1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
15 2007 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
151 * Running Protoize:: Automatically adding or removing function prototypes.
157 @section Option Summary
159 Here is a summary of all the options, grouped by type. Explanations are
160 in the following sections.
163 @item Overall Options
164 @xref{Overall Options,,Options Controlling the Kind of Output}.
165 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{]} --target-help @gol
167 --version @@@var{file}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
193 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
194 -fno-default-inline -fvisibility-inlines-hidden @gol
195 -fvisibility-ms-compat @gol
196 -Wabi -Wctor-dtor-privacy @gol
197 -Wnon-virtual-dtor -Wreorder @gol
198 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
199 -Wno-non-template-friend -Wold-style-cast @gol
200 -Woverloaded-virtual -Wno-pmf-conversions @gol
203 @item Objective-C and Objective-C++ Language Options
204 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
205 Objective-C and Objective-C++ Dialects}.
206 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
207 -fgnu-runtime -fnext-runtime @gol
208 -fno-nil-receivers @gol
209 -fobjc-call-cxx-cdtors @gol
210 -fobjc-direct-dispatch @gol
211 -fobjc-exceptions @gol
213 -freplace-objc-classes @gol
216 -Wassign-intercept @gol
217 -Wno-protocol -Wselector @gol
218 -Wstrict-selector-match @gol
219 -Wundeclared-selector}
221 @item Language Independent Options
222 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
223 @gccoptlist{-fmessage-length=@var{n} @gol
224 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
225 -fdiagnostics-show-option}
227 @item Warning Options
228 @xref{Warning Options,,Options to Request or Suppress Warnings}.
229 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
230 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
231 -Wno-attributes -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
232 -Wchar-subscripts -Wclobbered -Wcomment @gol
233 -Wconversion -Wcoverage-mismatch -Wno-deprecated-declarations @gol
234 -Wdisabled-optimization -Wno-div-by-zero @gol
235 -Wempty-body -Wno-endif-labels @gol
236 -Werror -Werror=* @gol
237 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
238 -Wno-format-extra-args -Wformat-nonliteral @gol
239 -Wformat-security -Wformat-y2k @gol
240 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
241 -Wimport -Wno-import -Winit-self -Winline @gol
242 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
243 -Winvalid-pch -Wlarger-than-@var{len} -Wunsafe-loop-optimizations @gol
244 -Wlogical-op -Wlong-long @gol
245 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
246 -Wmissing-format-attribute -Wmissing-include-dirs @gol
247 -Wmissing-noreturn @gol
248 -Wno-multichar -Wnonnull -Wno-overflow @gol
249 -Woverlength-strings -Wpacked -Wpadded @gol
250 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
251 -Wredundant-decls @gol
252 -Wreturn-type -Wsequence-point -Wshadow @gol
253 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
254 -Wstrict-aliasing -Wstrict-aliasing=n @gol
255 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
256 -Wswitch -Wswitch-default -Wswitch-enum @gol
257 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
258 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
259 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
260 -Wunused-value -Wunused-variable @gol
261 -Wvariadic-macros -Wvla @gol
262 -Wvolatile-register-var -Wwrite-strings}
264 @item C-only Warning Options
265 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
266 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
267 -Wold-style-declaration -Wold-style-definition @gol
268 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
269 -Wdeclaration-after-statement -Wpointer-sign}
271 @item Debugging Options
272 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
273 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
274 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
275 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
276 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
277 -fdump-ipa-all -fdump-ipa-cgraph @gol
279 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
284 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
289 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-nrv -fdump-tree-vect @gol
293 -fdump-tree-sink @gol
294 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-salias @gol
296 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
298 -ftree-vectorizer-verbose=@var{n} @gol
299 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
300 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
301 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
302 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
303 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
304 -ftest-coverage -ftime-report -fvar-tracking @gol
305 -g -g@var{level} -gcoff -gdwarf-2 @gol
306 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
307 -fdebug-prefix-map=@var{old}=@var{new} @gol
308 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
309 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
310 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
311 -print-multi-directory -print-multi-lib @gol
312 -print-prog-name=@var{program} -print-search-dirs -Q @gol
313 -print-sysroot-headers-suffix @gol
316 @item Optimization Options
317 @xref{Optimize Options,,Options that Control Optimization}.
318 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
319 -falign-labels=@var{n} -falign-loops=@var{n} @gol
320 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
321 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
322 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
323 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
324 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
325 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
326 -fexpensive-optimizations -ffast-math -ffloat-store @gol
327 -fforce-addr -fforward-propagate -ffunction-sections @gol
328 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
329 -fcrossjumping -fif-conversion -fif-conversion2 @gol
330 -finline-functions -finline-functions-called-once @gol
331 -finline-limit=@var{n} -fkeep-inline-functions @gol
332 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
333 -fmodulo-sched -fmodulo-sched-allow-regmoves -fno-branch-count-reg @gol
334 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
335 -fno-function-cse -fno-guess-branch-probability @gol
336 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
337 -funsafe-math-optimizations -funsafe-loop-optimizations @gol
338 -ffinite-math-only -fno-signed-zeros @gol
339 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
340 -fomit-frame-pointer -foptimize-register-move @gol
341 -foptimize-sibling-calls -fpredictive-commoning -fprefetch-loop-arrays @gol
342 -fprofile-generate -fprofile-use @gol
343 -fregmove -frename-registers @gol
344 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
345 -frerun-cse-after-loop @gol
346 -frounding-math -frtl-abstract-sequences @gol
347 -fschedule-insns -fschedule-insns2 @gol
348 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
349 -fsched-spec-load-dangerous @gol
350 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
351 -fsched2-use-superblocks @gol
352 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
353 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
354 -fno-split-wide-types -fstack-protector -fstack-protector-all @gol
355 -fstrict-aliasing -fstrict-overflow -ftracer -fthread-jumps @gol
356 -funroll-all-loops -funroll-loops -fpeel-loops @gol
357 -fsplit-ivs-in-unroller -funswitch-loops @gol
358 -fvariable-expansion-in-unroller -ftree-reassoc @gol
359 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
360 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
361 -fcheck-data-deps @gol
362 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
363 -ftree-ch -ftree-sra -ftree-ter -ftree-fre -ftree-vectorize @gol
364 -ftree-vect-loop-version -fvect-cost-model -ftree-salias -fipa-pta -fweb @gol
365 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
366 --param @var{name}=@var{value}
367 -O -O0 -O1 -O2 -O3 -Os}
369 @item Preprocessor Options
370 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
371 @gccoptlist{-A@var{question}=@var{answer} @gol
372 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
373 -C -dD -dI -dM -dN @gol
374 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
375 -idirafter @var{dir} @gol
376 -include @var{file} -imacros @var{file} @gol
377 -iprefix @var{file} -iwithprefix @var{dir} @gol
378 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
379 -imultilib @var{dir} -isysroot @var{dir} @gol
380 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
381 -P -fworking-directory -remap @gol
382 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
383 -Xpreprocessor @var{option}}
385 @item Assembler Option
386 @xref{Assembler Options,,Passing Options to the Assembler}.
387 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
390 @xref{Link Options,,Options for Linking}.
391 @gccoptlist{@var{object-file-name} -l@var{library} @gol
392 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
393 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
394 -Wl,@var{option} -Xlinker @var{option} @gol
397 @item Directory Options
398 @xref{Directory Options,,Options for Directory Search}.
399 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
400 -specs=@var{file} -I- --sysroot=@var{dir}}
403 @c I wrote this xref this way to avoid overfull hbox. -- rms
404 @xref{Target Options}.
405 @gccoptlist{-V @var{version} -b @var{machine}}
407 @item Machine Dependent Options
408 @xref{Submodel Options,,Hardware Models and Configurations}.
409 @c This list is ordered alphanumerically by subsection name.
410 @c Try and put the significant identifier (CPU or system) first,
411 @c so users have a clue at guessing where the ones they want will be.
414 @gccoptlist{-EB -EL @gol
415 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
416 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
419 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
420 -mabi=@var{name} @gol
421 -mapcs-stack-check -mno-apcs-stack-check @gol
422 -mapcs-float -mno-apcs-float @gol
423 -mapcs-reentrant -mno-apcs-reentrant @gol
424 -msched-prolog -mno-sched-prolog @gol
425 -mlittle-endian -mbig-endian -mwords-little-endian @gol
426 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
427 -mthumb-interwork -mno-thumb-interwork @gol
428 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
429 -mstructure-size-boundary=@var{n} @gol
430 -mabort-on-noreturn @gol
431 -mlong-calls -mno-long-calls @gol
432 -msingle-pic-base -mno-single-pic-base @gol
433 -mpic-register=@var{reg} @gol
434 -mnop-fun-dllimport @gol
435 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
436 -mpoke-function-name @gol
438 -mtpcs-frame -mtpcs-leaf-frame @gol
439 -mcaller-super-interworking -mcallee-super-interworking @gol
443 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
444 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
446 @emph{Blackfin Options}
447 @gccoptlist{-mcpu=@var{cpu} -msim -momit-leaf-frame-pointer @gol
448 -mno-omit-leaf-frame-pointer -mspecld-anomaly -mno-specld-anomaly @gol
449 -mcsync-anomaly -mno-csync-anomaly -mlow-64k -mno-low64k @gol
450 -mstack-check-l1 -mid-shared-library -mno-id-shared-library @gol
451 -mshared-library-id=@var{n} -mleaf-id-shared-library @gol
452 -mno-leaf-id-shared-library -msep-data -mno-sep-data -mlong-calls @gol
456 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
457 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
458 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
459 -mstack-align -mdata-align -mconst-align @gol
460 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
461 -melf -maout -melinux -mlinux -sim -sim2 @gol
462 -mmul-bug-workaround -mno-mul-bug-workaround}
465 @gccoptlist{-mmac -mpush-args}
467 @emph{Darwin Options}
468 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
469 -arch_only -bind_at_load -bundle -bundle_loader @gol
470 -client_name -compatibility_version -current_version @gol
472 -dependency-file -dylib_file -dylinker_install_name @gol
473 -dynamic -dynamiclib -exported_symbols_list @gol
474 -filelist -flat_namespace -force_cpusubtype_ALL @gol
475 -force_flat_namespace -headerpad_max_install_names @gol
477 -image_base -init -install_name -keep_private_externs @gol
478 -multi_module -multiply_defined -multiply_defined_unused @gol
479 -noall_load -no_dead_strip_inits_and_terms @gol
480 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
481 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
482 -private_bundle -read_only_relocs -sectalign @gol
483 -sectobjectsymbols -whyload -seg1addr @gol
484 -sectcreate -sectobjectsymbols -sectorder @gol
485 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
486 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
487 -segprot -segs_read_only_addr -segs_read_write_addr @gol
488 -single_module -static -sub_library -sub_umbrella @gol
489 -twolevel_namespace -umbrella -undefined @gol
490 -unexported_symbols_list -weak_reference_mismatches @gol
491 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
492 -mkernel -mone-byte-bool}
494 @emph{DEC Alpha Options}
495 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
496 -mieee -mieee-with-inexact -mieee-conformant @gol
497 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
498 -mtrap-precision=@var{mode} -mbuild-constants @gol
499 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
500 -mbwx -mmax -mfix -mcix @gol
501 -mfloat-vax -mfloat-ieee @gol
502 -mexplicit-relocs -msmall-data -mlarge-data @gol
503 -msmall-text -mlarge-text @gol
504 -mmemory-latency=@var{time}}
506 @emph{DEC Alpha/VMS Options}
507 @gccoptlist{-mvms-return-codes}
510 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
511 -mhard-float -msoft-float @gol
512 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
513 -mdouble -mno-double @gol
514 -mmedia -mno-media -mmuladd -mno-muladd @gol
515 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
516 -mlinked-fp -mlong-calls -malign-labels @gol
517 -mlibrary-pic -macc-4 -macc-8 @gol
518 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
519 -moptimize-membar -mno-optimize-membar @gol
520 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
521 -mvliw-branch -mno-vliw-branch @gol
522 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
523 -mno-nested-cond-exec -mtomcat-stats @gol
527 @emph{GNU/Linux Options}
528 @gccoptlist{-muclibc}
530 @emph{H8/300 Options}
531 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
534 @gccoptlist{-march=@var{architecture-type} @gol
535 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
536 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
537 -mfixed-range=@var{register-range} @gol
538 -mjump-in-delay -mlinker-opt -mlong-calls @gol
539 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
540 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
541 -mno-jump-in-delay -mno-long-load-store @gol
542 -mno-portable-runtime -mno-soft-float @gol
543 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
544 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
545 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
546 -munix=@var{unix-std} -nolibdld -static -threads}
548 @emph{i386 and x86-64 Options}
549 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
550 -mfpmath=@var{unit} @gol
551 -masm=@var{dialect} -mno-fancy-math-387 @gol
552 -mno-fp-ret-in-387 -msoft-float @gol
553 -mno-wide-multiply -mrtd -malign-double @gol
554 -mpreferred-stack-boundary=@var{num} -mcx16 -msahf -mrecip @gol
555 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 @gol
556 -msse4a -m3dnow -mpopcnt -mabm @gol
557 -mthreads -mno-align-stringops -minline-all-stringops @gol
558 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
559 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
560 -mpc32 -mpc64 -mpc80 mstackrealign @gol
561 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
562 -mcmodel=@var{code-model} @gol
563 -m32 -m64 -mlarge-data-threshold=@var{num}}
566 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
567 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
568 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
569 -minline-float-divide-max-throughput @gol
570 -minline-int-divide-min-latency @gol
571 -minline-int-divide-max-throughput @gol
572 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
573 -mno-dwarf2-asm -mearly-stop-bits @gol
574 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
575 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
576 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
577 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
578 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
579 -mno-sched-prefer-non-data-spec-insns @gol
580 -mno-sched-prefer-non-control-spec-insns @gol
581 -mno-sched-count-spec-in-critical-path}
583 @emph{M32R/D Options}
584 @gccoptlist{-m32r2 -m32rx -m32r @gol
586 -malign-loops -mno-align-loops @gol
587 -missue-rate=@var{number} @gol
588 -mbranch-cost=@var{number} @gol
589 -mmodel=@var{code-size-model-type} @gol
590 -msdata=@var{sdata-type} @gol
591 -mno-flush-func -mflush-func=@var{name} @gol
592 -mno-flush-trap -mflush-trap=@var{number} @gol
596 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
598 @emph{M680x0 Options}
599 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
600 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
601 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
602 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
603 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
604 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
605 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
606 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
608 @emph{M68hc1x Options}
609 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
610 -mauto-incdec -minmax -mlong-calls -mshort @gol
611 -msoft-reg-count=@var{count}}
614 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
615 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
616 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
617 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
618 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
621 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
622 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
623 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
624 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
625 -mfp32 -mfp64 -mhard-float -msoft-float @gol
626 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
627 -msmartmips -mno-smartmips @gol
628 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
629 -mips3d -mno-mips3d -mmt -mno-mt @gol
630 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
631 -G@var{num} -membedded-data -mno-embedded-data @gol
632 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
633 -mcode-readable=@var{setting} @gol
634 -msplit-addresses -mno-split-addresses @gol
635 -mexplicit-relocs -mno-explicit-relocs @gol
636 -mcheck-zero-division -mno-check-zero-division @gol
637 -mdivide-traps -mdivide-breaks @gol
638 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
639 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
640 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
641 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
642 -mfix-sb1 -mno-fix-sb1 @gol
643 -mflush-func=@var{func} -mno-flush-func @gol
644 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
645 -mfp-exceptions -mno-fp-exceptions @gol
646 -mvr4130-align -mno-vr4130-align}
649 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
650 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
651 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
652 -mno-base-addresses -msingle-exit -mno-single-exit}
654 @emph{MN10300 Options}
655 @gccoptlist{-mmult-bug -mno-mult-bug @gol
656 -mam33 -mno-am33 @gol
657 -mam33-2 -mno-am33-2 @gol
658 -mreturn-pointer-on-d0 @gol
662 @gccoptlist{-mno-crt0 -mbacc -msim @gol
663 -march=@var{cpu-type} }
665 @emph{PDP-11 Options}
666 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
667 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
668 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
669 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
670 -mbranch-expensive -mbranch-cheap @gol
671 -msplit -mno-split -munix-asm -mdec-asm}
673 @emph{PowerPC Options}
674 See RS/6000 and PowerPC Options.
676 @emph{RS/6000 and PowerPC Options}
677 @gccoptlist{-mcpu=@var{cpu-type} @gol
678 -mtune=@var{cpu-type} @gol
679 -mpower -mno-power -mpower2 -mno-power2 @gol
680 -mpowerpc -mpowerpc64 -mno-powerpc @gol
681 -maltivec -mno-altivec @gol
682 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
683 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
684 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
685 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mdfp -mno-dfp @gol
686 -mnew-mnemonics -mold-mnemonics @gol
687 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
688 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
689 -malign-power -malign-natural @gol
690 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
691 -mstring -mno-string -mupdate -mno-update @gol
692 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
693 -mstrict-align -mno-strict-align -mrelocatable @gol
694 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
695 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
696 -mdynamic-no-pic -maltivec -mswdiv @gol
697 -mprioritize-restricted-insns=@var{priority} @gol
698 -msched-costly-dep=@var{dependence_type} @gol
699 -minsert-sched-nops=@var{scheme} @gol
700 -mcall-sysv -mcall-netbsd @gol
701 -maix-struct-return -msvr4-struct-return @gol
702 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
703 -misel -mno-isel @gol
704 -misel=yes -misel=no @gol
706 -mspe=yes -mspe=no @gol
707 -mvrsave -mno-vrsave @gol
708 -mmulhw -mno-mulhw @gol
709 -mdlmzb -mno-dlmzb @gol
710 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
711 -mprototype -mno-prototype @gol
712 -msim -mmvme -mads -myellowknife -memb -msdata @gol
713 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
715 @emph{S/390 and zSeries Options}
716 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
717 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
718 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
719 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
720 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
721 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
722 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
725 @gccoptlist{-meb -mel @gol
729 -mscore5 -mscore5u -mscore7 -mscore7d}
732 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
733 -m4-nofpu -m4-single-only -m4-single -m4 @gol
734 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
735 -m5-64media -m5-64media-nofpu @gol
736 -m5-32media -m5-32media-nofpu @gol
737 -m5-compact -m5-compact-nofpu @gol
738 -mb -ml -mdalign -mrelax @gol
739 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
740 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
741 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
742 -mdivsi3_libfunc=@var{name} @gol
743 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
747 @gccoptlist{-mcpu=@var{cpu-type} @gol
748 -mtune=@var{cpu-type} @gol
749 -mcmodel=@var{code-model} @gol
750 -m32 -m64 -mapp-regs -mno-app-regs @gol
751 -mfaster-structs -mno-faster-structs @gol
752 -mfpu -mno-fpu -mhard-float -msoft-float @gol
753 -mhard-quad-float -msoft-quad-float @gol
754 -mimpure-text -mno-impure-text -mlittle-endian @gol
755 -mstack-bias -mno-stack-bias @gol
756 -munaligned-doubles -mno-unaligned-doubles @gol
757 -mv8plus -mno-v8plus -mvis -mno-vis
758 -threads -pthreads -pthread}
761 @gccoptlist{-mwarn-reloc -merror-reloc @gol
762 -msafe-dma -munsafe-dma @gol
764 -msmall-mem -mlarge-mem -mstdmain @gol
765 -mfixed-range=@var{register-range}}
767 @emph{System V Options}
768 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
770 @emph{TMS320C3x/C4x Options}
771 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
772 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
773 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
774 -mparallel-insns -mparallel-mpy -mpreserve-float}
777 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
778 -mprolog-function -mno-prolog-function -mspace @gol
779 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
780 -mapp-regs -mno-app-regs @gol
781 -mdisable-callt -mno-disable-callt @gol
787 @gccoptlist{-mg -mgnu -munix}
789 @emph{VxWorks Options}
790 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
791 -Xbind-lazy -Xbind-now}
793 @emph{x86-64 Options}
794 See i386 and x86-64 Options.
796 @emph{Xstormy16 Options}
799 @emph{Xtensa Options}
800 @gccoptlist{-mconst16 -mno-const16 @gol
801 -mfused-madd -mno-fused-madd @gol
802 -mtext-section-literals -mno-text-section-literals @gol
803 -mtarget-align -mno-target-align @gol
804 -mlongcalls -mno-longcalls}
806 @emph{zSeries Options}
807 See S/390 and zSeries Options.
809 @item Code Generation Options
810 @xref{Code Gen Options,,Options for Code Generation Conventions}.
811 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
812 -ffixed-@var{reg} -fexceptions @gol
813 -fnon-call-exceptions -funwind-tables @gol
814 -fasynchronous-unwind-tables @gol
815 -finhibit-size-directive -finstrument-functions @gol
816 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
817 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
818 -fno-common -fno-ident @gol
819 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
820 -fno-jump-tables @gol
821 -frecord-gcc-switches @gol
822 -freg-struct-return -fshort-enums @gol
823 -fshort-double -fshort-wchar @gol
824 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
825 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
826 -fargument-alias -fargument-noalias @gol
827 -fargument-noalias-global -fargument-noalias-anything
828 -fleading-underscore -ftls-model=@var{model} @gol
829 -ftrapv -fwrapv -fbounds-check @gol
834 * Overall Options:: Controlling the kind of output:
835 an executable, object files, assembler files,
836 or preprocessed source.
837 * C Dialect Options:: Controlling the variant of C language compiled.
838 * C++ Dialect Options:: Variations on C++.
839 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
841 * Language Independent Options:: Controlling how diagnostics should be
843 * Warning Options:: How picky should the compiler be?
844 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
845 * Optimize Options:: How much optimization?
846 * Preprocessor Options:: Controlling header files and macro definitions.
847 Also, getting dependency information for Make.
848 * Assembler Options:: Passing options to the assembler.
849 * Link Options:: Specifying libraries and so on.
850 * Directory Options:: Where to find header files and libraries.
851 Where to find the compiler executable files.
852 * Spec Files:: How to pass switches to sub-processes.
853 * Target Options:: Running a cross-compiler, or an old version of GCC.
856 @node Overall Options
857 @section Options Controlling the Kind of Output
859 Compilation can involve up to four stages: preprocessing, compilation
860 proper, assembly and linking, always in that order. GCC is capable of
861 preprocessing and compiling several files either into several
862 assembler input files, or into one assembler input file; then each
863 assembler input file produces an object file, and linking combines all
864 the object files (those newly compiled, and those specified as input)
865 into an executable file.
867 @cindex file name suffix
868 For any given input file, the file name suffix determines what kind of
873 C source code which must be preprocessed.
876 C source code which should not be preprocessed.
879 C++ source code which should not be preprocessed.
882 Objective-C source code. Note that you must link with the @file{libobjc}
883 library to make an Objective-C program work.
886 Objective-C source code which should not be preprocessed.
890 Objective-C++ source code. Note that you must link with the @file{libobjc}
891 library to make an Objective-C++ program work. Note that @samp{.M} refers
892 to a literal capital M@.
895 Objective-C++ source code which should not be preprocessed.
898 C, C++, Objective-C or Objective-C++ header file to be turned into a
903 @itemx @var{file}.cxx
904 @itemx @var{file}.cpp
905 @itemx @var{file}.CPP
906 @itemx @var{file}.c++
908 C++ source code which must be preprocessed. Note that in @samp{.cxx},
909 the last two letters must both be literally @samp{x}. Likewise,
910 @samp{.C} refers to a literal capital C@.
914 Objective-C++ source code which must be preprocessed.
917 Objective-C++ source code which should not be preprocessed.
922 @itemx @var{file}.hxx
923 @itemx @var{file}.hpp
924 @itemx @var{file}.HPP
925 @itemx @var{file}.h++
926 @itemx @var{file}.tcc
927 C++ header file to be turned into a precompiled header.
930 @itemx @var{file}.for
931 @itemx @var{file}.FOR
932 Fixed form Fortran source code which should not be preprocessed.
935 @itemx @var{file}.fpp
936 @itemx @var{file}.FPP
937 Fixed form Fortran source code which must be preprocessed (with the traditional
941 @itemx @var{file}.f95
942 Free form Fortran source code which should not be preprocessed.
945 @itemx @var{file}.F95
946 Free form Fortran source code which must be preprocessed (with the
947 traditional preprocessor).
949 @c FIXME: Descriptions of Java file types.
956 Ada source code file which contains a library unit declaration (a
957 declaration of a package, subprogram, or generic, or a generic
958 instantiation), or a library unit renaming declaration (a package,
959 generic, or subprogram renaming declaration). Such files are also
962 @itemx @var{file}.adb
963 Ada source code file containing a library unit body (a subprogram or
964 package body). Such files are also called @dfn{bodies}.
966 @c GCC also knows about some suffixes for languages not yet included:
977 Assembler code which must be preprocessed.
980 An object file to be fed straight into linking.
981 Any file name with no recognized suffix is treated this way.
985 You can specify the input language explicitly with the @option{-x} option:
988 @item -x @var{language}
989 Specify explicitly the @var{language} for the following input files
990 (rather than letting the compiler choose a default based on the file
991 name suffix). This option applies to all following input files until
992 the next @option{-x} option. Possible values for @var{language} are:
994 c c-header c-cpp-output
995 c++ c++-header c++-cpp-output
996 objective-c objective-c-header objective-c-cpp-output
997 objective-c++ objective-c++-header objective-c++-cpp-output
998 assembler assembler-with-cpp
1006 Turn off any specification of a language, so that subsequent files are
1007 handled according to their file name suffixes (as they are if @option{-x}
1008 has not been used at all).
1010 @item -pass-exit-codes
1011 @opindex pass-exit-codes
1012 Normally the @command{gcc} program will exit with the code of 1 if any
1013 phase of the compiler returns a non-success return code. If you specify
1014 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1015 numerically highest error produced by any phase that returned an error
1016 indication. The C, C++, and Fortran frontends return 4, if an internal
1017 compiler error is encountered.
1020 If you only want some of the stages of compilation, you can use
1021 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1022 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1023 @command{gcc} is to stop. Note that some combinations (for example,
1024 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1029 Compile or assemble the source files, but do not link. The linking
1030 stage simply is not done. The ultimate output is in the form of an
1031 object file for each source file.
1033 By default, the object file name for a source file is made by replacing
1034 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1036 Unrecognized input files, not requiring compilation or assembly, are
1041 Stop after the stage of compilation proper; do not assemble. The output
1042 is in the form of an assembler code file for each non-assembler input
1045 By default, the assembler file name for a source file is made by
1046 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1048 Input files that don't require compilation are ignored.
1052 Stop after the preprocessing stage; do not run the compiler proper. The
1053 output is in the form of preprocessed source code, which is sent to the
1056 Input files which don't require preprocessing are ignored.
1058 @cindex output file option
1061 Place output in file @var{file}. This applies regardless to whatever
1062 sort of output is being produced, whether it be an executable file,
1063 an object file, an assembler file or preprocessed C code.
1065 If @option{-o} is not specified, the default is to put an executable
1066 file in @file{a.out}, the object file for
1067 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1068 assembler file in @file{@var{source}.s}, a precompiled header file in
1069 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1074 Print (on standard error output) the commands executed to run the stages
1075 of compilation. Also print the version number of the compiler driver
1076 program and of the preprocessor and the compiler proper.
1080 Like @option{-v} except the commands are not executed and all command
1081 arguments are quoted. This is useful for shell scripts to capture the
1082 driver-generated command lines.
1086 Use pipes rather than temporary files for communication between the
1087 various stages of compilation. This fails to work on some systems where
1088 the assembler is unable to read from a pipe; but the GNU assembler has
1093 If you are compiling multiple source files, this option tells the driver
1094 to pass all the source files to the compiler at once (for those
1095 languages for which the compiler can handle this). This will allow
1096 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1097 language for which this is supported is C@. If you pass source files for
1098 multiple languages to the driver, using this option, the driver will invoke
1099 the compiler(s) that support IMA once each, passing each compiler all the
1100 source files appropriate for it. For those languages that do not support
1101 IMA this option will be ignored, and the compiler will be invoked once for
1102 each source file in that language. If you use this option in conjunction
1103 with @option{-save-temps}, the compiler will generate multiple
1105 (one for each source file), but only one (combined) @file{.o} or
1110 Print (on the standard output) a description of the command line options
1111 understood by @command{gcc}. If the @option{-v} option is also specified
1112 then @option{--help} will also be passed on to the various processes
1113 invoked by @command{gcc}, so that they can display the command line options
1114 they accept. If the @option{-Wextra} option has also been specified
1115 (prior to the @option{--help} option), then command line options which
1116 have no documentation associated with them will also be displayed.
1119 @opindex target-help
1120 Print (on the standard output) a description of target-specific command
1121 line options for each tool. For some targets extra target-specific
1122 information may also be printed.
1124 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1125 Print (on the standard output) a description of the command line
1126 options understood by the compiler that fit into a specific class.
1127 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1128 @samp{params}, or @var{language}:
1131 @item @samp{optimizers}
1132 This will display all of the optimization options supported by the
1135 @item @samp{warnings}
1136 This will display all of the options controlling warning messages
1137 produced by the compiler.
1140 This will display target-specific options. Unlike the
1141 @option{--target-help} option however, target-specific options of the
1142 linker and assembler will not be displayed. This is because those
1143 tools do not currently support the extended @option{--help=} syntax.
1146 This will display the values recognized by the @option{--param}
1149 @item @var{language}
1150 This will display the options supported for @var{language}, where
1151 @var{language} is the name of one of the languages supported in this
1155 This will display the options that are common to all languages.
1158 It is possible to further refine the output of the @option{--help=}
1159 option by adding a comma separated list of qualifiers after the
1160 class. These can be any from the following list:
1163 @item @samp{undocumented}
1164 Display only those options which are undocumented.
1167 Display options which take an argument that appears after an equal
1168 sign in the same continuous piece of text, such as:
1169 @samp{--help=target}.
1171 @item @samp{separate}
1172 Display options which take an argument that appears as a separate word
1173 following the original option, such as: @samp{-o output-file}.
1176 Thus for example to display all the undocumented target-specific
1177 switches supported by the compiler the following can be used:
1180 --help=target,undocumented
1183 The sense of a qualifier can be inverted by prefixing it with the
1184 @var{^} character, so for example to display all binary warning
1185 options (i.e. ones that are either on or off and that do not take an
1186 argument), which have a description the following can be used:
1189 --help=warnings,^joined,^undocumented
1192 A class can also be used as a qualifier, although this usually
1193 restricts the output by so much that there is nothing to display. One
1194 case where it does work however is when one of the classes is
1195 @var{target}. So for example to display all the target-specific
1196 optimization options the following can be used:
1199 --help=target,optimizers
1202 The @option{--help=} option can be repeated on the command line. Each
1203 successive use will display its requested class of options, skipping
1204 those that have already been displayed.
1206 If the @option{-Q} option appears on the command line before the
1207 @option{--help=} option, then the descriptive text displayed by
1208 @option{--help=} is changed. Instead of describing the displayed
1209 options, an indication is given as to whether the option is enabled,
1210 disabled or set to a specific value (assuming that the compiler
1211 knows this at the point where the @option{--help=} option is used).
1213 Here is a truncated example from the ARM port of @command{gcc}:
1216 % gcc -Q -mabi=2 --help=target -c
1217 The following options are target specific:
1219 -mabort-on-noreturn [disabled]
1223 The output is sensitive to the effects of previous command line
1224 options, so for example it is possible to find out which optimizations
1225 are enabled at @option{-O2} by using:
1228 -O2 --help=optimizers
1231 Alternatively you can discover which binary optimizations are enabled
1232 by @option{-O3} by using:
1235 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1236 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1237 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1242 Display the version number and copyrights of the invoked GCC@.
1244 @include @value{srcdir}/../libiberty/at-file.texi
1248 @section Compiling C++ Programs
1250 @cindex suffixes for C++ source
1251 @cindex C++ source file suffixes
1252 C++ source files conventionally use one of the suffixes @samp{.C},
1253 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1254 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1255 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1256 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1257 files with these names and compiles them as C++ programs even if you
1258 call the compiler the same way as for compiling C programs (usually
1259 with the name @command{gcc}).
1263 However, the use of @command{gcc} does not add the C++ library.
1264 @command{g++} is a program that calls GCC and treats @samp{.c},
1265 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1266 files unless @option{-x} is used, and automatically specifies linking
1267 against the C++ library. This program is also useful when
1268 precompiling a C header file with a @samp{.h} extension for use in C++
1269 compilations. On many systems, @command{g++} is also installed with
1270 the name @command{c++}.
1272 @cindex invoking @command{g++}
1273 When you compile C++ programs, you may specify many of the same
1274 command-line options that you use for compiling programs in any
1275 language; or command-line options meaningful for C and related
1276 languages; or options that are meaningful only for C++ programs.
1277 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1278 explanations of options for languages related to C@.
1279 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1280 explanations of options that are meaningful only for C++ programs.
1282 @node C Dialect Options
1283 @section Options Controlling C Dialect
1284 @cindex dialect options
1285 @cindex language dialect options
1286 @cindex options, dialect
1288 The following options control the dialect of C (or languages derived
1289 from C, such as C++, Objective-C and Objective-C++) that the compiler
1293 @cindex ANSI support
1297 In C mode, support all ISO C90 programs. In C++ mode,
1298 remove GNU extensions that conflict with ISO C++.
1300 This turns off certain features of GCC that are incompatible with ISO
1301 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1302 such as the @code{asm} and @code{typeof} keywords, and
1303 predefined macros such as @code{unix} and @code{vax} that identify the
1304 type of system you are using. It also enables the undesirable and
1305 rarely used ISO trigraph feature. For the C compiler,
1306 it disables recognition of C++ style @samp{//} comments as well as
1307 the @code{inline} keyword.
1309 The alternate keywords @code{__asm__}, @code{__extension__},
1310 @code{__inline__} and @code{__typeof__} continue to work despite
1311 @option{-ansi}. You would not want to use them in an ISO C program, of
1312 course, but it is useful to put them in header files that might be included
1313 in compilations done with @option{-ansi}. Alternate predefined macros
1314 such as @code{__unix__} and @code{__vax__} are also available, with or
1315 without @option{-ansi}.
1317 The @option{-ansi} option does not cause non-ISO programs to be
1318 rejected gratuitously. For that, @option{-pedantic} is required in
1319 addition to @option{-ansi}. @xref{Warning Options}.
1321 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1322 option is used. Some header files may notice this macro and refrain
1323 from declaring certain functions or defining certain macros that the
1324 ISO standard doesn't call for; this is to avoid interfering with any
1325 programs that might use these names for other things.
1327 Functions which would normally be built in but do not have semantics
1328 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1329 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1330 built-in functions provided by GCC}, for details of the functions
1335 Determine the language standard. This option is currently only
1336 supported when compiling C or C++. A value for this option must be
1337 provided; possible values are
1342 ISO C90 (same as @option{-ansi}).
1344 @item iso9899:199409
1345 ISO C90 as modified in amendment 1.
1351 ISO C99. Note that this standard is not yet fully supported; see
1352 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1353 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1356 Default, ISO C90 plus GNU extensions (including some C99 features).
1360 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1361 this will become the default. The name @samp{gnu9x} is deprecated.
1364 The 1998 ISO C++ standard plus amendments.
1367 The same as @option{-std=c++98} plus GNU extensions. This is the
1368 default for C++ code.
1371 The working draft of the upcoming ISO C++0x standard. This option
1372 enables experimental features that are likely to be included in
1373 C++0x. The working draft is constantly changing, and any feature that is
1374 enabled by this flag may be removed from future versions of GCC if it is
1375 not part of the C++0x standard.
1378 The same as @option{-std=c++0x} plus GNU extensions. As with
1379 @option{-std=c++0x}, this option enables experimental features that may
1380 be removed in future versions of GCC.
1383 Even when this option is not specified, you can still use some of the
1384 features of newer standards in so far as they do not conflict with
1385 previous C standards. For example, you may use @code{__restrict__} even
1386 when @option{-std=c99} is not specified.
1388 The @option{-std} options specifying some version of ISO C have the same
1389 effects as @option{-ansi}, except that features that were not in ISO C90
1390 but are in the specified version (for example, @samp{//} comments and
1391 the @code{inline} keyword in ISO C99) are not disabled.
1393 @xref{Standards,,Language Standards Supported by GCC}, for details of
1394 these standard versions.
1396 @item -fgnu89-inline
1397 @opindex fgnu89-inline
1398 The option @option{-fgnu89-inline} tells GCC to use the traditional
1399 GNU semantics for @code{inline} functions when in C99 mode.
1400 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1401 is accepted and ignored by GCC versions 4.1.3 up to but not including
1402 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1403 C99 mode. Using this option is roughly equivalent to adding the
1404 @code{gnu_inline} function attribute to all inline functions
1405 (@pxref{Function Attributes}).
1407 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1408 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1409 specifies the default behavior). This option was first supported in
1410 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1412 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1413 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1414 in effect for @code{inline} functions. @xref{Common Predefined
1415 Macros,,,cpp,The C Preprocessor}.
1417 @item -aux-info @var{filename}
1419 Output to the given filename prototyped declarations for all functions
1420 declared and/or defined in a translation unit, including those in header
1421 files. This option is silently ignored in any language other than C@.
1423 Besides declarations, the file indicates, in comments, the origin of
1424 each declaration (source file and line), whether the declaration was
1425 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1426 @samp{O} for old, respectively, in the first character after the line
1427 number and the colon), and whether it came from a declaration or a
1428 definition (@samp{C} or @samp{F}, respectively, in the following
1429 character). In the case of function definitions, a K&R-style list of
1430 arguments followed by their declarations is also provided, inside
1431 comments, after the declaration.
1435 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1436 keyword, so that code can use these words as identifiers. You can use
1437 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1438 instead. @option{-ansi} implies @option{-fno-asm}.
1440 In C++, this switch only affects the @code{typeof} keyword, since
1441 @code{asm} and @code{inline} are standard keywords. You may want to
1442 use the @option{-fno-gnu-keywords} flag instead, which has the same
1443 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1444 switch only affects the @code{asm} and @code{typeof} keywords, since
1445 @code{inline} is a standard keyword in ISO C99.
1448 @itemx -fno-builtin-@var{function}
1449 @opindex fno-builtin
1450 @cindex built-in functions
1451 Don't recognize built-in functions that do not begin with
1452 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1453 functions provided by GCC}, for details of the functions affected,
1454 including those which are not built-in functions when @option{-ansi} or
1455 @option{-std} options for strict ISO C conformance are used because they
1456 do not have an ISO standard meaning.
1458 GCC normally generates special code to handle certain built-in functions
1459 more efficiently; for instance, calls to @code{alloca} may become single
1460 instructions that adjust the stack directly, and calls to @code{memcpy}
1461 may become inline copy loops. The resulting code is often both smaller
1462 and faster, but since the function calls no longer appear as such, you
1463 cannot set a breakpoint on those calls, nor can you change the behavior
1464 of the functions by linking with a different library. In addition,
1465 when a function is recognized as a built-in function, GCC may use
1466 information about that function to warn about problems with calls to
1467 that function, or to generate more efficient code, even if the
1468 resulting code still contains calls to that function. For example,
1469 warnings are given with @option{-Wformat} for bad calls to
1470 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1471 known not to modify global memory.
1473 With the @option{-fno-builtin-@var{function}} option
1474 only the built-in function @var{function} is
1475 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1476 function is named this is not built-in in this version of GCC, this
1477 option is ignored. There is no corresponding
1478 @option{-fbuiltin-@var{function}} option; if you wish to enable
1479 built-in functions selectively when using @option{-fno-builtin} or
1480 @option{-ffreestanding}, you may define macros such as:
1483 #define abs(n) __builtin_abs ((n))
1484 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1489 @cindex hosted environment
1491 Assert that compilation takes place in a hosted environment. This implies
1492 @option{-fbuiltin}. A hosted environment is one in which the
1493 entire standard library is available, and in which @code{main} has a return
1494 type of @code{int}. Examples are nearly everything except a kernel.
1495 This is equivalent to @option{-fno-freestanding}.
1497 @item -ffreestanding
1498 @opindex ffreestanding
1499 @cindex hosted environment
1501 Assert that compilation takes place in a freestanding environment. This
1502 implies @option{-fno-builtin}. A freestanding environment
1503 is one in which the standard library may not exist, and program startup may
1504 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1505 This is equivalent to @option{-fno-hosted}.
1507 @xref{Standards,,Language Standards Supported by GCC}, for details of
1508 freestanding and hosted environments.
1512 @cindex openmp parallel
1513 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1514 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1515 compiler generates parallel code according to the OpenMP Application
1516 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
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++ only)}
1910 Warn when G++ generates code that is probably not compatible with the
1911 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1912 all such cases, there are probably some cases that are not warned about,
1913 even though G++ is generating incompatible code. There may also be
1914 cases where warnings are emitted even though the code that is generated
1917 You should rewrite your code to avoid these warnings if you are
1918 concerned about the fact that code generated by G++ may not be binary
1919 compatible with code generated by other compilers.
1921 The known incompatibilities at this point include:
1926 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1927 pack data into the same byte as a base class. For example:
1930 struct A @{ virtual void f(); int f1 : 1; @};
1931 struct B : public A @{ int f2 : 1; @};
1935 In this case, G++ will place @code{B::f2} into the same byte
1936 as@code{A::f1}; other compilers will not. You can avoid this problem
1937 by explicitly padding @code{A} so that its size is a multiple of the
1938 byte size on your platform; that will cause G++ and other compilers to
1939 layout @code{B} identically.
1942 Incorrect handling of tail-padding for virtual bases. G++ does not use
1943 tail padding when laying out virtual bases. For example:
1946 struct A @{ virtual void f(); char c1; @};
1947 struct B @{ B(); char c2; @};
1948 struct C : public A, public virtual B @{@};
1952 In this case, G++ will not place @code{B} into the tail-padding for
1953 @code{A}; other compilers will. You can avoid this problem by
1954 explicitly padding @code{A} so that its size is a multiple of its
1955 alignment (ignoring virtual base classes); that will cause G++ and other
1956 compilers to layout @code{C} identically.
1959 Incorrect handling of bit-fields with declared widths greater than that
1960 of their underlying types, when the bit-fields appear in a union. For
1964 union U @{ int i : 4096; @};
1968 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1969 union too small by the number of bits in an @code{int}.
1972 Empty classes can be placed at incorrect offsets. For example:
1982 struct C : public B, public A @{@};
1986 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1987 it should be placed at offset zero. G++ mistakenly believes that the
1988 @code{A} data member of @code{B} is already at offset zero.
1991 Names of template functions whose types involve @code{typename} or
1992 template template parameters can be mangled incorrectly.
1995 template <typename Q>
1996 void f(typename Q::X) @{@}
1998 template <template <typename> class Q>
1999 void f(typename Q<int>::X) @{@}
2003 Instantiations of these templates may be mangled incorrectly.
2007 @item -Wctor-dtor-privacy @r{(C++ only)}
2008 @opindex Wctor-dtor-privacy
2009 Warn when a class seems unusable because all the constructors or
2010 destructors in that class are private, and it has neither friends nor
2011 public static member functions.
2013 @item -Wnon-virtual-dtor @r{(C++ only)}
2014 @opindex Wnon-virtual-dtor
2015 Warn when a class has virtual functions and accessible non-virtual
2016 destructor, in which case it would be possible but unsafe to delete
2017 an instance of a derived class through a pointer to the base class.
2018 This warning is also enabled if -Weffc++ is specified.
2020 @item -Wreorder @r{(C++ only)}
2022 @cindex reordering, warning
2023 @cindex warning for reordering of member initializers
2024 Warn when the order of member initializers given in the code does not
2025 match the order in which they must be executed. For instance:
2031 A(): j (0), i (1) @{ @}
2035 The compiler will rearrange the member initializers for @samp{i}
2036 and @samp{j} to match the declaration order of the members, emitting
2037 a warning to that effect. This warning is enabled by @option{-Wall}.
2040 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2043 @item -Weffc++ @r{(C++ only)}
2045 Warn about violations of the following style guidelines from Scott Meyers'
2046 @cite{Effective C++} book:
2050 Item 11: Define a copy constructor and an assignment operator for classes
2051 with dynamically allocated memory.
2054 Item 12: Prefer initialization to assignment in constructors.
2057 Item 14: Make destructors virtual in base classes.
2060 Item 15: Have @code{operator=} return a reference to @code{*this}.
2063 Item 23: Don't try to return a reference when you must return an object.
2067 Also warn about violations of the following style guidelines from
2068 Scott Meyers' @cite{More Effective C++} book:
2072 Item 6: Distinguish between prefix and postfix forms of increment and
2073 decrement operators.
2076 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2080 When selecting this option, be aware that the standard library
2081 headers do not obey all of these guidelines; use @samp{grep -v}
2082 to filter out those warnings.
