and function declarations to specify the types of
15513 the arguments. The only ones not rewritten are those for varargs
15516 @code{protoize} optionally inserts prototype declarations at the
15517 beginning of the source file, to make them available for any calls that
15518 precede the function's definition. Or it can insert prototype
15519 declarations with block scope in the blocks where undeclared functions
15522 Basic conversion with @code{unprotoize} consists of rewriting most
15523 function declarations to remove any argument types, and rewriting
15524 function definitions to the old-style pre-ISO form.
15526 Both conversion programs print a warning for any function declaration or
15527 definition that they can't convert. You can suppress these warnings
15530 The output from @code{protoize} or @code{unprotoize} replaces the
15531 original source file. The original file is renamed to a name ending
15532 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15533 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15534 for DOS) file already exists, then the source file is simply discarded.
15536 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15537 scan the program and collect information about the functions it uses.
15538 So neither of these programs will work until GCC is installed.
15540 Here is a table of the options you can use with @code{protoize} and
15541 @code{unprotoize}. Each option works with both programs unless
15545 @item -B @var{directory}
15546 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15547 usual directory (normally @file{/usr/local/lib}). This file contains
15548 prototype information about standard system functions. This option
15549 applies only to @code{protoize}.
15551 @item -c @var{compilation-options}
15552 Use @var{compilation-options} as the options when running @command{gcc} to
15553 produce the @samp{.X} files. The special option @option{-aux-info} is
15554 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15556 Note that the compilation options must be given as a single argument to
15557 @code{protoize} or @code{unprotoize}. If you want to specify several
15558 @command{gcc} options, you must quote the entire set of compilation options
15559 to make them a single word in the shell.
15561 There are certain @command{gcc} arguments that you cannot use, because they
15562 would produce the wrong kind of output. These include @option{-g},
15563 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15564 the @var{compilation-options}, they are ignored.
15567 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15568 systems) instead of @samp{.c}. This is convenient if you are converting
15569 a C program to C++. This option applies only to @code{protoize}.
15572 Add explicit global declarations. This means inserting explicit
15573 declarations at the beginning of each source file for each function
15574 that is called in the file and was not declared. These declarations
15575 precede the first function definition that contains a call to an
15576 undeclared function. This option applies only to @code{protoize}.
15578 @item -i @var{string}
15579 Indent old-style parameter declarations with the string @var{string}.
15580 This option applies only to @code{protoize}.
15582 @code{unprotoize} converts prototyped function definitions to old-style
15583 function definitions, where the arguments are declared between the
15584 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15585 uses five spaces as the indentation. If you want to indent with just
15586 one space instead, use @option{-i " "}.
15589 Keep the @samp{.X} files. Normally, they are deleted after conversion
15593 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15594 a prototype declaration for each function in each block which calls the
15595 function without any declaration. This option applies only to
15599 Make no real changes. This mode just prints information about the conversions
15600 that would have been done without @option{-n}.
15603 Make no @samp{.save} files. The original files are simply deleted.
15604 Use this option with caution.
15606 @item -p @var{program}
15607 Use the program @var{program} as the compiler. Normally, the name
15608 @file{gcc} is used.
15611 Work quietly. Most warnings are suppressed.
15614 Print the version number, just like @option{-v} for @command{gcc}.
15617 If you need special compiler options to compile one of your program's
15618 source files, then you should generate that file's @samp{.X} file
15619 specially, by running @command{gcc} on that source file with the
15620 appropriate options and the option @option{-aux-info}. Then run
15621 @code{protoize} on the entire set of files. @code{protoize} will use
15622 the existing @samp{.X} file because it is newer than the source file.
15626 gcc -Dfoo=bar file1.c -aux-info file1.X
15631 You need to include the special files along with the rest in the
15632 @code{protoize} command, even though their @samp{.X} files already
15633 exist, because otherwise they won't get converted.
15635 @xref{Protoize Caveats}, for more information on how to use
15636 @code{protoize} successfully.