2084 @item -Wno-deprecated @r{(C++ only)}
2085 @opindex Wno-deprecated
2086 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
2088 @item -Wstrict-null-sentinel @r{(C++ only)}
2089 @opindex Wstrict-null-sentinel
2090 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2091 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2092 to @code{__null}. Although it is a null pointer constant not a null pointer,
2093 it is guaranteed to of the same size as a pointer. But this use is
2094 not portable across different compilers.
2096 @item -Wno-non-template-friend @r{(C++ only)}
2097 @opindex Wno-non-template-friend
2098 Disable warnings when non-templatized friend functions are declared
2099 within a template. Since the advent of explicit template specification
2100 support in G++, if the name of the friend is an unqualified-id (i.e.,
2101 @samp{friend foo(int)}), the C++ language specification demands that the
2102 friend declare or define an ordinary, nontemplate function. (Section
2103 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2104 could be interpreted as a particular specialization of a templatized
2105 function. Because this non-conforming behavior is no longer the default
2106 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2107 check existing code for potential trouble spots and is on by default.
2108 This new compiler behavior can be turned off with
2109 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2110 but disables the helpful warning.
2112 @item -Wold-style-cast @r{(C++ only)}
2113 @opindex Wold-style-cast
2114 Warn if an old-style (C-style) cast to a non-void type is used within
2115 a C++ program. The new-style casts (@samp{dynamic_cast},
2116 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2117 less vulnerable to unintended effects and much easier to search for.
2119 @item -Woverloaded-virtual @r{(C++ only)}
2120 @opindex Woverloaded-virtual
2121 @cindex overloaded virtual fn, warning
2122 @cindex warning for overloaded virtual fn
2123 Warn when a function declaration hides virtual functions from a
2124 base class. For example, in:
2131 struct B: public A @{
2136 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2144 will fail to compile.
2146 @item -Wno-pmf-conversions @r{(C++ only)}
2147 @opindex Wno-pmf-conversions
2148 Disable the diagnostic for converting a bound pointer to member function
2151 @item -Wsign-promo @r{(C++ only)}
2152 @opindex Wsign-promo
2153 Warn when overload resolution chooses a promotion from unsigned or
2154 enumerated type to a signed type, over a conversion to an unsigned type of
2155 the same size. Previous versions of G++ would try to preserve
2156 unsignedness, but the standard mandates the current behavior.
2161 A& operator = (int);
2171 In this example, G++ will synthesize a default @samp{A& operator =
2172 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2175 @node Objective-C and Objective-C++ Dialect Options
2176 @section Options Controlling Objective-C and Objective-C++ Dialects
2178 @cindex compiler options, Objective-C and Objective-C++
2179 @cindex Objective-C and Objective-C++ options, command line
2180 @cindex options, Objective-C and Objective-C++
2181 (NOTE: This manual does not describe the Objective-C and Objective-C++
2182 languages themselves. See @xref{Standards,,Language Standards
2183 Supported by GCC}, for references.)
2185 This section describes the command-line options that are only meaningful
2186 for Objective-C and Objective-C++ programs, but you can also use most of
2187 the language-independent GNU compiler options.
2188 For example, you might compile a file @code{some_class.m} like this:
2191 gcc -g -fgnu-runtime -O -c some_class.m
2195 In this example, @option{-fgnu-runtime} is an option meant only for
2196 Objective-C and Objective-C++ programs; you can use the other options with
2197 any language supported by GCC@.
2199 Note that since Objective-C is an extension of the C language, Objective-C
2200 compilations may also use options specific to the C front-end (e.g.,
2201 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2202 C++-specific options (e.g., @option{-Wabi}).
2204 Here is a list of options that are @emph{only} for compiling Objective-C
2205 and Objective-C++ programs:
2208 @item -fconstant-string-class=@var{class-name}
2209 @opindex fconstant-string-class
2210 Use @var{class-name} as the name of the class to instantiate for each
2211 literal string specified with the syntax @code{@@"@dots{}"}. The default
2212 class name is @code{NXConstantString} if the GNU runtime is being used, and
2213 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2214 @option{-fconstant-cfstrings} option, if also present, will override the
2215 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2216 to be laid out as constant CoreFoundation strings.
2219 @opindex fgnu-runtime
2220 Generate object code compatible with the standard GNU Objective-C
2221 runtime. This is the default for most types of systems.
2223 @item -fnext-runtime
2224 @opindex fnext-runtime
2225 Generate output compatible with the NeXT runtime. This is the default
2226 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2227 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2230 @item -fno-nil-receivers
2231 @opindex fno-nil-receivers
2232 Assume that all Objective-C message dispatches (e.g.,
2233 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2234 is not @code{nil}. This allows for more efficient entry points in the runtime
2235 to be used. Currently, this option is only available in conjunction with
2236 the NeXT runtime on Mac OS X 10.3 and later.
2238 @item -fobjc-call-cxx-cdtors
2239 @opindex fobjc-call-cxx-cdtors
2240 For each Objective-C class, check if any of its instance variables is a
2241 C++ object with a non-trivial default constructor. If so, synthesize a
2242 special @code{- (id) .cxx_construct} instance method that will run
2243 non-trivial default constructors on any such instance variables, in order,
2244 and then return @code{self}. Similarly, check if any instance variable
2245 is a C++ object with a non-trivial destructor, and if so, synthesize a
2246 special @code{- (void) .cxx_destruct} method that will run
2247 all such default destructors, in reverse order.
2249 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2250 thusly generated will only operate on instance variables declared in the
2251 current Objective-C class, and not those inherited from superclasses. It
2252 is the responsibility of the Objective-C runtime to invoke all such methods
2253 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2254 will be invoked by the runtime immediately after a new object
2255 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2256 be invoked immediately before the runtime deallocates an object instance.
2258 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2259 support for invoking the @code{- (id) .cxx_construct} and
2260 @code{- (void) .cxx_destruct} methods.
2262 @item -fobjc-direct-dispatch
2263 @opindex fobjc-direct-dispatch
2264 Allow fast jumps to the message dispatcher. On Darwin this is
2265 accomplished via the comm page.
2267 @item -fobjc-exceptions
2268 @opindex fobjc-exceptions
2269 Enable syntactic support for structured exception handling in Objective-C,
2270 similar to what is offered by C++ and Java. This option is
2271 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2280 @@catch (AnObjCClass *exc) @{
2287 @@catch (AnotherClass *exc) @{
2290 @@catch (id allOthers) @{
2300 The @code{@@throw} statement may appear anywhere in an Objective-C or
2301 Objective-C++ program; when used inside of a @code{@@catch} block, the
2302 @code{@@throw} may appear without an argument (as shown above), in which case
2303 the object caught by the @code{@@catch} will be rethrown.
2305 Note that only (pointers to) Objective-C objects may be thrown and
2306 caught using this scheme. When an object is thrown, it will be caught
2307 by the nearest @code{@@catch} clause capable of handling objects of that type,
2308 analogously to how @code{catch} blocks work in C++ and Java. A
2309 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2310 any and all Objective-C exceptions not caught by previous @code{@@catch}
2313 The @code{@@finally} clause, if present, will be executed upon exit from the
2314 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2315 regardless of whether any exceptions are thrown, caught or rethrown
2316 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2317 of the @code{finally} clause in Java.
2319 There are several caveats to using the new exception mechanism:
2323 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2324 idioms provided by the @code{NSException} class, the new
2325 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2326 systems, due to additional functionality needed in the (NeXT) Objective-C
2330 As mentioned above, the new exceptions do not support handling
2331 types other than Objective-C objects. Furthermore, when used from
2332 Objective-C++, the Objective-C exception model does not interoperate with C++
2333 exceptions at this time. This means you cannot @code{@@throw} an exception
2334 from Objective-C and @code{catch} it in C++, or vice versa
2335 (i.e., @code{throw @dots{} @@catch}).
2338 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2339 blocks for thread-safe execution:
2342 @@synchronized (ObjCClass *guard) @{
2347 Upon entering the @code{@@synchronized} block, a thread of execution shall
2348 first check whether a lock has been placed on the corresponding @code{guard}
2349 object by another thread. If it has, the current thread shall wait until
2350 the other thread relinquishes its lock. Once @code{guard} becomes available,
2351 the current thread will place its own lock on it, execute the code contained in
2352 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2353 making @code{guard} available to other threads).
2355 Unlike Java, Objective-C does not allow for entire methods to be marked
2356 @code{@@synchronized}. Note that throwing exceptions out of
2357 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2358 to be unlocked properly.
2362 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2364 @item -freplace-objc-classes
2365 @opindex freplace-objc-classes
2366 Emit a special marker instructing @command{ld(1)} not to statically link in
2367 the resulting object file, and allow @command{dyld(1)} to load it in at
2368 run time instead. This is used in conjunction with the Fix-and-Continue
2369 debugging mode, where the object file in question may be recompiled and
2370 dynamically reloaded in the course of program execution, without the need
2371 to restart the program itself. Currently, Fix-and-Continue functionality
2372 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2377 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2378 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2379 compile time) with static class references that get initialized at load time,
2380 which improves run-time performance. Specifying the @option{-fzero-link} flag
2381 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2382 to be retained. This is useful in Zero-Link debugging mode, since it allows
2383 for individual class implementations to be modified during program execution.
2387 Dump interface declarations for all classes seen in the source file to a
2388 file named @file{@var{sourcename}.decl}.
2390 @item -Wassign-intercept
2391 @opindex Wassign-intercept
2392 Warn whenever an Objective-C assignment is being intercepted by the
2396 @opindex Wno-protocol
2397 If a class is declared to implement a protocol, a warning is issued for
2398 every method in the protocol that is not implemented by the class. The
2399 default behavior is to issue a warning for every method not explicitly
2400 implemented in the class, even if a method implementation is inherited
2401 from the superclass. If you use the @option{-Wno-protocol} option, then
2402 methods inherited from the superclass are considered to be implemented,
2403 and no warning is issued for them.
2407 Warn if multiple methods of different types for the same selector are
2408 found during compilation. The check is performed on the list of methods
2409 in the final stage of compilation. Additionally, a check is performed
2410 for each selector appearing in a @code{@@selector(@dots{})}
2411 expression, and a corresponding method for that selector has been found
2412 during compilation. Because these checks scan the method table only at
2413 the end of compilation, these warnings are not produced if the final
2414 stage of compilation is not reached, for example because an error is
2415 found during compilation, or because the @option{-fsyntax-only} option is
2418 @item -Wstrict-selector-match
2419 @opindex Wstrict-selector-match
2420 Warn if multiple methods with differing argument and/or return types are
2421 found for a given selector when attempting to send a message using this
2422 selector to a receiver of type @code{id} or @code{Class}. When this flag
2423 is off (which is the default behavior), the compiler will omit such warnings
2424 if any differences found are confined to types which share the same size
2427 @item -Wundeclared-selector
2428 @opindex Wundeclared-selector
2429 Warn if a @code{@@selector(@dots{})} expression referring to an
2430 undeclared selector is found. A selector is considered undeclared if no
2431 method with that name has been declared before the
2432 @code{@@selector(@dots{})} expression, either explicitly in an
2433 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2434 an @code{@@implementation} section. This option always performs its
2435 checks as soon as a @code{@@selector(@dots{})} expression is found,
2436 while @option{-Wselector} only performs its checks in the final stage of
2437 compilation. This also enforces the coding style convention
2438 that methods and selectors must be declared before being used.
2440 @item -print-objc-runtime-info
2441 @opindex print-objc-runtime-info
2442 Generate C header describing the largest structure that is passed by
2447 @node Language Independent Options
2448 @section Options to Control Diagnostic Messages Formatting
2449 @cindex options to control diagnostics formatting
2450 @cindex diagnostic messages
2451 @cindex message formatting
2453 Traditionally, diagnostic messages have been formatted irrespective of
2454 the output device's aspect (e.g.@: its width, @dots{}). The options described
2455 below can be used to control the diagnostic messages formatting
2456 algorithm, e.g.@: how many characters per line, how often source location
2457 information should be reported. Right now, only the C++ front end can
2458 honor these options. However it is expected, in the near future, that
2459 the remaining front ends would be able to digest them correctly.
2462 @item -fmessage-length=@var{n}
2463 @opindex fmessage-length
2464 Try to format error messages so that they fit on lines of about @var{n}
2465 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2466 the front ends supported by GCC@. If @var{n} is zero, then no
2467 line-wrapping will be done; each error message will appear on a single
2470 @opindex fdiagnostics-show-location
2471 @item -fdiagnostics-show-location=once
2472 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2473 reporter to emit @emph{once} source location information; that is, in
2474 case the message is too long to fit on a single physical line and has to
2475 be wrapped, the source location won't be emitted (as prefix) again,
2476 over and over, in subsequent continuation lines. This is the default
2479 @item -fdiagnostics-show-location=every-line
2480 Only meaningful in line-wrapping mode. Instructs the diagnostic
2481 messages reporter to emit the same source location information (as
2482 prefix) for physical lines that result from the process of breaking
2483 a message which is too long to fit on a single line.
2485 @item -fdiagnostics-show-option
2486 @opindex fdiagnostics-show-option
2487 This option instructs the diagnostic machinery to add text to each
2488 diagnostic emitted, which indicates which command line option directly
2489 controls that diagnostic, when such an option is known to the
2490 diagnostic machinery.
2492 @item -Wcoverage-mismatch
2493 @opindex Wcoverage-mismatch
2494 Warn if feedback profiles do not match when using the
2495 @option{-fprofile-use} option.
2496 If a source file was changed between @option{-fprofile-gen} and
2497 @option{-fprofile-use}, the files with the profile feedback can fail
2498 to match the source file and GCC can not use the profile feedback
2499 information. By default, GCC emits an error message in this case.
2500 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2501 error. GCC does not use appropriate feedback profiles, so using this
2502 option can result in poorly optimized code. This option is useful
2503 only in the case of very minor changes such as bug fixes to an
2508 @node Warning Options
2509 @section Options to Request or Suppress Warnings
2510 @cindex options to control warnings
2511 @cindex warning messages
2512 @cindex messages, warning
2513 @cindex suppressing warnings
2515 Warnings are diagnostic messages that report constructions which
2516 are not inherently erroneous but which are risky or suggest there
2517 may have been an error.
2519 You can request many specific warnings with options beginning @samp{-W},
2520 for example @option{-Wimplicit} to request warnings on implicit
2521 declarations. Each of these specific warning options also has a
2522 negative form beginning @samp{-Wno-} to turn off warnings;
2523 for example, @option{-Wno-implicit}. This manual lists only one of the
2524 two forms, whichever is not the default.
2526 The following options control the amount and kinds of warnings produced
2527 by GCC; for further, language-specific options also refer to
2528 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2532 @cindex syntax checking
2534 @opindex fsyntax-only
2535 Check the code for syntax errors, but don't do anything beyond that.
2539 Issue all the warnings demanded by strict ISO C and ISO C++;
2540 reject all programs that use forbidden extensions, and some other
2541 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2542 version of the ISO C standard specified by any @option{-std} option used.
2544 Valid ISO C and ISO C++ programs should compile properly with or without
2545 this option (though a rare few will require @option{-ansi} or a
2546 @option{-std} option specifying the required version of ISO C)@. However,
2547 without this option, certain GNU extensions and traditional C and C++
2548 features are supported as well. With this option, they are rejected.
2550 @option{-pedantic} does not cause warning messages for use of the
2551 alternate keywords whose names begin and end with @samp{__}. Pedantic
2552 warnings are also disabled in the expression that follows
2553 @code{__extension__}. However, only system header files should use
2554 these escape routes; application programs should avoid them.
2555 @xref{Alternate Keywords}.
2557 Some users try to use @option{-pedantic} to check programs for strict ISO
2558 C conformance. They soon find that it does not do quite what they want:
2559 it finds some non-ISO practices, but not all---only those for which
2560 ISO C @emph{requires} a diagnostic, and some others for which
2561 diagnostics have been added.
2563 A feature to report any failure to conform to ISO C might be useful in
2564 some instances, but would require considerable additional work and would
2565 be quite different from @option{-pedantic}. We don't have plans to
2566 support such a feature in the near future.
2568 Where the standard specified with @option{-std} represents a GNU
2569 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2570 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2571 extended dialect is based. Warnings from @option{-pedantic} are given
2572 where they are required by the base standard. (It would not make sense
2573 for such warnings to be given only for features not in the specified GNU
2574 C dialect, since by definition the GNU dialects of C include all
2575 features the compiler supports with the given option, and there would be
2576 nothing to warn about.)
2578 @item -pedantic-errors
2579 @opindex pedantic-errors
2580 Like @option{-pedantic}, except that errors are produced rather than
2585 Inhibit all warning messages.
2589 Inhibit warning messages about the use of @samp{#import}.
2591 @item -Wchar-subscripts
2592 @opindex Wchar-subscripts
2593 Warn if an array subscript has type @code{char}. This is a common cause
2594 of error, as programmers often forget that this type is signed on some
2596 This warning is enabled by @option{-Wall}.
2600 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2601 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2602 This warning is enabled by @option{-Wall}.
2604 @item -Wfatal-errors
2605 @opindex Wfatal-errors
2606 This option causes the compiler to abort compilation on the first error
2607 occurred rather than trying to keep going and printing further error
2612 @opindex ffreestanding
2613 @opindex fno-builtin
2614 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2615 the arguments supplied have types appropriate to the format string
2616 specified, and that the conversions specified in the format string make
2617 sense. This includes standard functions, and others specified by format
2618 attributes (@pxref{Function Attributes}), in the @code{printf},
2619 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2620 not in the C standard) families (or other target-specific families).
2621 Which functions are checked without format attributes having been
2622 specified depends on the standard version selected, and such checks of
2623 functions without the attribute specified are disabled by
2624 @option{-ffreestanding} or @option{-fno-builtin}.
2626 The formats are checked against the format features supported by GNU
2627 libc version 2.2. These include all ISO C90 and C99 features, as well
2628 as features from the Single Unix Specification and some BSD and GNU
2629 extensions. Other library implementations may not support all these
2630 features; GCC does not support warning about features that go beyond a
2631 particular library's limitations. However, if @option{-pedantic} is used
2632 with @option{-Wformat}, warnings will be given about format features not
2633 in the selected standard version (but not for @code{strfmon} formats,
2634 since those are not in any version of the C standard). @xref{C Dialect
2635 Options,,Options Controlling C Dialect}.
2637 Since @option{-Wformat} also checks for null format arguments for
2638 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2640 @option{-Wformat} is included in @option{-Wall}. For more control over some
2641 aspects of format checking, the options @option{-Wformat-y2k},
2642 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2643 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2644 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2647 @opindex Wformat-y2k
2648 If @option{-Wformat} is specified, also warn about @code{strftime}
2649 formats which may yield only a two-digit year.
2651 @item -Wno-format-extra-args
2652 @opindex Wno-format-extra-args
2653 If @option{-Wformat} is specified, do not warn about excess arguments to a
2654 @code{printf} or @code{scanf} format function. The C standard specifies
2655 that such arguments are ignored.
2657 Where the unused arguments lie between used arguments that are
2658 specified with @samp{$} operand number specifications, normally
2659 warnings are still given, since the implementation could not know what
2660 type to pass to @code{va_arg} to skip the unused arguments. However,
2661 in the case of @code{scanf} formats, this option will suppress the
2662 warning if the unused arguments are all pointers, since the Single
2663 Unix Specification says that such unused arguments are allowed.
2665 @item -Wno-format-zero-length
2666 @opindex Wno-format-zero-length
2667 If @option{-Wformat} is specified, do not warn about zero-length formats.
2668 The C standard specifies that zero-length formats are allowed.
2670 @item -Wformat-nonliteral
2671 @opindex Wformat-nonliteral
2672 If @option{-Wformat} is specified, also warn if the format string is not a
2673 string literal and so cannot be checked, unless the format function
2674 takes its format arguments as a @code{va_list}.
2676 @item -Wformat-security
2677 @opindex Wformat-security
2678 If @option{-Wformat} is specified, also warn about uses of format
2679 functions that represent possible security problems. At present, this
2680 warns about calls to @code{printf} and @code{scanf} functions where the
2681 format string is not a string literal and there are no format arguments,
2682 as in @code{printf (foo);}. This may be a security hole if the format
2683 string came from untrusted input and contains @samp{%n}. (This is
2684 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2685 in future warnings may be added to @option{-Wformat-security} that are not
2686 included in @option{-Wformat-nonliteral}.)
2690 Enable @option{-Wformat} plus format checks not included in
2691 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2692 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2696 Warn about passing a null pointer for arguments marked as
2697 requiring a non-null value by the @code{nonnull} function attribute.
2699 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2700 can be disabled with the @option{-Wno-nonnull} option.
2702 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2704 Warn about uninitialized variables which are initialized with themselves.
2705 Note this option can only be used with the @option{-Wuninitialized} option,
2706 which in turn only works with @option{-O1} and above.
2708 For example, GCC will warn about @code{i} being uninitialized in the
2709 following snippet only when @option{-Winit-self} has been specified:
2720 @item -Wimplicit-int
2721 @opindex Wimplicit-int
2722 Warn when a declaration does not specify a type.
2723 This warning is enabled by @option{-Wall}.
2725 @item -Wimplicit-function-declaration
2726 @opindex Wimplicit-function-declaration
2727 @opindex Wno-implicit-function-declaration
2728 Give a warning whenever a function is used before being declared. In
2729 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2730 enabled by default and it is made into an error by
2731 @option{-pedantic-errors}. This warning is also enabled by
2736 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2737 This warning is enabled by @option{-Wall}.
2741 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2742 function with external linkage, returning int, taking either zero
2743 arguments, two, or three arguments of appropriate types.
2744 This warning is enabled by @option{-Wall}.
2746 @item -Wmissing-braces
2747 @opindex Wmissing-braces
2748 Warn if an aggregate or union initializer is not fully bracketed. In
2749 the following example, the initializer for @samp{a} is not fully
2750 bracketed, but that for @samp{b} is fully bracketed.
2753 int a[2][2] = @{ 0, 1, 2, 3 @};
2754 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2757 This warning is enabled by @option{-Wall}.
2759 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2760 @opindex Wmissing-include-dirs
2761 Warn if a user-supplied include directory does not exist.
2764 @opindex Wparentheses
2765 Warn if parentheses are omitted in certain contexts, such
2766 as when there is an assignment in a context where a truth value
2767 is expected, or when operators are nested whose precedence people
2768 often get confused about.
2770 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2771 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2772 interpretation from that of ordinary mathematical notation.
2774 Also warn about constructions where there may be confusion to which
2775 @code{if} statement an @code{else} branch belongs. Here is an example of
2790 In C/C++, every @code{else} branch belongs to the innermost possible
2791 @code{if} statement, which in this example is @code{if (b)}. This is
2792 often not what the programmer expected, as illustrated in the above
2793 example by indentation the programmer chose. When there is the
2794 potential for this confusion, GCC will issue a warning when this flag
2795 is specified. To eliminate the warning, add explicit braces around
2796 the innermost @code{if} statement so there is no way the @code{else}
2797 could belong to the enclosing @code{if}. The resulting code would
2814 This warning is enabled by @option{-Wall}.
2816 @item -Wsequence-point
2817 @opindex Wsequence-point
2818 Warn about code that may have undefined semantics because of violations
2819 of sequence point rules in the C and C++ standards.
2821 The C and C++ standards defines the order in which expressions in a C/C++
2822 program are evaluated in terms of @dfn{sequence points}, which represent
2823 a partial ordering between the execution of parts of the program: those
2824 executed before the sequence point, and those executed after it. These
2825 occur after the evaluation of a full expression (one which is not part
2826 of a larger expression), after the evaluation of the first operand of a
2827 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2828 function is called (but after the evaluation of its arguments and the
2829 expression denoting the called function), and in certain other places.
2830 Other than as expressed by the sequence point rules, the order of
2831 evaluation of subexpressions of an expression is not specified. All
2832 these rules describe only a partial order rather than a total order,
2833 since, for example, if two functions are called within one expression
2834 with no sequence point between them, the order in which the functions
2835 are called is not specified. However, the standards committee have
2836 ruled that function calls do not overlap.
2838 It is not specified when between sequence points modifications to the
2839 values of objects take effect. Programs whose behavior depends on this
2840 have undefined behavior; the C and C++ standards specify that ``Between
2841 the previous and next sequence point an object shall have its stored
2842 value modified at most once by the evaluation of an expression.
2843 Furthermore, the prior value shall be read only to determine the value
2844 to be stored.''. If a program breaks these rules, the results on any
2845 particular implementation are entirely unpredictable.
2847 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2848 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2849 diagnosed by this option, and it may give an occasional false positive
2850 result, but in general it has been found fairly effective at detecting
2851 this sort of problem in programs.
2853 The standard is worded confusingly, therefore there is some debate
2854 over the precise meaning of the sequence point rules in subtle cases.
2855 Links to discussions of the problem, including proposed formal
2856 definitions, may be found on the GCC readings page, at
2857 @w{@uref{http://gcc.gnu.org/readings.html}}.
2859 This warning is enabled by @option{-Wall} for C and C++.
2862 @opindex Wreturn-type
2863 @opindex Wno-return-type
2864 Warn whenever a function is defined with a return-type that defaults
2865 to @code{int}. Also warn about any @code{return} statement with no
2866 return-value in a function whose return-type is not @code{void}
2867 (falling off the end of the function body is considered returning
2868 without a value), and about a @code{return} statement with a
2869 expression in a function whose return-type is @code{void}.
2871 Also warn if the return type of a function has a type qualifier
2872 such as @code{const}. For ISO C such a type qualifier has no effect,
2873 since the value returned by a function is not an lvalue.
2874 For C++, the warning is only emitted for scalar types or @code{void}.
2875 ISO C prohibits qualified @code{void} return types on function
2876 definitions, so such return types always receive a warning
2877 even without this option.
2879 For C++, a function without return type always produces a diagnostic
2880 message, even when @option{-Wno-return-type} is specified. The only
2881 exceptions are @samp{main} and functions defined in system headers.
2883 This warning is enabled by @option{-Wall}.
2887 Warn whenever a @code{switch} statement has an index of enumerated type
2888 and lacks a @code{case} for one or more of the named codes of that
2889 enumeration. (The presence of a @code{default} label prevents this
2890 warning.) @code{case} labels outside the enumeration range also
2891 provoke warnings when this option is used.
2892 This warning is enabled by @option{-Wall}.
2894 @item -Wswitch-default
2895 @opindex Wswitch-switch
2896 Warn whenever a @code{switch} statement does not have a @code{default}
2900 @opindex Wswitch-enum
2901 Warn whenever a @code{switch} statement has an index of enumerated type
2902 and lacks a @code{case} for one or more of the named codes of that
2903 enumeration. @code{case} labels outside the enumeration range also
2904 provoke warnings when this option is used.
2908 Warn if any trigraphs are encountered that might change the meaning of
2909 the program (trigraphs within comments are not warned about).
2910 This warning is enabled by @option{-Wall}.
2912 @item -Wunused-function
2913 @opindex Wunused-function
2914 Warn whenever a static function is declared but not defined or a
2915 non-inline static function is unused.
2916 This warning is enabled by @option{-Wall}.
2918 @item -Wunused-label
2919 @opindex Wunused-label
2920 Warn whenever a label is declared but not used.
2921 This warning is enabled by @option{-Wall}.
2923 To suppress this warning use the @samp{unused} attribute
2924 (@pxref{Variable Attributes}).
2926 @item -Wunused-parameter
2927 @opindex Wunused-parameter
2928 Warn whenever a function parameter is unused aside from its declaration.
2930 To suppress this warning use the @samp{unused} attribute
2931 (@pxref{Variable Attributes}).
2933 @item -Wunused-variable
2934 @opindex Wunused-variable
2935 Warn whenever a local variable or non-constant static variable is unused
2936 aside from its declaration.
2937 This warning is enabled by @option{-Wall}.
2939 To suppress this warning use the @samp{unused} attribute
2940 (@pxref{Variable Attributes}).
2942 @item -Wunused-value
2943 @opindex Wunused-value
2944 Warn whenever a statement computes a result that is explicitly not
2945 used. To suppress this warning cast the unused expression to
2946 @samp{void}. This includes an expression-statement or the left-hand
2947 side of a comma expression that contains no side effects. For example,
2948 an expression such as @samp{x[i,j]} will cause a warning, while
2949 @samp{x[(void)i,j]} will not.
2951 This warning is enabled by @option{-Wall}.
2955 All the above @option{-Wunused} options combined.
2957 In order to get a warning about an unused function parameter, you must
2958 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2959 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2961 @item -Wuninitialized
2962 @opindex Wuninitialized
2963 Warn if an automatic variable is used without first being initialized or
2964 if a variable may be clobbered by a @code{setjmp} call.
2966 These warnings are possible only in optimizing compilation,
2967 because they require data flow information that is computed only
2968 when optimizing. If you do not specify @option{-O}, you will not get
2969 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2970 requiring @option{-O}.
2972 If you want to warn about code which uses the uninitialized value of the
2973 variable in its own initializer, use the @option{-Winit-self} option.
2975 These warnings occur for individual uninitialized or clobbered
2976 elements of structure, union or array variables as well as for
2977 variables which are uninitialized or clobbered as a whole. They do
2978 not occur for variables or elements declared @code{volatile}. Because
2979 these warnings depend on optimization, the exact variables or elements
2980 for which there are warnings will depend on the precise optimization
2981 options and version of GCC used.
2983 Note that there may be no warning about a variable that is used only
2984 to compute a value that itself is never used, because such
2985 computations may be deleted by data flow analysis before the warnings
2988 These warnings are made optional because GCC is not smart
2989 enough to see all the reasons why the code might be correct
2990 despite appearing to have an error. Here is one example of how
3011 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3012 always initialized, but GCC doesn't know this. Here is
3013 another common case:
3018 if (change_y) save_y = y, y = new_y;
3020 if (change_y) y = save_y;
3025 This has no bug because @code{save_y} is used only if it is set.
3027 @cindex @code{longjmp} warnings
3028 This option also warns when a non-volatile automatic variable might be
3029 changed by a call to @code{longjmp}. These warnings as well are possible
3030 only in optimizing compilation.
3032 The compiler sees only the calls to @code{setjmp}. It cannot know
3033 where @code{longjmp} will be called; in fact, a signal handler could
3034 call it at any point in the code. As a result, you may get a warning
3035 even when there is in fact no problem because @code{longjmp} cannot
3036 in fact be called at the place which would cause a problem.
3038 Some spurious warnings can be avoided if you declare all the functions
3039 you use that never return as @code{noreturn}. @xref{Function
3042 This warning is enabled by @option{-Wall}.
3044 @item -Wunknown-pragmas
3045 @opindex Wunknown-pragmas
3046 @cindex warning for unknown pragmas
3047 @cindex unknown pragmas, warning
3048 @cindex pragmas, warning of unknown
3049 Warn when a #pragma directive is encountered which is not understood by
3050 GCC@. If this command line option is used, warnings will even be issued
3051 for unknown pragmas in system header files. This is not the case if
3052 the warnings were only enabled by the @option{-Wall} command line option.
3055 @opindex Wno-pragmas
3057 Do not warn about misuses of pragmas, such as incorrect parameters,
3058 invalid syntax, or conflicts between pragmas. See also
3059 @samp{-Wunknown-pragmas}.
3061 @item -Wstrict-aliasing
3062 @opindex Wstrict-aliasing
3063 This option is only active when @option{-fstrict-aliasing} is active.
3064 It warns about code which might break the strict aliasing rules that the
3065 compiler is using for optimization. The warning does not catch all
3066 cases, but does attempt to catch the more common pitfalls. It is
3067 included in @option{-Wall}.
3068 It is equivalent to -Wstrict-aliasing=3
3070 @item -Wstrict-aliasing=n
3071 @opindex Wstrict-aliasing=n
3072 This option is only active when @option{-fstrict-aliasing} is active.
3073 It warns about code which might break the strict aliasing rules that the
3074 compiler is using for optimization.
3075 Higher levels correspond to higher accuracy (fewer false positives).
3076 Higher levels also correspond to more effort, similar to the way -O works.
3077 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3080 Level 1: Most aggressive, quick, least accurate.
3081 Possibly useful when higher levels
3082 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3083 false negatives. However, it has many false positives.
3084 Warns for all pointer conversions between possibly incompatible types,
3085 even if never dereferenced. Runs in the frontend only.
3087 Level 2: Aggressive, quick, not too precise.
3088 May still have many false positives (not as many as level 1 though),
3089 and few false negatives (but possibly more than level 1).
3090 Unlike level 1, it only warns when an address is taken. Warns about
3091 incomplete types. Runs in the frontend only.
3093 Level 3 (default for @option{-Wstrict-aliasing}):
3094 Should have very few false positives and few false
3095 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3096 Takes care of the common punn+dereference pattern in the frontend:
3097 @code{*(int*)&some_float}.
3098 If optimization is enabled, it also runs in the backend, where it deals
3099 with multiple statement cases using flow-sensitive points-to information.
3100 Only warns when the converted pointer is dereferenced.
3101 Does not warn about incomplete types.
3103 @item -Wstrict-overflow
3104 @item -Wstrict-overflow=@var{n}
3105 @opindex Wstrict-overflow
3106 This option is only active when @option{-fstrict-overflow} is active.
3107 It warns about cases where the compiler optimizes based on the
3108 assumption that signed overflow does not occur. Note that it does not
3109 warn about all cases where the code might overflow: it only warns
3110 about cases where the compiler implements some optimization. Thus
3111 this warning depends on the optimization level.
3113 An optimization which assumes that signed overflow does not occur is
3114 perfectly safe if the values of the variables involved are such that
3115 overflow never does, in fact, occur. Therefore this warning can
3116 easily give a false positive: a warning about code which is not
3117 actually a problem. To help focus on important issues, several
3118 warning levels are defined. No warnings are issued for the use of
3119 undefined signed overflow when estimating how many iterations a loop
3120 will require, in particular when determining whether a loop will be
3124 @item -Wstrict-overflow=1
3125 Warn about cases which are both questionable and easy to avoid. For
3126 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3127 compiler will simplify this to @code{1}. This level of
3128 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3129 are not, and must be explicitly requested.
3131 @item -Wstrict-overflow=2
3132 Also warn about other cases where a comparison is simplified to a
3133 constant. For example: @code{abs (x) >= 0}. This can only be
3134 simplified when @option{-fstrict-overflow} is in effect, because
3135 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3136 zero. @option{-Wstrict-overflow} (with no level) is the same as
3137 @option{-Wstrict-overflow=2}.
3139 @item -Wstrict-overflow=3
3140 Also warn about other cases where a comparison is simplified. For
3141 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3143 @item -Wstrict-overflow=4
3144 Also warn about other simplifications not covered by the above cases.
3145 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3147 @item -Wstrict-overflow=5
3148 Also warn about cases where the compiler reduces the magnitude of a
3149 constant involved in a comparison. For example: @code{x + 2 > y} will
3150 be simplified to @code{x + 1 >= y}. This is reported only at the
3151 highest warning level because this simplification applies to many
3152 comparisons, so this warning level will give a very large number of
3156 @item -Warray-bounds
3157 @opindex Wno-array-bounds
3158 @opindex Warray-bounds
3159 This option is only active when @option{-ftree-vrp} is active
3160 (default for -O2 and above). It warns about subscripts to arrays
3161 that are always out of bounds. This warning is enabled by @option{-Wall}.
3165 All of the above @samp{-W} options combined. This enables all the
3166 warnings about constructions that some users consider questionable, and
3167 that are easy to avoid (or modify to prevent the warning), even in
3168 conjunction with macros. This also enables some language-specific
3169 warnings described in @ref{C++ Dialect Options} and
3170 @ref{Objective-C and Objective-C++ Dialect Options}.
3173 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
3174 Some of them warn about constructions that users generally do not
3175 consider questionable, but which occasionally you might wish to check
3176 for; others warn about constructions that are necessary or hard to avoid
3177 in some cases, and there is no simple way to modify the code to suppress
3184 (This option used to be called @option{-W}. The older name is still
3185 supported, but the newer name is more descriptive.) Print extra warning
3186 messages for these events:
3190 Warn if a comparison is always true or always false due to the limited
3191 range of the data type, but do not warn for constant expressions. For
3192 example, warn if an unsigned variable is compared against zero with
3193 @samp{<} or @samp{>=}. This warning can be independently controlled
3194 by @option{-Wtype-limits}.
3197 Storage-class specifiers like @code{static} are not the first things
3198 in a declaration. According to the C Standard, this usage is
3199 obsolescent. This warning can be independently controlled by
3200 @option{-Wold-style-declaration}.
3203 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
3207 A comparison between signed and unsigned values could produce an
3208 incorrect result when the signed value is converted to unsigned.
3209 (But don't warn if @option{-Wno-sign-compare} is also specified.)
3212 An aggregate has an initializer which does not initialize all members.
3213 This warning can be independently controlled by
3214 @option{-Wmissing-field-initializers}.
3217 An initialized field without side effects is overridden when using
3218 designated initializers (@pxref{Designated Inits, , Designated
3219 Initializers}). This warning can be independently controlled by
3220 @option{-Woverride-init}.
3223 A function parameter is declared without a type specifier in K&R-style
3224 functions. This warning can be independently controlled by
3225 @option{-Wmissing-parameter-type}.
3228 An empty body occurs in an @samp{if}, @samp{else} or
3229 @samp{do while} statement. This warning can be independently
3230 controlled by @option{-Wempty-body}.
3232 @item @r{(C++ only)}
3233 An empty body occurs in a @samp{while} or @samp{for} statement with no
3234 whitespacing before the semicolon. This warning can be independently
3235 controlled by @option{-Wempty-body}.
3238 A pointer is compared against integer zero with @samp{<}, @samp{<=},
3239 @samp{>}, or @samp{>=}.
3242 A variable might be changed by @samp{longjmp} or @samp{vfork}.
3243 This warning can be independently controlled by @option{-Wclobbered}.
3245 @item @r{(C++ only)}
3246 An enumerator and a non-enumerator both appear in a conditional expression.
3248 @item @r{(C++ only)}
3249 A non-static reference or non-static @samp{const} member appears in a
3250 class without constructors.
3252 @item @r{(C++ only)}
3253 Ambiguous virtual bases.
3255 @item @r{(C++ only)}
3256 Subscripting an array which has been declared @samp{register}.
3258 @item @r{(C++ only)}
3259 Taking the address of a variable which has been declared @samp{register}.
3261 @item @r{(C++ only)}
3262 A base class is not initialized in a derived class' copy constructor.
3265 @item -Wno-div-by-zero
3266 @opindex Wno-div-by-zero
3267 @opindex Wdiv-by-zero
3268 Do not warn about compile-time integer division by zero. Floating point
3269 division by zero is not warned about, as it can be a legitimate way of
3270 obtaining infinities and NaNs.
3272 @item -Wsystem-headers
3273 @opindex Wsystem-headers
3274 @cindex warnings from system headers
3275 @cindex system headers, warnings from
3276 Print warning messages for constructs found in system header files.
3277 Warnings from system headers are normally suppressed, on the assumption
3278 that they usually do not indicate real problems and would only make the
3279 compiler output harder to read. Using this command line option tells
3280 GCC to emit warnings from system headers as if they occurred in user
3281 code. However, note that using @option{-Wall} in conjunction with this
3282 option will @emph{not} warn about unknown pragmas in system
3283 headers---for that, @option{-Wunknown-pragmas} must also be used.
3286 @opindex Wfloat-equal
3287 Warn if floating point values are used in equality comparisons.
3289 The idea behind this is that sometimes it is convenient (for the
3290 programmer) to consider floating-point values as approximations to
3291 infinitely precise real numbers. If you are doing this, then you need
3292 to compute (by analyzing the code, or in some other way) the maximum or
3293 likely maximum error that the computation introduces, and allow for it
3294 when performing comparisons (and when producing output, but that's a
3295 different problem). In particular, instead of testing for equality, you
3296 would check to see whether the two values have ranges that overlap; and
3297 this is done with the relational operators, so equality comparisons are
3300 @item -Wtraditional @r{(C only)}
3301 @opindex Wtraditional
3302 Warn about certain constructs that behave differently in traditional and
3303 ISO C@. Also warn about ISO C constructs that have no traditional C
3304 equivalent, and/or problematic constructs which should be avoided.
3308 Macro parameters that appear within string literals in the macro body.
3309 In traditional C macro replacement takes place within string literals,
3310 but does not in ISO C@.
3313 In traditional C, some preprocessor directives did not exist.
3314 Traditional preprocessors would only consider a line to be a directive
3315 if the @samp{#} appeared in column 1 on the line. Therefore
3316 @option{-Wtraditional} warns about directives that traditional C
3317 understands but would ignore because the @samp{#} does not appear as the
3318 first character on the line. It also suggests you hide directives like
3319 @samp{#pragma} not understood by traditional C by indenting them. Some
3320 traditional implementations would not recognize @samp{#elif}, so it
3321 suggests avoiding it altogether.
3324 A function-like macro that appears without arguments.
3327 The unary plus operator.
3330 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3331 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3332 constants.) Note, these suffixes appear in macros defined in the system
3333 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3334 Use of these macros in user code might normally lead to spurious
3335 warnings, however GCC's integrated preprocessor has enough context to
3336 avoid warning in these cases.
3339 A function declared external in one block and then used after the end of
3343 A @code{switch} statement has an operand of type @code{long}.
3346 A non-@code{static} function declaration follows a @code{static} one.
3347 This construct is not accepted by some traditional C compilers.
3350 The ISO type of an integer constant has a different width or
3351 signedness from its traditional type. This warning is only issued if
3352 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3353 typically represent bit patterns, are not warned about.
3356 Usage of ISO string concatenation is detected.
3359 Initialization of automatic aggregates.
3362 Identifier conflicts with labels. Traditional C lacks a separate
3363 namespace for labels.
3366 Initialization of unions. If the initializer is zero, the warning is
3367 omitted. This is done under the assumption that the zero initializer in
3368 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3369 initializer warnings and relies on default initialization to zero in the
3373 Conversions by prototypes between fixed/floating point values and vice
3374 versa. The absence of these prototypes when compiling with traditional
3375 C would cause serious problems. This is a subset of the possible
3376 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3379 Use of ISO C style function definitions. This warning intentionally is
3380 @emph{not} issued for prototype declarations or variadic functions
3381 because these ISO C features will appear in your code when using
3382 libiberty's traditional C compatibility macros, @code{PARAMS} and
3383 @code{VPARAMS}. This warning is also bypassed for nested functions
3384 because that feature is already a GCC extension and thus not relevant to
3385 traditional C compatibility.
3388 @item -Wtraditional-conversion @r{(C only)}
3389 @opindex Wtraditional-conversion
3390 Warn if a prototype causes a type conversion that is different from what
3391 would happen to the same argument in the absence of a prototype. This
3392 includes conversions of fixed point to floating and vice versa, and
3393 conversions changing the width or signedness of a fixed point argument
3394 except when the same as the default promotion.
3396 @item -Wdeclaration-after-statement @r{(C only)}
3397 @opindex Wdeclaration-after-statement
3398 Warn when a declaration is found after a statement in a block. This
3399 construct, known from C++, was introduced with ISO C99 and is by default
3400 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3401 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3405 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3407 @item -Wno-endif-labels
3408 @opindex Wno-endif-labels
3409 @opindex Wendif-labels
3410 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3414 Warn whenever a local variable shadows another local variable, parameter or
3415 global variable or whenever a built-in function is shadowed.
3417 @item -Wlarger-than-@var{len}
3418 @opindex Wlarger-than
3419 Warn whenever an object of larger than @var{len} bytes is defined.
3421 @item -Wunsafe-loop-optimizations
3422 @opindex Wunsafe-loop-optimizations
3423 Warn if the loop cannot be optimized because the compiler could not
3424 assume anything on the bounds of the loop indices. With
3425 @option{-funsafe-loop-optimizations} warn if the compiler made
3428 @item -Wpointer-arith
3429 @opindex Wpointer-arith
3430 Warn about anything that depends on the ``size of'' a function type or
3431 of @code{void}. GNU C assigns these types a size of 1, for
3432 convenience in calculations with @code{void *} pointers and pointers
3433 to functions. In C++, warn also when an arithmetic operation involves
3434 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3437 @opindex Wtype-limits
3438 @opindex Wno-type-limits
3439 Warn if a comparison is always true or always false due to the limited
3440 range of the data type, but do not warn for constant expressions. For
3441 example, warn if an unsigned variable is compared against zero with
3442 @samp{<} or @samp{>=}. This warning is also enabled by
3445 @item -Wbad-function-cast @r{(C only)}
3446 @opindex Wbad-function-cast
3447 Warn whenever a function call is cast to a non-matching type.
3448 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3451 Warn about ISO C constructs that are outside of the common subset of
3452 ISO C and ISO C++, e.g.@: request for implicit conversion from
3453 @code{void *} to a pointer to non-@code{void} type.
3455 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3456 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3457 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3458 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3462 Warn whenever a pointer is cast so as to remove a type qualifier from
3463 the target type. For example, warn if a @code{const char *} is cast
3464 to an ordinary @code{char *}.
3467 @opindex Wcast-align
3468 Warn whenever a pointer is cast such that the required alignment of the
3469 target is increased. For example, warn if a @code{char *} is cast to
3470 an @code{int *} on machines where integers can only be accessed at
3471 two- or four-byte boundaries.
3473 @item -Wwrite-strings
3474 @opindex Wwrite-strings
3475 When compiling C, give string constants the type @code{const
3476 char[@var{length}]} so that
3477 copying the address of one into a non-@code{const} @code{char *}
3478 pointer will get a warning; when compiling C++, warn about the
3479 deprecated conversion from string literals to @code{char *}. This
3480 warning, by default, is enabled for C++ programs.
3481 These warnings will help you find at
3482 compile time code that can try to write into a string constant, but
3483 only if you have been very careful about using @code{const} in
3484 declarations and prototypes. Otherwise, it will just be a nuisance;
3485 this is why we did not make @option{-Wall} request these warnings.
3489 Warn for variables that might be changed by @samp{longjmp} or
3490 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3493 @opindex Wconversion
3494 @opindex Wno-conversion
3495 Warn for implicit conversions that may alter a value. This includes
3496 conversions between real and integer, like @code{abs (x)} when
3497 @code{x} is @code{double}; conversions between signed and unsigned,
3498 like @code{unsigned ui = -1}; and conversions to smaller types, like
3499 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3500 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3501 changed by the conversion like in @code{abs (2.0)}. Warnings about
3502 conversions between signed and unsigned integers can be disabled by
3503 using @option{-Wno-sign-conversion}.
3505 For C++, also warn for conversions between @code{NULL} and non-pointer
3506 types; confusing overload resolution for user-defined conversions; and
3507 conversions that will never use a type conversion operator:
3508 conversions to @code{void}, the same type, a base class or a reference
3509 to them. Warnings about conversions between signed and unsigned
3510 integers are disabled by default in C++ unless
3511 @option{-Wsign-conversion} is explicitly enabled.
3514 @opindex Wempty-body
3515 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3516 while} statement. Additionally, in C++, warn when an empty body occurs
3517 in a @samp{while} or @samp{for} statement with no whitespacing before
3518 the semicolon. This warning is also enabled by @option{-Wextra}.
3520 @item -Wsign-compare
3521 @opindex Wsign-compare
3522 @cindex warning for comparison of signed and unsigned values
3523 @cindex comparison of signed and unsigned values, warning
3524 @cindex signed and unsigned values, comparison warning
3525 Warn when a comparison between signed and unsigned values could produce
3526 an incorrect result when the signed value is converted to unsigned.
3527 This warning is also enabled by @option{-Wextra}; to get the other warnings
3528 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3530 @item -Wsign-conversion
3531 @opindex Wsign-conversion
3532 @opindex Wno-sign-conversion
3533 Warn for implicit conversions that may change the sign of an integer
3534 value, like assigning a signed integer expression to an unsigned
3535 integer variable. An explicit cast silences the warning. In C, this
3536 option is enabled also by @option{-Wconversion}.
3540 @opindex Wno-address
3541 Warn about suspicious uses of memory addresses. These include using
3542 the address of a function in a conditional expression, such as
3543 @code{void func(void); if (func)}, and comparisons against the memory
3544 address of a string literal, such as @code{if (x == "abc")}. Such
3545 uses typically indicate a programmer error: the address of a function
3546 always evaluates to true, so their use in a conditional usually
3547 indicate that the programmer forgot the parentheses in a function
3548 call; and comparisons against string literals result in unspecified
3549 behavior and are not portable in C, so they usually indicate that the
3550 programmer intended to use @code{strcmp}. This warning is enabled by
3554 @opindex Wlogical-op
3555 @opindex Wno-logical-op
3556 Warn about suspicious uses of logical operators in expressions.
3557 This includes using logical operators in contexts where a
3558 bit-wise operator is likely to be expected.
3560 @item -Waggregate-return
3561 @opindex Waggregate-return
3562 Warn if any functions that return structures or unions are defined or
3563 called. (In languages where you can return an array, this also elicits
3566 @item -Wno-attributes
3567 @opindex Wno-attributes
3568 @opindex Wattributes
3569 Do not warn if an unexpected @code{__attribute__} is used, such as
3570 unrecognized attributes, function attributes applied to variables,
3571 etc. This will not stop errors for incorrect use of supported
3574 @item -Wstrict-prototypes @r{(C only)}
3575 @opindex Wstrict-prototypes
3576 Warn if a function is declared or defined without specifying the
3577 argument types. (An old-style function definition is permitted without
3578 a warning if preceded by a declaration which specifies the argument
3581 @item -Wold-style-declaration @r{(C only)}
3582 @opindex Wold-style-declaration
3583 Warn for obsolescent usages, according to the C Standard, in a
3584 declaration. For example, warn if storage-class specifiers like
3585 @code{static} are not the first things in a declaration. This warning
3586 is also enabled by @option{-Wextra}.
3588 @item -Wold-style-definition @r{(C only)}
3589 @opindex Wold-style-definition
3590 Warn if an old-style function definition is used. A warning is given
3591 even if there is a previous prototype.
3593 @item -Wmissing-parameter-type @r{(C only)}
3594 @opindex Wmissing-parameter-type
3595 A function parameter is declared without a type specifier in K&R-style
3602 This warning is also enabled by @option{-Wextra}.
3604 @item -Wmissing-prototypes @r{(C only)}
3605 @opindex Wmissing-prototypes
3606 Warn if a global function is defined without a previous prototype
3607 declaration. This warning is issued even if the definition itself
3608 provides a prototype. The aim is to detect global functions that fail
3609 to be declared in header files.
3611 @item -Wmissing-declarations @r{(C and C++ only)}
3612 @opindex Wmissing-declarations
3613 Warn if a global function is defined without a previous declaration.
3614 Do so even if the definition itself provides a prototype.
3615 Use this option to detect global functions that are not declared in
3616 header files. In C++, no warnings are issued for function templates,
3617 or for inline functions, or for functions in anonymous namespaces.
3619 @item -Wmissing-field-initializers
3620 @opindex Wmissing-field-initializers
3623 Warn if a structure's initializer has some fields missing. For
3624 example, the following code would cause such a warning, because
3625 @code{x.h} is implicitly zero:
3628 struct s @{ int f, g, h; @};
3629 struct s x = @{ 3, 4 @};
3632 This option does not warn about designated initializers, so the following
3633 modification would not trigger a warning:
3636 struct s @{ int f, g, h; @};
3637 struct s x = @{ .f = 3, .g = 4 @};
3640 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3641 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3643 @item -Wmissing-noreturn
3644 @opindex Wmissing-noreturn
3645 Warn about functions which might be candidates for attribute @code{noreturn}.
3646 Note these are only possible candidates, not absolute ones. Care should
3647 be taken to manually verify functions actually do not ever return before
3648 adding the @code{noreturn} attribute, otherwise subtle code generation
3649 bugs could be introduced. You will not get a warning for @code{main} in
3650 hosted C environments.
3652 @item -Wmissing-format-attribute
3653 @opindex Wmissing-format-attribute
3655 Warn about function pointers which might be candidates for @code{format}
3656 attributes. Note these are only possible candidates, not absolute ones.
3657 GCC will guess that function pointers with @code{format} attributes that
3658 are used in assignment, initialization, parameter passing or return
3659 statements should have a corresponding @code{format} attribute in the
3660 resulting type. I.e.@: the left-hand side of the assignment or
3661 initialization, the type of the parameter variable, or the return type
3662 of the containing function respectively should also have a @code{format}
3663 attribute to avoid the warning.
3665 GCC will also warn about function definitions which might be
3666 candidates for @code{format} attributes. Again, these are only
3667 possible candidates. GCC will guess that @code{format} attributes
3668 might be appropriate for any function that calls a function like
3669 @code{vprintf} or @code{vscanf}, but this might not always be the
3670 case, and some functions for which @code{format} attributes are
3671 appropriate may not be detected.
3673 @item -Wno-multichar
3674 @opindex Wno-multichar
3676 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3677 Usually they indicate a typo in the user's code, as they have
3678 implementation-defined values, and should not be used in portable code.
3680 @item -Wnormalized=<none|id|nfc|nfkc>
3681 @opindex Wnormalized
3684 @cindex character set, input normalization
3685 In ISO C and ISO C++, two identifiers are different if they are
3686 different sequences of characters. However, sometimes when characters
3687 outside the basic ASCII character set are used, you can have two
3688 different character sequences that look the same. To avoid confusion,
3689 the ISO 10646 standard sets out some @dfn{normalization rules} which
3690 when applied ensure that two sequences that look the same are turned into
3691 the same sequence. GCC can warn you if you are using identifiers which
3692 have not been normalized; this option controls that warning.
3694 There are four levels of warning that GCC supports. The default is
3695 @option{-Wnormalized=nfc}, which warns about any identifier which is
3696 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3697 recommended form for most uses.
3699 Unfortunately, there are some characters which ISO C and ISO C++ allow
3700 in identifiers that when turned into NFC aren't allowable as
3701 identifiers. That is, there's no way to use these symbols in portable
3702 ISO C or C++ and have all your identifiers in NFC.
3703 @option{-Wnormalized=id} suppresses the warning for these characters.
3704 It is hoped that future versions of the standards involved will correct
3705 this, which is why this option is not the default.
3707 You can switch the warning off for all characters by writing
3708 @option{-Wnormalized=none}. You would only want to do this if you
3709 were using some other normalization scheme (like ``D''), because
3710 otherwise you can easily create bugs that are literally impossible to see.
3712 Some characters in ISO 10646 have distinct meanings but look identical
3713 in some fonts or display methodologies, especially once formatting has
3714 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3715 LETTER N'', will display just like a regular @code{n} which has been
3716 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3717 normalization scheme to convert all these into a standard form as
3718 well, and GCC will warn if your code is not in NFKC if you use
3719 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3720 about every identifier that contains the letter O because it might be
3721 confused with the digit 0, and so is not the default, but may be
3722 useful as a local coding convention if the programming environment is
3723 unable to be fixed to display these characters distinctly.
3725 @item -Wno-deprecated-declarations
3726 @opindex Wno-deprecated-declarations
3727 Do not warn about uses of functions (@pxref{Function Attributes}),
3728 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3729 Attributes}) marked as deprecated by using the @code{deprecated}
3733 @opindex Wno-overflow
3734 Do not warn about compile-time overflow in constant expressions.
3736 @item -Woverride-init
3737 @opindex Woverride-init
3740 Warn if an initialized field without side effects is overridden when
3741 using designated initializers (@pxref{Designated Inits, , Designated
3744 This warning is included in @option{-Wextra}. To get other
3745 @option{-Wextra} warnings without this one, use @samp{-Wextra
3746 -Wno-override-init}.
3750 Warn if a structure is given the packed attribute, but the packed
3751 attribute has no effect on the layout or size of the structure.
3752 Such structures may be mis-aligned for little benefit. For
3753 instance, in this code, the variable @code{f.x} in @code{struct bar}
3754 will be misaligned even though @code{struct bar} does not itself
3755 have the packed attribute:
3762 @} __attribute__((packed));
3772 Warn if padding is included in a structure, either to align an element
3773 of the structure or to align the whole structure. Sometimes when this
3774 happens it is possible to rearrange the fields of the structure to
3775 reduce the padding and so make the structure smaller.
3777 @item -Wredundant-decls
3778 @opindex Wredundant-decls
3779 Warn if anything is declared more than once in the same scope, even in
3780 cases where multiple declaration is valid and changes nothing.
3782 @item -Wnested-externs @r{(C only)}
3783 @opindex Wnested-externs
3784 Warn if an @code{extern} declaration is encountered within a function.
3786 @item -Wunreachable-code
3787 @opindex Wunreachable-code
3788 Warn if the compiler detects that code will never be executed.
3790 This option is intended to warn when the compiler detects that at
3791 least a whole line of source code will never be executed, because
3792 some condition is never satisfied or because it is after a
3793 procedure that never returns.
3795 It is possible for this option to produce a warning even though there
3796 are circumstances under which part of the affected line can be executed,
3797 so care should be taken when removing apparently-unreachable code.
3799 For instance, when a function is inlined, a warning may mean that the
3800 line is unreachable in only one inlined copy of the function.
3802 This option is not made part of @option{-Wall} because in a debugging
3803 version of a program there is often substantial code which checks
3804 correct functioning of the program and is, hopefully, unreachable
3805 because the program does work. Another common use of unreachable
3806 code is to provide behavior which is selectable at compile-time.
3810 Warn if a function can not be inlined and it was declared as inline.
3811 Even with this option, the compiler will not warn about failures to
3812 inline functions declared in system headers.
3814 The compiler uses a variety of heuristics to determine whether or not
3815 to inline a function. For example, the compiler takes into account
3816 the size of the function being inlined and the amount of inlining
3817 that has already been done in the current function. Therefore,
3818 seemingly insignificant changes in the source program can cause the
3819 warnings produced by @option{-Winline} to appear or disappear.
3821 @item -Wno-invalid-offsetof @r{(C++ only)}
3822 @opindex Wno-invalid-offsetof
3823 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3824 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3825 to a non-POD type is undefined. In existing C++ implementations,
3826 however, @samp{offsetof} typically gives meaningful results even when
3827 applied to certain kinds of non-POD types. (Such as a simple
3828 @samp{struct} that fails to be a POD type only by virtue of having a
3829 constructor.) This flag is for users who are aware that they are
3830 writing nonportable code and who have deliberately chosen to ignore the
3833 The restrictions on @samp{offsetof} may be relaxed in a future version
3834 of the C++ standard.
3836 @item -Wno-int-to-pointer-cast @r{(C only)}
3837 @opindex Wno-int-to-pointer-cast
3838 Suppress warnings from casts to pointer type of an integer of a
3841 @item -Wno-pointer-to-int-cast @r{(C only)}
3842 @opindex Wno-pointer-to-int-cast
3843 Suppress warnings from casts from a pointer to an integer type of a
3847 @opindex Winvalid-pch
3848 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3849 the search path but can't be used.
3853 @opindex Wno-long-long
3854 Warn if @samp{long long} type is used. This is default. To inhibit
3855 the warning messages, use @option{-Wno-long-long}. Flags
3856 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3857 only when @option{-pedantic} flag is used.
3859 @item -Wvariadic-macros
3860 @opindex Wvariadic-macros
3861 @opindex Wno-variadic-macros
3862 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3863 alternate syntax when in pedantic ISO C99 mode. This is default.
3864 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3869 Warn if variable length array is used in the code.
3870 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
3871 the variable length array.
3873 @item -Wvolatile-register-var
3874 @opindex Wvolatile-register-var
3875 @opindex Wno-volatile-register-var
3876 Warn if a register variable is declared volatile. The volatile
3877 modifier does not inhibit all optimizations that may eliminate reads
3878 and/or writes to register variables.
3880 @item -Wdisabled-optimization
3881 @opindex Wdisabled-optimization
3882 Warn if a requested optimization pass is disabled. This warning does
3883 not generally indicate that there is anything wrong with your code; it
3884 merely indicates that GCC's optimizers were unable to handle the code
3885 effectively. Often, the problem is that your code is too big or too
3886 complex; GCC will refuse to optimize programs when the optimization
3887 itself is likely to take inordinate amounts of time.
3889 @item -Wpointer-sign
3890 @opindex Wpointer-sign
3891 @opindex Wno-pointer-sign
3892 Warn for pointer argument passing or assignment with different signedness.
3893 This option is only supported for C and Objective-C@. It is implied by
3894 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3895 @option{-Wno-pointer-sign}.
3899 Make all warnings into errors.
3903 Make the specified warning into an errors. The specifier for a
3904 warning is appended, for example @option{-Werror=switch} turns the
3905 warnings controlled by @option{-Wswitch} into errors. This switch
3906 takes a negative form, to be used to negate @option{-Werror} for
3907 specific warnings, for example @option{-Wno-error=switch} makes
3908 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3909 is in effect. You can use the @option{-fdiagnostics-show-option}
3910 option to have each controllable warning amended with the option which
3911 controls it, to determine what to use with this option.
3913 Note that specifying @option{-Werror=}@var{foo} automatically implies
3914 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3917 @item -Wstack-protector
3918 @opindex Wstack-protector
3919 This option is only active when @option{-fstack-protector} is active. It
3920 warns about functions that will not be protected against stack smashing.
3922 @item -Woverlength-strings
3923 @opindex Woverlength-strings
3924 Warn about string constants which are longer than the ``minimum
3925 maximum'' length specified in the C standard. Modern compilers
3926 generally allow string constants which are much longer than the
3927 standard's minimum limit, but very portable programs should avoid
3928 using longer strings.
3930 The limit applies @emph{after} string constant concatenation, and does
3931 not count the trailing NUL@. In C89, the limit was 509 characters; in
3932 C99, it was raised to 4095. C++98 does not specify a normative
3933 minimum maximum, so we do not diagnose overlength strings in C++@.
3935 This option is implied by @option{-pedantic}, and can be disabled with
3936 @option{-Wno-overlength-strings}.
3939 @node Debugging Options
3940 @section Options for Debugging Your Program or GCC
3941 @cindex options, debugging
3942 @cindex debugging information options
3944 GCC has various special options that are used for debugging
3945 either your program or GCC:
3950 Produce debugging information in the operating system's native format
3951 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3954 On most systems that use stabs format, @option{-g} enables use of extra
3955 debugging information that only GDB can use; this extra information
3956 makes debugging work better in GDB but will probably make other debuggers
3958 refuse to read the program. If you want to control for certain whether
3959 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3960 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3962 GCC allows you to use @option{-g} with
3963 @option{-O}. The shortcuts taken by optimized code may occasionally
3964 produce surprising results: some variables you declared may not exist
3965 at all; flow of control may briefly move where you did not expect it;
3966 some statements may not be executed because they compute constant
3967 results or their values were already at hand; some statements may
3968 execute in different places because they were moved out of loops.
3970 Nevertheless it proves possible to debug optimized output. This makes
3971 it reasonable to use the optimizer for programs that might have bugs.
3973 The following options are useful when GCC is generated with the
3974 capability for more than one debugging format.
3978 Produce debugging information for use by GDB@. This means to use the
3979 most expressive format available (DWARF 2, stabs, or the native format
3980 if neither of those are supported), including GDB extensions if at all
3985 Produce debugging information in stabs format (if that is supported),
3986 without GDB extensions. This is the format used by DBX on most BSD
3987 systems. On MIPS, Alpha and System V Release 4 systems this option
3988 produces stabs debugging output which is not understood by DBX or SDB@.
3989 On System V Release 4 systems this option requires the GNU assembler.
3991 @item -feliminate-unused-debug-symbols
3992 @opindex feliminate-unused-debug-symbols
3993 Produce debugging information in stabs format (if that is supported),
3994 for only symbols that are actually used.
3996 @item -femit-class-debug-always
3997 Instead of emitting debugging information for a C++ class in only one
3998 object file, emit it in all object files using the class. This option
3999 should be used only with debuggers that are unable to handle the way GCC
4000 normally emits debugging information for classes because using this
4001 option will increase the size of debugging information by as much as a
4006 Produce debugging information in stabs format (if that is supported),
4007 using GNU extensions understood only by the GNU debugger (GDB)@. The
4008 use of these extensions is likely to make other debuggers crash or
4009 refuse to read the program.
4013 Produce debugging information in COFF format (if that is supported).
4014 This is the format used by SDB on most System V systems prior to
4019 Produce debugging information in XCOFF format (if that is supported).
4020 This is the format used by the DBX debugger on IBM RS/6000 systems.
4024 Produce debugging information in XCOFF format (if that is supported),
4025 using GNU extensions understood only by the GNU debugger (GDB)@. The
4026 use of these extensions is likely to make other debuggers crash or
4027 refuse to read the program, and may cause assemblers other than the GNU
4028 assembler (GAS) to fail with an error.
4032 Produce debugging information in DWARF version 2 format (if that is
4033 supported). This is the format used by DBX on IRIX 6. With this
4034 option, GCC uses features of DWARF version 3 when they are useful;
4035 version 3 is upward compatible with version 2, but may still cause
4036 problems for older debuggers.
4040 Produce debugging information in VMS debug format (if that is
4041 supported). This is the format used by DEBUG on VMS systems.
4044 @itemx -ggdb@var{level}
4045 @itemx -gstabs@var{level}
4046 @itemx -gcoff@var{level}
4047 @itemx -gxcoff@var{level}
4048 @itemx -gvms@var{level}
4049 Request debugging information and also use @var{level} to specify how
4050 much information. The default level is 2.
4052 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4055 Level 1 produces minimal information, enough for making backtraces in
4056 parts of the program that you don't plan to debug. This includes
4057 descriptions of functions and external variables, but no information
4058 about local variables and no line numbers.
4060 Level 3 includes extra information, such as all the macro definitions
4061 present in the program. Some debuggers support macro expansion when
4062 you use @option{-g3}.
4064 @option{-gdwarf-2} does not accept a concatenated debug level, because
4065 GCC used to support an option @option{-gdwarf} that meant to generate
4066 debug information in version 1 of the DWARF format (which is very
4067 different from version 2), and it would have been too confusing. That
4068 debug format is long obsolete, but the option cannot be changed now.
4069 Instead use an additional @option{-g@var{level}} option to change the
4070 debug level for DWARF2.
4072 @item -feliminate-dwarf2-dups
4073 @opindex feliminate-dwarf2-dups
4074 Compress DWARF2 debugging information by eliminating duplicated
4075 information about each symbol. This option only makes sense when
4076 generating DWARF2 debugging information with @option{-gdwarf-2}.
4078 @item -femit-struct-debug-baseonly
4079 Emit debug information for struct-like types
4080 only when the base name of the compilation source file
4081 matches the base name of file in which the struct was defined.
4083 This option substantially reduces the size of debugging information,
4084 but at significant potential loss in type information to the debugger.
4085 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4086 See @option{-femit-struct-debug-detailed} for more detailed control.
4088 This option works only with DWARF 2.
4090 @item -femit-struct-debug-reduced
4091 Emit debug information for struct-like types
4092 only when the base name of the compilation source file
4093 matches the base name of file in which the type was defined,
4094 unless the struct is a template or defined in a system header.
4096 This option significantly reduces the size of debugging information,
4097 with some potential loss in type information to the debugger.
4098 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4099 See @option{-femit-struct-debug-detailed} for more detailed control.
4101 This option works only with DWARF 2.
4103 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4104 Specify the struct-like types
4105 for which the compiler will generate debug information.
4106 The intent is to reduce duplicate struct debug information
4107 between different object files within the same program.
4109 This option is a detailed version of
4110 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4111 which will serve for most needs.
4113 A specification has the syntax
4114 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4116 The optional first word limits the specification to
4117 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4118 A struct type is used directly when it is the type of a variable, member.
4119 Indirect uses arise through pointers to structs.
4120 That is, when use of an incomplete struct would be legal, the use is indirect.
4122 @samp{struct one direct; struct two * indirect;}.
4124 The optional second word limits the specification to
4125 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4126 Generic structs are a bit complicated to explain.
4127 For C++, these are non-explicit specializations of template classes,
4128 or non-template classes within the above.
4129 Other programming languages have generics,
4130 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4132 The third word specifies the source files for those
4133 structs for which the compiler will emit debug information.
4134 The values @samp{none} and @samp{any} have the normal meaning.
4135 The value @samp{base} means that
4136 the base of name of the file in which the type declaration appears
4137 must match the base of the name of the main compilation file.
4138 In practice, this means that
4139 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4140 but types declared in other header will not.
4141 The value @samp{sys} means those types satisfying @samp{base}
4142 or declared in system or compiler headers.
4144 You may need to experiment to determine the best settings for your application.
4146 The default is @samp{-femit-struct-debug-detailed=all}.
4148 This option works only with DWARF 2.
4150 @item -fdebug-prefix-map=@var{old}=@var{new}
4151 @opindex fdebug-prefix-map
4152 When compiling files in directory @file{@var{old}}, record debugging
4153 information describing them as in @file{@var{new}} instead.
4155 @cindex @command{prof}
4158 Generate extra code to write profile information suitable for the
4159 analysis program @command{prof}. You must use this option when compiling
4160 the source files you want data about, and you must also use it when
4163 @cindex @command{gprof}
4166 Generate extra code to write profile information suitable for the
4167 analysis program @command{gprof}. You must use this option when compiling
4168 the source files you want data about, and you must also use it when
4173 Makes the compiler print out each function name as it is compiled, and
4174 print some statistics about each pass when it finishes.
4177 @opindex ftime-report
4178 Makes the compiler print some statistics about the time consumed by each
4179 pass when it finishes.
4182 @opindex fmem-report
4183 Makes the compiler print some statistics about permanent memory
4184 allocation when it finishes.
4186 @item -fpre-ipa-mem-report
4187 @opindex fpre-ipa-mem-report
4188 @item -fpost-ipa-mem-report
4189 @opindex fpost-ipa-mem-report
4190 Makes the compiler print some statistics about permanent memory
4191 allocation before or after interprocedural optimization.
4193 @item -fprofile-arcs
4194 @opindex fprofile-arcs
4195 Add code so that program flow @dfn{arcs} are instrumented. During
4196 execution the program records how many times each branch and call is
4197 executed and how many times it is taken or returns. When the compiled
4198 program exits it saves this data to a file called
4199 @file{@var{auxname}.gcda} for each source file. The data may be used for
4200 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4201 test coverage analysis (@option{-ftest-coverage}). Each object file's
4202 @var{auxname} is generated from the name of the output file, if
4203 explicitly specified and it is not the final executable, otherwise it is
4204 the basename of the source file. In both cases any suffix is removed
4205 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4206 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4207 @xref{Cross-profiling}.
4209 @cindex @command{gcov}
4213 This option is used to compile and link code instrumented for coverage
4214 analysis. The option is a synonym for @option{-fprofile-arcs}
4215 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4216 linking). See the documentation for those options for more details.
4221 Compile the source files with @option{-fprofile-arcs} plus optimization
4222 and code generation options. For test coverage analysis, use the
4223 additional @option{-ftest-coverage} option. You do not need to profile
4224 every source file in a program.
4227 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4228 (the latter implies the former).
4231 Run the program on a representative workload to generate the arc profile
4232 information. This may be repeated any number of times. You can run
4233 concurrent instances of your program, and provided that the file system
4234 supports locking, the data files will be correctly updated. Also
4235 @code{fork} calls are detected and correctly handled (double counting
4239 For profile-directed optimizations, compile the source files again with
4240 the same optimization and code generation options plus
4241 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4242 Control Optimization}).
4245 For test coverage analysis, use @command{gcov} to produce human readable
4246 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4247 @command{gcov} documentation for further information.
4251 With @option{-fprofile-arcs}, for each function of your program GCC
4252 creates a program flow graph, then finds a spanning tree for the graph.
4253 Only arcs that are not on the spanning tree have to be instrumented: the
4254 compiler adds code to count the number of times that these arcs are
4255 executed. When an arc is the only exit or only entrance to a block, the
4256 instrumentation code can be added to the block; otherwise, a new basic
4257 block must be created to hold the instrumentation code.
4260 @item -ftest-coverage
4261 @opindex ftest-coverage
4262 Produce a notes file that the @command{gcov} code-coverage utility
4263 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4264 show program coverage. Each source file's note file is called
4265 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4266 above for a description of @var{auxname} and instructions on how to
4267 generate test coverage data. Coverage data will match the source files
4268 more closely, if you do not optimize.
4270 @item -fdbg-cnt-list
4271 @opindex fdbg-cnt-list
4272 Print the name and the counter upperbound for all debug counters.
4274 @item -fdbg-cnt=@var{counter-value-list}
4276 Set the internal debug counter upperbound. @var{counter-value-list}
4277 is a comma-separated list of @var{name}:@var{value} pairs
4278 which sets the upperbound of each debug counter @var{name} to @var{value}.
4279 All debug counters have the initial upperbound of @var{UINT_MAX},
4280 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4281 e.g. With -fdbg-cnt=dce:10,tail_call:0
4282 dbg_cnt(dce) will return true only for first 10 invocations
4283 and dbg_cnt(tail_call) will return false always.
4285 @item -d@var{letters}
4286 @item -fdump-rtl-@var{pass}
4288 Says to make debugging dumps during compilation at times specified by
4289 @var{letters}. This is used for debugging the RTL-based passes of the
4290 compiler. The file names for most of the dumps are made by appending a
4291 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4292 from the name of the output file, if explicitly specified and it is not
4293 an executable, otherwise it is the basename of the source file. These
4294 switches may have different effects when @option{-E} is used for
4297 Most debug dumps can be enabled either passing a letter to the @option{-d}
4298 option, or with a long @option{-fdump-rtl} switch; here are the possible
4299 letters for use in @var{letters} and @var{pass}, and their meanings:
4304 Annotate the assembler output with miscellaneous debugging information.
4307 @itemx -fdump-rtl-bbro
4309 @opindex fdump-rtl-bbro
4310 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4313 @itemx -fdump-rtl-combine
4315 @opindex fdump-rtl-combine
4316 Dump after the RTL instruction combination pass, to the file
4317 @file{@var{file}.129r.combine}.
4320 @itemx -fdump-rtl-ce1
4321 @itemx -fdump-rtl-ce2
4323 @opindex fdump-rtl-ce1
4324 @opindex fdump-rtl-ce2
4325 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4326 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4327 and @option{-fdump-rtl-ce2} enable dumping after the second if
4328 conversion, to the file @file{@var{file}.130r.ce2}.
4331 @itemx -fdump-rtl-btl
4332 @itemx -fdump-rtl-dbr
4334 @opindex fdump-rtl-btl
4335 @opindex fdump-rtl-dbr
4336 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4337 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4338 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4339 scheduling, to @file{@var{file}.36.dbr}.
4343 Dump all macro definitions, at the end of preprocessing, in addition to
4347 @itemx -fdump-rtl-ce3
4349 @opindex fdump-rtl-ce3
4350 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4353 @itemx -fdump-rtl-cfg
4354 @itemx -fdump-rtl-life
4356 @opindex fdump-rtl-cfg
4357 @opindex fdump-rtl-life
4358 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4359 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4360 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4361 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4364 @itemx -fdump-rtl-greg
4366 @opindex fdump-rtl-greg
4367 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4370 @itemx -fdump-rtl-gcse
4371 @itemx -fdump-rtl-bypass
4373 @opindex fdump-rtl-gcse
4374 @opindex fdump-rtl-bypass
4375 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4376 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4377 enable dumping after jump bypassing and control flow optimizations, to
4378 @file{@var{file}.115r.bypass}.
4381 @itemx -fdump-rtl-eh
4383 @opindex fdump-rtl-eh
4384 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4387 @itemx -fdump-rtl-sibling
4389 @opindex fdump-rtl-sibling
4390 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4393 @itemx -fdump-rtl-jump
4395 @opindex fdump-rtl-jump
4396 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4399 @itemx -fdump-rtl-stack
4401 @opindex fdump-rtl-stack
4402 Dump after conversion from GCC's "flat register file" registers to the
4403 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4406 @itemx -fdump-rtl-lreg
4408 @opindex fdump-rtl-lreg
4409 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4412 @itemx -fdump-rtl-loop2
4414 @opindex fdump-rtl-loop2
4415 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4416 loop optimization pass, to @file{@var{file}.119r.loop2},
4417 @file{@var{file}.120r.loop2_init},
4418 @file{@var{file}.121r.loop2_invariant}, and
4419 @file{@var{file}.125r.loop2_done}.
4422 @itemx -fdump-rtl-sms
4424 @opindex fdump-rtl-sms
4425 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4428 @itemx -fdump-rtl-mach
4430 @opindex fdump-rtl-mach
4431 Dump after performing the machine dependent reorganization pass, to
4432 @file{@var{file}.155r.mach} if that pass exists.
4435 @itemx -fdump-rtl-rnreg
4437 @opindex fdump-rtl-rnreg
4438 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4441 @itemx -fdump-rtl-regmove
4443 @opindex fdump-rtl-regmove
4444 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4447 @itemx -fdump-rtl-postreload
4449 @opindex fdump-rtl-postreload
4450 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4453 @itemx -fdump-rtl-expand
4455 @opindex fdump-rtl-expand
4456 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4459 @itemx -fdump-rtl-sched2
4461 @opindex fdump-rtl-sched2
4462 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4465 @itemx -fdump-rtl-cse
4467 @opindex fdump-rtl-cse
4468 Dump after CSE (including the jump optimization that sometimes follows
4469 CSE), to @file{@var{file}.113r.cse}.
4472 @itemx -fdump-rtl-sched1
4474 @opindex fdump-rtl-sched1
4475 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4478 @itemx -fdump-rtl-cse2
4480 @opindex fdump-rtl-cse2
4481 Dump after the second CSE pass (including the jump optimization that
4482 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4485 @itemx -fdump-rtl-tracer
4487 @opindex fdump-rtl-tracer
4488 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4491 @itemx -fdump-rtl-vpt
4492 @itemx -fdump-rtl-vartrack
4494 @opindex fdump-rtl-vpt
4495 @opindex fdump-rtl-vartrack
4496 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4497 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4498 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4499 to @file{@var{file}.154r.vartrack}.
4502 @itemx -fdump-rtl-flow2
4504 @opindex fdump-rtl-flow2
4505 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4508 @itemx -fdump-rtl-peephole2
4510 @opindex fdump-rtl-peephole2
4511 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4514 @itemx -fdump-rtl-web
4516 @opindex fdump-rtl-web
4517 Dump after live range splitting, to @file{@var{file}.126r.web}.
4520 @itemx -fdump-rtl-all
4522 @opindex fdump-rtl-all
4523 Produce all the dumps listed above.
4527 Produce a core dump whenever an error occurs.
4531 Print statistics on memory usage, at the end of the run, to
4536 Annotate the assembler output with a comment indicating which
4537 pattern and alternative was used. The length of each instruction is
4542 Dump the RTL in the assembler output as a comment before each instruction.
4543 Also turns on @option{-dp} annotation.
4547 For each of the other indicated dump files (either with @option{-d} or
4548 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4549 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4553 Just generate RTL for a function instead of compiling it. Usually used
4554 with @samp{r} (@option{-fdump-rtl-expand}).
4558 Dump debugging information during parsing, to standard error.
4562 @opindex fdump-noaddr
4563 When doing debugging dumps (see @option{-d} option above), suppress
4564 address output. This makes it more feasible to use diff on debugging
4565 dumps for compiler invocations with different compiler binaries and/or
4566 different text / bss / data / heap / stack / dso start locations.
4568 @item -fdump-unnumbered
4569 @opindex fdump-unnumbered
4570 When doing debugging dumps (see @option{-d} option above), suppress instruction
4571 numbers and address output. This makes it more feasible to
4572 use diff on debugging dumps for compiler invocations with different
4573 options, in particular with and without @option{-g}.
4575 @item -fdump-translation-unit @r{(C++ only)}
4576 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4577 @opindex fdump-translation-unit
4578 Dump a representation of the tree structure for the entire translation
4579 unit to a file. The file name is made by appending @file{.tu} to the
4580 source file name. If the @samp{-@var{options}} form is used, @var{options}
4581 controls the details of the dump as described for the
4582 @option{-fdump-tree} options.
4584 @item -fdump-class-hierarchy @r{(C++ only)}
4585 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4586 @opindex fdump-class-hierarchy
4587 Dump a representation of each class's hierarchy and virtual function
4588 table layout to a file. The file name is made by appending @file{.class}
4589 to the source file name. If the @samp{-@var{options}} form is used,
4590 @var{options} controls the details of the dump as described for the
4591 @option{-fdump-tree} options.
4593 @item -fdump-ipa-@var{switch}
4595 Control the dumping at various stages of inter-procedural analysis
4596 language tree to a file. The file name is generated by appending a switch
4597 specific suffix to the source file name. The following dumps are possible:
4601 Enables all inter-procedural analysis dumps; currently the only produced
4602 dump is the @samp{cgraph} dump.
4605 Dumps information about call-graph optimization, unused function removal,
4606 and inlining decisions.
4609 @item -fdump-tree-@var{switch}
4610 @itemx -fdump-tree-@var{switch}-@var{options}
4612 Control the dumping at various stages of processing the intermediate
4613 language tree to a file. The file name is generated by appending a switch
4614 specific suffix to the source file name. If the @samp{-@var{options}}
4615 form is used, @var{options} is a list of @samp{-} separated options that
4616 control the details of the dump. Not all options are applicable to all
4617 dumps, those which are not meaningful will be ignored. The following
4618 options are available
4622 Print the address of each node. Usually this is not meaningful as it
4623 changes according to the environment and source file. Its primary use
4624 is for tying up a dump file with a debug environment.
4626 Inhibit dumping of members of a scope or body of a function merely
4627 because that scope has been reached. Only dump such items when they
4628 are directly reachable by some other path. When dumping pretty-printed
4629 trees, this option inhibits dumping the bodies of control structures.
4631 Print a raw representation of the tree. By default, trees are
4632 pretty-printed into a C-like representation.
4634 Enable more detailed dumps (not honored by every dump option).
4636 Enable dumping various statistics about the pass (not honored by every dump
4639 Enable showing basic block boundaries (disabled in raw dumps).
4641 Enable showing virtual operands for every statement.
4643 Enable showing line numbers for statements.
4645 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4647 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4650 The following tree dumps are possible:
4654 Dump before any tree based optimization, to @file{@var{file}.original}.
4657 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4660 Dump after function inlining, to @file{@var{file}.inlined}.
4663 @opindex fdump-tree-gimple
4664 Dump each function before and after the gimplification pass to a file. The
4665 file name is made by appending @file{.gimple} to the source file name.
4668 @opindex fdump-tree-cfg
4669 Dump the control flow graph of each function to a file. The file name is
4670 made by appending @file{.cfg} to the source file name.
4673 @opindex fdump-tree-vcg
4674 Dump the control flow graph of each function to a file in VCG format. The
4675 file name is made by appending @file{.vcg} to the source file name. Note
4676 that if the file contains more than one function, the generated file cannot
4677 be used directly by VCG@. You will need to cut and paste each function's
4678 graph into its own separate file first.
4681 @opindex fdump-tree-ch
4682 Dump each function after copying loop headers. The file name is made by
4683 appending @file{.ch} to the source file name.
4686 @opindex fdump-tree-ssa
4687 Dump SSA related information to a file. The file name is made by appending
4688 @file{.ssa} to the source file name.
4691 @opindex fdump-tree-salias
4692 Dump structure aliasing variable information to a file. This file name
4693 is made by appending @file{.salias} to the source file name.
4696 @opindex fdump-tree-alias
4697 Dump aliasing information for each function. The file name is made by
4698 appending @file{.alias} to the source file name.
4701 @opindex fdump-tree-ccp
4702 Dump each function after CCP@. The file name is made by appending
4703 @file{.ccp} to the source file name.
4706 @opindex fdump-tree-storeccp
4707 Dump each function after STORE-CCP. The file name is made by appending
4708 @file{.storeccp} to the source file name.
4711 @opindex fdump-tree-pre
4712 Dump trees after partial redundancy elimination. The file name is made
4713 by appending @file{.pre} to the source file name.
4716 @opindex fdump-tree-fre
4717 Dump trees after full redundancy elimination. The file name is made
4718 by appending @file{.fre} to the source file name.
4721 @opindex fdump-tree-copyprop
4722 Dump trees after copy propagation. The file name is made
4723 by appending @file{.copyprop} to the source file name.
4725 @item store_copyprop
4726 @opindex fdump-tree-store_copyprop
4727 Dump trees after store copy-propagation. The file name is made
4728 by appending @file{.store_copyprop} to the source file name.
4731 @opindex fdump-tree-dce
4732 Dump each function after dead code elimination. The file name is made by
4733 appending @file{.dce} to the source file name.
4736 @opindex fdump-tree-mudflap
4737 Dump each function after adding mudflap instrumentation. The file name is
4738 made by appending @file{.mudflap} to the source file name.
4741 @opindex fdump-tree-sra
4742 Dump each function after performing scalar replacement of aggregates. The
4743 file name is made by appending @file{.sra} to the source file name.
4746 @opindex fdump-tree-sink
4747 Dump each function after performing code sinking. The file name is made
4748 by appending @file{.sink} to the source file name.
4751 @opindex fdump-tree-dom
4752 Dump each function after applying dominator tree optimizations. The file
4753 name is made by appending @file{.dom} to the source file name.
4756 @opindex fdump-tree-dse
4757 Dump each function after applying dead store elimination. The file
4758 name is made by appending @file{.dse} to the source file name.
4761 @opindex fdump-tree-phiopt
4762 Dump each function after optimizing PHI nodes into straightline code. The file
4763 name is made by appending @file{.phiopt} to the source file name.
4766 @opindex fdump-tree-forwprop
4767 Dump each function after forward propagating single use variables. The file
4768 name is made by appending @file{.forwprop} to the source file name.
4771 @opindex fdump-tree-copyrename
4772 Dump each function after applying the copy rename optimization. The file
4773 name is made by appending @file{.copyrename} to the source file name.
4776 @opindex fdump-tree-nrv
4777 Dump each function after applying the named return value optimization on
4778 generic trees. The file name is made by appending @file{.nrv} to the source
4782 @opindex fdump-tree-vect
4783 Dump each function after applying vectorization of loops. The file name is
4784 made by appending @file{.vect} to the source file name.
4787 @opindex fdump-tree-vrp
4788 Dump each function after Value Range Propagation (VRP). The file name
4789 is made by appending @file{.vrp} to the source file name.
4792 @opindex fdump-tree-all
4793 Enable all the available tree dumps with the flags provided in this option.
4796 @item -ftree-vectorizer-verbose=@var{n}
4797 @opindex ftree-vectorizer-verbose
4798 This option controls the amount of debugging output the vectorizer prints.
4799 This information is written to standard error, unless
4800 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4801 in which case it is output to the usual dump listing file, @file{.vect}.
4802 For @var{n}=0 no diagnostic information is reported.
4803 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4804 and the total number of loops that got vectorized.
4805 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4806 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4807 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4808 level that @option{-fdump-tree-vect-stats} uses.
4809 Higher verbosity levels mean either more information dumped for each
4810 reported loop, or same amount of information reported for more loops:
4811 If @var{n}=3, alignment related information is added to the reports.
4812 If @var{n}=4, data-references related information (e.g. memory dependences,
4813 memory access-patterns) is added to the reports.
4814 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4815 that did not pass the first analysis phase (i.e. may not be countable, or
4816 may have complicated control-flow).
4817 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4818 For @var{n}=7, all the information the vectorizer generates during its
4819 analysis and transformation is reported. This is the same verbosity level
4820 that @option{-fdump-tree-vect-details} uses.
4822 @item -frandom-seed=@var{string}
4823 @opindex frandom-string
4824 This option provides a seed that GCC uses when it would otherwise use
4825 random numbers. It is used to generate certain symbol names
4826 that have to be different in every compiled file. It is also used to
4827 place unique stamps in coverage data files and the object files that
4828 produce them. You can use the @option{-frandom-seed} option to produce
4829 reproducibly identical object files.
4831 The @var{string} should be different for every file you compile.
4833 @item -fsched-verbose=@var{n}
4834 @opindex fsched-verbose
4835 On targets that use instruction scheduling, this option controls the
4836 amount of debugging output the scheduler prints. This information is
4837 written to standard error, unless @option{-dS} or @option{-dR} is
4838 specified, in which case it is output to the usual dump
4839 listing file, @file{.sched} or @file{.sched2} respectively. However
4840 for @var{n} greater than nine, the output is always printed to standard
4843 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4844 same information as @option{-dRS}. For @var{n} greater than one, it
4845 also output basic block probabilities, detailed ready list information
4846 and unit/insn info. For @var{n} greater than two, it includes RTL
4847 at abort point, control-flow and regions info. And for @var{n} over
4848 four, @option{-fsched-verbose} also includes dependence info.
4852 Store the usual ``temporary'' intermediate files permanently; place them
4853 in the current directory and name them based on the source file. Thus,
4854 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4855 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4856 preprocessed @file{foo.i} output file even though the compiler now
4857 normally uses an integrated preprocessor.
4859 When used in combination with the @option{-x} command line option,
4860 @option{-save-temps} is sensible enough to avoid over writing an
4861 input source file with the same extension as an intermediate file.
4862 The corresponding intermediate file may be obtained by renaming the
4863 source file before using @option{-save-temps}.
4867 Report the CPU time taken by each subprocess in the compilation
4868 sequence. For C source files, this is the compiler proper and assembler
4869 (plus the linker if linking is done). The output looks like this:
4876 The first number on each line is the ``user time'', that is time spent
4877 executing the program itself. The second number is ``system time'',
4878 time spent executing operating system routines on behalf of the program.
4879 Both numbers are in seconds.
4881 @item -fvar-tracking
4882 @opindex fvar-tracking
4883 Run variable tracking pass. It computes where variables are stored at each
4884 position in code. Better debugging information is then generated
4885 (if the debugging information format supports this information).
4887 It is enabled by default when compiling with optimization (@option{-Os},
4888 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4889 the debug info format supports it.
4891 @item -print-file-name=@var{library}
4892 @opindex print-file-name
4893 Print the full absolute name of the library file @var{library} that
4894 would be used when linking---and don't do anything else. With this
4895 option, GCC does not compile or link anything; it just prints the
4898 @item -print-multi-directory
4899 @opindex print-multi-directory
4900 Print the directory name corresponding to the multilib selected by any
4901 other switches present in the command line. This directory is supposed
4902 to exist in @env{GCC_EXEC_PREFIX}.
4904 @item -print-multi-lib
4905 @opindex print-multi-lib
4906 Print the mapping from multilib directory names to compiler switches
4907 that enable them. The directory name is separated from the switches by
4908 @samp{;}, and each switch starts with an @samp{@@} instead of the
4909 @samp{-}, without spaces between multiple switches. This is supposed to
4910 ease shell-processing.
4912 @item -print-prog-name=@var{program}
4913 @opindex print-prog-name
4914 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4916 @item -print-libgcc-file-name
4917 @opindex print-libgcc-file-name
4918 Same as @option{-print-file-name=libgcc.a}.
4920 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4921 but you do want to link with @file{libgcc.a}. You can do
4924 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4927 @item -print-search-dirs
4928 @opindex print-search-dirs
4929 Print the name of the configured installation directory and a list of
4930 program and library directories @command{gcc} will search---and don't do anything else.
4932 This is useful when @command{gcc} prints the error message
4933 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4934 To resolve this you either need to put @file{cpp0} and the other compiler
4935 components where @command{gcc} expects to find them, or you can set the environment
4936 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4937 Don't forget the trailing @samp{/}.
4938 @xref{Environment Variables}.
4940 @item -print-sysroot-headers-suffix
4941 @opindex print-sysroot-headers-suffix
4942 Print the suffix added to the target sysroot when searching for
4943 headers, or give an error if the compiler is not configured with such
4944 a suffix---and don't do anything else.
4947 @opindex dumpmachine
4948 Print the compiler's target machine (for example,
4949 @samp{i686-pc-linux-gnu})---and don't do anything else.
4952 @opindex dumpversion
4953 Print the compiler version (for example, @samp{3.0})---and don't do
4958 Print the compiler's built-in specs---and don't do anything else. (This
4959 is used when GCC itself is being built.) @xref{Spec Files}.
4961 @item -feliminate-unused-debug-types
4962 @opindex feliminate-unused-debug-types
4963 Normally, when producing DWARF2 output, GCC will emit debugging
4964 information for all types declared in a compilation
4965 unit, regardless of whether or not they are actually used
4966 in that compilation unit. Sometimes this is useful, such as
4967 if, in the debugger, you want to cast a value to a type that is
4968 not actually used in your program (but is declared). More often,
4969 however, this results in a significant amount of wasted space.
4970 With this option, GCC will avoid producing debug symbol output
4971 for types that are nowhere used in the source file being compiled.
4974 @node Optimize Options
4975 @section Options That Control Optimization
4976 @cindex optimize options
4977 @cindex options, optimization
4979 These options control various sorts of optimizations.
4981 Without any optimization option, the compiler's goal is to reduce the
4982 cost of compilation and to make debugging produce the expected
4983 results. Statements are independent: if you stop the program with a
4984 breakpoint between statements, you can then assign a new value to any
4985 variable or change the program counter to any other statement in the
4986 function and get exactly the results you would expect from the source
4989 Turning on optimization flags makes the compiler attempt to improve
4990 the performance and/or code size at the expense of compilation time
4991 and possibly the ability to debug the program.
4993 The compiler performs optimization based on the knowledge it has of
4994 the program. Optimization levels @option{-O} and above, in
4995 particular, enable @emph{unit-at-a-time} mode, which allows the
4996 compiler to consider information gained from later functions in
4997 the file when compiling a function. Compiling multiple files at
4998 once to a single output file in @emph{unit-at-a-time} mode allows
4999 the compiler to use information gained from all of the files when
5000 compiling each of them.
5002 Not all optimizations are controlled directly by a flag. Only
5003 optimizations that have a flag are listed.
5010 Optimize. Optimizing compilation takes somewhat more time, and a lot
5011 more memory for a large function.
5013 With @option{-O}, the compiler tries to reduce code size and execution
5014 time, without performing any optimizations that take a great deal of
5017 @option{-O} turns on the following optimization flags:
5018 @gccoptlist{-fdefer-pop @gol
5019 -fdelayed-branch @gol
5020 -fguess-branch-probability @gol
5021 -fcprop-registers @gol
5022 -fif-conversion @gol
5023 -fif-conversion2 @gol
5024 -fsplit-wide-types @gol
5027 -ftree-dominator-opts @gol
5031 -ftree-copyrename @gol
5034 -funit-at-a-time @gol
5037 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5038 where doing so does not interfere with debugging.
5042 Optimize even more. GCC performs nearly all supported optimizations
5043 that do not involve a space-speed tradeoff. The compiler does not
5044 perform loop unrolling or function inlining when you specify @option{-O2}.
5045 As compared to @option{-O}, this option increases both compilation time
5046 and the performance of the generated code.
5048 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5049 also turns on the following optimization flags:
5050 @gccoptlist{-fthread-jumps @gol
5052 -foptimize-sibling-calls @gol
5053 -fcse-follow-jumps -fcse-skip-blocks @gol
5054 -fgcse -fgcse-lm @gol
5055 -fexpensive-optimizations @gol
5056 -frerun-cse-after-loop @gol
5059 -fschedule-insns -fschedule-insns2 @gol
5060 -fsched-interblock -fsched-spec @gol
5062 -fstrict-aliasing -fstrict-overflow @gol
5063 -fdelete-null-pointer-checks @gol
5064 -freorder-blocks -freorder-functions @gol
5065 -falign-functions -falign-jumps @gol
5066 -falign-loops -falign-labels @gol
5070 Please note the warning under @option{-fgcse} about
5071 invoking @option{-O2} on programs that use computed gotos.
5075 Optimize yet more. @option{-O3} turns on all optimizations specified by
5076 @option{-O2} and also turns on the @option{-finline-functions},
5077 @option{-funswitch-loops}, @option{-fpredictive-commoning} and
5078 @option{-fgcse-after-reload} options.
5082 Reduce compilation time and make debugging produce the expected
5083 results. This is the default.
5087 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5088 do not typically increase code size. It also performs further
5089 optimizations designed to reduce code size.
5091 @option{-Os} disables the following optimization flags:
5092 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5093 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5094 -fprefetch-loop-arrays -ftree-vect-loop-version}
5096 If you use multiple @option{-O} options, with or without level numbers,
5097 the last such option is the one that is effective.
5100 Options of the form @option{-f@var{flag}} specify machine-independent
5101 flags. Most flags have both positive and negative forms; the negative
5102 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5103 below, only one of the forms is listed---the one you typically will
5104 use. You can figure out the other form by either removing @samp{no-}
5107 The following options control specific optimizations. They are either
5108 activated by @option{-O} options or are related to ones that are. You
5109 can use the following flags in the rare cases when ``fine-tuning'' of
5110 optimizations to be performed is desired.
5113 @item -fno-default-inline
5114 @opindex fno-default-inline
5115 Do not make member functions inline by default merely because they are
5116 defined inside the class scope (C++ only). Otherwise, when you specify
5117 @w{@option{-O}}, member functions defined inside class scope are compiled
5118 inline by default; i.e., you don't need to add @samp{inline} in front of
5119 the member function name.
5121 @item -fno-defer-pop
5122 @opindex fno-defer-pop
5123 Always pop the arguments to each function call as soon as that function
5124 returns. For machines which must pop arguments after a function call,
5125 the compiler normally lets arguments accumulate on the stack for several
5126 function calls and pops them all at once.
5128 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5131 @opindex fforce-addr
5132 Force memory address constants to be copied into registers before
5133 doing arithmetic on them.
5135 @item -fforward-propagate
5136 @opindex fforward-propagate
5137 Perform a forward propagation pass on RTL. The pass tries to combine two
5138 instructions and checks if the result can be simplified. If loop unrolling
5139 is active, two passes are performed and the second is scheduled after
5142 This option is enabled by default at optimization levels @option{-O2},
5143 @option{-O3}, @option{-Os}.
5145 @item -fomit-frame-pointer
5146 @opindex fomit-frame-pointer
5147 Don't keep the frame pointer in a register for functions that
5148 don't need one. This avoids the instructions to save, set up and
5149 restore frame pointers; it also makes an extra register available
5150 in many functions. @strong{It also makes debugging impossible on
5153 On some machines, such as the VAX, this flag has no effect, because
5154 the standard calling sequence automatically handles the frame pointer
5155 and nothing is saved by pretending it doesn't exist. The
5156 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5157 whether a target machine supports this flag. @xref{Registers,,Register
5158 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5160 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5162 @item -foptimize-sibling-calls
5163 @opindex foptimize-sibling-calls
5164 Optimize sibling and tail recursive calls.
5166 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5170 Don't pay attention to the @code{inline} keyword. Normally this option
5171 is used to keep the compiler from expanding any functions inline.
5172 Note that if you are not optimizing, no functions can be expanded inline.
5174 @item -finline-functions
5175 @opindex finline-functions
5176 Integrate all simple functions into their callers. The compiler
5177 heuristically decides which functions are simple enough to be worth
5178 integrating in this way.
5180 If all calls to a given function are integrated, and the function is
5181 declared @code{static}, then the function is normally not output as
5182 assembler code in its own right.
5184 Enabled at level @option{-O3}.
5186 @item -finline-functions-called-once
5187 @opindex finline-functions-called-once
5188 Consider all @code{static} functions called once for inlining into their
5189 caller even if they are not marked @code{inline}. If a call to a given
5190 function is integrated, then the function is not output as assembler code
5193 Enabled if @option{-funit-at-a-time} is enabled.
5195 @item -fearly-inlining
5196 @opindex fearly-inlining
5197 Inline functions marked by @code{always_inline} and functions whose body seems
5198 smaller than the function call overhead early before doing
5199 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5200 makes profiling significantly cheaper and usually inlining faster on programs
5201 having large chains of nested wrapper functions.
5205 @item -finline-limit=@var{n}
5206 @opindex finline-limit
5207 By default, GCC limits the size of functions that can be inlined. This flag
5208 allows the control of this limit for functions that are explicitly marked as
5209 inline (i.e., marked with the inline keyword or defined within the class
5210 definition in c++). @var{n} is the size of functions that can be inlined in
5211 number of pseudo instructions (not counting parameter handling). The default
5212 value of @var{n} is 600.
5213 Increasing this value can result in more inlined code at
5214 the cost of compilation time and memory consumption. Decreasing usually makes
5215 the compilation faster and less code will be inlined (which presumably
5216 means slower programs). This option is particularly useful for programs that
5217 use inlining heavily such as those based on recursive templates with C++.
5219 Inlining is actually controlled by a number of parameters, which may be
5220 specified individually by using @option{--param @var{name}=@var{value}}.
5221 The @option{-finline-limit=@var{n}} option sets some of these parameters
5225 @item max-inline-insns-single
5226 is set to @var{n}/2.
5227 @item max-inline-insns-auto
5228 is set to @var{n}/2.
5229 @item min-inline-insns
5230 is set to 130 or @var{n}/4, whichever is smaller.
5231 @item max-inline-insns-rtl
5235 See below for a documentation of the individual
5236 parameters controlling inlining.
5238 @emph{Note:} pseudo instruction represents, in this particular context, an
5239 abstract measurement of function's size. In no way does it represent a count
5240 of assembly instructions and as such its exact meaning might change from one
5241 release to an another.
5243 @item -fkeep-inline-functions
5244 @opindex fkeep-inline-functions
5245 In C, emit @code{static} functions that are declared @code{inline}
5246 into the object file, even if the function has been inlined into all
5247 of its callers. This switch does not affect functions using the
5248 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5249 inline functions into the object file.
5251 @item -fkeep-static-consts
5252 @opindex fkeep-static-consts
5253 Emit variables declared @code{static const} when optimization isn't turned
5254 on, even if the variables aren't referenced.
5256 GCC enables this option by default. If you want to force the compiler to
5257 check if the variable was referenced, regardless of whether or not
5258 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5260 @item -fmerge-constants
5261 Attempt to merge identical constants (string constants and floating point
5262 constants) across compilation units.
5264 This option is the default for optimized compilation if the assembler and
5265 linker support it. Use @option{-fno-merge-constants} to inhibit this
5268 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5270 @item -fmerge-all-constants
5271 Attempt to merge identical constants and identical variables.
5273 This option implies @option{-fmerge-constants}. In addition to
5274 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5275 arrays or initialized constant variables with integral or floating point
5276 types. Languages like C or C++ require each non-automatic variable to
5277 have distinct location, so using this option will result in non-conforming
5280 @item -fmodulo-sched
5281 @opindex fmodulo-sched
5282 Perform swing modulo scheduling immediately before the first scheduling
5283 pass. This pass looks at innermost loops and reorders their
5284 instructions by overlapping different iterations.
5286 @item -fmodulo-sched-allow-regmoves
5287 @opindex fmodulo-sched-allow-regmoves
5288 Perform more aggressive SMS based modulo scheduling with register moves
5289 allowed. By setting this flag certain anti-dependences edges will be
5290 deleted which will trigger the generation of reg-moves based on the
5291 life-range analysis.
5293 @item -fno-branch-count-reg
5294 @opindex fno-branch-count-reg
5295 Do not use ``decrement and branch'' instructions on a count register,
5296 but instead generate a sequence of instructions that decrement a
5297 register, compare it against zero, then branch based upon the result.
5298 This option is only meaningful on architectures that support such
5299 instructions, which include x86, PowerPC, IA-64 and S/390.
5301 The default is @option{-fbranch-count-reg}.
5303 @item -fno-function-cse
5304 @opindex fno-function-cse
5305 Do not put function addresses in registers; make each instruction that
5306 calls a constant function contain the function's address explicitly.
5308 This option results in less efficient code, but some strange hacks
5309 that alter the assembler output may be confused by the optimizations
5310 performed when this option is not used.
5312 The default is @option{-ffunction-cse}
5314 @item -fno-zero-initialized-in-bss
5315 @opindex fno-zero-initialized-in-bss
5316 If the target supports a BSS section, GCC by default puts variables that
5317 are initialized to zero into BSS@. This can save space in the resulting
5320 This option turns off this behavior because some programs explicitly
5321 rely on variables going to the data section. E.g., so that the
5322 resulting executable can find the beginning of that section and/or make
5323 assumptions based on that.
5325 The default is @option{-fzero-initialized-in-bss}.
5327 @item -fbounds-check
5328 @opindex fbounds-check
5329 For front-ends that support it, generate additional code to check that
5330 indices used to access arrays are within the declared range. This is
5331 currently only supported by the Java and Fortran front-ends, where
5332 this option defaults to true and false respectively.
5334 @item -fmudflap -fmudflapth -fmudflapir
5338 @cindex bounds checking
5340 For front-ends that support it (C and C++), instrument all risky
5341 pointer/array dereferencing operations, some standard library
5342 string/heap functions, and some other associated constructs with
5343 range/validity tests. Modules so instrumented should be immune to
5344 buffer overflows, invalid heap use, and some other classes of C/C++
5345 programming errors. The instrumentation relies on a separate runtime
5346 library (@file{libmudflap}), which will be linked into a program if
5347 @option{-fmudflap} is given at link time. Run-time behavior of the
5348 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5349 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5352 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5353 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5354 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5355 instrumentation should ignore pointer reads. This produces less
5356 instrumentation (and therefore faster execution) and still provides
5357 some protection against outright memory corrupting writes, but allows
5358 erroneously read data to propagate within a program.
5360 @item -fthread-jumps
5361 @opindex fthread-jumps
5362 Perform optimizations where we check to see if a jump branches to a
5363 location where another comparison subsumed by the first is found. If
5364 so, the first branch is redirected to either the destination of the
5365 second branch or a point immediately following it, depending on whether
5366 the condition is known to be true or false.
5368 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5370 @item -fsplit-wide-types
5371 @opindex fsplit-wide-types
5372 When using a type that occupies multiple registers, such as @code{long
5373 long} on a 32-bit system, split the registers apart and allocate them
5374 independently. This normally generates better code for those types,
5375 but may make debugging more difficult.
5377 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5380 @item -fcse-follow-jumps
5381 @opindex fcse-follow-jumps
5382 In common subexpression elimination, scan through jump instructions
5383 when the target of the jump is not reached by any other path. For
5384 example, when CSE encounters an @code{if} statement with an
5385 @code{else} clause, CSE will follow the jump when the condition
5388 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5390 @item -fcse-skip-blocks
5391 @opindex fcse-skip-blocks
5392 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5393 follow jumps which conditionally skip over blocks. When CSE
5394 encounters a simple @code{if} statement with no else clause,
5395 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5396 body of the @code{if}.
5398 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5400 @item -frerun-cse-after-loop
5401 @opindex frerun-cse-after-loop
5402 Re-run common subexpression elimination after loop optimizations has been
5405 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5409 Perform a global common subexpression elimination pass.
5410 This pass also performs global constant and copy propagation.
5412 @emph{Note:} When compiling a program using computed gotos, a GCC
5413 extension, you may get better runtime performance if you disable
5414 the global common subexpression elimination pass by adding
5415 @option{-fno-gcse} to the command line.
5417 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5421 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5422 attempt to move loads which are only killed by stores into themselves. This
5423 allows a loop containing a load/store sequence to be changed to a load outside
5424 the loop, and a copy/store within the loop.
5426 Enabled by default when gcse is enabled.
5430 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5431 global common subexpression elimination. This pass will attempt to move
5432 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5433 loops containing a load/store sequence can be changed to a load before
5434 the loop and a store after the loop.
5436 Not enabled at any optimization level.
5440 When @option{-fgcse-las} is enabled, the global common subexpression
5441 elimination pass eliminates redundant loads that come after stores to the
5442 same memory location (both partial and full redundancies).
5444 Not enabled at any optimization level.
5446 @item -fgcse-after-reload
5447 @opindex fgcse-after-reload
5448 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5449 pass is performed after reload. The purpose of this pass is to cleanup
5452 @item -funsafe-loop-optimizations
5453 @opindex funsafe-loop-optimizations
5454 If given, the loop optimizer will assume that loop indices do not
5455 overflow, and that the loops with nontrivial exit condition are not
5456 infinite. This enables a wider range of loop optimizations even if
5457 the loop optimizer itself cannot prove that these assumptions are valid.
5458 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5459 if it finds this kind of loop.
5461 @item -fcrossjumping
5462 @opindex crossjumping
5463 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5464 resulting code may or may not perform better than without cross-jumping.
5466 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5468 @item -fif-conversion
5469 @opindex if-conversion
5470 Attempt to transform conditional jumps into branch-less equivalents. This
5471 include use of conditional moves, min, max, set flags and abs instructions, and
5472 some tricks doable by standard arithmetics. The use of conditional execution
5473 on chips where it is available is controlled by @code{if-conversion2}.
5475 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5477 @item -fif-conversion2
5478 @opindex if-conversion2
5479 Use conditional execution (where available) to transform conditional jumps into
5480 branch-less equivalents.
5482 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5484 @item -fdelete-null-pointer-checks
5485 @opindex fdelete-null-pointer-checks
5486 Use global dataflow analysis to identify and eliminate useless checks
5487 for null pointers. The compiler assumes that dereferencing a null
5488 pointer would have halted the program. If a pointer is checked after
5489 it has already been dereferenced, it cannot be null.
5491 In some environments, this assumption is not true, and programs can
5492 safely dereference null pointers. Use
5493 @option{-fno-delete-null-pointer-checks} to disable this optimization
5494 for programs which depend on that behavior.
5496 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5498 @item -fexpensive-optimizations
5499 @opindex fexpensive-optimizations
5500 Perform a number of minor optimizations that are relatively expensive.
5502 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5504 @item -foptimize-register-move
5506 @opindex foptimize-register-move
5508 Attempt to reassign register numbers in move instructions and as
5509 operands of other simple instructions in order to maximize the amount of
5510 register tying. This is especially helpful on machines with two-operand
5513 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5516 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5518 @item -fdelayed-branch
5519 @opindex fdelayed-branch
5520 If supported for the target machine, attempt to reorder instructions
5521 to exploit instruction slots available after delayed branch
5524 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5526 @item -fschedule-insns
5527 @opindex fschedule-insns
5528 If supported for the target machine, attempt to reorder instructions to
5529 eliminate execution stalls due to required data being unavailable. This
5530 helps machines that have slow floating point or memory load instructions
5531 by allowing other instructions to be issued until the result of the load
5532 or floating point instruction is required.
5534 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5536 @item -fschedule-insns2
5537 @opindex fschedule-insns2
5538 Similar to @option{-fschedule-insns}, but requests an additional pass of
5539 instruction scheduling after register allocation has been done. This is
5540 especially useful on machines with a relatively small number of
5541 registers and where memory load instructions take more than one cycle.
5543 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5545 @item -fno-sched-interblock
5546 @opindex fno-sched-interblock
5547 Don't schedule instructions across basic blocks. This is normally
5548 enabled by default when scheduling before register allocation, i.e.@:
5549 with @option{-fschedule-insns} or at @option{-O2} or higher.
5551 @item -fno-sched-spec
5552 @opindex fno-sched-spec
5553 Don't allow speculative motion of non-load instructions. This is normally
5554 enabled by default when scheduling before register allocation, i.e.@:
5555 with @option{-fschedule-insns} or at @option{-O2} or higher.
5557 @item -fsched-spec-load
5558 @opindex fsched-spec-load
5559 Allow speculative motion of some load instructions. This only makes
5560 sense when scheduling before register allocation, i.e.@: with
5561 @option{-fschedule-insns} or at @option{-O2} or higher.
5563 @item -fsched-spec-load-dangerous
5564 @opindex fsched-spec-load-dangerous
5565 Allow speculative motion of more load instructions. This only makes
5566 sense when scheduling before register allocation, i.e.@: with
5567 @option{-fschedule-insns} or at @option{-O2} or higher.
5569 @item -fsched-stalled-insns=@var{n}
5570 @opindex fsched-stalled-insns
5571 Define how many insns (if any) can be moved prematurely from the queue
5572 of stalled insns into the ready list, during the second scheduling pass.
5574 @item -fsched-stalled-insns-dep=@var{n}
5575 @opindex fsched-stalled-insns-dep
5576 Define how many insn groups (cycles) will be examined for a dependency
5577 on a stalled insn that is candidate for premature removal from the queue
5578 of stalled insns. Has an effect only during the second scheduling pass,
5579 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5581 @item -fsched2-use-superblocks
5582 @opindex fsched2-use-superblocks
5583 When scheduling after register allocation, do use superblock scheduling
5584 algorithm. Superblock scheduling allows motion across basic block boundaries
5585 resulting on faster schedules. This option is experimental, as not all machine
5586 descriptions used by GCC model the CPU closely enough to avoid unreliable
5587 results from the algorithm.
5589 This only makes sense when scheduling after register allocation, i.e.@: with
5590 @option{-fschedule-insns2} or at @option{-O2} or higher.
5592 @item -fsched2-use-traces
5593 @opindex fsched2-use-traces
5594 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5595 allocation and additionally perform code duplication in order to increase the
5596 size of superblocks using tracer pass. See @option{-ftracer} for details on
5599 This mode should produce faster but significantly longer programs. Also
5600 without @option{-fbranch-probabilities} the traces constructed may not
5601 match the reality and hurt the performance. This only makes
5602 sense when scheduling after register allocation, i.e.@: with
5603 @option{-fschedule-insns2} or at @option{-O2} or higher.
5607 Eliminates redundant extension instructions and move the non redundant
5608 ones to optimal placement using LCM.
5610 @item -freschedule-modulo-scheduled-loops
5611 @opindex fscheduling-in-modulo-scheduled-loops
5612 The modulo scheduling comes before the traditional scheduling, if a loop
5613 was modulo scheduled we may want to prevent the later scheduling passes
5614 from changing its schedule, we use this option to control that.
5616 @item -fcaller-saves
5617 @opindex fcaller-saves
5618 Enable values to be allocated in registers that will be clobbered by
5619 function calls, by emitting extra instructions to save and restore the
5620 registers around such calls. Such allocation is done only when it
5621 seems to result in better code than would otherwise be produced.
5623 This option is always enabled by default on certain machines, usually
5624 those which have no call-preserved registers to use instead.
5626 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5628 @item -ftree-reassoc
5629 Perform Reassociation on trees This flag is enabled by default
5630 at @option{-O} and higher.
5633 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5634 enabled by default at @option{-O2} and @option{-O3}.
5637 Perform Full Redundancy Elimination (FRE) on trees. The difference
5638 between FRE and PRE is that FRE only considers expressions
5639 that are computed on all paths leading to the redundant computation.
5640 This analysis is faster than PRE, though it exposes fewer redundancies.
5641 This flag is enabled by default at @option{-O} and higher.
5643 @item -ftree-copy-prop
5644 Perform copy propagation on trees. This pass eliminates unnecessary
5645 copy operations. This flag is enabled by default at @option{-O} and
5648 @item -ftree-store-copy-prop
5649 Perform copy propagation of memory loads and stores. This pass
5650 eliminates unnecessary copy operations in memory references
5651 (structures, global variables, arrays, etc). This flag is enabled by
5652 default at @option{-O2} and higher.
5655 Perform structural alias analysis on trees. This flag
5656 is enabled by default at @option{-O} and higher.
5659 Perform interprocedural pointer analysis.
5662 Perform forward store motion on trees. This flag is
5663 enabled by default at @option{-O} and higher.
5666 Perform sparse conditional constant propagation (CCP) on trees. This
5667 pass only operates on local scalar variables and is enabled by default
5668 at @option{-O} and higher.
5670 @item -ftree-store-ccp
5671 Perform sparse conditional constant propagation (CCP) on trees. This
5672 pass operates on both local scalar variables and memory stores and
5673 loads (global variables, structures, arrays, etc). This flag is
5674 enabled by default at @option{-O2} and higher.
5677 Perform dead code elimination (DCE) on trees. This flag is enabled by
5678 default at @option{-O} and higher.
5680 @item -ftree-dominator-opts
5681 Perform a variety of simple scalar cleanups (constant/copy
5682 propagation, redundancy elimination, range propagation and expression
5683 simplification) based on a dominator tree traversal. This also
5684 performs jump threading (to reduce jumps to jumps). This flag is
5685 enabled by default at @option{-O} and higher.
5688 Perform loop header copying on trees. This is beneficial since it increases
5689 effectiveness of code motion optimizations. It also saves one jump. This flag
5690 is enabled by default at @option{-O} and higher. It is not enabled
5691 for @option{-Os}, since it usually increases code size.
5693 @item -ftree-loop-optimize
5694 Perform loop optimizations on trees. This flag is enabled by default
5695 at @option{-O} and higher.
5697 @item -ftree-loop-linear
5698 Perform linear loop transformations on tree. This flag can improve cache
5699 performance and allow further loop optimizations to take place.
5701 @item -fcheck-data-deps
5702 Compare the results of several data dependence analyzers. This option
5703 is used for debugging the data dependence analyzers.
5705 @item -ftree-loop-im
5706 Perform loop invariant motion on trees. This pass moves only invariants that
5707 would be hard to handle at RTL level (function calls, operations that expand to
5708 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5709 operands of conditions that are invariant out of the loop, so that we can use
5710 just trivial invariantness analysis in loop unswitching. The pass also includes
5713 @item -ftree-loop-ivcanon
5714 Create a canonical counter for number of iterations in the loop for that
5715 determining number of iterations requires complicated analysis. Later
5716 optimizations then may determine the number easily. Useful especially
5717 in connection with unrolling.
5720 Perform induction variable optimizations (strength reduction, induction
5721 variable merging and induction variable elimination) on trees.
5724 Perform scalar replacement of aggregates. This pass replaces structure
5725 references with scalars to prevent committing structures to memory too
5726 early. This flag is enabled by default at @option{-O} and higher.
5728 @item -ftree-copyrename
5729 Perform copy renaming on trees. This pass attempts to rename compiler
5730 temporaries to other variables at copy locations, usually resulting in
5731 variable names which more closely resemble the original variables. This flag
5732 is enabled by default at @option{-O} and higher.
5735 Perform temporary expression replacement during the SSA->normal phase. Single
5736 use/single def temporaries are replaced at their use location with their
5737 defining expression. This results in non-GIMPLE code, but gives the expanders
5738 much more complex trees to work on resulting in better RTL generation. This is
5739 enabled by default at @option{-O} and higher.
5741 @item -ftree-vectorize
5742 Perform loop vectorization on trees.
5744 @item -ftree-vect-loop-version
5745 @opindex ftree-vect-loop-version
5746 Perform loop versioning when doing loop vectorization on trees. When a loop
5747 appears to be vectorizable except that data alignment or data dependence cannot
5748 be determined at compile time then vectorized and non-vectorized versions of
5749 the loop are generated along with runtime checks for alignment or dependence
5750 to control which version is executed. This option is enabled by default
5751 except at level @option{-Os} where it is disabled.
5753 @item -fvect-cost-model
5754 Enable cost model for vectorization.
5757 Perform Value Range Propagation on trees. This is similar to the
5758 constant propagation pass, but instead of values, ranges of values are
5759 propagated. This allows the optimizers to remove unnecessary range
5760 checks like array bound checks and null pointer checks. This is
5761 enabled by default at @option{-O2} and higher. Null pointer check
5762 elimination is only done if @option{-fdelete-null-pointer-checks} is
5767 Perform tail duplication to enlarge superblock size. This transformation
5768 simplifies the control flow of the function allowing other optimizations to do
5771 @item -funroll-loops
5772 @opindex funroll-loops
5773 Unroll loops whose number of iterations can be determined at compile
5774 time or upon entry to the loop. @option{-funroll-loops} implies
5775 @option{-frerun-cse-after-loop}. This option makes code larger,
5776 and may or may not make it run faster.
5778 @item -funroll-all-loops
5779 @opindex funroll-all-loops
5780 Unroll all loops, even if their number of iterations is uncertain when
5781 the loop is entered. This usually makes programs run more slowly.
5782 @option{-funroll-all-loops} implies the same options as
5783 @option{-funroll-loops},
5785 @item -fsplit-ivs-in-unroller
5786 @opindex fsplit-ivs-in-unroller
5787 Enables expressing of values of induction variables in later iterations
5788 of the unrolled loop using the value in the first iteration. This breaks
5789 long dependency chains, thus improving efficiency of the scheduling passes.
5791 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5792 same effect. However in cases the loop body is more complicated than
5793 a single basic block, this is not reliable. It also does not work at all
5794 on some of the architectures due to restrictions in the CSE pass.
5796 This optimization is enabled by default.
5798 @item -fvariable-expansion-in-unroller
5799 @opindex fvariable-expansion-in-unroller
5800 With this option, the compiler will create multiple copies of some
5801 local variables when unrolling a loop which can result in superior code.
5803 @item -fpredictive-commoning
5804 @opindex fpredictive-commoning
5805 Perform predictive commoning optimization, i.e., reusing computations
5806 (especially memory loads and stores) performed in previous
5807 iterations of loops.
5809 This option is enabled at level @option{-O3}.
5811 @item -fprefetch-loop-arrays
5812 @opindex fprefetch-loop-arrays
5813 If supported by the target machine, generate instructions to prefetch
5814 memory to improve the performance of loops that access large arrays.
5816 This option may generate better or worse code; results are highly
5817 dependent on the structure of loops within the source code.
5819 Disabled at level @option{-Os}.
5822 @itemx -fno-peephole2
5823 @opindex fno-peephole
5824 @opindex fno-peephole2
5825 Disable any machine-specific peephole optimizations. The difference
5826 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5827 are implemented in the compiler; some targets use one, some use the
5828 other, a few use both.
5830 @option{-fpeephole} is enabled by default.
5831 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5833 @item -fno-guess-branch-probability
5834 @opindex fno-guess-branch-probability
5835 Do not guess branch probabilities using heuristics.
5837 GCC will use heuristics to guess branch probabilities if they are
5838 not provided by profiling feedback (@option{-fprofile-arcs}). These
5839 heuristics are based on the control flow graph. If some branch probabilities
5840 are specified by @samp{__builtin_expect}, then the heuristics will be
5841 used to guess branch probabilities for the rest of the control flow graph,
5842 taking the @samp{__builtin_expect} info into account. The interactions
5843 between the heuristics and @samp{__builtin_expect} can be complex, and in
5844 some cases, it may be useful to disable the heuristics so that the effects
5845 of @samp{__builtin_expect} are easier to understand.
5847 The default is @option{-fguess-branch-probability} at levels
5848 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5850 @item -freorder-blocks
5851 @opindex freorder-blocks
5852 Reorder basic blocks in the compiled function in order to reduce number of
5853 taken branches and improve code locality.
5855 Enabled at levels @option{-O2}, @option{-O3}.
5857 @item -freorder-blocks-and-partition
5858 @opindex freorder-blocks-and-partition
5859 In addition to reordering basic blocks in the compiled function, in order
5860 to reduce number of taken branches, partitions hot and cold basic blocks
5861 into separate sections of the assembly and .o files, to improve
5862 paging and cache locality performance.
5864 This optimization is automatically turned off in the presence of
5865 exception handling, for linkonce sections, for functions with a user-defined
5866 section attribute and on any architecture that does not support named
5869 @item -freorder-functions
5870 @opindex freorder-functions
5871 Reorder functions in the object file in order to
5872 improve code locality. This is implemented by using special
5873 subsections @code{.text.hot} for most frequently executed functions and
5874 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5875 the linker so object file format must support named sections and linker must
5876 place them in a reasonable way.
5878 Also profile feedback must be available in to make this option effective. See
5879 @option{-fprofile-arcs} for details.
5881 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5883 @item -fstrict-aliasing
5884 @opindex fstrict-aliasing
5885 Allows the compiler to assume the strictest aliasing rules applicable to
5886 the language being compiled. For C (and C++), this activates
5887 optimizations based on the type of expressions. In particular, an
5888 object of one type is assumed never to reside at the same address as an
5889 object of a different type, unless the types are almost the same. For
5890 example, an @code{unsigned int} can alias an @code{int}, but not a
5891 @code{void*} or a @code{double}. A character type may alias any other
5894 Pay special attention to code like this:
5907 The practice of reading from a different union member than the one most
5908 recently written to (called ``type-punning'') is common. Even with
5909 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5910 is accessed through the union type. So, the code above will work as
5911 expected. However, this code might not:
5922 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5924 @item -fstrict-overflow
5925 @opindex fstrict-overflow
5926 Allow the compiler to assume strict signed overflow rules, depending
5927 on the language being compiled. For C (and C++) this means that
5928 overflow when doing arithmetic with signed numbers is undefined, which
5929 means that the compiler may assume that it will not happen. This
5930 permits various optimizations. For example, the compiler will assume
5931 that an expression like @code{i + 10 > i} will always be true for
5932 signed @code{i}. This assumption is only valid if signed overflow is
5933 undefined, as the expression is false if @code{i + 10} overflows when
5934 using twos complement arithmetic. When this option is in effect any
5935 attempt to determine whether an operation on signed numbers will
5936 overflow must be written carefully to not actually involve overflow.
5938 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
5939 that signed overflow is fully defined: it wraps. When
5940 @option{-fwrapv} is used, there is no difference between
5941 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
5942 @option{-fwrapv} certain types of overflow are permitted. For
5943 example, if the compiler gets an overflow when doing arithmetic on
5944 constants, the overflowed value can still be used with
5945 @option{-fwrapv}, but not otherwise.
5947 The @option{-fstrict-overflow} option is enabled at levels
5948 @option{-O2}, @option{-O3}, @option{-Os}.
5950 @item -falign-functions
5951 @itemx -falign-functions=@var{n}
5952 @opindex falign-functions
5953 Align the start of functions to the next power-of-two greater than
5954 @var{n}, skipping up to @var{n} bytes. For instance,
5955 @option{-falign-functions=32} aligns functions to the next 32-byte
5956 boundary, but @option{-falign-functions=24} would align to the next
5957 32-byte boundary only if this can be done by skipping 23 bytes or less.
5959 @option{-fno-align-functions} and @option{-falign-functions=1} are
5960 equivalent and mean that functions will not be aligned.
5962 Some assemblers only support this flag when @var{n} is a power of two;
5963 in that case, it is rounded up.
5965 If @var{n} is not specified or is zero, use a machine-dependent default.
5967 Enabled at levels @option{-O2}, @option{-O3}.
5969 @item -falign-labels
5970 @itemx -falign-labels=@var{n}
5971 @opindex falign-labels
5972 Align all branch targets to a power-of-two boundary, skipping up to
5973 @var{n} bytes like @option{-falign-functions}. This option can easily
5974 make code slower, because it must insert dummy operations for when the
5975 branch target is reached in the usual flow of the code.
5977 @option{-fno-align-labels} and @option{-falign-labels=1} are
5978 equivalent and mean that labels will not be aligned.
5980 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5981 are greater than this value, then their values are used instead.
5983 If @var{n} is not specified or is zero, use a machine-dependent default
5984 which is very likely to be @samp{1}, meaning no alignment.
5986 Enabled at levels @option{-O2}, @option{-O3}.
5989 @itemx -falign-loops=@var{n}
5990 @opindex falign-loops
5991 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5992 like @option{-falign-functions}. The hope is that the loop will be
5993 executed many times, which will make up for any execution of the dummy
5996 @option{-fno-align-loops} and @option{-falign-loops=1} are
5997 equivalent and mean that loops will not be aligned.
5999 If @var{n} is not specified or is zero, use a machine-dependent default.
6001 Enabled at levels @option{-O2}, @option{-O3}.
6004 @itemx -falign-jumps=@var{n}
6005 @opindex falign-jumps
6006 Align branch targets to a power-of-two boundary, for branch targets
6007 where the targets can only be reached by jumping, skipping up to @var{n}
6008 bytes like @option{-falign-functions}. In this case, no dummy operations
6011 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6012 equivalent and mean that loops will not be aligned.
6014 If @var{n} is not specified or is zero, use a machine-dependent default.
6016 Enabled at levels @option{-O2}, @option{-O3}.
6018 @item -funit-at-a-time
6019 @opindex funit-at-a-time
6020 Parse the whole compilation unit before starting to produce code.
6021 This allows some extra optimizations to take place but consumes
6022 more memory (in general). There are some compatibility issues
6023 with @emph{unit-at-a-time} mode:
6026 enabling @emph{unit-at-a-time} mode may change the order
6027 in which functions, variables, and top-level @code{asm} statements
6028 are emitted, and will likely break code relying on some particular
6029 ordering. The majority of such top-level @code{asm} statements,
6030 though, can be replaced by @code{section} attributes. The
6031 @option{fno-toplevel-reorder} option may be used to keep the ordering
6032 used in the input file, at the cost of some optimizations.
6035 @emph{unit-at-a-time} mode removes unreferenced static variables
6036 and functions. This may result in undefined references
6037 when an @code{asm} statement refers directly to variables or functions
6038 that are otherwise unused. In that case either the variable/function
6039 shall be listed as an operand of the @code{asm} statement operand or,
6040 in the case of top-level @code{asm} statements the attribute @code{used}
6041 shall be used on the declaration.
6044 Static functions now can use non-standard passing conventions that
6045 may break @code{asm} statements calling functions directly. Again,
6046 attribute @code{used} will prevent this behavior.
6049 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
6050 but this scheme may not be supported by future releases of GCC@.
6052 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6054 @item -fno-toplevel-reorder
6055 Do not reorder top-level functions, variables, and @code{asm}
6056 statements. Output them in the same order that they appear in the
6057 input file. When this option is used, unreferenced static variables
6058 will not be removed. This option is intended to support existing code
6059 which relies on a particular ordering. For new code, it is better to
6064 Constructs webs as commonly used for register allocation purposes and assign
6065 each web individual pseudo register. This allows the register allocation pass
6066 to operate on pseudos directly, but also strengthens several other optimization
6067 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6068 however, make debugging impossible, since variables will no longer stay in a
6071 Enabled by default with @option{-funroll-loops}.
6073 @item -fwhole-program
6074 @opindex fwhole-program
6075 Assume that the current compilation unit represents whole program being
6076 compiled. All public functions and variables with the exception of @code{main}
6077 and those merged by attribute @code{externally_visible} become static functions
6078 and in a affect gets more aggressively optimized by interprocedural optimizers.
6079 While this option is equivalent to proper use of @code{static} keyword for
6080 programs consisting of single file, in combination with option
6081 @option{--combine} this flag can be used to compile most of smaller scale C
6082 programs since the functions and variables become local for the whole combined
6083 compilation unit, not for the single source file itself.
6086 @item -fno-cprop-registers
6087 @opindex fno-cprop-registers
6088 After register allocation and post-register allocation instruction splitting,
6089 we perform a copy-propagation pass to try to reduce scheduling dependencies
6090 and occasionally eliminate the copy.
6092 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6094 @item -fprofile-generate
6095 @opindex fprofile-generate
6097 Enable options usually used for instrumenting application to produce
6098 profile useful for later recompilation with profile feedback based
6099 optimization. You must use @option{-fprofile-generate} both when
6100 compiling and when linking your program.
6102 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6105 @opindex fprofile-use
6106 Enable profile feedback directed optimizations, and optimizations
6107 generally profitable only with profile feedback available.
6109 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6110 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6112 By default, GCC emits an error message if the feedback profiles do not
6113 match the source code. This error can be turned into a warning by using
6114 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6118 The following options control compiler behavior regarding floating
6119 point arithmetic. These options trade off between speed and
6120 correctness. All must be specifically enabled.
6124 @opindex ffloat-store
6125 Do not store floating point variables in registers, and inhibit other
6126 options that might change whether a floating point value is taken from a
6129 @cindex floating point precision
6130 This option prevents undesirable excess precision on machines such as
6131 the 68000 where the floating registers (of the 68881) keep more
6132 precision than a @code{double} is supposed to have. Similarly for the
6133 x86 architecture. For most programs, the excess precision does only
6134 good, but a few programs rely on the precise definition of IEEE floating
6135 point. Use @option{-ffloat-store} for such programs, after modifying
6136 them to store all pertinent intermediate computations into variables.
6140 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
6141 @option{-fno-trapping-math}, @option{-ffinite-math-only},
6142 @option{-fno-rounding-math}, @option{-fno-signaling-nans},
6143 @option{-fno-signed-zeros} and @option{fcx-limited-range}.
6145 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6147 This option is not turned on by any @option{-O} option since
6148 it can result in incorrect output for programs which depend on
6149 an exact implementation of IEEE or ISO rules/specifications for
6150 math functions. It may, however, yield faster code for programs
6151 that do not require the guarantees of these specifications.
6153 @item -fno-math-errno
6154 @opindex fno-math-errno
6155 Do not set ERRNO after calling math functions that are executed
6156 with a single instruction, e.g., sqrt. A program that relies on
6157 IEEE exceptions for math error handling may want to use this flag
6158 for speed while maintaining IEEE arithmetic compatibility.
6160 This option is not turned on by any @option{-O} option since
6161 it can result in incorrect output for programs which depend on
6162 an exact implementation of IEEE or ISO rules/specifications for
6163 math functions. It may, however, yield faster code for programs
6164 that do not require the guarantees of these specifications.
6166 The default is @option{-fmath-errno}.
6168 On Darwin systems, the math library never sets @code{errno}. There is
6169 therefore no reason for the compiler to consider the possibility that
6170 it might, and @option{-fno-math-errno} is the default.
6172 @item -funsafe-math-optimizations
6173 @opindex funsafe-math-optimizations
6174 Allow optimizations for floating-point arithmetic that (a) assume
6175 that arguments and results are valid and (b) may violate IEEE or
6176 ANSI standards. When used at link-time, it may include libraries
6177 or startup files that change the default FPU control word or other
6178 similar optimizations.
6180 This option is not turned on by any @option{-O} option since
6181 it can result in incorrect output for programs which depend on
6182 an exact implementation of IEEE or ISO rules/specifications for
6183 math functions. It may, however, yield faster code for programs
6184 that do not require the guarantees of these specifications.
6186 The default is @option{-fno-unsafe-math-optimizations}.
6188 @item -ffinite-math-only
6189 @opindex ffinite-math-only
6190 Allow optimizations for floating-point arithmetic that assume
6191 that arguments and results are not NaNs or +-Infs.
6193 This option is not turned on by any @option{-O} option since
6194 it can result in incorrect output for programs which depend on
6195 an exact implementation of IEEE or ISO rules/specifications for
6196 math functions. It may, however, yield faster code for programs
6197 that do not require the guarantees of these specifications.
6199 The default is @option{-fno-finite-math-only}.
6201 @item -fno-signed-zeros
6202 @opindex fno-signed-zeros
6203 Allow optimizations for floating point arithmetic that ignore the
6204 signedness of zero. IEEE arithmetic specifies the behavior of
6205 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6206 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6207 This option implies that the sign of a zero result isn't significant.
6209 The default is @option{-fsigned-zeros}.
6211 @item -fno-trapping-math
6212 @opindex fno-trapping-math
6213 Compile code assuming that floating-point operations cannot generate
6214 user-visible traps. These traps include division by zero, overflow,
6215 underflow, inexact result and invalid operation. This option implies
6216 @option{-fno-signaling-nans}. Setting this option may allow faster
6217 code if one relies on ``non-stop'' IEEE arithmetic, for example.
6219 This option should never be turned on by any @option{-O} option since
6220 it can result in incorrect output for programs which depend on
6221 an exact implementation of IEEE or ISO rules/specifications for
6224 The default is @option{-ftrapping-math}.
6226 @item -frounding-math
6227 @opindex frounding-math
6228 Disable transformations and optimizations that assume default floating
6229 point rounding behavior. This is round-to-zero for all floating point
6230 to integer conversions, and round-to-nearest for all other arithmetic
6231 truncations. This option should be specified for programs that change
6232 the FP rounding mode dynamically, or that may be executed with a
6233 non-default rounding mode. This option disables constant folding of
6234 floating point expressions at compile-time (which may be affected by
6235 rounding mode) and arithmetic transformations that are unsafe in the
6236 presence of sign-dependent rounding modes.
6238 The default is @option{-fno-rounding-math}.
6240 This option is experimental and does not currently guarantee to
6241 disable all GCC optimizations that are affected by rounding mode.
6242 Future versions of GCC may provide finer control of this setting
6243 using C99's @code{FENV_ACCESS} pragma. This command line option
6244 will be used to specify the default state for @code{FENV_ACCESS}.
6246 @item -frtl-abstract-sequences
6247 @opindex frtl-abstract-sequences
6248 It is a size optimization method. This option is to find identical
6249 sequences of code, which can be turned into pseudo-procedures and
6250 then replace all occurrences with calls to the newly created
6251 subroutine. It is kind of an opposite of @option{-finline-functions}.
6252 This optimization runs at RTL level.
6254 @item -fsignaling-nans
6255 @opindex fsignaling-nans
6256 Compile code assuming that IEEE signaling NaNs may generate user-visible
6257 traps during floating-point operations. Setting this option disables
6258 optimizations that may change the number of exceptions visible with
6259 signaling NaNs. This option implies @option{-ftrapping-math}.
6261 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6264 The default is @option{-fno-signaling-nans}.
6266 This option is experimental and does not currently guarantee to
6267 disable all GCC optimizations that affect signaling NaN behavior.
6269 @item -fsingle-precision-constant
6270 @opindex fsingle-precision-constant
6271 Treat floating point constant as single precision constant instead of
6272 implicitly converting it to double precision constant.
6274 @item -fcx-limited-range
6275 @itemx -fno-cx-limited-range
6276 @opindex fcx-limited-range
6277 @opindex fno-cx-limited-range
6278 When enabled, this option states that a range reduction step is not
6279 needed when performing complex division. The default is
6280 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
6282 This option controls the default setting of the ISO C99
6283 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6288 The following options control optimizations that may improve
6289 performance, but are not enabled by any @option{-O} options. This
6290 section includes experimental options that may produce broken code.
6293 @item -fbranch-probabilities
6294 @opindex fbranch-probabilities
6295 After running a program compiled with @option{-fprofile-arcs}
6296 (@pxref{Debugging Options,, Options for Debugging Your Program or
6297 @command{gcc}}), you can compile it a second time using
6298 @option{-fbranch-probabilities}, to improve optimizations based on
6299 the number of times each branch was taken. When the program
6300 compiled with @option{-fprofile-arcs} exits it saves arc execution
6301 counts to a file called @file{@var{sourcename}.gcda} for each source
6302 file. The information in this data file is very dependent on the
6303 structure of the generated code, so you must use the same source code
6304 and the same optimization options for both compilations.
6306 With @option{-fbranch-probabilities}, GCC puts a
6307 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6308 These can be used to improve optimization. Currently, they are only
6309 used in one place: in @file{reorg.c}, instead of guessing which path a
6310 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6311 exactly determine which path is taken more often.
6313 @item -fprofile-values
6314 @opindex fprofile-values
6315 If combined with @option{-fprofile-arcs}, it adds code so that some
6316 data about values of expressions in the program is gathered.
6318 With @option{-fbranch-probabilities}, it reads back the data gathered
6319 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6320 notes to instructions for their later usage in optimizations.
6322 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6326 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6327 a code to gather information about values of expressions.
6329 With @option{-fbranch-probabilities}, it reads back the data gathered
6330 and actually performs the optimizations based on them.
6331 Currently the optimizations include specialization of division operation
6332 using the knowledge about the value of the denominator.
6334 @item -frename-registers
6335 @opindex frename-registers
6336 Attempt to avoid false dependencies in scheduled code by making use
6337 of registers left over after register allocation. This optimization
6338 will most benefit processors with lots of registers. Depending on the
6339 debug information format adopted by the target, however, it can
6340 make debugging impossible, since variables will no longer stay in
6341 a ``home register''.
6343 Enabled by default with @option{-funroll-loops}.
6347 Perform tail duplication to enlarge superblock size. This transformation
6348 simplifies the control flow of the function allowing other optimizations to do
6351 Enabled with @option{-fprofile-use}.
6353 @item -funroll-loops
6354 @opindex funroll-loops
6355 Unroll loops whose number of iterations can be determined at compile time or
6356 upon entry to the loop. @option{-funroll-loops} implies
6357 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6358 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6359 small constant number of iterations). This option makes code larger, and may
6360 or may not make it run faster.
6362 Enabled with @option{-fprofile-use}.
6364 @item -funroll-all-loops
6365 @opindex funroll-all-loops
6366 Unroll all loops, even if their number of iterations is uncertain when
6367 the loop is entered. This usually makes programs run more slowly.
6368 @option{-funroll-all-loops} implies the same options as
6369 @option{-funroll-loops}.
6372 @opindex fpeel-loops
6373 Peels the loops for that there is enough information that they do not
6374 roll much (from profile feedback). It also turns on complete loop peeling
6375 (i.e.@: complete removal of loops with small constant number of iterations).
6377 Enabled with @option{-fprofile-use}.
6379 @item -fmove-loop-invariants
6380 @opindex fmove-loop-invariants
6381 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6382 at level @option{-O1}
6384 @item -funswitch-loops
6385 @opindex funswitch-loops
6386 Move branches with loop invariant conditions out of the loop, with duplicates
6387 of the loop on both branches (modified according to result of the condition).
6389 @item -ffunction-sections
6390 @itemx -fdata-sections
6391 @opindex ffunction-sections
6392 @opindex fdata-sections
6393 Place each function or data item into its own section in the output
6394 file if the target supports arbitrary sections. The name of the
6395 function or the name of the data item determines the section's name
6398 Use these options on systems where the linker can perform optimizations
6399 to improve locality of reference in the instruction space. Most systems
6400 using the ELF object format and SPARC processors running Solaris 2 have
6401 linkers with such optimizations. AIX may have these optimizations in
6404 Only use these options when there are significant benefits from doing
6405 so. When you specify these options, the assembler and linker will
6406 create larger object and executable files and will also be slower.
6407 You will not be able to use @code{gprof} on all systems if you
6408 specify this option and you may have problems with debugging if
6409 you specify both this option and @option{-g}.
6411 @item -fbranch-target-load-optimize
6412 @opindex fbranch-target-load-optimize
6413 Perform branch target register load optimization before prologue / epilogue
6415 The use of target registers can typically be exposed only during reload,
6416 thus hoisting loads out of loops and doing inter-block scheduling needs
6417 a separate optimization pass.
6419 @item -fbranch-target-load-optimize2
6420 @opindex fbranch-target-load-optimize2
6421 Perform branch target register load optimization after prologue / epilogue
6424 @item -fbtr-bb-exclusive
6425 @opindex fbtr-bb-exclusive
6426 When performing branch target register load optimization, don't reuse
6427 branch target registers in within any basic block.
6429 @item -fstack-protector
6430 Emit extra code to check for buffer overflows, such as stack smashing
6431 attacks. This is done by adding a guard variable to functions with
6432 vulnerable objects. This includes functions that call alloca, and
6433 functions with buffers larger than 8 bytes. The guards are initialized
6434 when a function is entered and then checked when the function exits.
6435 If a guard check fails, an error message is printed and the program exits.
6437 @item -fstack-protector-all
6438 Like @option{-fstack-protector} except that all functions are protected.
6440 @item -fsection-anchors
6441 @opindex fsection-anchors
6442 Try to reduce the number of symbolic address calculations by using
6443 shared ``anchor'' symbols to address nearby objects. This transformation
6444 can help to reduce the number of GOT entries and GOT accesses on some
6447 For example, the implementation of the following function @code{foo}:
6451 int foo (void) @{ return a + b + c; @}
6454 would usually calculate the addresses of all three variables, but if you
6455 compile it with @option{-fsection-anchors}, it will access the variables
6456 from a common anchor point instead. The effect is similar to the
6457 following pseudocode (which isn't valid C):
6462 register int *xr = &x;
6463 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6467 Not all targets support this option.
6469 @item --param @var{name}=@var{value}
6471 In some places, GCC uses various constants to control the amount of
6472 optimization that is done. For example, GCC will not inline functions
6473 that contain more that a certain number of instructions. You can
6474 control some of these constants on the command-line using the
6475 @option{--param} option.
6477 The names of specific parameters, and the meaning of the values, are
6478 tied to the internals of the compiler, and are subject to change
6479 without notice in future releases.
6481 In each case, the @var{value} is an integer. The allowable choices for
6482 @var{name} are given in the following table:
6485 @item salias-max-implicit-fields
6486 The maximum number of fields in a variable without direct
6487 structure accesses for which structure aliasing will consider trying
6488 to track each field. The default is 5
6490 @item salias-max-array-elements
6491 The maximum number of elements an array can have and its elements
6492 still be tracked individually by structure aliasing. The default is 4
6494 @item sra-max-structure-size
6495 The maximum structure size, in bytes, at which the scalar replacement
6496 of aggregates (SRA) optimization will perform block copies. The
6497 default value, 0, implies that GCC will select the most appropriate
6500 @item sra-field-structure-ratio
6501 The threshold ratio (as a percentage) between instantiated fields and
6502 the complete structure size. We say that if the ratio of the number
6503 of bytes in instantiated fields to the number of bytes in the complete
6504 structure exceeds this parameter, then block copies are not used. The
6507 @item max-crossjump-edges
6508 The maximum number of incoming edges to consider for crossjumping.
6509 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6510 the number of edges incoming to each block. Increasing values mean
6511 more aggressive optimization, making the compile time increase with
6512 probably small improvement in executable size.
6514 @item min-crossjump-insns
6515 The minimum number of instructions which must be matched at the end
6516 of two blocks before crossjumping will be performed on them. This
6517 value is ignored in the case where all instructions in the block being
6518 crossjumped from are matched. The default value is 5.
6520 @item max-grow-copy-bb-insns
6521 The maximum code size expansion factor when copying basic blocks
6522 instead of jumping. The expansion is relative to a jump instruction.
6523 The default value is 8.
6525 @item max-goto-duplication-insns
6526 The maximum number of instructions to duplicate to a block that jumps
6527 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6528 passes, GCC factors computed gotos early in the compilation process,
6529 and unfactors them as late as possible. Only computed jumps at the
6530 end of a basic blocks with no more than max-goto-duplication-insns are
6531 unfactored. The default value is 8.
6533 @item max-delay-slot-insn-search
6534 The maximum number of instructions to consider when looking for an
6535 instruction to fill a delay slot. If more than this arbitrary number of
6536 instructions is searched, the time savings from filling the delay slot
6537 will be minimal so stop searching. Increasing values mean more
6538 aggressive optimization, making the compile time increase with probably
6539 small improvement in executable run time.
6541 @item max-delay-slot-live-search
6542 When trying to fill delay slots, the maximum number of instructions to
6543 consider when searching for a block with valid live register
6544 information. Increasing this arbitrarily chosen value means more
6545 aggressive optimization, increasing the compile time. This parameter
6546 should be removed when the delay slot code is rewritten to maintain the
6549 @item max-gcse-memory
6550 The approximate maximum amount of memory that will be allocated in
6551 order to perform the global common subexpression elimination
6552 optimization. If more memory than specified is required, the
6553 optimization will not be done.
6555 @item max-gcse-passes
6556 The maximum number of passes of GCSE to run. The default is 1.
6558 @item max-pending-list-length
6559 The maximum number of pending dependencies scheduling will allow
6560 before flushing the current state and starting over. Large functions
6561 with few branches or calls can create excessively large lists which
6562 needlessly consume memory and resources.
6564 @item max-inline-insns-single
6565 Several parameters control the tree inliner used in gcc.
6566 This number sets the maximum number of instructions (counted in GCC's
6567 internal representation) in a single function that the tree inliner
6568 will consider for inlining. This only affects functions declared
6569 inline and methods implemented in a class declaration (C++).
6570 The default value is 450.
6572 @item max-inline-insns-auto
6573 When you use @option{-finline-functions} (included in @option{-O3}),
6574 a lot of functions that would otherwise not be considered for inlining
6575 by the compiler will be investigated. To those functions, a different
6576 (more restrictive) limit compared to functions declared inline can
6578 The default value is 90.
6580 @item large-function-insns
6581 The limit specifying really large functions. For functions larger than this
6582 limit after inlining inlining is constrained by
6583 @option{--param large-function-growth}. This parameter is useful primarily
6584 to avoid extreme compilation time caused by non-linear algorithms used by the
6586 This parameter is ignored when @option{-funit-at-a-time} is not used.
6587 The default value is 2700.
6589 @item large-function-growth
6590 Specifies maximal growth of large function caused by inlining in percents.
6591 This parameter is ignored when @option{-funit-at-a-time} is not used.
6592 The default value is 100 which limits large function growth to 2.0 times
6595 @item large-unit-insns
6596 The limit specifying large translation unit. Growth caused by inlining of
6597 units larger than this limit is limited by @option{--param inline-unit-growth}.
6598 For small units this might be too tight (consider unit consisting of function A
6599 that is inline and B that just calls A three time. If B is small relative to
6600 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6601 large units consisting of small inlininable functions however the overall unit
6602 growth limit is needed to avoid exponential explosion of code size. Thus for
6603 smaller units, the size is increased to @option{--param large-unit-insns}
6604 before applying @option{--param inline-unit-growth}. The default is 10000
6606 @item inline-unit-growth
6607 Specifies maximal overall growth of the compilation unit caused by inlining.
6608 This parameter is ignored when @option{-funit-at-a-time} is not used.
6609 The default value is 30 which limits unit growth to 1.3 times the original
6612 @item large-stack-frame
6613 The limit specifying large stack frames. While inlining the algorithm is trying
6614 to not grow past this limit too much. Default value is 256 bytes.
6616 @item large-stack-frame-growth
6617 Specifies maximal growth of large stack frames caused by inlining in percents.
6618 The default value is 1000 which limits large stack frame growth to 11 times
6621 @item max-inline-insns-recursive
6622 @itemx max-inline-insns-recursive-auto
6623 Specifies maximum number of instructions out-of-line copy of self recursive inline
6624 function can grow into by performing recursive inlining.
6626 For functions declared inline @option{--param max-inline-insns-recursive} is
6627 taken into account. For function not declared inline, recursive inlining
6628 happens only when @option{-finline-functions} (included in @option{-O3}) is
6629 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6630 default value is 450.
6632 @item max-inline-recursive-depth
6633 @itemx max-inline-recursive-depth-auto
6634 Specifies maximum recursion depth used by the recursive inlining.
6636 For functions declared inline @option{--param max-inline-recursive-depth} is
6637 taken into account. For function not declared inline, recursive inlining
6638 happens only when @option{-finline-functions} (included in @option{-O3}) is
6639 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6640 default value is 450.
6642 @item min-inline-recursive-probability
6643 Recursive inlining is profitable only for function having deep recursion
6644 in average and can hurt for function having little recursion depth by
6645 increasing the prologue size or complexity of function body to other
6648 When profile feedback is available (see @option{-fprofile-generate}) the actual
6649 recursion depth can be guessed from probability that function will recurse via
6650 given call expression. This parameter limits inlining only to call expression
6651 whose probability exceeds given threshold (in percents). The default value is
6654 @item inline-call-cost
6655 Specify cost of call instruction relative to simple arithmetics operations
6656 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6657 functions and at the same time increases size of leaf function that is believed to
6658 reduce function size by being inlined. In effect it increases amount of
6659 inlining for code having large abstraction penalty (many functions that just
6660 pass the arguments to other functions) and decrease inlining for code with low
6661 abstraction penalty. The default value is 16.
6663 @item min-vect-loop-bound
6664 The minimum number of iterations under which a loop will not get vectorized
6665 when @option{-ftree-vectorize} is used. The number of iterations after
6666 vectorization needs to be greater than the value specified by this option
6667 to allow vectorization. The default value is 0.
6669 @item max-unrolled-insns
6670 The maximum number of instructions that a loop should have if that loop
6671 is unrolled, and if the loop is unrolled, it determines how many times
6672 the loop code is unrolled.
6674 @item max-average-unrolled-insns
6675 The maximum number of instructions biased by probabilities of their execution
6676 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6677 it determines how many times the loop code is unrolled.
6679 @item max-unroll-times
6680 The maximum number of unrollings of a single loop.
6682 @item max-peeled-insns
6683 The maximum number of instructions that a loop should have if that loop
6684 is peeled, and if the loop is peeled, it determines how many times
6685 the loop code is peeled.
6687 @item max-peel-times
6688 The maximum number of peelings of a single loop.
6690 @item max-completely-peeled-insns
6691 The maximum number of insns of a completely peeled loop.
6693 @item max-completely-peel-times
6694 The maximum number of iterations of a loop to be suitable for complete peeling.
6696 @item max-unswitch-insns
6697 The maximum number of insns of an unswitched loop.
6699 @item max-unswitch-level
6700 The maximum number of branches unswitched in a single loop.
6703 The minimum cost of an expensive expression in the loop invariant motion.
6705 @item iv-consider-all-candidates-bound
6706 Bound on number of candidates for induction variables below that
6707 all candidates are considered for each use in induction variable
6708 optimizations. Only the most relevant candidates are considered
6709 if there are more candidates, to avoid quadratic time complexity.
6711 @item iv-max-considered-uses
6712 The induction variable optimizations give up on loops that contain more
6713 induction variable uses.
6715 @item iv-always-prune-cand-set-bound
6716 If number of candidates in the set is smaller than this value,
6717 we always try to remove unnecessary ivs from the set during its
6718 optimization when a new iv is added to the set.
6720 @item scev-max-expr-size
6721 Bound on size of expressions used in the scalar evolutions analyzer.
6722 Large expressions slow the analyzer.
6724 @item omega-max-vars
6725 The maximum number of variables in an Omega constraint system.
6726 The default value is 128.
6728 @item omega-max-geqs
6729 The maximum number of inequalities in an Omega constraint system.
6730 The default value is 256.
6733 The maximum number of equalities in an Omega constraint system.
6734 The default value is 128.
6736 @item omega-max-wild-cards
6737 The maximum number of wildcard variables that the Omega solver will
6738 be able to insert. The default value is 18.
6740 @item omega-hash-table-size
6741 The size of the hash table in the Omega solver. The default value is
6744 @item omega-max-keys
6745 The maximal number of keys used by the Omega solver. The default
6748 @item omega-eliminate-redundant-constraints
6749 When set to 1, use expensive methods to eliminate all redundant
6750 constraints. The default value is 0.
6752 @item vect-max-version-for-alignment-checks
6753 The maximum number of runtime checks that can be performed when
6754 doing loop versioning for alignment in the vectorizer. See option
6755 ftree-vect-loop-version for more information.
6757 @item vect-max-version-for-alias-checks
6758 The maximum number of runtime checks that can be performed when
6759 doing loop versioning for alias in the vectorizer. See option
6760 ftree-vect-loop-version for more information.
6762 @item max-iterations-to-track
6764 The maximum number of iterations of a loop the brute force algorithm
6765 for analysis of # of iterations of the loop tries to evaluate.
6767 @item hot-bb-count-fraction
6768 Select fraction of the maximal count of repetitions of basic block in program
6769 given basic block needs to have to be considered hot.
6771 @item hot-bb-frequency-fraction
6772 Select fraction of the maximal frequency of executions of basic block in
6773 function given basic block needs to have to be considered hot
6775 @item max-predicted-iterations
6776 The maximum number of loop iterations we predict statically. This is useful
6777 in cases where function contain single loop with known bound and other loop
6778 with unknown. We predict the known number of iterations correctly, while
6779 the unknown number of iterations average to roughly 10. This means that the
6780 loop without bounds would appear artificially cold relative to the other one.
6782 @item tracer-dynamic-coverage
6783 @itemx tracer-dynamic-coverage-feedback
6785 This value is used to limit superblock formation once the given percentage of
6786 executed instructions is covered. This limits unnecessary code size
6789 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6790 feedback is available. The real profiles (as opposed to statically estimated
6791 ones) are much less balanced allowing the threshold to be larger value.
6793 @item tracer-max-code-growth
6794 Stop tail duplication once code growth has reached given percentage. This is
6795 rather hokey argument, as most of the duplicates will be eliminated later in
6796 cross jumping, so it may be set to much higher values than is the desired code
6799 @item tracer-min-branch-ratio
6801 Stop reverse growth when the reverse probability of best edge is less than this
6802 threshold (in percent).
6804 @item tracer-min-branch-ratio
6805 @itemx tracer-min-branch-ratio-feedback
6807 Stop forward growth if the best edge do have probability lower than this
6810 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6811 compilation for profile feedback and one for compilation without. The value
6812 for compilation with profile feedback needs to be more conservative (higher) in
6813 order to make tracer effective.
6815 @item max-cse-path-length
6817 Maximum number of basic blocks on path that cse considers. The default is 10.
6820 The maximum instructions CSE process before flushing. The default is 1000.
6822 @item max-aliased-vops
6824 Maximum number of virtual operands per function allowed to represent
6825 aliases before triggering the alias partitioning heuristic. Alias
6826 partitioning reduces compile times and memory consumption needed for
6827 aliasing at the expense of precision loss in alias information. The
6828 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
6831 Notice that if a function contains more memory statements than the
6832 value of this parameter, it is not really possible to achieve this
6833 reduction. In this case, the compiler will use the number of memory
6834 statements as the value for @option{max-aliased-vops}.
6836 @item avg-aliased-vops
6838 Average number of virtual operands per statement allowed to represent
6839 aliases before triggering the alias partitioning heuristic. This
6840 works in conjunction with @option{max-aliased-vops}. If a function
6841 contains more than @option{max-aliased-vops} virtual operators, then
6842 memory symbols will be grouped into memory partitions until either the
6843 total number of virtual operators is below @option{max-aliased-vops}
6844 or the average number of virtual operators per memory statement is
6845 below @option{avg-aliased-vops}. The default value for this parameter
6846 is 1 for -O1 and -O2, and 3 for -O3.
6848 @item ggc-min-expand
6850 GCC uses a garbage collector to manage its own memory allocation. This
6851 parameter specifies the minimum percentage by which the garbage
6852 collector's heap should be allowed to expand between collections.
6853 Tuning this may improve compilation speed; it has no effect on code
6856 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6857 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6858 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6859 GCC is not able to calculate RAM on a particular platform, the lower
6860 bound of 30% is used. Setting this parameter and
6861 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6862 every opportunity. This is extremely slow, but can be useful for
6865 @item ggc-min-heapsize
6867 Minimum size of the garbage collector's heap before it begins bothering
6868 to collect garbage. The first collection occurs after the heap expands
6869 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6870 tuning this may improve compilation speed, and has no effect on code
6873 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6874 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6875 with a lower bound of 4096 (four megabytes) and an upper bound of
6876 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6877 particular platform, the lower bound is used. Setting this parameter
6878 very large effectively disables garbage collection. Setting this
6879 parameter and @option{ggc-min-expand} to zero causes a full collection
6880 to occur at every opportunity.
6882 @item max-reload-search-insns
6883 The maximum number of instruction reload should look backward for equivalent
6884 register. Increasing values mean more aggressive optimization, making the
6885 compile time increase with probably slightly better performance. The default
6888 @item max-cselib-memory-locations
6889 The maximum number of memory locations cselib should take into account.
6890 Increasing values mean more aggressive optimization, making the compile time
6891 increase with probably slightly better performance. The default value is 500.
6893 @item max-flow-memory-locations
6894 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6895 The default value is 100.
6897 @item reorder-blocks-duplicate
6898 @itemx reorder-blocks-duplicate-feedback
6900 Used by basic block reordering pass to decide whether to use unconditional
6901 branch or duplicate the code on its destination. Code is duplicated when its
6902 estimated size is smaller than this value multiplied by the estimated size of
6903 unconditional jump in the hot spots of the program.
6905 The @option{reorder-block-duplicate-feedback} is used only when profile
6906 feedback is available and may be set to higher values than
6907 @option{reorder-block-duplicate} since information about the hot spots is more
6910 @item max-sched-ready-insns
6911 The maximum number of instructions ready to be issued the scheduler should
6912 consider at any given time during the first scheduling pass. Increasing
6913 values mean more thorough searches, making the compilation time increase
6914 with probably little benefit. The default value is 100.
6916 @item max-sched-region-blocks
6917 The maximum number of blocks in a region to be considered for
6918 interblock scheduling. The default value is 10.
6920 @item max-sched-region-insns
6921 The maximum number of insns in a region to be considered for
6922 interblock scheduling. The default value is 100.
6925 The minimum probability (in percents) of reaching a source block
6926 for interblock speculative scheduling. The default value is 40.
6928 @item max-sched-extend-regions-iters
6929 The maximum number of iterations through CFG to extend regions.
6930 0 - disable region extension,
6931 N - do at most N iterations.
6932 The default value is 0.
6934 @item max-sched-insn-conflict-delay
6935 The maximum conflict delay for an insn to be considered for speculative motion.
6936 The default value is 3.
6938 @item sched-spec-prob-cutoff
6939 The minimal probability of speculation success (in percents), so that
6940 speculative insn will be scheduled.
6941 The default value is 40.
6943 @item max-last-value-rtl
6945 The maximum size measured as number of RTLs that can be recorded in an expression
6946 in combiner for a pseudo register as last known value of that register. The default
6949 @item integer-share-limit
6950 Small integer constants can use a shared data structure, reducing the
6951 compiler's memory usage and increasing its speed. This sets the maximum
6952 value of a shared integer constant's. The default value is 256.
6954 @item min-virtual-mappings
6955 Specifies the minimum number of virtual mappings in the incremental
6956 SSA updater that should be registered to trigger the virtual mappings
6957 heuristic defined by virtual-mappings-ratio. The default value is
6960 @item virtual-mappings-ratio
6961 If the number of virtual mappings is virtual-mappings-ratio bigger
6962 than the number of virtual symbols to be updated, then the incremental
6963 SSA updater switches to a full update for those symbols. The default
6966 @item ssp-buffer-size
6967 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6968 protection when @option{-fstack-protection} is used.
6970 @item max-jump-thread-duplication-stmts
6971 Maximum number of statements allowed in a block that needs to be
6972 duplicated when threading jumps.
6974 @item max-fields-for-field-sensitive
6975 Maximum number of fields in a structure we will treat in
6976 a field sensitive manner during pointer analysis.
6978 @item prefetch-latency
6979 Estimate on average number of instructions that are executed before
6980 prefetch finishes. The distance we prefetch ahead is proportional
6981 to this constant. Increasing this number may also lead to less
6982 streams being prefetched (see @option{simultaneous-prefetches}).
6984 @item simultaneous-prefetches
6985 Maximum number of prefetches that can run at the same time.
6987 @item l1-cache-line-size
6988 The size of cache line in L1 cache, in bytes.
6991 The size of L1 cache, in kilobytes.
6994 The size of L2 cache, in kilobytes.
6996 @item use-canonical-types
6997 Whether the compiler should use the ``canonical'' type system. By
6998 default, this should always be 1, which uses a more efficient internal
6999 mechanism for comparing types in C++ and Objective-C++. However, if
7000 bugs in the canonical type system are causing compilation failures,
7001 set this value to 0 to disable canonical types.
7006 @node Preprocessor Options
7007 @section Options Controlling the Preprocessor
7008 @cindex preprocessor options
7009 @cindex options, preprocessor
7011 These options control the C preprocessor, which is run on each C source
7012 file before actual compilation.
7014 If you use the @option{-E} option, nothing is done except preprocessing.
7015 Some of these options make sense only together with @option{-E} because
7016 they cause the preprocessor output to be unsuitable for actual
7021 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7022 and pass @var{option} directly through to the preprocessor. If
7023 @var{option} contains commas, it is split into multiple options at the
7024 commas. However, many options are modified, translated or interpreted
7025 by the compiler driver before being passed to the preprocessor, and
7026 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7027 interface is undocumented and subject to change, so whenever possible
7028 you should avoid using @option{-Wp} and let the driver handle the
7031 @item -Xpreprocessor @var{option}
7032 @opindex preprocessor
7033 Pass @var{option} as an option to the preprocessor. You can use this to
7034 supply system-specific preprocessor options which GCC does not know how to
7037 If you want to pass an option that takes an argument, you must use
7038 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7041 @include cppopts.texi
7043 @node Assembler Options
7044 @section Passing Options to the Assembler
7046 @c prevent bad page break with this line
7047 You can pass options to the assembler.
7050 @item -Wa,@var{option}
7052 Pass @var{option} as an option to the assembler. If @var{option}
7053 contains commas, it is split into multiple options at the commas.
7055 @item -Xassembler @var{option}
7057 Pass @var{option} as an option to the assembler. You can use this to
7058 supply system-specific assembler options which GCC does not know how to
7061 If you want to pass an option that takes an argument, you must use
7062 @option{-Xassembler} twice, once for the option and once for the argument.
7067 @section Options for Linking
7068 @cindex link options
7069 @cindex options, linking
7071 These options come into play when the compiler links object files into
7072 an executable output file. They are meaningless if the compiler is
7073 not doing a link step.
7077 @item @var{object-file-name}
7078 A file name that does not end in a special recognized suffix is
7079 considered to name an object file or library. (Object files are
7080 distinguished from libraries by the linker according to the file
7081 contents.) If linking is done, these object files are used as input
7090 If any of these options is used, then the linker is not run, and
7091 object file names should not be used as arguments. @xref{Overall
7095 @item -l@var{library}
7096 @itemx -l @var{library}
7098 Search the library named @var{library} when linking. (The second
7099 alternative with the library as a separate argument is only for
7100 POSIX compliance and is not recommended.)
7102 It makes a difference where in the command you write this option; the
7103 linker searches and processes libraries and object files in the order they
7104 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7105 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7106 to functions in @samp{z}, those functions may not be loaded.
7108 The linker searches a standard list of directories for the library,
7109 which is actually a file named @file{lib@var{library}.a}. The linker
7110 then uses this file as if it had been specified precisely by name.
7112 The directories searched include several standard system directories
7113 plus any that you specify with @option{-L}.
7115 Normally the files found this way are library files---archive files
7116 whose members are object files. The linker handles an archive file by
7117 scanning through it for members which define symbols that have so far
7118 been referenced but not defined. But if the file that is found is an
7119 ordinary object file, it is linked in the usual fashion. The only
7120 difference between using an @option{-l} option and specifying a file name
7121 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7122 and searches several directories.
7126 You need this special case of the @option{-l} option in order to
7127 link an Objective-C or Objective-C++ program.
7130 @opindex nostartfiles
7131 Do not use the standard system startup files when linking.
7132 The standard system libraries are used normally, unless @option{-nostdlib}
7133 or @option{-nodefaultlibs} is used.
7135 @item -nodefaultlibs
7136 @opindex nodefaultlibs
7137 Do not use the standard system libraries when linking.
7138 Only the libraries you specify will be passed to the linker.
7139 The standard startup files are used normally, unless @option{-nostartfiles}
7140 is used. The compiler may generate calls to @code{memcmp},
7141 @code{memset}, @code{memcpy} and @code{memmove}.
7142 These entries are usually resolved by entries in
7143 libc. These entry points should be supplied through some other
7144 mechanism when this option is specified.
7148 Do not use the standard system startup files or libraries when linking.
7149 No startup files and only the libraries you specify will be passed to
7150 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7151 @code{memcpy} and @code{memmove}.
7152 These entries are usually resolved by entries in
7153 libc. These entry points should be supplied through some other
7154 mechanism when this option is specified.
7156 @cindex @option{-lgcc}, use with @option{-nostdlib}
7157 @cindex @option{-nostdlib} and unresolved references
7158 @cindex unresolved references and @option{-nostdlib}
7159 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7160 @cindex @option{-nodefaultlibs} and unresolved references
7161 @cindex unresolved references and @option{-nodefaultlibs}
7162 One of the standard libraries bypassed by @option{-nostdlib} and
7163 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7164 that GCC uses to overcome shortcomings of particular machines, or special
7165 needs for some languages.
7166 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7167 Collection (GCC) Internals},
7168 for more discussion of @file{libgcc.a}.)
7169 In most cases, you need @file{libgcc.a} even when you want to avoid
7170 other standard libraries. In other words, when you specify @option{-nostdlib}
7171 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7172 This ensures that you have no unresolved references to internal GCC
7173 library subroutines. (For example, @samp{__main}, used to ensure C++
7174 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7175 GNU Compiler Collection (GCC) Internals}.)
7179 Produce a position independent executable on targets which support it.
7180 For predictable results, you must also specify the same set of options
7181 that were used to generate code (@option{-fpie}, @option{-fPIE},
7182 or model suboptions) when you specify this option.
7186 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7187 that support it. This instructs the linker to add all symbols, not
7188 only used ones, to the dynamic symbol table. This option is needed
7189 for some uses of @code{dlopen} or to allow obtaining backtraces
7190 from within a program.
7194 Remove all symbol table and relocation information from the executable.
7198 On systems that support dynamic linking, this prevents linking with the shared
7199 libraries. On other systems, this option has no effect.
7203 Produce a shared object which can then be linked with other objects to
7204 form an executable. Not all systems support this option. For predictable
7205 results, you must also specify the same set of options that were used to
7206 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7207 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7208 needs to build supplementary stub code for constructors to work. On
7209 multi-libbed systems, @samp{gcc -shared} must select the correct support
7210 libraries to link against. Failing to supply the correct flags may lead
7211 to subtle defects. Supplying them in cases where they are not necessary
7214 @item -shared-libgcc
7215 @itemx -static-libgcc
7216 @opindex shared-libgcc
7217 @opindex static-libgcc
7218 On systems that provide @file{libgcc} as a shared library, these options
7219 force the use of either the shared or static version respectively.
7220 If no shared version of @file{libgcc} was built when the compiler was
7221 configured, these options have no effect.
7223 There are several situations in which an application should use the
7224 shared @file{libgcc} instead of the static version. The most common
7225 of these is when the application wishes to throw and catch exceptions
7226 across different shared libraries. In that case, each of the libraries
7227 as well as the application itself should use the shared @file{libgcc}.
7229 Therefore, the G++ and GCJ drivers automatically add
7230 @option{-shared-libgcc} whenever you build a shared library or a main
7231 executable, because C++ and Java programs typically use exceptions, so
7232 this is the right thing to do.
7234 If, instead, you use the GCC driver to create shared libraries, you may
7235 find that they will not always be linked with the shared @file{libgcc}.
7236 If GCC finds, at its configuration time, that you have a non-GNU linker
7237 or a GNU linker that does not support option @option{--eh-frame-hdr},
7238 it will link the shared version of @file{libgcc} into shared libraries
7239 by default. Otherwise, it will take advantage of the linker and optimize
7240 away the linking with the shared version of @file{libgcc}, linking with
7241 the static version of libgcc by default. This allows exceptions to
7242 propagate through such shared libraries, without incurring relocation
7243 costs at library load time.
7245 However, if a library or main executable is supposed to throw or catch
7246 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7247 for the languages used in the program, or using the option
7248 @option{-shared-libgcc}, such that it is linked with the shared
7253 Bind references to global symbols when building a shared object. Warn
7254 about any unresolved references (unless overridden by the link editor
7255 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7258 @item -Xlinker @var{option}
7260 Pass @var{option} as an option to the linker. You can use this to
7261 supply system-specific linker options which GCC does not know how to
7264 If you want to pass an option that takes an argument, you must use
7265 @option{-Xlinker} twice, once for the option and once for the argument.
7266 For example, to pass @option{-assert definitions}, you must write
7267 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7268 @option{-Xlinker "-assert definitions"}, because this passes the entire
7269 string as a single argument, which is not what the linker expects.
7271 @item -Wl,@var{option}
7273 Pass @var{option} as an option to the linker. If @var{option} contains
7274 commas, it is split into multiple options at the commas.
7276 @item -u @var{symbol}
7278 Pretend the symbol @var{symbol} is undefined, to force linking of
7279 library modules to define it. You can use @option{-u} multiple times with
7280 different symbols to force loading of additional library modules.
7283 @node Directory Options
7284 @section Options for Directory Search
7285 @cindex directory options
7286 @cindex options, directory search
7289 These options specify directories to search for header files, for
7290 libraries and for parts of the compiler:
7295 Add the directory @var{dir} to the head of the list of directories to be
7296 searched for header files. This can be used to override a system header
7297 file, substituting your own version, since these directories are
7298 searched before the system header file directories. However, you should
7299 not use this option to add directories that contain vendor-supplied
7300 system header files (use @option{-isystem} for that). If you use more than
7301 one @option{-I} option, the directories are scanned in left-to-right
7302 order; the standard system directories come after.
7304 If a standard system include directory, or a directory specified with
7305 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7306 option will be ignored. The directory will still be searched but as a
7307 system directory at its normal position in the system include chain.
7308 This is to ensure that GCC's procedure to fix buggy system headers and
7309 the ordering for the include_next directive are not inadvertently changed.
7310 If you really need to change the search order for system directories,
7311 use the @option{-nostdinc} and/or @option{-isystem} options.
7313 @item -iquote@var{dir}
7315 Add the directory @var{dir} to the head of the list of directories to
7316 be searched for header files only for the case of @samp{#include
7317 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7318 otherwise just like @option{-I}.
7322 Add directory @var{dir} to the list of directories to be searched
7325 @item -B@var{prefix}
7327 This option specifies where to find the executables, libraries,
7328 include files, and data files of the compiler itself.
7330 The compiler driver program runs one or more of the subprograms
7331 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7332 @var{prefix} as a prefix for each program it tries to run, both with and
7333 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7335 For each subprogram to be run, the compiler driver first tries the
7336 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7337 was not specified, the driver tries two standard prefixes, which are
7338 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7339 those results in a file name that is found, the unmodified program
7340 name is searched for using the directories specified in your
7341 @env{PATH} environment variable.
7343 The compiler will check to see if the path provided by the @option{-B}
7344 refers to a directory, and if necessary it will add a directory
7345 separator character at the end of the path.
7347 @option{-B} prefixes that effectively specify directory names also apply
7348 to libraries in the linker, because the compiler translates these
7349 options into @option{-L} options for the linker. They also apply to
7350 includes files in the preprocessor, because the compiler translates these
7351 options into @option{-isystem} options for the preprocessor. In this case,
7352 the compiler appends @samp{include} to the prefix.
7354 The run-time support file @file{libgcc.a} can also be searched for using
7355 the @option{-B} prefix, if needed. If it is not found there, the two
7356 standard prefixes above are tried, and that is all. The file is left
7357 out of the link if it is not found by those means.
7359 Another way to specify a prefix much like the @option{-B} prefix is to use
7360 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7363 As a special kludge, if the path provided by @option{-B} is
7364 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7365 9, then it will be replaced by @file{[dir/]include}. This is to help
7366 with boot-strapping the compiler.
7368 @item -specs=@var{file}
7370 Process @var{file} after the compiler reads in the standard @file{specs}
7371 file, in order to override the defaults that the @file{gcc} driver
7372 program uses when determining what switches to pass to @file{cc1},
7373 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7374 @option{-specs=@var{file}} can be specified on the command line, and they
7375 are processed in order, from left to right.
7377 @item --sysroot=@var{dir}
7379 Use @var{dir} as the logical root directory for headers and libraries.
7380 For example, if the compiler would normally search for headers in
7381 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7382 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7384 If you use both this option and the @option{-isysroot} option, then
7385 the @option{--sysroot} option will apply to libraries, but the
7386 @option{-isysroot} option will apply to header files.
7388 The GNU linker (beginning with version 2.16) has the necessary support
7389 for this option. If your linker does not support this option, the
7390 header file aspect of @option{--sysroot} will still work, but the
7391 library aspect will not.
7395 This option has been deprecated. Please use @option{-iquote} instead for
7396 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7397 Any directories you specify with @option{-I} options before the @option{-I-}
7398 option are searched only for the case of @samp{#include "@var{file}"};
7399 they are not searched for @samp{#include <@var{file}>}.
7401 If additional directories are specified with @option{-I} options after
7402 the @option{-I-}, these directories are searched for all @samp{#include}
7403 directives. (Ordinarily @emph{all} @option{-I} directories are used
7406 In addition, the @option{-I-} option inhibits the use of the current
7407 directory (where the current input file came from) as the first search
7408 directory for @samp{#include "@var{file}"}. There is no way to
7409 override this effect of @option{-I-}. With @option{-I.} you can specify
7410 searching the directory which was current when the compiler was
7411 invoked. That is not exactly the same as what the preprocessor does
7412 by default, but it is often satisfactory.
7414 @option{-I-} does not inhibit the use of the standard system directories
7415 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7422 @section Specifying subprocesses and the switches to pass to them
7425 @command{gcc} is a driver program. It performs its job by invoking a
7426 sequence of other programs to do the work of compiling, assembling and
7427 linking. GCC interprets its command-line parameters and uses these to
7428 deduce which programs it should invoke, and which command-line options
7429 it ought to place on their command lines. This behavior is controlled
7430 by @dfn{spec strings}. In most cases there is one spec string for each
7431 program that GCC can invoke, but a few programs have multiple spec
7432 strings to control their behavior. The spec strings built into GCC can
7433 be overridden by using the @option{-specs=} command-line switch to specify
7436 @dfn{Spec files} are plaintext files that are used to construct spec
7437 strings. They consist of a sequence of directives separated by blank
7438 lines. The type of directive is determined by the first non-whitespace
7439 character on the line and it can be one of the following:
7442 @item %@var{command}
7443 Issues a @var{command} to the spec file processor. The commands that can
7447 @item %include <@var{file}>
7449 Search for @var{file} and insert its text at the current point in the
7452 @item %include_noerr <@var{file}>
7453 @cindex %include_noerr
7454 Just like @samp{%include}, but do not generate an error message if the include
7455 file cannot be found.
7457 @item %rename @var{old_name} @var{new_name}
7459 Rename the spec string @var{old_name} to @var{new_name}.
7463 @item *[@var{spec_name}]:
7464 This tells the compiler to create, override or delete the named spec
7465 string. All lines after this directive up to the next directive or
7466 blank line are considered to be the text for the spec string. If this
7467 results in an empty string then the spec will be deleted. (Or, if the
7468 spec did not exist, then nothing will happened.) Otherwise, if the spec
7469 does not currently exist a new spec will be created. If the spec does
7470 exist then its contents will be overridden by the text of this
7471 directive, unless the first character of that text is the @samp{+}
7472 character, in which case the text will be appended to the spec.
7474 @item [@var{suffix}]:
7475 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7476 and up to the next directive or blank line are considered to make up the
7477 spec string for the indicated suffix. When the compiler encounters an
7478 input file with the named suffix, it will processes the spec string in
7479 order to work out how to compile that file. For example:
7486 This says that any input file whose name ends in @samp{.ZZ} should be
7487 passed to the program @samp{z-compile}, which should be invoked with the
7488 command-line switch @option{-input} and with the result of performing the
7489 @samp{%i} substitution. (See below.)
7491 As an alternative to providing a spec string, the text that follows a
7492 suffix directive can be one of the following:
7495 @item @@@var{language}
7496 This says that the suffix is an alias for a known @var{language}. This is
7497 similar to using the @option{-x} command-line switch to GCC to specify a
7498 language explicitly. For example:
7505 Says that .ZZ files are, in fact, C++ source files.
7508 This causes an error messages saying:
7511 @var{name} compiler not installed on this system.
7515 GCC already has an extensive list of suffixes built into it.
7516 This directive will add an entry to the end of the list of suffixes, but
7517 since the list is searched from the end backwards, it is effectively
7518 possible to override earlier entries using this technique.
7522 GCC has the following spec strings built into it. Spec files can
7523 override these strings or create their own. Note that individual
7524 targets can also add their own spec strings to this list.
7527 asm Options to pass to the assembler
7528 asm_final Options to pass to the assembler post-processor
7529 cpp Options to pass to the C preprocessor
7530 cc1 Options to pass to the C compiler
7531 cc1plus Options to pass to the C++ compiler
7532 endfile Object files to include at the end of the link
7533 link Options to pass to the linker
7534 lib Libraries to include on the command line to the linker
7535 libgcc Decides which GCC support library to pass to the linker
7536 linker Sets the name of the linker
7537 predefines Defines to be passed to the C preprocessor
7538 signed_char Defines to pass to CPP to say whether @code{char} is signed
7540 startfile Object files to include at the start of the link
7543 Here is a small example of a spec file:
7549 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7552 This example renames the spec called @samp{lib} to @samp{old_lib} and
7553 then overrides the previous definition of @samp{lib} with a new one.
7554 The new definition adds in some extra command-line options before
7555 including the text of the old definition.
7557 @dfn{Spec strings} are a list of command-line options to be passed to their
7558 corresponding program. In addition, the spec strings can contain
7559 @samp{%}-prefixed sequences to substitute variable text or to
7560 conditionally insert text into the command line. Using these constructs
7561 it is possible to generate quite complex command lines.
7563 Here is a table of all defined @samp{%}-sequences for spec
7564 strings. Note that spaces are not generated automatically around the
7565 results of expanding these sequences. Therefore you can concatenate them
7566 together or combine them with constant text in a single argument.
7570 Substitute one @samp{%} into the program name or argument.
7573 Substitute the name of the input file being processed.
7576 Substitute the basename of the input file being processed.
7577 This is the substring up to (and not including) the last period
7578 and not including the directory.
7581 This is the same as @samp{%b}, but include the file suffix (text after
7585 Marks the argument containing or following the @samp{%d} as a
7586 temporary file name, so that that file will be deleted if GCC exits
7587 successfully. Unlike @samp{%g}, this contributes no text to the
7590 @item %g@var{suffix}
7591 Substitute a file name that has suffix @var{suffix} and is chosen
7592 once per compilation, and mark the argument in the same way as
7593 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7594 name is now chosen in a way that is hard to predict even when previously
7595 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7596 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7597 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7598 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7599 was simply substituted with a file name chosen once per compilation,
7600 without regard to any appended suffix (which was therefore treated
7601 just like ordinary text), making such attacks more likely to succeed.
7603 @item %u@var{suffix}
7604 Like @samp{%g}, but generates a new temporary file name even if
7605 @samp{%u@var{suffix}} was already seen.
7607 @item %U@var{suffix}
7608 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7609 new one if there is no such last file name. In the absence of any
7610 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7611 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7612 would involve the generation of two distinct file names, one
7613 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7614 simply substituted with a file name chosen for the previous @samp{%u},
7615 without regard to any appended suffix.
7617 @item %j@var{suffix}
7618 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7619 writable, and if save-temps is off; otherwise, substitute the name
7620 of a temporary file, just like @samp{%u}. This temporary file is not
7621 meant for communication between processes, but rather as a junk
7624 @item %|@var{suffix}
7625 @itemx %m@var{suffix}
7626 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7627 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7628 all. These are the two most common ways to instruct a program that it
7629 should read from standard input or write to standard output. If you
7630 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7631 construct: see for example @file{f/lang-specs.h}.
7633 @item %.@var{SUFFIX}
7634 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7635 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7636 terminated by the next space or %.
7639 Marks the argument containing or following the @samp{%w} as the
7640 designated output file of this compilation. This puts the argument
7641 into the sequence of arguments that @samp{%o} will substitute later.
7644 Substitutes the names of all the output files, with spaces
7645 automatically placed around them. You should write spaces
7646 around the @samp{%o} as well or the results are undefined.
7647 @samp{%o} is for use in the specs for running the linker.
7648 Input files whose names have no recognized suffix are not compiled
7649 at all, but they are included among the output files, so they will
7653 Substitutes the suffix for object files. Note that this is
7654 handled specially when it immediately follows @samp{%g, %u, or %U},
7655 because of the need for those to form complete file names. The
7656 handling is such that @samp{%O} is treated exactly as if it had already
7657 been substituted, except that @samp{%g, %u, and %U} do not currently
7658 support additional @var{suffix} characters following @samp{%O} as they would
7659 following, for example, @samp{.o}.
7662 Substitutes the standard macro predefinitions for the
7663 current target machine. Use this when running @code{cpp}.
7666 Like @samp{%p}, but puts @samp{__} before and after the name of each
7667 predefined macro, except for macros that start with @samp{__} or with
7668 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7672 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7673 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7674 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7675 and @option{-imultilib} as necessary.
7678 Current argument is the name of a library or startup file of some sort.
7679 Search for that file in a standard list of directories and substitute
7680 the full name found.
7683 Print @var{str} as an error message. @var{str} is terminated by a newline.
7684 Use this when inconsistent options are detected.
7687 Substitute the contents of spec string @var{name} at this point.
7690 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7692 @item %x@{@var{option}@}
7693 Accumulate an option for @samp{%X}.
7696 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7700 Output the accumulated assembler options specified by @option{-Wa}.
7703 Output the accumulated preprocessor options specified by @option{-Wp}.
7706 Process the @code{asm} spec. This is used to compute the
7707 switches to be passed to the assembler.
7710 Process the @code{asm_final} spec. This is a spec string for
7711 passing switches to an assembler post-processor, if such a program is
7715 Process the @code{link} spec. This is the spec for computing the
7716 command line passed to the linker. Typically it will make use of the
7717 @samp{%L %G %S %D and %E} sequences.
7720 Dump out a @option{-L} option for each directory that GCC believes might
7721 contain startup files. If the target supports multilibs then the
7722 current multilib directory will be prepended to each of these paths.
7725 Process the @code{lib} spec. This is a spec string for deciding which
7726 libraries should be included on the command line to the linker.
7729 Process the @code{libgcc} spec. This is a spec string for deciding
7730 which GCC support library should be included on the command line to the linker.
7733 Process the @code{startfile} spec. This is a spec for deciding which
7734 object files should be the first ones passed to the linker. Typically
7735 this might be a file named @file{crt0.o}.
7738 Process the @code{endfile} spec. This is a spec string that specifies
7739 the last object files that will be passed to the linker.
7742 Process the @code{cpp} spec. This is used to construct the arguments
7743 to be passed to the C preprocessor.
7746 Process the @code{cc1} spec. This is used to construct the options to be
7747 passed to the actual C compiler (@samp{cc1}).
7750 Process the @code{cc1plus} spec. This is used to construct the options to be
7751 passed to the actual C++ compiler (@samp{cc1plus}).
7754 Substitute the variable part of a matched option. See below.
7755 Note that each comma in the substituted string is replaced by
7759 Remove all occurrences of @code{-S} from the command line. Note---this
7760 command is position dependent. @samp{%} commands in the spec string
7761 before this one will see @code{-S}, @samp{%} commands in the spec string
7762 after this one will not.
7764 @item %:@var{function}(@var{args})
7765 Call the named function @var{function}, passing it @var{args}.
7766 @var{args} is first processed as a nested spec string, then split
7767 into an argument vector in the usual fashion. The function returns
7768 a string which is processed as if it had appeared literally as part
7769 of the current spec.
7771 The following built-in spec functions are provided:
7775 The @code{getenv} spec function takes two arguments: an environment
7776 variable name and a string. If the environment variable is not
7777 defined, a fatal error is issued. Otherwise, the return value is the
7778 value of the environment variable concatenated with the string. For
7779 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
7782 %:getenv(TOPDIR /include)
7785 expands to @file{/path/to/top/include}.
7787 @item @code{if-exists}
7788 The @code{if-exists} spec function takes one argument, an absolute
7789 pathname to a file. If the file exists, @code{if-exists} returns the
7790 pathname. Here is a small example of its usage:
7794 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7797 @item @code{if-exists-else}
7798 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7799 spec function, except that it takes two arguments. The first argument is
7800 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7801 returns the pathname. If it does not exist, it returns the second argument.
7802 This way, @code{if-exists-else} can be used to select one file or another,
7803 based on the existence of the first. Here is a small example of its usage:
7807 crt0%O%s %:if-exists(crti%O%s) \
7808 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7811 @item @code{replace-outfile}
7812 The @code{replace-outfile} spec function takes two arguments. It looks for the
7813 first argument in the outfiles array and replaces it with the second argument. Here
7814 is a small example of its usage:
7817 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7820 @item @code{print-asm-header}
7821 The @code{print-asm-header} function takes no arguments and simply
7822 prints a banner like:
7828 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
7831 It is used to separate compiler options from assembler options
7832 in the @option{--target-help} output.
7836 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7837 If that switch was not specified, this substitutes nothing. Note that
7838 the leading dash is omitted when specifying this option, and it is
7839 automatically inserted if the substitution is performed. Thus the spec
7840 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7841 and would output the command line option @option{-foo}.
7843 @item %W@{@code{S}@}
7844 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7847 @item %@{@code{S}*@}
7848 Substitutes all the switches specified to GCC whose names start
7849 with @code{-S}, but which also take an argument. This is used for
7850 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7851 GCC considers @option{-o foo} as being
7852 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7853 text, including the space. Thus two arguments would be generated.
7855 @item %@{@code{S}*&@code{T}*@}
7856 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7857 (the order of @code{S} and @code{T} in the spec is not significant).
7858 There can be any number of ampersand-separated variables; for each the
7859 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7861 @item %@{@code{S}:@code{X}@}
7862 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7864 @item %@{!@code{S}:@code{X}@}
7865 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7867 @item %@{@code{S}*:@code{X}@}
7868 Substitutes @code{X} if one or more switches whose names start with
7869 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7870 once, no matter how many such switches appeared. However, if @code{%*}
7871 appears somewhere in @code{X}, then @code{X} will be substituted once
7872 for each matching switch, with the @code{%*} replaced by the part of
7873 that switch that matched the @code{*}.
7875 @item %@{.@code{S}:@code{X}@}
7876 Substitutes @code{X}, if processing a file with suffix @code{S}.
7878 @item %@{!.@code{S}:@code{X}@}
7879 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7881 @item %@{,@code{S}:@code{X}@}
7882 Substitutes @code{X}, if processing a file for language @code{S}.
7884 @item %@{!,@code{S}:@code{X}@}
7885 Substitutes @code{X}, if not processing a file for language @code{S}.
7887 @item %@{@code{S}|@code{P}:@code{X}@}
7888 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
7889 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
7890 @code{*} sequences as well, although they have a stronger binding than
7891 the @samp{|}. If @code{%*} appears in @code{X}, all of the
7892 alternatives must be starred, and only the first matching alternative
7895 For example, a spec string like this:
7898 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7901 will output the following command-line options from the following input
7902 command-line options:
7907 -d fred.c -foo -baz -boggle
7908 -d jim.d -bar -baz -boggle
7911 @item %@{S:X; T:Y; :D@}
7913 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7914 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7915 be as many clauses as you need. This may be combined with @code{.},
7916 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
7921 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7922 construct may contain other nested @samp{%} constructs or spaces, or
7923 even newlines. They are processed as usual, as described above.
7924 Trailing white space in @code{X} is ignored. White space may also
7925 appear anywhere on the left side of the colon in these constructs,
7926 except between @code{.} or @code{*} and the corresponding word.
7928 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7929 handled specifically in these constructs. If another value of
7930 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7931 @option{-W} switch is found later in the command line, the earlier
7932 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7933 just one letter, which passes all matching options.
7935 The character @samp{|} at the beginning of the predicate text is used to
7936 indicate that a command should be piped to the following command, but
7937 only if @option{-pipe} is specified.
7939 It is built into GCC which switches take arguments and which do not.
7940 (You might think it would be useful to generalize this to allow each
7941 compiler's spec to say which switches take arguments. But this cannot
7942 be done in a consistent fashion. GCC cannot even decide which input
7943 files have been specified without knowing which switches take arguments,
7944 and it must know which input files to compile in order to tell which
7947 GCC also knows implicitly that arguments starting in @option{-l} are to be
7948 treated as compiler output files, and passed to the linker in their
7949 proper position among the other output files.
7951 @c man begin OPTIONS
7953 @node Target Options
7954 @section Specifying Target Machine and Compiler Version
7955 @cindex target options
7956 @cindex cross compiling
7957 @cindex specifying machine version
7958 @cindex specifying compiler version and target machine
7959 @cindex compiler version, specifying
7960 @cindex target machine, specifying
7962 The usual way to run GCC is to run the executable called @file{gcc}, or
7963 @file{<machine>-gcc} when cross-compiling, or
7964 @file{<machine>-gcc-<version>} to run a version other than the one that
7965 was installed last. Sometimes this is inconvenient, so GCC provides
7966 options that will switch to another cross-compiler or version.
7969 @item -b @var{machine}
7971 The argument @var{machine} specifies the target machine for compilation.
7973 The value to use for @var{machine} is the same as was specified as the
7974 machine type when configuring GCC as a cross-compiler. For
7975 example, if a cross-compiler was configured with @samp{configure
7976 arm-elf}, meaning to compile for an arm processor with elf binaries,
7977 then you would specify @option{-b arm-elf} to run that cross compiler.
7978 Because there are other options beginning with @option{-b}, the
7979 configuration must contain a hyphen.
7981 @item -V @var{version}
7983 The argument @var{version} specifies which version of GCC to run.
7984 This is useful when multiple versions are installed. For example,
7985 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7988 The @option{-V} and @option{-b} options work by running the
7989 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7990 use them if you can just run that directly.
7992 @node Submodel Options
7993 @section Hardware Models and Configurations
7994 @cindex submodel options
7995 @cindex specifying hardware config
7996 @cindex hardware models and configurations, specifying
7997 @cindex machine dependent options
7999 Earlier we discussed the standard option @option{-b} which chooses among
8000 different installed compilers for completely different target
8001 machines, such as VAX vs.@: 68000 vs.@: 80386.
8003 In addition, each of these target machine types can have its own
8004 special options, starting with @samp{-m}, to choose among various
8005 hardware models or configurations---for example, 68010 vs 68020,
8006 floating coprocessor or none. A single installed version of the
8007 compiler can compile for any model or configuration, according to the
8010 Some configurations of the compiler also support additional special
8011 options, usually for compatibility with other compilers on the same
8014 @c This list is ordered alphanumerically by subsection name.
8015 @c It should be the same order and spelling as these options are listed
8016 @c in Machine Dependent Options
8022 * Blackfin Options::
8026 * DEC Alpha Options::
8027 * DEC Alpha/VMS Options::
8029 * GNU/Linux Options::
8032 * i386 and x86-64 Options::
8045 * RS/6000 and PowerPC Options::
8046 * S/390 and zSeries Options::
8051 * System V Options::
8052 * TMS320C3x/C4x Options::
8057 * Xstormy16 Options::
8063 @subsection ARC Options
8066 These options are defined for ARC implementations:
8071 Compile code for little endian mode. This is the default.
8075 Compile code for big endian mode.
8078 @opindex mmangle-cpu
8079 Prepend the name of the cpu to all public symbol names.
8080 In multiple-processor systems, there are many ARC variants with different
8081 instruction and register set characteristics. This flag prevents code
8082 compiled for one cpu to be linked with code compiled for another.
8083 No facility exists for handling variants that are ``almost identical''.
8084 This is an all or nothing option.
8086 @item -mcpu=@var{cpu}
8088 Compile code for ARC variant @var{cpu}.
8089 Which variants are supported depend on the configuration.
8090 All variants support @option{-mcpu=base}, this is the default.
8092 @item -mtext=@var{text-section}
8093 @itemx -mdata=@var{data-section}
8094 @itemx -mrodata=@var{readonly-data-section}
8098 Put functions, data, and readonly data in @var{text-section},
8099 @var{data-section}, and @var{readonly-data-section} respectively
8100 by default. This can be overridden with the @code{section} attribute.
8101 @xref{Variable Attributes}.
8106 @subsection ARM Options
8109 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8113 @item -mabi=@var{name}
8115 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8116 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8119 @opindex mapcs-frame
8120 Generate a stack frame that is compliant with the ARM Procedure Call
8121 Standard for all functions, even if this is not strictly necessary for
8122 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8123 with this option will cause the stack frames not to be generated for
8124 leaf functions. The default is @option{-mno-apcs-frame}.
8128 This is a synonym for @option{-mapcs-frame}.
8131 @c not currently implemented
8132 @item -mapcs-stack-check
8133 @opindex mapcs-stack-check
8134 Generate code to check the amount of stack space available upon entry to
8135 every function (that actually uses some stack space). If there is
8136 insufficient space available then either the function
8137 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8138 called, depending upon the amount of stack space required. The run time
8139 system is required to provide these functions. The default is
8140 @option{-mno-apcs-stack-check}, since this produces smaller code.
8142 @c not currently implemented
8144 @opindex mapcs-float
8145 Pass floating point arguments using the float point registers. This is
8146 one of the variants of the APCS@. This option is recommended if the
8147 target hardware has a floating point unit or if a lot of floating point
8148 arithmetic is going to be performed by the code. The default is
8149 @option{-mno-apcs-float}, since integer only code is slightly increased in
8150 size if @option{-mapcs-float} is used.
8152 @c not currently implemented
8153 @item -mapcs-reentrant
8154 @opindex mapcs-reentrant
8155 Generate reentrant, position independent code. The default is
8156 @option{-mno-apcs-reentrant}.
8159 @item -mthumb-interwork
8160 @opindex mthumb-interwork
8161 Generate code which supports calling between the ARM and Thumb
8162 instruction sets. Without this option the two instruction sets cannot
8163 be reliably used inside one program. The default is
8164 @option{-mno-thumb-interwork}, since slightly larger code is generated
8165 when @option{-mthumb-interwork} is specified.
8167 @item -mno-sched-prolog
8168 @opindex mno-sched-prolog
8169 Prevent the reordering of instructions in the function prolog, or the
8170 merging of those instruction with the instructions in the function's
8171 body. This means that all functions will start with a recognizable set
8172 of instructions (or in fact one of a choice from a small set of
8173 different function prologues), and this information can be used to
8174 locate the start if functions inside an executable piece of code. The
8175 default is @option{-msched-prolog}.
8178 @opindex mhard-float
8179 Generate output containing floating point instructions. This is the
8183 @opindex msoft-float
8184 Generate output containing library calls for floating point.
8185 @strong{Warning:} the requisite libraries are not available for all ARM
8186 targets. Normally the facilities of the machine's usual C compiler are
8187 used, but this cannot be done directly in cross-compilation. You must make
8188 your own arrangements to provide suitable library functions for
8191 @option{-msoft-float} changes the calling convention in the output file;
8192 therefore, it is only useful if you compile @emph{all} of a program with
8193 this option. In particular, you need to compile @file{libgcc.a}, the
8194 library that comes with GCC, with @option{-msoft-float} in order for
8197 @item -mfloat-abi=@var{name}
8199 Specifies which ABI to use for floating point values. Permissible values
8200 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8202 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8203 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8204 of floating point instructions, but still uses the soft-float calling
8207 @item -mlittle-endian
8208 @opindex mlittle-endian
8209 Generate code for a processor running in little-endian mode. This is
8210 the default for all standard configurations.
8213 @opindex mbig-endian
8214 Generate code for a processor running in big-endian mode; the default is
8215 to compile code for a little-endian processor.
8217 @item -mwords-little-endian
8218 @opindex mwords-little-endian
8219 This option only applies when generating code for big-endian processors.
8220 Generate code for a little-endian word order but a big-endian byte
8221 order. That is, a byte order of the form @samp{32107654}. Note: this
8222 option should only be used if you require compatibility with code for
8223 big-endian ARM processors generated by versions of the compiler prior to
8226 @item -mcpu=@var{name}
8228 This specifies the name of the target ARM processor. GCC uses this name
8229 to determine what kind of instructions it can emit when generating
8230 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8231 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8232 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8233 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8234 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8235 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8236 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8237 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8238 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8239 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8240 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8241 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8242 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8243 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8244 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
8245 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8247 @itemx -mtune=@var{name}
8249 This option is very similar to the @option{-mcpu=} option, except that
8250 instead of specifying the actual target processor type, and hence
8251 restricting which instructions can be used, it specifies that GCC should
8252 tune the performance of the code as if the target were of the type
8253 specified in this option, but still choosing the instructions that it
8254 will generate based on the cpu specified by a @option{-mcpu=} option.
8255 For some ARM implementations better performance can be obtained by using
8258 @item -march=@var{name}
8260 This specifies the name of the target ARM architecture. GCC uses this
8261 name to determine what kind of instructions it can emit when generating
8262 assembly code. This option can be used in conjunction with or instead
8263 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8264 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8265 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8266 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
8267 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
8269 @item -mfpu=@var{name}
8270 @itemx -mfpe=@var{number}
8271 @itemx -mfp=@var{number}
8275 This specifies what floating point hardware (or hardware emulation) is
8276 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8277 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
8278 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8279 with older versions of GCC@.
8281 If @option{-msoft-float} is specified this specifies the format of
8282 floating point values.
8284 @item -mstructure-size-boundary=@var{n}
8285 @opindex mstructure-size-boundary
8286 The size of all structures and unions will be rounded up to a multiple
8287 of the number of bits set by this option. Permissible values are 8, 32
8288 and 64. The default value varies for different toolchains. For the COFF
8289 targeted toolchain the default value is 8. A value of 64 is only allowed
8290 if the underlying ABI supports it.
8292 Specifying the larger number can produce faster, more efficient code, but
8293 can also increase the size of the program. Different values are potentially
8294 incompatible. Code compiled with one value cannot necessarily expect to
8295 work with code or libraries compiled with another value, if they exchange
8296 information using structures or unions.
8298 @item -mabort-on-noreturn
8299 @opindex mabort-on-noreturn
8300 Generate a call to the function @code{abort} at the end of a
8301 @code{noreturn} function. It will be executed if the function tries to
8305 @itemx -mno-long-calls
8306 @opindex mlong-calls
8307 @opindex mno-long-calls
8308 Tells the compiler to perform function calls by first loading the
8309 address of the function into a register and then performing a subroutine
8310 call on this register. This switch is needed if the target function
8311 will lie outside of the 64 megabyte addressing range of the offset based
8312 version of subroutine call instruction.
8314 Even if this switch is enabled, not all function calls will be turned
8315 into long calls. The heuristic is that static functions, functions
8316 which have the @samp{short-call} attribute, functions that are inside
8317 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8318 definitions have already been compiled within the current compilation
8319 unit, will not be turned into long calls. The exception to this rule is
8320 that weak function definitions, functions with the @samp{long-call}
8321 attribute or the @samp{section} attribute, and functions that are within
8322 the scope of a @samp{#pragma long_calls} directive, will always be
8323 turned into long calls.
8325 This feature is not enabled by default. Specifying
8326 @option{-mno-long-calls} will restore the default behavior, as will
8327 placing the function calls within the scope of a @samp{#pragma
8328 long_calls_off} directive. Note these switches have no effect on how
8329 the compiler generates code to handle function calls via function
8332 @item -mnop-fun-dllimport
8333 @opindex mnop-fun-dllimport
8334 Disable support for the @code{dllimport} attribute.
8336 @item -msingle-pic-base
8337 @opindex msingle-pic-base
8338 Treat the register used for PIC addressing as read-only, rather than
8339 loading it in the prologue for each function. The run-time system is
8340 responsible for initializing this register with an appropriate value
8341 before execution begins.
8343 @item -mpic-register=@var{reg}
8344 @opindex mpic-register
8345 Specify the register to be used for PIC addressing. The default is R10
8346 unless stack-checking is enabled, when R9 is used.
8348 @item -mcirrus-fix-invalid-insns
8349 @opindex mcirrus-fix-invalid-insns
8350 @opindex mno-cirrus-fix-invalid-insns
8351 Insert NOPs into the instruction stream to in order to work around
8352 problems with invalid Maverick instruction combinations. This option
8353 is only valid if the @option{-mcpu=ep9312} option has been used to
8354 enable generation of instructions for the Cirrus Maverick floating
8355 point co-processor. This option is not enabled by default, since the
8356 problem is only present in older Maverick implementations. The default
8357 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8360 @item -mpoke-function-name
8361 @opindex mpoke-function-name
8362 Write the name of each function into the text section, directly
8363 preceding the function prologue. The generated code is similar to this:
8367 .ascii "arm_poke_function_name", 0
8370 .word 0xff000000 + (t1 - t0)
8371 arm_poke_function_name
8373 stmfd sp!, @{fp, ip, lr, pc@}
8377 When performing a stack backtrace, code can inspect the value of
8378 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8379 location @code{pc - 12} and the top 8 bits are set, then we know that
8380 there is a function name embedded immediately preceding this location
8381 and has length @code{((pc[-3]) & 0xff000000)}.
8385 Generate code for the Thumb instruction set. The default is to
8386 use the 32-bit ARM instruction set.
8387 This option automatically enables either 16-bit Thumb-1 or
8388 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8389 and @option{-march=@var{name}} options.
8392 @opindex mtpcs-frame
8393 Generate a stack frame that is compliant with the Thumb Procedure Call
8394 Standard for all non-leaf functions. (A leaf function is one that does
8395 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8397 @item -mtpcs-leaf-frame
8398 @opindex mtpcs-leaf-frame
8399 Generate a stack frame that is compliant with the Thumb Procedure Call
8400 Standard for all leaf functions. (A leaf function is one that does
8401 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8403 @item -mcallee-super-interworking
8404 @opindex mcallee-super-interworking
8405 Gives all externally visible functions in the file being compiled an ARM
8406 instruction set header which switches to Thumb mode before executing the
8407 rest of the function. This allows these functions to be called from
8408 non-interworking code.
8410 @item -mcaller-super-interworking
8411 @opindex mcaller-super-interworking
8412 Allows calls via function pointers (including virtual functions) to
8413 execute correctly regardless of whether the target code has been
8414 compiled for interworking or not. There is a small overhead in the cost
8415 of executing a function pointer if this option is enabled.
8417 @item -mtp=@var{name}
8419 Specify the access model for the thread local storage pointer. The valid
8420 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8421 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8422 (supported in the arm6k architecture), and @option{auto}, which uses the
8423 best available method for the selected processor. The default setting is
8429 @subsection AVR Options
8432 These options are defined for AVR implementations:
8435 @item -mmcu=@var{mcu}
8437 Specify ATMEL AVR instruction set or MCU type.
8439 Instruction set avr1 is for the minimal AVR core, not supported by the C
8440 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8441 attiny11, attiny12, attiny15, attiny28).
8443 Instruction set avr2 (default) is for the classic AVR core with up to
8444 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8445 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8446 at90c8534, at90s8535).
8448 Instruction set avr3 is for the classic AVR core with up to 128K program
8449 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8451 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8452 memory space (MCU types: atmega8, atmega83, atmega85).
8454 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8455 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8456 atmega64, atmega128, at43usb355, at94k).
8460 Output instruction sizes to the asm file.
8462 @item -minit-stack=@var{N}
8463 @opindex minit-stack
8464 Specify the initial stack address, which may be a symbol or numeric value,
8465 @samp{__stack} is the default.
8467 @item -mno-interrupts
8468 @opindex mno-interrupts
8469 Generated code is not compatible with hardware interrupts.
8470 Code size will be smaller.
8472 @item -mcall-prologues
8473 @opindex mcall-prologues
8474 Functions prologues/epilogues expanded as call to appropriate
8475 subroutines. Code size will be smaller.
8477 @item -mno-tablejump
8478 @opindex mno-tablejump
8479 Do not generate tablejump insns which sometimes increase code size.
8482 @opindex mtiny-stack
8483 Change only the low 8 bits of the stack pointer.
8487 Assume int to be 8 bit integer. This affects the sizes of all types: A
8488 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8489 and long long will be 4 bytes. Please note that this option does not
8490 comply to the C standards, but it will provide you with smaller code
8494 @node Blackfin Options
8495 @subsection Blackfin Options
8496 @cindex Blackfin Options
8499 @item -mcpu=@var{cpu}
8501 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8502 can be one of @samp{bf531}, @samp{bf532}, @samp{bf533},
8503 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf561}.
8504 Without this option, @samp{bf532} is used as the processor by default.
8505 The corresponding predefined processor macros for @var{cpu} is to
8506 be defined. For the @samp{bfin-elf} toolchain, this causes the hardware
8507 BSP provided by libgloss to be linked in if @samp{-msim} is not given.
8508 Support for @samp{bf561} is incomplete; only the processor macro is defined.
8512 Specifies that the program will be run on the simulator. This causes
8513 the simulator BSP provided by libgloss to be linked in. This option
8514 has effect only for @samp{bfin-elf} toolchain.
8516 @item -momit-leaf-frame-pointer
8517 @opindex momit-leaf-frame-pointer
8518 Don't keep the frame pointer in a register for leaf functions. This
8519 avoids the instructions to save, set up and restore frame pointers and
8520 makes an extra register available in leaf functions. The option
8521 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8522 which might make debugging harder.
8524 @item -mspecld-anomaly
8525 @opindex mspecld-anomaly
8526 When enabled, the compiler will ensure that the generated code does not
8527 contain speculative loads after jump instructions. This option is enabled
8530 @item -mno-specld-anomaly
8531 @opindex mno-specld-anomaly
8532 Don't generate extra code to prevent speculative loads from occurring.
8534 @item -mcsync-anomaly
8535 @opindex mcsync-anomaly
8536 When enabled, the compiler will ensure that the generated code does not
8537 contain CSYNC or SSYNC instructions too soon after conditional branches.
8538 This option is enabled by default.
8540 @item -mno-csync-anomaly
8541 @opindex mno-csync-anomaly
8542 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8543 occurring too soon after a conditional branch.
8547 When enabled, the compiler is free to take advantage of the knowledge that
8548 the entire program fits into the low 64k of memory.
8551 @opindex mno-low-64k
8552 Assume that the program is arbitrarily large. This is the default.
8554 @item -mstack-check-l1
8555 @opindex mstack-check-l1
8556 Do stack checking using information placed into L1 scratchpad memory by the
8559 @item -mid-shared-library
8560 @opindex mid-shared-library
8561 Generate code that supports shared libraries via the library ID method.
8562 This allows for execute in place and shared libraries in an environment
8563 without virtual memory management. This option implies @option{-fPIC}.
8565 @item -mno-id-shared-library
8566 @opindex mno-id-shared-library
8567 Generate code that doesn't assume ID based shared libraries are being used.
8568 This is the default.
8570 @item -mleaf-id-shared-library
8571 @opindex mleaf-id-shared-library
8572 Generate code that supports shared libraries via the library ID method,
8573 but assumes that this library or executable won't link against any other
8574 ID shared libraries. That allows the compiler to use faster code for jumps
8577 @item -mno-leaf-id-shared-library
8578 @opindex mno-leaf-id-shared-library
8579 Do not assume that the code being compiled won't link against any ID shared
8580 libraries. Slower code will be generated for jump and call insns.
8582 @item -mshared-library-id=n
8583 @opindex mshared-library-id
8584 Specified the identification number of the ID based shared library being
8585 compiled. Specifying a value of 0 will generate more compact code, specifying
8586 other values will force the allocation of that number to the current
8587 library but is no more space or time efficient than omitting this option.
8591 Generate code that allows the data segment to be located in a different
8592 area of memory from the text segment. This allows for execute in place in
8593 an environment without virtual memory management by eliminating relocations
8594 against the text section.
8597 @opindex mno-sep-data
8598 Generate code that assumes that the data segment follows the text segment.
8599 This is the default.
8602 @itemx -mno-long-calls
8603 @opindex mlong-calls
8604 @opindex mno-long-calls
8605 Tells the compiler to perform function calls by first loading the
8606 address of the function into a register and then performing a subroutine
8607 call on this register. This switch is needed if the target function
8608 will lie outside of the 24 bit addressing range of the offset based
8609 version of subroutine call instruction.
8611 This feature is not enabled by default. Specifying
8612 @option{-mno-long-calls} will restore the default behavior. Note these
8613 switches have no effect on how the compiler generates code to handle
8614 function calls via function pointers.
8618 @subsection CRIS Options
8619 @cindex CRIS Options
8621 These options are defined specifically for the CRIS ports.
8624 @item -march=@var{architecture-type}
8625 @itemx -mcpu=@var{architecture-type}
8628 Generate code for the specified architecture. The choices for
8629 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8630 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8631 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8634 @item -mtune=@var{architecture-type}
8636 Tune to @var{architecture-type} everything applicable about the generated
8637 code, except for the ABI and the set of available instructions. The
8638 choices for @var{architecture-type} are the same as for
8639 @option{-march=@var{architecture-type}}.
8641 @item -mmax-stack-frame=@var{n}
8642 @opindex mmax-stack-frame
8643 Warn when the stack frame of a function exceeds @var{n} bytes.
8645 @item -melinux-stacksize=@var{n}
8646 @opindex melinux-stacksize
8647 Only available with the @samp{cris-axis-aout} target. Arranges for
8648 indications in the program to the kernel loader that the stack of the
8649 program should be set to @var{n} bytes.
8655 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8656 @option{-march=v3} and @option{-march=v8} respectively.
8658 @item -mmul-bug-workaround
8659 @itemx -mno-mul-bug-workaround
8660 @opindex mmul-bug-workaround
8661 @opindex mno-mul-bug-workaround
8662 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8663 models where it applies. This option is active by default.
8667 Enable CRIS-specific verbose debug-related information in the assembly
8668 code. This option also has the effect to turn off the @samp{#NO_APP}
8669 formatted-code indicator to the assembler at the beginning of the
8674 Do not use condition-code results from previous instruction; always emit
8675 compare and test instructions before use of condition codes.
8677 @item -mno-side-effects
8678 @opindex mno-side-effects
8679 Do not emit instructions with side-effects in addressing modes other than
8683 @itemx -mno-stack-align
8685 @itemx -mno-data-align
8686 @itemx -mconst-align
8687 @itemx -mno-const-align
8688 @opindex mstack-align
8689 @opindex mno-stack-align
8690 @opindex mdata-align
8691 @opindex mno-data-align
8692 @opindex mconst-align
8693 @opindex mno-const-align
8694 These options (no-options) arranges (eliminate arrangements) for the
8695 stack-frame, individual data and constants to be aligned for the maximum
8696 single data access size for the chosen CPU model. The default is to
8697 arrange for 32-bit alignment. ABI details such as structure layout are
8698 not affected by these options.
8706 Similar to the stack- data- and const-align options above, these options
8707 arrange for stack-frame, writable data and constants to all be 32-bit,
8708 16-bit or 8-bit aligned. The default is 32-bit alignment.
8710 @item -mno-prologue-epilogue
8711 @itemx -mprologue-epilogue
8712 @opindex mno-prologue-epilogue
8713 @opindex mprologue-epilogue
8714 With @option{-mno-prologue-epilogue}, the normal function prologue and
8715 epilogue that sets up the stack-frame are omitted and no return
8716 instructions or return sequences are generated in the code. Use this
8717 option only together with visual inspection of the compiled code: no
8718 warnings or errors are generated when call-saved registers must be saved,
8719 or storage for local variable needs to be allocated.
8725 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8726 instruction sequences that load addresses for functions from the PLT part
8727 of the GOT rather than (traditional on other architectures) calls to the
8728 PLT@. The default is @option{-mgotplt}.
8732 Legacy no-op option only recognized with the cris-axis-aout target.
8736 Legacy no-op option only recognized with the cris-axis-elf and
8737 cris-axis-linux-gnu targets.
8741 Only recognized with the cris-axis-aout target, where it selects a
8742 GNU/linux-like multilib, include files and instruction set for
8747 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8751 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8752 to link with input-output functions from a simulator library. Code,
8753 initialized data and zero-initialized data are allocated consecutively.
8757 Like @option{-sim}, but pass linker options to locate initialized data at
8758 0x40000000 and zero-initialized data at 0x80000000.
8762 @subsection CRX Options
8765 These options are defined specifically for the CRX ports.
8771 Enable the use of multiply-accumulate instructions. Disabled by default.
8775 Push instructions will be used to pass outgoing arguments when functions
8776 are called. Enabled by default.
8779 @node Darwin Options
8780 @subsection Darwin Options
8781 @cindex Darwin options
8783 These options are defined for all architectures running the Darwin operating
8786 FSF GCC on Darwin does not create ``fat'' object files; it will create
8787 an object file for the single architecture that it was built to
8788 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8789 @option{-arch} options are used; it does so by running the compiler or
8790 linker multiple times and joining the results together with
8793 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8794 @samp{i686}) is determined by the flags that specify the ISA
8795 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8796 @option{-force_cpusubtype_ALL} option can be used to override this.
8798 The Darwin tools vary in their behavior when presented with an ISA
8799 mismatch. The assembler, @file{as}, will only permit instructions to
8800 be used that are valid for the subtype of the file it is generating,
8801 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8802 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8803 and print an error if asked to create a shared library with a less
8804 restrictive subtype than its input files (for instance, trying to put
8805 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8806 for executables, @file{ld}, will quietly give the executable the most
8807 restrictive subtype of any of its input files.
8812 Add the framework directory @var{dir} to the head of the list of
8813 directories to be searched for header files. These directories are
8814 interleaved with those specified by @option{-I} options and are
8815 scanned in a left-to-right order.
8817 A framework directory is a directory with frameworks in it. A
8818 framework is a directory with a @samp{"Headers"} and/or
8819 @samp{"PrivateHeaders"} directory contained directly in it that ends
8820 in @samp{".framework"}. The name of a framework is the name of this
8821 directory excluding the @samp{".framework"}. Headers associated with
8822 the framework are found in one of those two directories, with
8823 @samp{"Headers"} being searched first. A subframework is a framework
8824 directory that is in a framework's @samp{"Frameworks"} directory.
8825 Includes of subframework headers can only appear in a header of a
8826 framework that contains the subframework, or in a sibling subframework
8827 header. Two subframeworks are siblings if they occur in the same
8828 framework. A subframework should not have the same name as a
8829 framework, a warning will be issued if this is violated. Currently a
8830 subframework cannot have subframeworks, in the future, the mechanism
8831 may be extended to support this. The standard frameworks can be found
8832 in @samp{"/System/Library/Frameworks"} and
8833 @samp{"/Library/Frameworks"}. An example include looks like
8834 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8835 the name of the framework and header.h is found in the
8836 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8838 @item -iframework@var{dir}
8840 Like @option{-F} except the directory is a treated as a system
8841 directory. The main difference between this @option{-iframework} and
8842 @option{-F} is that with @option{-iframework} the compiler does not
8843 warn about constructs contained within header files found via
8844 @var{dir}. This option is valid only for the C family of languages.
8848 Emit debugging information for symbols that are used. For STABS
8849 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8850 This is by default ON@.
8854 Emit debugging information for all symbols and types.
8856 @item -mmacosx-version-min=@var{version}
8857 The earliest version of MacOS X that this executable will run on
8858 is @var{version}. Typical values of @var{version} include @code{10.1},
8859 @code{10.2}, and @code{10.3.9}.
8861 If the compiler was built to use the system's headers by default,
8862 then the default for this option is the system version on which the
8863 compiler is running, otherwise the default is to make choices which
8864 are compatible with as many systems and code bases as possible.
8868 Enable kernel development mode. The @option{-mkernel} option sets
8869 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8870 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8871 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8872 applicable. This mode also sets @option{-mno-altivec},
8873 @option{-msoft-float}, @option{-fno-builtin} and
8874 @option{-mlong-branch} for PowerPC targets.
8876 @item -mone-byte-bool
8877 @opindex mone-byte-bool
8878 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8879 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8880 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8881 option has no effect on x86.
8883 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8884 to generate code that is not binary compatible with code generated
8885 without that switch. Using this switch may require recompiling all
8886 other modules in a program, including system libraries. Use this
8887 switch to conform to a non-default data model.
8889 @item -mfix-and-continue
8890 @itemx -ffix-and-continue
8891 @itemx -findirect-data
8892 @opindex mfix-and-continue
8893 @opindex ffix-and-continue
8894 @opindex findirect-data
8895 Generate code suitable for fast turn around development. Needed to
8896 enable gdb to dynamically load @code{.o} files into already running
8897 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8898 are provided for backwards compatibility.
8902 Loads all members of static archive libraries.
8903 See man ld(1) for more information.
8905 @item -arch_errors_fatal
8906 @opindex arch_errors_fatal
8907 Cause the errors having to do with files that have the wrong architecture
8911 @opindex bind_at_load
8912 Causes the output file to be marked such that the dynamic linker will
8913 bind all undefined references when the file is loaded or launched.
8917 Produce a Mach-o bundle format file.
8918 See man ld(1) for more information.
8920 @item -bundle_loader @var{executable}
8921 @opindex bundle_loader
8922 This option specifies the @var{executable} that will be loading the build
8923 output file being linked. See man ld(1) for more information.
8927 When passed this option, GCC will produce a dynamic library instead of
8928 an executable when linking, using the Darwin @file{libtool} command.
8930 @item -force_cpusubtype_ALL
8931 @opindex force_cpusubtype_ALL
8932 This causes GCC's output file to have the @var{ALL} subtype, instead of
8933 one controlled by the @option{-mcpu} or @option{-march} option.
8935 @item -allowable_client @var{client_name}
8937 @itemx -compatibility_version
8938 @itemx -current_version
8940 @itemx -dependency-file
8942 @itemx -dylinker_install_name
8944 @itemx -exported_symbols_list
8946 @itemx -flat_namespace
8947 @itemx -force_flat_namespace
8948 @itemx -headerpad_max_install_names
8951 @itemx -install_name
8952 @itemx -keep_private_externs
8953 @itemx -multi_module
8954 @itemx -multiply_defined
8955 @itemx -multiply_defined_unused
8957 @itemx -no_dead_strip_inits_and_terms
8958 @itemx -nofixprebinding
8961 @itemx -noseglinkedit
8962 @itemx -pagezero_size
8964 @itemx -prebind_all_twolevel_modules
8965 @itemx -private_bundle
8966 @itemx -read_only_relocs
8968 @itemx -sectobjectsymbols
8972 @itemx -sectobjectsymbols
8975 @itemx -segs_read_only_addr
8976 @itemx -segs_read_write_addr
8977 @itemx -seg_addr_table
8978 @itemx -seg_addr_table_filename
8981 @itemx -segs_read_only_addr
8982 @itemx -segs_read_write_addr
8983 @itemx -single_module
8986 @itemx -sub_umbrella
8987 @itemx -twolevel_namespace
8990 @itemx -unexported_symbols_list
8991 @itemx -weak_reference_mismatches
8994 @opindex allowable_client
8995 @opindex client_name
8996 @opindex compatibility_version
8997 @opindex current_version
8999 @opindex dependency-file
9001 @opindex dylinker_install_name
9003 @opindex exported_symbols_list
9005 @opindex flat_namespace
9006 @opindex force_flat_namespace
9007 @opindex headerpad_max_install_names
9010 @opindex install_name
9011 @opindex keep_private_externs
9012 @opindex multi_module
9013 @opindex multiply_defined
9014 @opindex multiply_defined_unused
9016 @opindex no_dead_strip_inits_and_terms
9017 @opindex nofixprebinding
9018 @opindex nomultidefs
9020 @opindex noseglinkedit
9021 @opindex pagezero_size
9023 @opindex prebind_all_twolevel_modules
9024 @opindex private_bundle
9025 @opindex read_only_relocs
9027 @opindex sectobjectsymbols
9031 @opindex sectobjectsymbols
9034 @opindex segs_read_only_addr
9035 @opindex segs_read_write_addr
9036 @opindex seg_addr_table
9037 @opindex seg_addr_table_filename
9038 @opindex seglinkedit
9040 @opindex segs_read_only_addr
9041 @opindex segs_read_write_addr
9042 @opindex single_module
9044 @opindex sub_library
9045 @opindex sub_umbrella
9046 @opindex twolevel_namespace
9049 @opindex unexported_symbols_list
9050 @opindex weak_reference_mismatches
9051 @opindex whatsloaded
9053 These options are passed to the Darwin linker. The Darwin linker man page
9054 describes them in detail.
9057 @node DEC Alpha Options
9058 @subsection DEC Alpha Options
9060 These @samp{-m} options are defined for the DEC Alpha implementations:
9063 @item -mno-soft-float
9065 @opindex mno-soft-float
9066 @opindex msoft-float
9067 Use (do not use) the hardware floating-point instructions for
9068 floating-point operations. When @option{-msoft-float} is specified,
9069 functions in @file{libgcc.a} will be used to perform floating-point
9070 operations. Unless they are replaced by routines that emulate the
9071 floating-point operations, or compiled in such a way as to call such
9072 emulations routines, these routines will issue floating-point
9073 operations. If you are compiling for an Alpha without floating-point
9074 operations, you must ensure that the library is built so as not to call
9077 Note that Alpha implementations without floating-point operations are
9078 required to have floating-point registers.
9083 @opindex mno-fp-regs
9084 Generate code that uses (does not use) the floating-point register set.
9085 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9086 register set is not used, floating point operands are passed in integer
9087 registers as if they were integers and floating-point results are passed
9088 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9089 so any function with a floating-point argument or return value called by code
9090 compiled with @option{-mno-fp-regs} must also be compiled with that
9093 A typical use of this option is building a kernel that does not use,
9094 and hence need not save and restore, any floating-point registers.
9098 The Alpha architecture implements floating-point hardware optimized for
9099 maximum performance. It is mostly compliant with the IEEE floating
9100 point standard. However, for full compliance, software assistance is
9101 required. This option generates code fully IEEE compliant code
9102 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9103 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9104 defined during compilation. The resulting code is less efficient but is
9105 able to correctly support denormalized numbers and exceptional IEEE
9106 values such as not-a-number and plus/minus infinity. Other Alpha
9107 compilers call this option @option{-ieee_with_no_inexact}.
9109 @item -mieee-with-inexact
9110 @opindex mieee-with-inexact
9111 This is like @option{-mieee} except the generated code also maintains
9112 the IEEE @var{inexact-flag}. Turning on this option causes the
9113 generated code to implement fully-compliant IEEE math. In addition to
9114 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9115 macro. On some Alpha implementations the resulting code may execute
9116 significantly slower than the code generated by default. Since there is
9117 very little code that depends on the @var{inexact-flag}, you should
9118 normally not specify this option. Other Alpha compilers call this
9119 option @option{-ieee_with_inexact}.
9121 @item -mfp-trap-mode=@var{trap-mode}
9122 @opindex mfp-trap-mode
9123 This option controls what floating-point related traps are enabled.
9124 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9125 The trap mode can be set to one of four values:
9129 This is the default (normal) setting. The only traps that are enabled
9130 are the ones that cannot be disabled in software (e.g., division by zero
9134 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9138 Like @samp{u}, but the instructions are marked to be safe for software
9139 completion (see Alpha architecture manual for details).
9142 Like @samp{su}, but inexact traps are enabled as well.
9145 @item -mfp-rounding-mode=@var{rounding-mode}
9146 @opindex mfp-rounding-mode
9147 Selects the IEEE rounding mode. Other Alpha compilers call this option
9148 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9153 Normal IEEE rounding mode. Floating point numbers are rounded towards
9154 the nearest machine number or towards the even machine number in case
9158 Round towards minus infinity.
9161 Chopped rounding mode. Floating point numbers are rounded towards zero.
9164 Dynamic rounding mode. A field in the floating point control register
9165 (@var{fpcr}, see Alpha architecture reference manual) controls the
9166 rounding mode in effect. The C library initializes this register for
9167 rounding towards plus infinity. Thus, unless your program modifies the
9168 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9171 @item -mtrap-precision=@var{trap-precision}
9172 @opindex mtrap-precision
9173 In the Alpha architecture, floating point traps are imprecise. This
9174 means without software assistance it is impossible to recover from a
9175 floating trap and program execution normally needs to be terminated.
9176 GCC can generate code that can assist operating system trap handlers
9177 in determining the exact location that caused a floating point trap.
9178 Depending on the requirements of an application, different levels of
9179 precisions can be selected:
9183 Program precision. This option is the default and means a trap handler
9184 can only identify which program caused a floating point exception.
9187 Function precision. The trap handler can determine the function that
9188 caused a floating point exception.
9191 Instruction precision. The trap handler can determine the exact
9192 instruction that caused a floating point exception.
9195 Other Alpha compilers provide the equivalent options called
9196 @option{-scope_safe} and @option{-resumption_safe}.
9198 @item -mieee-conformant
9199 @opindex mieee-conformant
9200 This option marks the generated code as IEEE conformant. You must not
9201 use this option unless you also specify @option{-mtrap-precision=i} and either
9202 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9203 is to emit the line @samp{.eflag 48} in the function prologue of the
9204 generated assembly file. Under DEC Unix, this has the effect that
9205 IEEE-conformant math library routines will be linked in.
9207 @item -mbuild-constants
9208 @opindex mbuild-constants
9209 Normally GCC examines a 32- or 64-bit integer constant to
9210 see if it can construct it from smaller constants in two or three
9211 instructions. If it cannot, it will output the constant as a literal and
9212 generate code to load it from the data segment at runtime.
9214 Use this option to require GCC to construct @emph{all} integer constants
9215 using code, even if it takes more instructions (the maximum is six).
9217 You would typically use this option to build a shared library dynamic
9218 loader. Itself a shared library, it must relocate itself in memory
9219 before it can find the variables and constants in its own data segment.
9225 Select whether to generate code to be assembled by the vendor-supplied
9226 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9244 Indicate whether GCC should generate code to use the optional BWX,
9245 CIX, FIX and MAX instruction sets. The default is to use the instruction
9246 sets supported by the CPU type specified via @option{-mcpu=} option or that
9247 of the CPU on which GCC was built if none was specified.
9252 @opindex mfloat-ieee
9253 Generate code that uses (does not use) VAX F and G floating point
9254 arithmetic instead of IEEE single and double precision.
9256 @item -mexplicit-relocs
9257 @itemx -mno-explicit-relocs
9258 @opindex mexplicit-relocs
9259 @opindex mno-explicit-relocs
9260 Older Alpha assemblers provided no way to generate symbol relocations
9261 except via assembler macros. Use of these macros does not allow
9262 optimal instruction scheduling. GNU binutils as of version 2.12
9263 supports a new syntax that allows the compiler to explicitly mark
9264 which relocations should apply to which instructions. This option
9265 is mostly useful for debugging, as GCC detects the capabilities of
9266 the assembler when it is built and sets the default accordingly.
9270 @opindex msmall-data
9271 @opindex mlarge-data
9272 When @option{-mexplicit-relocs} is in effect, static data is
9273 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9274 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9275 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9276 16-bit relocations off of the @code{$gp} register. This limits the
9277 size of the small data area to 64KB, but allows the variables to be
9278 directly accessed via a single instruction.
9280 The default is @option{-mlarge-data}. With this option the data area
9281 is limited to just below 2GB@. Programs that require more than 2GB of
9282 data must use @code{malloc} or @code{mmap} to allocate the data in the
9283 heap instead of in the program's data segment.
9285 When generating code for shared libraries, @option{-fpic} implies
9286 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9290 @opindex msmall-text
9291 @opindex mlarge-text
9292 When @option{-msmall-text} is used, the compiler assumes that the
9293 code of the entire program (or shared library) fits in 4MB, and is
9294 thus reachable with a branch instruction. When @option{-msmall-data}
9295 is used, the compiler can assume that all local symbols share the
9296 same @code{$gp} value, and thus reduce the number of instructions
9297 required for a function call from 4 to 1.
9299 The default is @option{-mlarge-text}.
9301 @item -mcpu=@var{cpu_type}
9303 Set the instruction set and instruction scheduling parameters for
9304 machine type @var{cpu_type}. You can specify either the @samp{EV}
9305 style name or the corresponding chip number. GCC supports scheduling
9306 parameters for the EV4, EV5 and EV6 family of processors and will
9307 choose the default values for the instruction set from the processor
9308 you specify. If you do not specify a processor type, GCC will default
9309 to the processor on which the compiler was built.
9311 Supported values for @var{cpu_type} are
9317 Schedules as an EV4 and has no instruction set extensions.
9321 Schedules as an EV5 and has no instruction set extensions.
9325 Schedules as an EV5 and supports the BWX extension.
9330 Schedules as an EV5 and supports the BWX and MAX extensions.
9334 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9338 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9341 @item -mtune=@var{cpu_type}
9343 Set only the instruction scheduling parameters for machine type
9344 @var{cpu_type}. The instruction set is not changed.
9346 @item -mmemory-latency=@var{time}
9347 @opindex mmemory-latency
9348 Sets the latency the scheduler should assume for typical memory
9349 references as seen by the application. This number is highly
9350 dependent on the memory access patterns used by the application
9351 and the size of the external cache on the machine.
9353 Valid options for @var{time} are
9357 A decimal number representing clock cycles.
9363 The compiler contains estimates of the number of clock cycles for
9364 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9365 (also called Dcache, Scache, and Bcache), as well as to main memory.
9366 Note that L3 is only valid for EV5.
9371 @node DEC Alpha/VMS Options
9372 @subsection DEC Alpha/VMS Options
9374 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9377 @item -mvms-return-codes
9378 @opindex mvms-return-codes
9379 Return VMS condition codes from main. The default is to return POSIX
9380 style condition (e.g.@: error) codes.
9384 @subsection FRV Options
9391 Only use the first 32 general purpose registers.
9396 Use all 64 general purpose registers.
9401 Use only the first 32 floating point registers.
9406 Use all 64 floating point registers
9409 @opindex mhard-float
9411 Use hardware instructions for floating point operations.
9414 @opindex msoft-float
9416 Use library routines for floating point operations.
9421 Dynamically allocate condition code registers.
9426 Do not try to dynamically allocate condition code registers, only
9427 use @code{icc0} and @code{fcc0}.
9432 Change ABI to use double word insns.
9437 Do not use double word instructions.
9442 Use floating point double instructions.
9447 Do not use floating point double instructions.
9452 Use media instructions.
9457 Do not use media instructions.
9462 Use multiply and add/subtract instructions.
9467 Do not use multiply and add/subtract instructions.
9472 Select the FDPIC ABI, that uses function descriptors to represent
9473 pointers to functions. Without any PIC/PIE-related options, it
9474 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9475 assumes GOT entries and small data are within a 12-bit range from the
9476 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9477 are computed with 32 bits.
9480 @opindex minline-plt
9482 Enable inlining of PLT entries in function calls to functions that are
9483 not known to bind locally. It has no effect without @option{-mfdpic}.
9484 It's enabled by default if optimizing for speed and compiling for
9485 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9486 optimization option such as @option{-O3} or above is present in the
9492 Assume a large TLS segment when generating thread-local code.
9497 Do not assume a large TLS segment when generating thread-local code.
9502 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9503 that is known to be in read-only sections. It's enabled by default,
9504 except for @option{-fpic} or @option{-fpie}: even though it may help
9505 make the global offset table smaller, it trades 1 instruction for 4.
9506 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9507 one of which may be shared by multiple symbols, and it avoids the need
9508 for a GOT entry for the referenced symbol, so it's more likely to be a
9509 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9511 @item -multilib-library-pic
9512 @opindex multilib-library-pic
9514 Link with the (library, not FD) pic libraries. It's implied by
9515 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9516 @option{-fpic} without @option{-mfdpic}. You should never have to use
9522 Follow the EABI requirement of always creating a frame pointer whenever
9523 a stack frame is allocated. This option is enabled by default and can
9524 be disabled with @option{-mno-linked-fp}.
9527 @opindex mlong-calls
9529 Use indirect addressing to call functions outside the current
9530 compilation unit. This allows the functions to be placed anywhere
9531 within the 32-bit address space.
9533 @item -malign-labels
9534 @opindex malign-labels
9536 Try to align labels to an 8-byte boundary by inserting nops into the
9537 previous packet. This option only has an effect when VLIW packing
9538 is enabled. It doesn't create new packets; it merely adds nops to
9542 @opindex mlibrary-pic
9544 Generate position-independent EABI code.
9549 Use only the first four media accumulator registers.
9554 Use all eight media accumulator registers.
9559 Pack VLIW instructions.
9564 Do not pack VLIW instructions.
9569 Do not mark ABI switches in e_flags.
9574 Enable the use of conditional-move instructions (default).
9576 This switch is mainly for debugging the compiler and will likely be removed
9577 in a future version.
9579 @item -mno-cond-move
9580 @opindex mno-cond-move
9582 Disable the use of conditional-move instructions.
9584 This switch is mainly for debugging the compiler and will likely be removed
9585 in a future version.
9590 Enable the use of conditional set instructions (default).
9592 This switch is mainly for debugging the compiler and will likely be removed
9593 in a future version.
9598 Disable the use of conditional set instructions.
9600 This switch is mainly for debugging the compiler and will likely be removed
9601 in a future version.
9606 Enable the use of conditional execution (default).
9608 This switch is mainly for debugging the compiler and will likely be removed
9609 in a future version.
9611 @item -mno-cond-exec
9612 @opindex mno-cond-exec
9614 Disable the use of conditional execution.
9616 This switch is mainly for debugging the compiler and will likely be removed
9617 in a future version.
9620 @opindex mvliw-branch
9622 Run a pass to pack branches into VLIW instructions (default).
9624 This switch is mainly for debugging the compiler and will likely be removed
9625 in a future version.
9627 @item -mno-vliw-branch
9628 @opindex mno-vliw-branch
9630 Do not run a pass to pack branches into VLIW instructions.
9632 This switch is mainly for debugging the compiler and will likely be removed
9633 in a future version.
9635 @item -mmulti-cond-exec
9636 @opindex mmulti-cond-exec
9638 Enable optimization of @code{&&} and @code{||} in conditional execution
9641 This switch is mainly for debugging the compiler and will likely be removed
9642 in a future version.
9644 @item -mno-multi-cond-exec
9645 @opindex mno-multi-cond-exec
9647 Disable optimization of @code{&&} and @code{||} in conditional execution.
9649 This switch is mainly for debugging the compiler and will likely be removed
9650 in a future version.
9652 @item -mnested-cond-exec
9653 @opindex mnested-cond-exec
9655 Enable nested conditional execution optimizations (default).
9657 This switch is mainly for debugging the compiler and will likely be removed
9658 in a future version.
9660 @item -mno-nested-cond-exec
9661 @opindex mno-nested-cond-exec
9663 Disable nested conditional execution optimizations.
9665 This switch is mainly for debugging the compiler and will likely be removed
9666 in a future version.
9668 @item -moptimize-membar
9669 @opindex moptimize-membar
9671 This switch removes redundant @code{membar} instructions from the
9672 compiler generated code. It is enabled by default.
9674 @item -mno-optimize-membar
9675 @opindex mno-optimize-membar
9677 This switch disables the automatic removal of redundant @code{membar}
9678 instructions from the generated code.
9680 @item -mtomcat-stats
9681 @opindex mtomcat-stats
9683 Cause gas to print out tomcat statistics.
9685 @item -mcpu=@var{cpu}
9688 Select the processor type for which to generate code. Possible values are
9689 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9690 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9694 @node GNU/Linux Options
9695 @subsection GNU/Linux Options
9697 These @samp{-m} options are defined for GNU/Linux targets:
9702 Use the GNU C library instead of uClibc. This is the default except
9703 on @samp{*-*-linux-*uclibc*} targets.
9707 Use uClibc instead of the GNU C library. This is the default on
9708 @samp{*-*-linux-*uclibc*} targets.
9711 @node H8/300 Options
9712 @subsection H8/300 Options
9714 These @samp{-m} options are defined for the H8/300 implementations:
9719 Shorten some address references at link time, when possible; uses the
9720 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9721 ld, Using ld}, for a fuller description.
9725 Generate code for the H8/300H@.
9729 Generate code for the H8S@.
9733 Generate code for the H8S and H8/300H in the normal mode. This switch
9734 must be used either with @option{-mh} or @option{-ms}.
9738 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9742 Make @code{int} data 32 bits by default.
9746 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9747 The default for the H8/300H and H8S is to align longs and floats on 4
9749 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9750 This option has no effect on the H8/300.
9754 @subsection HPPA Options
9755 @cindex HPPA Options
9757 These @samp{-m} options are defined for the HPPA family of computers:
9760 @item -march=@var{architecture-type}
9762 Generate code for the specified architecture. The choices for
9763 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9764 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9765 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9766 architecture option for your machine. Code compiled for lower numbered
9767 architectures will run on higher numbered architectures, but not the
9771 @itemx -mpa-risc-1-1
9772 @itemx -mpa-risc-2-0
9773 @opindex mpa-risc-1-0
9774 @opindex mpa-risc-1-1
9775 @opindex mpa-risc-2-0
9776 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9779 @opindex mbig-switch
9780 Generate code suitable for big switch tables. Use this option only if
9781 the assembler/linker complain about out of range branches within a switch
9784 @item -mjump-in-delay
9785 @opindex mjump-in-delay
9786 Fill delay slots of function calls with unconditional jump instructions
9787 by modifying the return pointer for the function call to be the target
9788 of the conditional jump.
9790 @item -mdisable-fpregs
9791 @opindex mdisable-fpregs
9792 Prevent floating point registers from being used in any manner. This is
9793 necessary for compiling kernels which perform lazy context switching of
9794 floating point registers. If you use this option and attempt to perform
9795 floating point operations, the compiler will abort.
9797 @item -mdisable-indexing
9798 @opindex mdisable-indexing
9799 Prevent the compiler from using indexing address modes. This avoids some
9800 rather obscure problems when compiling MIG generated code under MACH@.
9802 @item -mno-space-regs
9803 @opindex mno-space-regs
9804 Generate code that assumes the target has no space registers. This allows
9805 GCC to generate faster indirect calls and use unscaled index address modes.
9807 Such code is suitable for level 0 PA systems and kernels.
9809 @item -mfast-indirect-calls
9810 @opindex mfast-indirect-calls
9811 Generate code that assumes calls never cross space boundaries. This
9812 allows GCC to emit code which performs faster indirect calls.
9814 This option will not work in the presence of shared libraries or nested
9817 @item -mfixed-range=@var{register-range}
9818 @opindex mfixed-range
9819 Generate code treating the given register range as fixed registers.
9820 A fixed register is one that the register allocator can not use. This is
9821 useful when compiling kernel code. A register range is specified as
9822 two registers separated by a dash. Multiple register ranges can be
9823 specified separated by a comma.
9825 @item -mlong-load-store
9826 @opindex mlong-load-store
9827 Generate 3-instruction load and store sequences as sometimes required by
9828 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9831 @item -mportable-runtime
9832 @opindex mportable-runtime
9833 Use the portable calling conventions proposed by HP for ELF systems.
9837 Enable the use of assembler directives only GAS understands.
9839 @item -mschedule=@var{cpu-type}
9841 Schedule code according to the constraints for the machine type
9842 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9843 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9844 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9845 proper scheduling option for your machine. The default scheduling is
9849 @opindex mlinker-opt
9850 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9851 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9852 linkers in which they give bogus error messages when linking some programs.
9855 @opindex msoft-float
9856 Generate output containing library calls for floating point.
9857 @strong{Warning:} the requisite libraries are not available for all HPPA
9858 targets. Normally the facilities of the machine's usual C compiler are
9859 used, but this cannot be done directly in cross-compilation. You must make
9860 your own arrangements to provide suitable library functions for
9861 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9862 does provide software floating point support.
9864 @option{-msoft-float} changes the calling convention in the output file;
9865 therefore, it is only useful if you compile @emph{all} of a program with
9866 this option. In particular, you need to compile @file{libgcc.a}, the
9867 library that comes with GCC, with @option{-msoft-float} in order for
9872 Generate the predefine, @code{_SIO}, for server IO@. The default is
9873 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9874 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9875 options are available under HP-UX and HI-UX@.
9879 Use GNU ld specific options. This passes @option{-shared} to ld when
9880 building a shared library. It is the default when GCC is configured,
9881 explicitly or implicitly, with the GNU linker. This option does not
9882 have any affect on which ld is called, it only changes what parameters
9883 are passed to that ld. The ld that is called is determined by the
9884 @option{--with-ld} configure option, GCC's program search path, and
9885 finally by the user's @env{PATH}. The linker used by GCC can be printed
9886 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9887 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9891 Use HP ld specific options. This passes @option{-b} to ld when building
9892 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9893 links. It is the default when GCC is configured, explicitly or
9894 implicitly, with the HP linker. This option does not have any affect on
9895 which ld is called, it only changes what parameters are passed to that
9896 ld. The ld that is called is determined by the @option{--with-ld}
9897 configure option, GCC's program search path, and finally by the user's
9898 @env{PATH}. The linker used by GCC can be printed using @samp{which
9899 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9900 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9903 @opindex mno-long-calls
9904 Generate code that uses long call sequences. This ensures that a call
9905 is always able to reach linker generated stubs. The default is to generate
9906 long calls only when the distance from the call site to the beginning
9907 of the function or translation unit, as the case may be, exceeds a
9908 predefined limit set by the branch type being used. The limits for
9909 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9910 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9913 Distances are measured from the beginning of functions when using the
9914 @option{-ffunction-sections} option, or when using the @option{-mgas}
9915 and @option{-mno-portable-runtime} options together under HP-UX with
9918 It is normally not desirable to use this option as it will degrade
9919 performance. However, it may be useful in large applications,
9920 particularly when partial linking is used to build the application.
9922 The types of long calls used depends on the capabilities of the
9923 assembler and linker, and the type of code being generated. The
9924 impact on systems that support long absolute calls, and long pic
9925 symbol-difference or pc-relative calls should be relatively small.
9926 However, an indirect call is used on 32-bit ELF systems in pic code
9927 and it is quite long.
9929 @item -munix=@var{unix-std}
9931 Generate compiler predefines and select a startfile for the specified
9932 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9933 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9934 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9935 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9936 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9939 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9940 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9941 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9942 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9943 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9944 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9946 It is @emph{important} to note that this option changes the interfaces
9947 for various library routines. It also affects the operational behavior
9948 of the C library. Thus, @emph{extreme} care is needed in using this
9951 Library code that is intended to operate with more than one UNIX
9952 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9953 as appropriate. Most GNU software doesn't provide this capability.
9957 Suppress the generation of link options to search libdld.sl when the
9958 @option{-static} option is specified on HP-UX 10 and later.
9962 The HP-UX implementation of setlocale in libc has a dependency on
9963 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9964 when the @option{-static} option is specified, special link options
9965 are needed to resolve this dependency.
9967 On HP-UX 10 and later, the GCC driver adds the necessary options to
9968 link with libdld.sl when the @option{-static} option is specified.
9969 This causes the resulting binary to be dynamic. On the 64-bit port,
9970 the linkers generate dynamic binaries by default in any case. The
9971 @option{-nolibdld} option can be used to prevent the GCC driver from
9972 adding these link options.
9976 Add support for multithreading with the @dfn{dce thread} library
9977 under HP-UX@. This option sets flags for both the preprocessor and
9981 @node i386 and x86-64 Options
9982 @subsection Intel 386 and AMD x86-64 Options
9983 @cindex i386 Options
9984 @cindex x86-64 Options
9985 @cindex Intel 386 Options
9986 @cindex AMD x86-64 Options
9988 These @samp{-m} options are defined for the i386 and x86-64 family of
9992 @item -mtune=@var{cpu-type}
9994 Tune to @var{cpu-type} everything applicable about the generated code, except
9995 for the ABI and the set of available instructions. The choices for
9999 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10000 If you know the CPU on which your code will run, then you should use
10001 the corresponding @option{-mtune} option instead of
10002 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10003 of your application will have, then you should use this option.
10005 As new processors are deployed in the marketplace, the behavior of this
10006 option will change. Therefore, if you upgrade to a newer version of
10007 GCC, the code generated option will change to reflect the processors
10008 that were most common when that version of GCC was released.
10010 There is no @option{-march=generic} option because @option{-march}
10011 indicates the instruction set the compiler can use, and there is no
10012 generic instruction set applicable to all processors. In contrast,
10013 @option{-mtune} indicates the processor (or, in this case, collection of
10014 processors) for which the code is optimized.
10016 This selects the CPU to tune for at compilation time by determining
10017 the processor type of the compiling machine. Using @option{-mtune=native}
10018 will produce code optimized for the local machine under the constraints
10019 of the selected instruction set. Using @option{-march=native} will
10020 enable all instruction subsets supported by the local machine (hence
10021 the result might not run on different machines).
10023 Original Intel's i386 CPU@.
10025 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10026 @item i586, pentium
10027 Intel Pentium CPU with no MMX support.
10029 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10031 Intel PentiumPro CPU@.
10033 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10034 instruction set will be used, so the code will run on all i686 family chips.
10036 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10037 @item pentium3, pentium3m
10038 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10041 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10042 support. Used by Centrino notebooks.
10043 @item pentium4, pentium4m
10044 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10046 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10049 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10050 SSE2 and SSE3 instruction set support.
10052 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10053 instruction set support.
10055 AMD K6 CPU with MMX instruction set support.
10057 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
10058 @item athlon, athlon-tbird
10059 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
10061 @item athlon-4, athlon-xp, athlon-mp
10062 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
10063 instruction set support.
10064 @item k8, opteron, athlon64, athlon-fx
10065 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10066 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
10067 @item k8-sse3, opteron-sse3, athlon64-sse3
10068 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10069 @item amdfam10, barcelona
10070 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10071 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10072 instruction set extensions.)
10074 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10077 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
10078 instruction set support.
10080 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
10081 implemented for this chip.)
10083 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10084 implemented for this chip.)
10086 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10089 While picking a specific @var{cpu-type} will schedule things appropriately
10090 for that particular chip, the compiler will not generate any code that
10091 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10094 @item -march=@var{cpu-type}
10096 Generate instructions for the machine type @var{cpu-type}. The choices
10097 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10098 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10100 @item -mcpu=@var{cpu-type}
10102 A deprecated synonym for @option{-mtune}.
10104 @item -mfpmath=@var{unit}
10106 Generate floating point arithmetics for selected unit @var{unit}. The choices
10107 for @var{unit} are:
10111 Use the standard 387 floating point coprocessor present majority of chips and
10112 emulated otherwise. Code compiled with this option will run almost everywhere.
10113 The temporary results are computed in 80bit precision instead of precision
10114 specified by the type resulting in slightly different results compared to most
10115 of other chips. See @option{-ffloat-store} for more detailed description.
10117 This is the default choice for i386 compiler.
10120 Use scalar floating point instructions present in the SSE instruction set.
10121 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10122 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10123 instruction set supports only single precision arithmetics, thus the double and
10124 extended precision arithmetics is still done using 387. Later version, present
10125 only in Pentium4 and the future AMD x86-64 chips supports double precision
10128 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10129 or @option{-msse2} switches to enable SSE extensions and make this option
10130 effective. For the x86-64 compiler, these extensions are enabled by default.
10132 The resulting code should be considerably faster in the majority of cases and avoid
10133 the numerical instability problems of 387 code, but may break some existing
10134 code that expects temporaries to be 80bit.
10136 This is the default choice for the x86-64 compiler.
10139 Attempt to utilize both instruction sets at once. This effectively double the
10140 amount of available registers and on chips with separate execution units for
10141 387 and SSE the execution resources too. Use this option with care, as it is
10142 still experimental, because the GCC register allocator does not model separate
10143 functional units well resulting in instable performance.
10146 @item -masm=@var{dialect}
10147 @opindex masm=@var{dialect}
10148 Output asm instructions using selected @var{dialect}. Supported
10149 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10150 not support @samp{intel}.
10153 @itemx -mno-ieee-fp
10155 @opindex mno-ieee-fp
10156 Control whether or not the compiler uses IEEE floating point
10157 comparisons. These handle correctly the case where the result of a
10158 comparison is unordered.
10161 @opindex msoft-float
10162 Generate output containing library calls for floating point.
10163 @strong{Warning:} the requisite libraries are not part of GCC@.
10164 Normally the facilities of the machine's usual C compiler are used, but
10165 this can't be done directly in cross-compilation. You must make your
10166 own arrangements to provide suitable library functions for
10169 On machines where a function returns floating point results in the 80387
10170 register stack, some floating point opcodes may be emitted even if
10171 @option{-msoft-float} is used.
10173 @item -mno-fp-ret-in-387
10174 @opindex mno-fp-ret-in-387
10175 Do not use the FPU registers for return values of functions.
10177 The usual calling convention has functions return values of types
10178 @code{float} and @code{double} in an FPU register, even if there
10179 is no FPU@. The idea is that the operating system should emulate
10182 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10183 in ordinary CPU registers instead.
10185 @item -mno-fancy-math-387
10186 @opindex mno-fancy-math-387
10187 Some 387 emulators do not support the @code{sin}, @code{cos} and
10188 @code{sqrt} instructions for the 387. Specify this option to avoid
10189 generating those instructions. This option is the default on FreeBSD,
10190 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10191 indicates that the target cpu will always have an FPU and so the
10192 instruction will not need emulation. As of revision 2.6.1, these
10193 instructions are not generated unless you also use the
10194 @option{-funsafe-math-optimizations} switch.
10196 @item -malign-double
10197 @itemx -mno-align-double
10198 @opindex malign-double
10199 @opindex mno-align-double
10200 Control whether GCC aligns @code{double}, @code{long double}, and
10201 @code{long long} variables on a two word boundary or a one word
10202 boundary. Aligning @code{double} variables on a two word boundary will
10203 produce code that runs somewhat faster on a @samp{Pentium} at the
10204 expense of more memory.
10206 On x86-64, @option{-malign-double} is enabled by default.
10208 @strong{Warning:} if you use the @option{-malign-double} switch,
10209 structures containing the above types will be aligned differently than
10210 the published application binary interface specifications for the 386
10211 and will not be binary compatible with structures in code compiled
10212 without that switch.
10214 @item -m96bit-long-double
10215 @itemx -m128bit-long-double
10216 @opindex m96bit-long-double
10217 @opindex m128bit-long-double
10218 These switches control the size of @code{long double} type. The i386
10219 application binary interface specifies the size to be 96 bits,
10220 so @option{-m96bit-long-double} is the default in 32 bit mode.
10222 Modern architectures (Pentium and newer) would prefer @code{long double}
10223 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10224 conforming to the ABI, this would not be possible. So specifying a
10225 @option{-m128bit-long-double} will align @code{long double}
10226 to a 16 byte boundary by padding the @code{long double} with an additional
10229 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10230 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10232 Notice that neither of these options enable any extra precision over the x87
10233 standard of 80 bits for a @code{long double}.
10235 @strong{Warning:} if you override the default value for your target ABI, the
10236 structures and arrays containing @code{long double} variables will change
10237 their size as well as function calling convention for function taking
10238 @code{long double} will be modified. Hence they will not be binary
10239 compatible with arrays or structures in code compiled without that switch.
10241 @item -mmlarge-data-threshold=@var{number}
10242 @opindex mlarge-data-threshold=@var{number}
10243 When @option{-mcmodel=medium} is specified, the data greater than
10244 @var{threshold} are placed in large data section. This value must be the
10245 same across all object linked into the binary and defaults to 65535.
10249 Use a different function-calling convention, in which functions that
10250 take a fixed number of arguments return with the @code{ret} @var{num}
10251 instruction, which pops their arguments while returning. This saves one
10252 instruction in the caller since there is no need to pop the arguments
10255 You can specify that an individual function is called with this calling
10256 sequence with the function attribute @samp{stdcall}. You can also
10257 override the @option{-mrtd} option by using the function attribute
10258 @samp{cdecl}. @xref{Function Attributes}.
10260 @strong{Warning:} this calling convention is incompatible with the one
10261 normally used on Unix, so you cannot use it if you need to call
10262 libraries compiled with the Unix compiler.
10264 Also, you must provide function prototypes for all functions that
10265 take variable numbers of arguments (including @code{printf});
10266 otherwise incorrect code will be generated for calls to those
10269 In addition, seriously incorrect code will result if you call a
10270 function with too many arguments. (Normally, extra arguments are
10271 harmlessly ignored.)
10273 @item -mregparm=@var{num}
10275 Control how many registers are used to pass integer arguments. By
10276 default, no registers are used to pass arguments, and at most 3
10277 registers can be used. You can control this behavior for a specific
10278 function by using the function attribute @samp{regparm}.
10279 @xref{Function Attributes}.
10281 @strong{Warning:} if you use this switch, and
10282 @var{num} is nonzero, then you must build all modules with the same
10283 value, including any libraries. This includes the system libraries and
10287 @opindex msseregparm
10288 Use SSE register passing conventions for float and double arguments
10289 and return values. You can control this behavior for a specific
10290 function by using the function attribute @samp{sseregparm}.
10291 @xref{Function Attributes}.
10293 @strong{Warning:} if you use this switch then you must build all
10294 modules with the same value, including any libraries. This includes
10295 the system libraries and startup modules.
10304 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10305 is specified, the significands of results of floating-point operations are
10306 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10307 significands of results of floating-point operations to 53 bits (double
10308 precision) and @option{-mpc80} rounds the significands of results of
10309 floating-point operations to 64 bits (extended double precision), which is
10310 the default. When this option is used, floating-point operations in higher
10311 precisions are not available to the programmer without setting the FPU
10312 control word explicitly.
10314 Setting the rounding of floating-point operations to less than the default
10315 80 bits can speed some programs by 2% or more. Note that some mathematical
10316 libraries assume that extended precision (80 bit) floating-point operations
10317 are enabled by default; routines in such libraries could suffer significant
10318 loss of accuracy, typically through so-called "catastrophic cancellation",
10319 when this option is used to set the precision to less than extended precision.
10321 @item -mstackrealign
10322 @opindex mstackrealign
10323 Realign the stack at entry. On the Intel x86, the
10324 @option{-mstackrealign} option will generate an alternate prologue and
10325 epilogue that realigns the runtime stack. This supports mixing legacy
10326 codes that keep a 4-byte aligned stack with modern codes that keep a
10327 16-byte stack for SSE compatibility. The alternate prologue and
10328 epilogue are slower and bigger than the regular ones, and the
10329 alternate prologue requires an extra scratch register; this lowers the
10330 number of registers available if used in conjunction with the
10331 @code{regparm} attribute. The @option{-mstackrealign} option is
10332 incompatible with the nested function prologue; this is considered a
10333 hard error. See also the attribute @code{force_align_arg_pointer},
10334 applicable to individual functions.
10336 @item -mpreferred-stack-boundary=@var{num}
10337 @opindex mpreferred-stack-boundary
10338 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10339 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10340 the default is 4 (16 bytes or 128 bits).
10342 On Pentium and PentiumPro, @code{double} and @code{long double} values
10343 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10344 suffer significant run time performance penalties. On Pentium III, the
10345 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10346 properly if it is not 16 byte aligned.
10348 To ensure proper alignment of this values on the stack, the stack boundary
10349 must be as aligned as that required by any value stored on the stack.
10350 Further, every function must be generated such that it keeps the stack
10351 aligned. Thus calling a function compiled with a higher preferred
10352 stack boundary from a function compiled with a lower preferred stack
10353 boundary will most likely misalign the stack. It is recommended that
10354 libraries that use callbacks always use the default setting.
10356 This extra alignment does consume extra stack space, and generally
10357 increases code size. Code that is sensitive to stack space usage, such
10358 as embedded systems and operating system kernels, may want to reduce the
10359 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10391 These switches enable or disable the use of instructions in the MMX,
10392 SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4A, ABM or 3DNow! extended
10394 These extensions are also available as built-in functions: see
10395 @ref{X86 Built-in Functions}, for details of the functions enabled and
10396 disabled by these switches.
10398 To have SSE/SSE2 instructions generated automatically from floating-point
10399 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10401 These options will enable GCC to use these extended instructions in
10402 generated code, even without @option{-mfpmath=sse}. Applications which
10403 perform runtime CPU detection must compile separate files for each
10404 supported architecture, using the appropriate flags. In particular,
10405 the file containing the CPU detection code should be compiled without
10410 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10411 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10412 data types. This is useful for high resolution counters that could be updated
10413 by multiple processors (or cores). This instruction is generated as part of
10414 atomic built-in functions: see @ref{Atomic Builtins} for details.
10418 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10419 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10420 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10421 SAHF are load and store instructions, respectively, for certain status flags.
10422 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10423 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10427 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10428 vectorized variants RCPPS and RSQRTPS) instead of DIVSS and SQRTSS (and their
10429 vectorized variants). These instructions will be generated only when
10430 @option{-funsafe-math-optimizations} is enabled.
10433 @itemx -mno-push-args
10434 @opindex mpush-args
10435 @opindex mno-push-args
10436 Use PUSH operations to store outgoing parameters. This method is shorter
10437 and usually equally fast as method using SUB/MOV operations and is enabled
10438 by default. In some cases disabling it may improve performance because of
10439 improved scheduling and reduced dependencies.
10441 @item -maccumulate-outgoing-args
10442 @opindex maccumulate-outgoing-args
10443 If enabled, the maximum amount of space required for outgoing arguments will be
10444 computed in the function prologue. This is faster on most modern CPUs
10445 because of reduced dependencies, improved scheduling and reduced stack usage
10446 when preferred stack boundary is not equal to 2. The drawback is a notable
10447 increase in code size. This switch implies @option{-mno-push-args}.
10451 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10452 on thread-safe exception handling must compile and link all code with the
10453 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10454 @option{-D_MT}; when linking, it links in a special thread helper library
10455 @option{-lmingwthrd} which cleans up per thread exception handling data.
10457 @item -mno-align-stringops
10458 @opindex mno-align-stringops
10459 Do not align destination of inlined string operations. This switch reduces
10460 code size and improves performance in case the destination is already aligned,
10461 but GCC doesn't know about it.
10463 @item -minline-all-stringops
10464 @opindex minline-all-stringops
10465 By default GCC inlines string operations only when destination is known to be
10466 aligned at least to 4 byte boundary. This enables more inlining, increase code
10467 size, but may improve performance of code that depends on fast memcpy, strlen
10468 and memset for short lengths.
10470 @item -minline-stringops-dynamically
10471 @opindex minline-stringops-dynamically
10472 For string operation of unknown size, inline runtime checks so for small
10473 blocks inline code is used, while for large blocks library call is used.
10475 @item -mstringop-strategy=@var{alg}
10476 @opindex mstringop-strategy=@var{alg}
10477 Overwrite internal decision heuristic about particular algorithm to inline
10478 string operation with. The allowed values are @code{rep_byte},
10479 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10480 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10481 expanding inline loop, @code{libcall} for always expanding library call.
10483 @item -momit-leaf-frame-pointer
10484 @opindex momit-leaf-frame-pointer
10485 Don't keep the frame pointer in a register for leaf functions. This
10486 avoids the instructions to save, set up and restore frame pointers and
10487 makes an extra register available in leaf functions. The option
10488 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10489 which might make debugging harder.
10491 @item -mtls-direct-seg-refs
10492 @itemx -mno-tls-direct-seg-refs
10493 @opindex mtls-direct-seg-refs
10494 Controls whether TLS variables may be accessed with offsets from the
10495 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10496 or whether the thread base pointer must be added. Whether or not this
10497 is legal depends on the operating system, and whether it maps the
10498 segment to cover the entire TLS area.
10500 For systems that use GNU libc, the default is on.
10503 These @samp{-m} switches are supported in addition to the above
10504 on AMD x86-64 processors in 64-bit environments.
10511 Generate code for a 32-bit or 64-bit environment.
10512 The 32-bit environment sets int, long and pointer to 32 bits and
10513 generates code that runs on any i386 system.
10514 The 64-bit environment sets int to 32 bits and long and pointer
10515 to 64 bits and generates code for AMD's x86-64 architecture. For
10516 darwin only the -m64 option turns off the @option{-fno-pic} and
10517 @option{-mdynamic-no-pic} options.
10519 @item -mno-red-zone
10520 @opindex no-red-zone
10521 Do not use a so called red zone for x86-64 code. The red zone is mandated
10522 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10523 stack pointer that will not be modified by signal or interrupt handlers
10524 and therefore can be used for temporary data without adjusting the stack
10525 pointer. The flag @option{-mno-red-zone} disables this red zone.
10527 @item -mcmodel=small
10528 @opindex mcmodel=small
10529 Generate code for the small code model: the program and its symbols must
10530 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10531 Programs can be statically or dynamically linked. This is the default
10534 @item -mcmodel=kernel
10535 @opindex mcmodel=kernel
10536 Generate code for the kernel code model. The kernel runs in the
10537 negative 2 GB of the address space.
10538 This model has to be used for Linux kernel code.
10540 @item -mcmodel=medium
10541 @opindex mcmodel=medium
10542 Generate code for the medium model: The program is linked in the lower 2
10543 GB of the address space but symbols can be located anywhere in the
10544 address space. Programs can be statically or dynamically linked, but
10545 building of shared libraries are not supported with the medium model.
10547 @item -mcmodel=large
10548 @opindex mcmodel=large
10549 Generate code for the large model: This model makes no assumptions
10550 about addresses and sizes of sections.
10553 @node IA-64 Options
10554 @subsection IA-64 Options
10555 @cindex IA-64 Options
10557 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10561 @opindex mbig-endian
10562 Generate code for a big endian target. This is the default for HP-UX@.
10564 @item -mlittle-endian
10565 @opindex mlittle-endian
10566 Generate code for a little endian target. This is the default for AIX5
10572 @opindex mno-gnu-as
10573 Generate (or don't) code for the GNU assembler. This is the default.
10574 @c Also, this is the default if the configure option @option{--with-gnu-as}
10580 @opindex mno-gnu-ld
10581 Generate (or don't) code for the GNU linker. This is the default.
10582 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10587 Generate code that does not use a global pointer register. The result
10588 is not position independent code, and violates the IA-64 ABI@.
10590 @item -mvolatile-asm-stop
10591 @itemx -mno-volatile-asm-stop
10592 @opindex mvolatile-asm-stop
10593 @opindex mno-volatile-asm-stop
10594 Generate (or don't) a stop bit immediately before and after volatile asm
10597 @item -mregister-names
10598 @itemx -mno-register-names
10599 @opindex mregister-names
10600 @opindex mno-register-names
10601 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10602 the stacked registers. This may make assembler output more readable.
10608 Disable (or enable) optimizations that use the small data section. This may
10609 be useful for working around optimizer bugs.
10611 @item -mconstant-gp
10612 @opindex mconstant-gp
10613 Generate code that uses a single constant global pointer value. This is
10614 useful when compiling kernel code.
10618 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10619 This is useful when compiling firmware code.
10621 @item -minline-float-divide-min-latency
10622 @opindex minline-float-divide-min-latency
10623 Generate code for inline divides of floating point values
10624 using the minimum latency algorithm.
10626 @item -minline-float-divide-max-throughput
10627 @opindex minline-float-divide-max-throughput
10628 Generate code for inline divides of floating point values
10629 using the maximum throughput algorithm.
10631 @item -minline-int-divide-min-latency
10632 @opindex minline-int-divide-min-latency
10633 Generate code for inline divides of integer values
10634 using the minimum latency algorithm.
10636 @item -minline-int-divide-max-throughput
10637 @opindex minline-int-divide-max-throughput
10638 Generate code for inline divides of integer values
10639 using the maximum throughput algorithm.
10641 @item -minline-sqrt-min-latency
10642 @opindex minline-sqrt-min-latency
10643 Generate code for inline square roots
10644 using the minimum latency algorithm.
10646 @item -minline-sqrt-max-throughput
10647 @opindex minline-sqrt-max-throughput
10648 Generate code for inline square roots
10649 using the maximum throughput algorithm.
10651 @item -mno-dwarf2-asm
10652 @itemx -mdwarf2-asm
10653 @opindex mno-dwarf2-asm
10654 @opindex mdwarf2-asm
10655 Don't (or do) generate assembler code for the DWARF2 line number debugging
10656 info. This may be useful when not using the GNU assembler.
10658 @item -mearly-stop-bits
10659 @itemx -mno-early-stop-bits
10660 @opindex mearly-stop-bits
10661 @opindex mno-early-stop-bits
10662 Allow stop bits to be placed earlier than immediately preceding the
10663 instruction that triggered the stop bit. This can improve instruction
10664 scheduling, but does not always do so.
10666 @item -mfixed-range=@var{register-range}
10667 @opindex mfixed-range
10668 Generate code treating the given register range as fixed registers.
10669 A fixed register is one that the register allocator can not use. This is
10670 useful when compiling kernel code. A register range is specified as
10671 two registers separated by a dash. Multiple register ranges can be
10672 specified separated by a comma.
10674 @item -mtls-size=@var{tls-size}
10676 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
10679 @item -mtune=@var{cpu-type}
10681 Tune the instruction scheduling for a particular CPU, Valid values are
10682 itanium, itanium1, merced, itanium2, and mckinley.
10688 Add support for multithreading using the POSIX threads library. This
10689 option sets flags for both the preprocessor and linker. It does
10690 not affect the thread safety of object code produced by the compiler or
10691 that of libraries supplied with it. These are HP-UX specific flags.
10697 Generate code for a 32-bit or 64-bit environment.
10698 The 32-bit environment sets int, long and pointer to 32 bits.
10699 The 64-bit environment sets int to 32 bits and long and pointer
10700 to 64 bits. These are HP-UX specific flags.
10702 @item -mno-sched-br-data-spec
10703 @itemx -msched-br-data-spec
10704 @opindex mno-sched-br-data-spec
10705 @opindex msched-br-data-spec
10706 (Dis/En)able data speculative scheduling before reload.
10707 This will result in generation of the ld.a instructions and
10708 the corresponding check instructions (ld.c / chk.a).
10709 The default is 'disable'.
10711 @item -msched-ar-data-spec
10712 @itemx -mno-sched-ar-data-spec
10713 @opindex msched-ar-data-spec
10714 @opindex mno-sched-ar-data-spec
10715 (En/Dis)able data speculative scheduling after reload.
10716 This will result in generation of the ld.a instructions and
10717 the corresponding check instructions (ld.c / chk.a).
10718 The default is 'enable'.
10720 @item -mno-sched-control-spec
10721 @itemx -msched-control-spec
10722 @opindex mno-sched-control-spec
10723 @opindex msched-control-spec
10724 (Dis/En)able control speculative scheduling. This feature is
10725 available only during region scheduling (i.e. before reload).
10726 This will result in generation of the ld.s instructions and
10727 the corresponding check instructions chk.s .
10728 The default is 'disable'.
10730 @item -msched-br-in-data-spec
10731 @itemx -mno-sched-br-in-data-spec
10732 @opindex msched-br-in-data-spec
10733 @opindex mno-sched-br-in-data-spec
10734 (En/Dis)able speculative scheduling of the instructions that
10735 are dependent on the data speculative loads before reload.
10736 This is effective only with @option{-msched-br-data-spec} enabled.
10737 The default is 'enable'.
10739 @item -msched-ar-in-data-spec
10740 @itemx -mno-sched-ar-in-data-spec
10741 @opindex msched-ar-in-data-spec
10742 @opindex mno-sched-ar-in-data-spec
10743 (En/Dis)able speculative scheduling of the instructions that
10744 are dependent on the data speculative loads after reload.
10745 This is effective only with @option{-msched-ar-data-spec} enabled.
10746 The default is 'enable'.
10748 @item -msched-in-control-spec
10749 @itemx -mno-sched-in-control-spec
10750 @opindex msched-in-control-spec
10751 @opindex mno-sched-in-control-spec
10752 (En/Dis)able speculative scheduling of the instructions that
10753 are dependent on the control speculative loads.
10754 This is effective only with @option{-msched-control-spec} enabled.
10755 The default is 'enable'.
10758 @itemx -mno-sched-ldc
10759 @opindex msched-ldc
10760 @opindex mno-sched-ldc
10761 (En/Dis)able use of simple data speculation checks ld.c .
10762 If disabled, only chk.a instructions will be emitted to check
10763 data speculative loads.
10764 The default is 'enable'.
10766 @item -mno-sched-control-ldc
10767 @itemx -msched-control-ldc
10768 @opindex mno-sched-control-ldc
10769 @opindex msched-control-ldc
10770 (Dis/En)able use of ld.c instructions to check control speculative loads.
10771 If enabled, in case of control speculative load with no speculatively
10772 scheduled dependent instructions this load will be emitted as ld.sa and
10773 ld.c will be used to check it.
10774 The default is 'disable'.
10776 @item -mno-sched-spec-verbose
10777 @itemx -msched-spec-verbose
10778 @opindex mno-sched-spec-verbose
10779 @opindex msched-spec-verbose
10780 (Dis/En)able printing of the information about speculative motions.
10782 @item -mno-sched-prefer-non-data-spec-insns
10783 @itemx -msched-prefer-non-data-spec-insns
10784 @opindex mno-sched-prefer-non-data-spec-insns
10785 @opindex msched-prefer-non-data-spec-insns
10786 If enabled, data speculative instructions will be chosen for schedule
10787 only if there are no other choices at the moment. This will make
10788 the use of the data speculation much more conservative.
10789 The default is 'disable'.
10791 @item -mno-sched-prefer-non-control-spec-insns
10792 @itemx -msched-prefer-non-control-spec-insns
10793 @opindex mno-sched-prefer-non-control-spec-insns
10794 @opindex msched-prefer-non-control-spec-insns
10795 If enabled, control speculative instructions will be chosen for schedule
10796 only if there are no other choices at the moment. This will make
10797 the use of the control speculation much more conservative.
10798 The default is 'disable'.
10800 @item -mno-sched-count-spec-in-critical-path
10801 @itemx -msched-count-spec-in-critical-path
10802 @opindex mno-sched-count-spec-in-critical-path
10803 @opindex msched-count-spec-in-critical-path
10804 If enabled, speculative dependencies will be considered during
10805 computation of the instructions priorities. This will make the use of the
10806 speculation a bit more conservative.
10807 The default is 'disable'.
10812 @subsection M32C Options
10813 @cindex M32C options
10816 @item -mcpu=@var{name}
10818 Select the CPU for which code is generated. @var{name} may be one of
10819 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
10820 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
10821 the M32C/80 series.
10825 Specifies that the program will be run on the simulator. This causes
10826 an alternate runtime library to be linked in which supports, for
10827 example, file I/O. You must not use this option when generating
10828 programs that will run on real hardware; you must provide your own
10829 runtime library for whatever I/O functions are needed.
10831 @item -memregs=@var{number}
10833 Specifies the number of memory-based pseudo-registers GCC will use
10834 during code generation. These pseudo-registers will be used like real
10835 registers, so there is a tradeoff between GCC's ability to fit the
10836 code into available registers, and the performance penalty of using
10837 memory instead of registers. Note that all modules in a program must
10838 be compiled with the same value for this option. Because of that, you
10839 must not use this option with the default runtime libraries gcc
10844 @node M32R/D Options
10845 @subsection M32R/D Options
10846 @cindex M32R/D options
10848 These @option{-m} options are defined for Renesas M32R/D architectures:
10853 Generate code for the M32R/2@.
10857 Generate code for the M32R/X@.
10861 Generate code for the M32R@. This is the default.
10863 @item -mmodel=small
10864 @opindex mmodel=small
10865 Assume all objects live in the lower 16MB of memory (so that their addresses
10866 can be loaded with the @code{ld24} instruction), and assume all subroutines
10867 are reachable with the @code{bl} instruction.
10868 This is the default.
10870 The addressability of a particular object can be set with the
10871 @code{model} attribute.
10873 @item -mmodel=medium
10874 @opindex mmodel=medium
10875 Assume objects may be anywhere in the 32-bit address space (the compiler
10876 will generate @code{seth/add3} instructions to load their addresses), and
10877 assume all subroutines are reachable with the @code{bl} instruction.
10879 @item -mmodel=large
10880 @opindex mmodel=large
10881 Assume objects may be anywhere in the 32-bit address space (the compiler
10882 will generate @code{seth/add3} instructions to load their addresses), and
10883 assume subroutines may not be reachable with the @code{bl} instruction
10884 (the compiler will generate the much slower @code{seth/add3/jl}
10885 instruction sequence).
10888 @opindex msdata=none
10889 Disable use of the small data area. Variables will be put into
10890 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10891 @code{section} attribute has been specified).
10892 This is the default.
10894 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10895 Objects may be explicitly put in the small data area with the
10896 @code{section} attribute using one of these sections.
10898 @item -msdata=sdata
10899 @opindex msdata=sdata
10900 Put small global and static data in the small data area, but do not
10901 generate special code to reference them.
10904 @opindex msdata=use
10905 Put small global and static data in the small data area, and generate
10906 special instructions to reference them.
10910 @cindex smaller data references
10911 Put global and static objects less than or equal to @var{num} bytes
10912 into the small data or bss sections instead of the normal data or bss
10913 sections. The default value of @var{num} is 8.
10914 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10915 for this option to have any effect.
10917 All modules should be compiled with the same @option{-G @var{num}} value.
10918 Compiling with different values of @var{num} may or may not work; if it
10919 doesn't the linker will give an error message---incorrect code will not be
10924 Makes the M32R specific code in the compiler display some statistics
10925 that might help in debugging programs.
10927 @item -malign-loops
10928 @opindex malign-loops
10929 Align all loops to a 32-byte boundary.
10931 @item -mno-align-loops
10932 @opindex mno-align-loops
10933 Do not enforce a 32-byte alignment for loops. This is the default.
10935 @item -missue-rate=@var{number}
10936 @opindex missue-rate=@var{number}
10937 Issue @var{number} instructions per cycle. @var{number} can only be 1
10940 @item -mbranch-cost=@var{number}
10941 @opindex mbranch-cost=@var{number}
10942 @var{number} can only be 1 or 2. If it is 1 then branches will be
10943 preferred over conditional code, if it is 2, then the opposite will
10946 @item -mflush-trap=@var{number}
10947 @opindex mflush-trap=@var{number}
10948 Specifies the trap number to use to flush the cache. The default is
10949 12. Valid numbers are between 0 and 15 inclusive.
10951 @item -mno-flush-trap
10952 @opindex mno-flush-trap
10953 Specifies that the cache cannot be flushed by using a trap.
10955 @item -mflush-func=@var{name}
10956 @opindex mflush-func=@var{name}
10957 Specifies the name of the operating system function to call to flush
10958 the cache. The default is @emph{_flush_cache}, but a function call
10959 will only be used if a trap is not available.
10961 @item -mno-flush-func
10962 @opindex mno-flush-func
10963 Indicates that there is no OS function for flushing the cache.
10967 @node M680x0 Options
10968 @subsection M680x0 Options
10969 @cindex M680x0 options
10971 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
10972 The default settings depend on which architecture was selected when
10973 the compiler was configured; the defaults for the most common choices
10977 @item -march=@var{arch}
10979 Generate code for a specific M680x0 or ColdFire instruction set
10980 architecture. Permissible values of @var{arch} for M680x0
10981 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
10982 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
10983 architectures are selected according to Freescale's ISA classification
10984 and the permissible values are: @samp{isaa}, @samp{isaaplus},
10985 @samp{isab} and @samp{isac}.
10987 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
10988 code for a ColdFire target. The @var{arch} in this macro is one of the
10989 @option{-march} arguments given above.
10991 When used together, @option{-march} and @option{-mtune} select code
10992 that runs on a family of similar processors but that is optimized
10993 for a particular microarchitecture.
10995 @item -mcpu=@var{cpu}
10997 Generate code for a specific M680x0 or ColdFire processor.
10998 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
10999 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11000 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11001 below, which also classifies the CPUs into families:
11003 @multitable @columnfractions 0.20 0.80
11004 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11005 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11006 @item @samp{5206e} @tab @samp{5206e}
11007 @item @samp{5208} @tab @samp{5207} @samp{5208}
11008 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11009 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11010 @item @samp{5216} @tab @samp{5214} @samp{5216}
11011 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11012 @item @samp{5225} @tab @samp{5224} @samp{5225}
11013 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11014 @item @samp{5249} @tab @samp{5249}
11015 @item @samp{5250} @tab @samp{5250}
11016 @item @samp{5271} @tab @samp{5270} @samp{5271}
11017 @item @samp{5272} @tab @samp{5272}
11018 @item @samp{5275} @tab @samp{5274} @samp{5275}
11019 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11020 @item @samp{5307} @tab @samp{5307}
11021 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11022 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11023 @item @samp{5407} @tab @samp{5407}
11024 @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}
11027 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11028 @var{arch} is compatible with @var{cpu}. Other combinations of
11029 @option{-mcpu} and @option{-march} are rejected.
11031 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11032 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11033 where the value of @var{family} is given by the table above.
11035 @item -mtune=@var{tune}
11037 Tune the code for a particular microarchitecture, within the
11038 constraints set by @option{-march} and @option{-mcpu}.
11039 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11040 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11041 and @samp{cpu32}. The ColdFire microarchitectures
11042 are: @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11044 You can also use @option{-mtune=68020-40} for code that needs
11045 to run relatively well on 68020, 68030 and 68040 targets.
11046 @option{-mtune=68020-60} is similar but includes 68060 targets
11047 as well. These two options select the same tuning decisions as
11048 @option{-m68020-40} and @option{-m68020-60} respectively.
11050 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11051 when tuning for 680x0 architecture @var{arch}. It also defines
11052 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11053 option is used. If gcc is tuning for a range of architectures,
11054 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11055 it defines the macros for every architecture in the range.
11057 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11058 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11059 of the arguments given above.
11065 Generate output for a 68000. This is the default
11066 when the compiler is configured for 68000-based systems.
11067 It is equivalent to @option{-march=68000}.
11069 Use this option for microcontrollers with a 68000 or EC000 core,
11070 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11074 Generate output for a 68010. This is the default
11075 when the compiler is configured for 68010-based systems.
11076 It is equivalent to @option{-march=68010}.
11082 Generate output for a 68020. This is the default
11083 when the compiler is configured for 68020-based systems.
11084 It is equivalent to @option{-march=68020}.
11088 Generate output for a 68030. This is the default when the compiler is
11089 configured for 68030-based systems. It is equivalent to
11090 @option{-march=68030}.
11094 Generate output for a 68040. This is the default when the compiler is
11095 configured for 68040-based systems. It is equivalent to
11096 @option{-march=68040}.
11098 This option inhibits the use of 68881/68882 instructions that have to be
11099 emulated by software on the 68040. Use this option if your 68040 does not
11100 have code to emulate those instructions.
11104 Generate output for a 68060. This is the default when the compiler is
11105 configured for 68060-based systems. It is equivalent to
11106 @option{-march=68060}.
11108 This option inhibits the use of 68020 and 68881/68882 instructions that
11109 have to be emulated by software on the 68060. Use this option if your 68060
11110 does not have code to emulate those instructions.
11114 Generate output for a CPU32. This is the default
11115 when the compiler is configured for CPU32-based systems.
11116 It is equivalent to @option{-march=cpu32}.
11118 Use this option for microcontrollers with a
11119 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11120 68336, 68340, 68341, 68349 and 68360.
11124 Generate output for a 520X ColdFire CPU. This is the default
11125 when the compiler is configured for 520X-based systems.
11126 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11127 in favor of that option.
11129 Use this option for microcontroller with a 5200 core, including
11130 the MCF5202, MCF5203, MCF5204 and MCF5206.
11134 Generate output for a 5206e ColdFire CPU. The option is now
11135 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11139 Generate output for a member of the ColdFire 528X family.
11140 The option is now deprecated in favor of the equivalent
11141 @option{-mcpu=528x}.
11145 Generate output for a ColdFire 5307 CPU. The option is now deprecated
11146 in favor of the equivalent @option{-mcpu=5307}.
11150 Generate output for a ColdFire 5407 CPU. The option is now deprecated
11151 in favor of the equivalent @option{-mcpu=5407}.
11155 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11156 This includes use of hardware floating point instructions.
11157 The option is equivalent to @option{-mcpu=547x}, and is now
11158 deprecated in favor of that option.
11162 Generate output for a 68040, without using any of the new instructions.
11163 This results in code which can run relatively efficiently on either a
11164 68020/68881 or a 68030 or a 68040. The generated code does use the
11165 68881 instructions that are emulated on the 68040.
11167 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11171 Generate output for a 68060, without using any of the new instructions.
11172 This results in code which can run relatively efficiently on either a
11173 68020/68881 or a 68030 or a 68040. The generated code does use the
11174 68881 instructions that are emulated on the 68060.
11176 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11180 @opindex mhard-float
11182 Generate floating-point instructions. This is the default for 68020
11183 and above, and for ColdFire devices that have an FPU. It defines the
11184 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11185 on ColdFire targets.
11188 @opindex msoft-float
11189 Do not generate floating-point instructions; use library calls instead.
11190 This is the default for 68000, 68010, and 68832 targets. It is also
11191 the default for ColdFire devices that have no FPU.
11197 Generate (do not generate) ColdFire hardware divide and remainder
11198 instructions. If @option{-march} is used without @option{-mcpu},
11199 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11200 architectures. Otherwise, the default is taken from the target CPU
11201 (either the default CPU, or the one specified by @option{-mcpu}). For
11202 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11203 @option{-mcpu=5206e}.
11205 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11209 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11210 Additionally, parameters passed on the stack are also aligned to a
11211 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11215 Do not consider type @code{int} to be 16 bits wide. This is the default.
11218 @itemx -mno-bitfield
11219 @opindex mnobitfield
11220 @opindex mno-bitfield
11221 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11222 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11226 Do use the bit-field instructions. The @option{-m68020} option implies
11227 @option{-mbitfield}. This is the default if you use a configuration
11228 designed for a 68020.
11232 Use a different function-calling convention, in which functions
11233 that take a fixed number of arguments return with the @code{rtd}
11234 instruction, which pops their arguments while returning. This
11235 saves one instruction in the caller since there is no need to pop
11236 the arguments there.
11238 This calling convention is incompatible with the one normally
11239 used on Unix, so you cannot use it if you need to call libraries
11240 compiled with the Unix compiler.
11242 Also, you must provide function prototypes for all functions that
11243 take variable numbers of arguments (including @code{printf});
11244 otherwise incorrect code will be generated for calls to those
11247 In addition, seriously incorrect code will result if you call a
11248 function with too many arguments. (Normally, extra arguments are
11249 harmlessly ignored.)
11251 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11252 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11256 Do not use the calling conventions selected by @option{-mrtd}.
11257 This is the default.
11260 @itemx -mno-align-int
11261 @opindex malign-int
11262 @opindex mno-align-int
11263 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11264 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11265 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11266 Aligning variables on 32-bit boundaries produces code that runs somewhat
11267 faster on processors with 32-bit busses at the expense of more memory.
11269 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11270 align structures containing the above types differently than
11271 most published application binary interface specifications for the m68k.
11275 Use the pc-relative addressing mode of the 68000 directly, instead of
11276 using a global offset table. At present, this option implies @option{-fpic},
11277 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11278 not presently supported with @option{-mpcrel}, though this could be supported for
11279 68020 and higher processors.
11281 @item -mno-strict-align
11282 @itemx -mstrict-align
11283 @opindex mno-strict-align
11284 @opindex mstrict-align
11285 Do not (do) assume that unaligned memory references will be handled by
11289 Generate code that allows the data segment to be located in a different
11290 area of memory from the text segment. This allows for execute in place in
11291 an environment without virtual memory management. This option implies
11294 @item -mno-sep-data
11295 Generate code that assumes that the data segment follows the text segment.
11296 This is the default.
11298 @item -mid-shared-library
11299 Generate code that supports shared libraries via the library ID method.
11300 This allows for execute in place and shared libraries in an environment
11301 without virtual memory management. This option implies @option{-fPIC}.
11303 @item -mno-id-shared-library
11304 Generate code that doesn't assume ID based shared libraries are being used.
11305 This is the default.
11307 @item -mshared-library-id=n
11308 Specified the identification number of the ID based shared library being
11309 compiled. Specifying a value of 0 will generate more compact code, specifying
11310 other values will force the allocation of that number to the current
11311 library but is no more space or time efficient than omitting this option.
11315 @node M68hc1x Options
11316 @subsection M68hc1x Options
11317 @cindex M68hc1x options
11319 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11320 microcontrollers. The default values for these options depends on
11321 which style of microcontroller was selected when the compiler was configured;
11322 the defaults for the most common choices are given below.
11329 Generate output for a 68HC11. This is the default
11330 when the compiler is configured for 68HC11-based systems.
11336 Generate output for a 68HC12. This is the default
11337 when the compiler is configured for 68HC12-based systems.
11343 Generate output for a 68HCS12.
11345 @item -mauto-incdec
11346 @opindex mauto-incdec
11347 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11354 Enable the use of 68HC12 min and max instructions.
11357 @itemx -mno-long-calls
11358 @opindex mlong-calls
11359 @opindex mno-long-calls
11360 Treat all calls as being far away (near). If calls are assumed to be
11361 far away, the compiler will use the @code{call} instruction to
11362 call a function and the @code{rtc} instruction for returning.
11366 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11368 @item -msoft-reg-count=@var{count}
11369 @opindex msoft-reg-count
11370 Specify the number of pseudo-soft registers which are used for the
11371 code generation. The maximum number is 32. Using more pseudo-soft
11372 register may or may not result in better code depending on the program.
11373 The default is 4 for 68HC11 and 2 for 68HC12.
11377 @node MCore Options
11378 @subsection MCore Options
11379 @cindex MCore options
11381 These are the @samp{-m} options defined for the Motorola M*Core
11387 @itemx -mno-hardlit
11389 @opindex mno-hardlit
11390 Inline constants into the code stream if it can be done in two
11391 instructions or less.
11397 Use the divide instruction. (Enabled by default).
11399 @item -mrelax-immediate
11400 @itemx -mno-relax-immediate
11401 @opindex mrelax-immediate
11402 @opindex mno-relax-immediate
11403 Allow arbitrary sized immediates in bit operations.
11405 @item -mwide-bitfields
11406 @itemx -mno-wide-bitfields
11407 @opindex mwide-bitfields
11408 @opindex mno-wide-bitfields
11409 Always treat bit-fields as int-sized.
11411 @item -m4byte-functions
11412 @itemx -mno-4byte-functions
11413 @opindex m4byte-functions
11414 @opindex mno-4byte-functions
11415 Force all functions to be aligned to a four byte boundary.
11417 @item -mcallgraph-data
11418 @itemx -mno-callgraph-data
11419 @opindex mcallgraph-data
11420 @opindex mno-callgraph-data
11421 Emit callgraph information.
11424 @itemx -mno-slow-bytes
11425 @opindex mslow-bytes
11426 @opindex mno-slow-bytes
11427 Prefer word access when reading byte quantities.
11429 @item -mlittle-endian
11430 @itemx -mbig-endian
11431 @opindex mlittle-endian
11432 @opindex mbig-endian
11433 Generate code for a little endian target.
11439 Generate code for the 210 processor.
11443 @subsection MIPS Options
11444 @cindex MIPS options
11450 Generate big-endian code.
11454 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11457 @item -march=@var{arch}
11459 Generate code that will run on @var{arch}, which can be the name of a
11460 generic MIPS ISA, or the name of a particular processor.
11462 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11463 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11464 The processor names are:
11465 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11466 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11467 @samp{5kc}, @samp{5kf},
11469 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11470 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11471 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11472 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11475 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11476 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11477 @samp{rm7000}, @samp{rm9000},
11480 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11481 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11482 The special value @samp{from-abi} selects the
11483 most compatible architecture for the selected ABI (that is,
11484 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11486 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11487 (for example, @samp{-march=r2k}). Prefixes are optional, and
11488 @samp{vr} may be written @samp{r}.
11490 Names of the form @samp{@var{n}f2_1} refer to processors with
11491 FPUs clocked at half the rate of the core, names of the form
11492 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
11493 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
11494 processors with FPUs clocked a ratio of 3:2 with respect to the core.
11495 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
11496 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
11497 accepted as synonyms for @samp{@var{n}f1_1}.
11499 GCC defines two macros based on the value of this option. The first
11500 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11501 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11502 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11503 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11504 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11506 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11507 above. In other words, it will have the full prefix and will not
11508 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11509 the macro names the resolved architecture (either @samp{"mips1"} or
11510 @samp{"mips3"}). It names the default architecture when no
11511 @option{-march} option is given.
11513 @item -mtune=@var{arch}
11515 Optimize for @var{arch}. Among other things, this option controls
11516 the way instructions are scheduled, and the perceived cost of arithmetic
11517 operations. The list of @var{arch} values is the same as for
11520 When this option is not used, GCC will optimize for the processor
11521 specified by @option{-march}. By using @option{-march} and
11522 @option{-mtune} together, it is possible to generate code that will
11523 run on a family of processors, but optimize the code for one
11524 particular member of that family.
11526 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11527 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11528 @samp{-march} ones described above.
11532 Equivalent to @samp{-march=mips1}.
11536 Equivalent to @samp{-march=mips2}.
11540 Equivalent to @samp{-march=mips3}.
11544 Equivalent to @samp{-march=mips4}.
11548 Equivalent to @samp{-march=mips32}.
11552 Equivalent to @samp{-march=mips32r2}.
11556 Equivalent to @samp{-march=mips64}.
11561 @opindex mno-mips16
11562 Generate (do not generate) MIPS16 code. If GCC is targetting a
11563 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11575 Generate code for the given ABI@.
11577 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11578 generates 64-bit code when you select a 64-bit architecture, but you
11579 can use @option{-mgp32} to get 32-bit code instead.
11581 For information about the O64 ABI, see
11582 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11584 GCC supports a variant of the o32 ABI in which floating-point registers
11585 are 64 rather than 32 bits wide. You can select this combination with
11586 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11587 and @samp{mfhc1} instructions and is therefore only supported for
11588 MIPS32R2 processors.
11590 The register assignments for arguments and return values remain the
11591 same, but each scalar value is passed in a single 64-bit register
11592 rather than a pair of 32-bit registers. For example, scalar
11593 floating-point values are returned in @samp{$f0} only, not a
11594 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
11595 remains the same, but all 64 bits are saved.
11598 @itemx -mno-abicalls
11600 @opindex mno-abicalls
11601 Generate (do not generate) code that is suitable for SVR4-style
11602 dynamic objects. @option{-mabicalls} is the default for SVR4-based
11607 Generate (do not generate) code that is fully position-independent,
11608 and that can therefore be linked into shared libraries. This option
11609 only affects @option{-mabicalls}.
11611 All @option{-mabicalls} code has traditionally been position-independent,
11612 regardless of options like @option{-fPIC} and @option{-fpic}. However,
11613 as an extension, the GNU toolchain allows executables to use absolute
11614 accesses for locally-binding symbols. It can also use shorter GP
11615 initialization sequences and generate direct calls to locally-defined
11616 functions. This mode is selected by @option{-mno-shared}.
11618 @option{-mno-shared} depends on binutils 2.16 or higher and generates
11619 objects that can only be linked by the GNU linker. However, the option
11620 does not affect the ABI of the final executable; it only affects the ABI
11621 of relocatable objects. Using @option{-mno-shared} will generally make
11622 executables both smaller and quicker.
11624 @option{-mshared} is the default.
11630 Lift (do not lift) the usual restrictions on the size of the global
11633 GCC normally uses a single instruction to load values from the GOT@.
11634 While this is relatively efficient, it will only work if the GOT
11635 is smaller than about 64k. Anything larger will cause the linker
11636 to report an error such as:
11638 @cindex relocation truncated to fit (MIPS)
11640 relocation truncated to fit: R_MIPS_GOT16 foobar
11643 If this happens, you should recompile your code with @option{-mxgot}.
11644 It should then work with very large GOTs, although it will also be
11645 less efficient, since it will take three instructions to fetch the
11646 value of a global symbol.
11648 Note that some linkers can create multiple GOTs. If you have such a
11649 linker, you should only need to use @option{-mxgot} when a single object
11650 file accesses more than 64k's worth of GOT entries. Very few do.
11652 These options have no effect unless GCC is generating position
11657 Assume that general-purpose registers are 32 bits wide.
11661 Assume that general-purpose registers are 64 bits wide.
11665 Assume that floating-point registers are 32 bits wide.
11669 Assume that floating-point registers are 64 bits wide.
11672 @opindex mhard-float
11673 Use floating-point coprocessor instructions.
11676 @opindex msoft-float
11677 Do not use floating-point coprocessor instructions. Implement
11678 floating-point calculations using library calls instead.
11680 @item -msingle-float
11681 @opindex msingle-float
11682 Assume that the floating-point coprocessor only supports single-precision
11685 @item -mdouble-float
11686 @opindex mdouble-float
11687 Assume that the floating-point coprocessor supports double-precision
11688 operations. This is the default.
11694 Use (do not use) revision 1 of the MIPS DSP ASE.
11695 @xref{MIPS DSP Built-in Functions}. This option defines the
11696 preprocessor macro @samp{__mips_dsp}. It also defines
11697 @samp{__mips_dsp_rev} to 1.
11703 Use (do not use) revision 2 of the MIPS DSP ASE.
11704 @xref{MIPS DSP Built-in Functions}. This option defines the
11705 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
11706 It also defines @samp{__mips_dsp_rev} to 2.
11709 @itemx -mno-smartmips
11710 @opindex msmartmips
11711 @opindex mno-smartmips
11712 Use (do not use) the MIPS SmartMIPS ASE.
11714 @item -mpaired-single
11715 @itemx -mno-paired-single
11716 @opindex mpaired-single
11717 @opindex mno-paired-single
11718 Use (do not use) paired-single floating-point instructions.
11719 @xref{MIPS Paired-Single Support}. This option can only be used
11720 when generating 64-bit code and requires hardware floating-point
11721 support to be enabled.
11727 Use (do not use) MIPS Digital Media Extension instructions.
11728 This option can only be used when generating 64-bit code and requires
11729 hardware floating-point support to be enabled.
11734 @opindex mno-mips3d
11735 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
11736 The option @option{-mips3d} implies @option{-mpaired-single}.
11742 Use (do not use) MT Multithreading instructions.
11746 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
11747 an explanation of the default and the way that the pointer size is
11752 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
11754 The default size of @code{int}s, @code{long}s and pointers depends on
11755 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
11756 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
11757 32-bit @code{long}s. Pointers are the same size as @code{long}s,
11758 or the same size as integer registers, whichever is smaller.
11764 Assume (do not assume) that all symbols have 32-bit values, regardless
11765 of the selected ABI@. This option is useful in combination with
11766 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
11767 to generate shorter and faster references to symbolic addresses.
11771 @cindex smaller data references (MIPS)
11772 @cindex gp-relative references (MIPS)
11773 Put global and static items less than or equal to @var{num} bytes into
11774 the small data or bss section instead of the normal data or bss section.
11775 This allows the data to be accessed using a single instruction.
11777 All modules should be compiled with the same @option{-G @var{num}}
11780 @item -membedded-data
11781 @itemx -mno-embedded-data
11782 @opindex membedded-data
11783 @opindex mno-embedded-data
11784 Allocate variables to the read-only data section first if possible, then
11785 next in the small data section if possible, otherwise in data. This gives
11786 slightly slower code than the default, but reduces the amount of RAM required
11787 when executing, and thus may be preferred for some embedded systems.
11789 @item -muninit-const-in-rodata
11790 @itemx -mno-uninit-const-in-rodata
11791 @opindex muninit-const-in-rodata
11792 @opindex mno-uninit-const-in-rodata
11793 Put uninitialized @code{const} variables in the read-only data section.
11794 This option is only meaningful in conjunction with @option{-membedded-data}.
11796 @item -mcode-readable=@var{setting}
11797 @opindex mcode-readable
11798 Specify whether GCC may generate code that reads from executable sections.
11799 There are three possible settings:
11802 @item -mcode-readable=yes
11803 Instructions may freely access executable sections. This is the
11806 @item -mcode-readable=pcrel
11807 MIPS16 PC-relative load instructions can access executable sections,
11808 but other instructions must not do so. This option is useful on 4KSc
11809 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
11810 It is also useful on processors that can be configured to have a dual
11811 instruction/data SRAM interface and that, like the M4K, automatically
11812 redirect PC-relative loads to the instruction RAM.
11814 @item -mcode-readable=no
11815 Instructions must not access executable sections. This option can be
11816 useful on targets that are configured to have a dual instruction/data
11817 SRAM interface but that (unlike the M4K) do not automatically redirect
11818 PC-relative loads to the instruction RAM.
11821 @item -msplit-addresses
11822 @itemx -mno-split-addresses
11823 @opindex msplit-addresses
11824 @opindex mno-split-addresses
11825 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
11826 relocation operators. This option has been superseded by
11827 @option{-mexplicit-relocs} but is retained for backwards compatibility.
11829 @item -mexplicit-relocs
11830 @itemx -mno-explicit-relocs
11831 @opindex mexplicit-relocs
11832 @opindex mno-explicit-relocs
11833 Use (do not use) assembler relocation operators when dealing with symbolic
11834 addresses. The alternative, selected by @option{-mno-explicit-relocs},
11835 is to use assembler macros instead.
11837 @option{-mexplicit-relocs} is the default if GCC was configured
11838 to use an assembler that supports relocation operators.
11840 @item -mcheck-zero-division
11841 @itemx -mno-check-zero-division
11842 @opindex mcheck-zero-division
11843 @opindex mno-check-zero-division
11844 Trap (do not trap) on integer division by zero.
11846 The default is @option{-mcheck-zero-division}.
11848 @item -mdivide-traps
11849 @itemx -mdivide-breaks
11850 @opindex mdivide-traps
11851 @opindex mdivide-breaks
11852 MIPS systems check for division by zero by generating either a
11853 conditional trap or a break instruction. Using traps results in
11854 smaller code, but is only supported on MIPS II and later. Also, some
11855 versions of the Linux kernel have a bug that prevents trap from
11856 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
11857 allow conditional traps on architectures that support them and
11858 @option{-mdivide-breaks} to force the use of breaks.
11860 The default is usually @option{-mdivide-traps}, but this can be
11861 overridden at configure time using @option{--with-divide=breaks}.
11862 Divide-by-zero checks can be completely disabled using
11863 @option{-mno-check-zero-division}.
11868 @opindex mno-memcpy
11869 Force (do not force) the use of @code{memcpy()} for non-trivial block
11870 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
11871 most constant-sized copies.
11874 @itemx -mno-long-calls
11875 @opindex mlong-calls
11876 @opindex mno-long-calls
11877 Disable (do not disable) use of the @code{jal} instruction. Calling
11878 functions using @code{jal} is more efficient but requires the caller
11879 and callee to be in the same 256 megabyte segment.
11881 This option has no effect on abicalls code. The default is
11882 @option{-mno-long-calls}.
11888 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
11889 instructions, as provided by the R4650 ISA@.
11892 @itemx -mno-fused-madd
11893 @opindex mfused-madd
11894 @opindex mno-fused-madd
11895 Enable (disable) use of the floating point multiply-accumulate
11896 instructions, when they are available. The default is
11897 @option{-mfused-madd}.
11899 When multiply-accumulate instructions are used, the intermediate
11900 product is calculated to infinite precision and is not subject to
11901 the FCSR Flush to Zero bit. This may be undesirable in some
11906 Tell the MIPS assembler to not run its preprocessor over user
11907 assembler files (with a @samp{.s} suffix) when assembling them.
11910 @itemx -mno-fix-r4000
11911 @opindex mfix-r4000
11912 @opindex mno-fix-r4000
11913 Work around certain R4000 CPU errata:
11916 A double-word or a variable shift may give an incorrect result if executed
11917 immediately after starting an integer division.
11919 A double-word or a variable shift may give an incorrect result if executed
11920 while an integer multiplication is in progress.
11922 An integer division may give an incorrect result if started in a delay slot
11923 of a taken branch or a jump.
11927 @itemx -mno-fix-r4400
11928 @opindex mfix-r4400
11929 @opindex mno-fix-r4400
11930 Work around certain R4400 CPU errata:
11933 A double-word or a variable shift may give an incorrect result if executed
11934 immediately after starting an integer division.
11938 @itemx -mno-fix-vr4120
11939 @opindex mfix-vr4120
11940 Work around certain VR4120 errata:
11943 @code{dmultu} does not always produce the correct result.
11945 @code{div} and @code{ddiv} do not always produce the correct result if one
11946 of the operands is negative.
11948 The workarounds for the division errata rely on special functions in
11949 @file{libgcc.a}. At present, these functions are only provided by
11950 the @code{mips64vr*-elf} configurations.
11952 Other VR4120 errata require a nop to be inserted between certain pairs of
11953 instructions. These errata are handled by the assembler, not by GCC itself.
11956 @opindex mfix-vr4130
11957 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
11958 workarounds are implemented by the assembler rather than by GCC,
11959 although GCC will avoid using @code{mflo} and @code{mfhi} if the
11960 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
11961 instructions are available instead.
11964 @itemx -mno-fix-sb1
11966 Work around certain SB-1 CPU core errata.
11967 (This flag currently works around the SB-1 revision 2
11968 ``F1'' and ``F2'' floating point errata.)
11970 @item -mflush-func=@var{func}
11971 @itemx -mno-flush-func
11972 @opindex mflush-func
11973 Specifies the function to call to flush the I and D caches, or to not
11974 call any such function. If called, the function must take the same
11975 arguments as the common @code{_flush_func()}, that is, the address of the
11976 memory range for which the cache is being flushed, the size of the
11977 memory range, and the number 3 (to flush both caches). The default
11978 depends on the target GCC was configured for, but commonly is either
11979 @samp{_flush_func} or @samp{__cpu_flush}.
11981 @item mbranch-cost=@var{num}
11982 @opindex mbranch-cost
11983 Set the cost of branches to roughly @var{num} ``simple'' instructions.
11984 This cost is only a heuristic and is not guaranteed to produce
11985 consistent results across releases. A zero cost redundantly selects
11986 the default, which is based on the @option{-mtune} setting.
11988 @item -mbranch-likely
11989 @itemx -mno-branch-likely
11990 @opindex mbranch-likely
11991 @opindex mno-branch-likely
11992 Enable or disable use of Branch Likely instructions, regardless of the
11993 default for the selected architecture. By default, Branch Likely
11994 instructions may be generated if they are supported by the selected
11995 architecture. An exception is for the MIPS32 and MIPS64 architectures
11996 and processors which implement those architectures; for those, Branch
11997 Likely instructions will not be generated by default because the MIPS32
11998 and MIPS64 architectures specifically deprecate their use.
12000 @item -mfp-exceptions
12001 @itemx -mno-fp-exceptions
12002 @opindex mfp-exceptions
12003 Specifies whether FP exceptions are enabled. This affects how we schedule
12004 FP instructions for some processors. The default is that FP exceptions are
12007 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12008 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12011 @item -mvr4130-align
12012 @itemx -mno-vr4130-align
12013 @opindex mvr4130-align
12014 The VR4130 pipeline is two-way superscalar, but can only issue two
12015 instructions together if the first one is 8-byte aligned. When this
12016 option is enabled, GCC will align pairs of instructions that it
12017 thinks should execute in parallel.
12019 This option only has an effect when optimizing for the VR4130.
12020 It normally makes code faster, but at the expense of making it bigger.
12021 It is enabled by default at optimization level @option{-O3}.
12025 @subsection MMIX Options
12026 @cindex MMIX Options
12028 These options are defined for the MMIX:
12032 @itemx -mno-libfuncs
12034 @opindex mno-libfuncs
12035 Specify that intrinsic library functions are being compiled, passing all
12036 values in registers, no matter the size.
12039 @itemx -mno-epsilon
12041 @opindex mno-epsilon
12042 Generate floating-point comparison instructions that compare with respect
12043 to the @code{rE} epsilon register.
12045 @item -mabi=mmixware
12047 @opindex mabi-mmixware
12049 Generate code that passes function parameters and return values that (in
12050 the called function) are seen as registers @code{$0} and up, as opposed to
12051 the GNU ABI which uses global registers @code{$231} and up.
12053 @item -mzero-extend
12054 @itemx -mno-zero-extend
12055 @opindex mzero-extend
12056 @opindex mno-zero-extend
12057 When reading data from memory in sizes shorter than 64 bits, use (do not
12058 use) zero-extending load instructions by default, rather than
12059 sign-extending ones.
12062 @itemx -mno-knuthdiv
12064 @opindex mno-knuthdiv
12065 Make the result of a division yielding a remainder have the same sign as
12066 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12067 remainder follows the sign of the dividend. Both methods are
12068 arithmetically valid, the latter being almost exclusively used.
12070 @item -mtoplevel-symbols
12071 @itemx -mno-toplevel-symbols
12072 @opindex mtoplevel-symbols
12073 @opindex mno-toplevel-symbols
12074 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12075 code can be used with the @code{PREFIX} assembly directive.
12079 Generate an executable in the ELF format, rather than the default
12080 @samp{mmo} format used by the @command{mmix} simulator.
12082 @item -mbranch-predict
12083 @itemx -mno-branch-predict
12084 @opindex mbranch-predict
12085 @opindex mno-branch-predict
12086 Use (do not use) the probable-branch instructions, when static branch
12087 prediction indicates a probable branch.
12089 @item -mbase-addresses
12090 @itemx -mno-base-addresses
12091 @opindex mbase-addresses
12092 @opindex mno-base-addresses
12093 Generate (do not generate) code that uses @emph{base addresses}. Using a
12094 base address automatically generates a request (handled by the assembler
12095 and the linker) for a constant to be set up in a global register. The
12096 register is used for one or more base address requests within the range 0
12097 to 255 from the value held in the register. The generally leads to short
12098 and fast code, but the number of different data items that can be
12099 addressed is limited. This means that a program that uses lots of static
12100 data may require @option{-mno-base-addresses}.
12102 @item -msingle-exit
12103 @itemx -mno-single-exit
12104 @opindex msingle-exit
12105 @opindex mno-single-exit
12106 Force (do not force) generated code to have a single exit point in each
12110 @node MN10300 Options
12111 @subsection MN10300 Options
12112 @cindex MN10300 options
12114 These @option{-m} options are defined for Matsushita MN10300 architectures:
12119 Generate code to avoid bugs in the multiply instructions for the MN10300
12120 processors. This is the default.
12122 @item -mno-mult-bug
12123 @opindex mno-mult-bug
12124 Do not generate code to avoid bugs in the multiply instructions for the
12125 MN10300 processors.
12129 Generate code which uses features specific to the AM33 processor.
12133 Do not generate code which uses features specific to the AM33 processor. This
12136 @item -mreturn-pointer-on-d0
12137 @opindex mreturn-pointer-on-d0
12138 When generating a function which returns a pointer, return the pointer
12139 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12140 only in a0, and attempts to call such functions without a prototype
12141 would result in errors. Note that this option is on by default; use
12142 @option{-mno-return-pointer-on-d0} to disable it.
12146 Do not link in the C run-time initialization object file.
12150 Indicate to the linker that it should perform a relaxation optimization pass
12151 to shorten branches, calls and absolute memory addresses. This option only
12152 has an effect when used on the command line for the final link step.
12154 This option makes symbolic debugging impossible.
12158 @subsection MT Options
12161 These @option{-m} options are defined for Morpho MT architectures:
12165 @item -march=@var{cpu-type}
12167 Generate code that will run on @var{cpu-type}, which is the name of a system
12168 representing a certain processor type. Possible values for
12169 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
12170 @samp{ms1-16-003} and @samp{ms2}.
12172 When this option is not used, the default is @option{-march=ms1-16-002}.
12176 Use byte loads and stores when generating code.
12180 Do not use byte loads and stores when generating code.
12184 Use simulator runtime
12188 Do not link in the C run-time initialization object file
12189 @file{crti.o}. Other run-time initialization and termination files
12190 such as @file{startup.o} and @file{exit.o} are still included on the
12191 linker command line.
12195 @node PDP-11 Options
12196 @subsection PDP-11 Options
12197 @cindex PDP-11 Options
12199 These options are defined for the PDP-11:
12204 Use hardware FPP floating point. This is the default. (FIS floating
12205 point on the PDP-11/40 is not supported.)
12208 @opindex msoft-float
12209 Do not use hardware floating point.
12213 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12217 Return floating-point results in memory. This is the default.
12221 Generate code for a PDP-11/40.
12225 Generate code for a PDP-11/45. This is the default.
12229 Generate code for a PDP-11/10.
12231 @item -mbcopy-builtin
12232 @opindex bcopy-builtin
12233 Use inline @code{movmemhi} patterns for copying memory. This is the
12238 Do not use inline @code{movmemhi} patterns for copying memory.
12244 Use 16-bit @code{int}. This is the default.
12250 Use 32-bit @code{int}.
12253 @itemx -mno-float32
12255 @opindex mno-float32
12256 Use 64-bit @code{float}. This is the default.
12259 @itemx -mno-float64
12261 @opindex mno-float64
12262 Use 32-bit @code{float}.
12266 Use @code{abshi2} pattern. This is the default.
12270 Do not use @code{abshi2} pattern.
12272 @item -mbranch-expensive
12273 @opindex mbranch-expensive
12274 Pretend that branches are expensive. This is for experimenting with
12275 code generation only.
12277 @item -mbranch-cheap
12278 @opindex mbranch-cheap
12279 Do not pretend that branches are expensive. This is the default.
12283 Generate code for a system with split I&D@.
12287 Generate code for a system without split I&D@. This is the default.
12291 Use Unix assembler syntax. This is the default when configured for
12292 @samp{pdp11-*-bsd}.
12296 Use DEC assembler syntax. This is the default when configured for any
12297 PDP-11 target other than @samp{pdp11-*-bsd}.
12300 @node PowerPC Options
12301 @subsection PowerPC Options
12302 @cindex PowerPC options
12304 These are listed under @xref{RS/6000 and PowerPC Options}.
12306 @node RS/6000 and PowerPC Options
12307 @subsection IBM RS/6000 and PowerPC Options
12308 @cindex RS/6000 and PowerPC Options
12309 @cindex IBM RS/6000 and PowerPC Options
12311 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12318 @itemx -mno-powerpc
12319 @itemx -mpowerpc-gpopt
12320 @itemx -mno-powerpc-gpopt
12321 @itemx -mpowerpc-gfxopt
12322 @itemx -mno-powerpc-gfxopt
12324 @itemx -mno-powerpc64
12328 @itemx -mno-popcntb
12340 @opindex mno-power2
12342 @opindex mno-powerpc
12343 @opindex mpowerpc-gpopt
12344 @opindex mno-powerpc-gpopt
12345 @opindex mpowerpc-gfxopt
12346 @opindex mno-powerpc-gfxopt
12347 @opindex mpowerpc64
12348 @opindex mno-powerpc64
12352 @opindex mno-popcntb
12358 @opindex mno-mfpgpr
12361 GCC supports two related instruction set architectures for the
12362 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12363 instructions supported by the @samp{rios} chip set used in the original
12364 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12365 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12366 the IBM 4xx, 6xx, and follow-on microprocessors.
12368 Neither architecture is a subset of the other. However there is a
12369 large common subset of instructions supported by both. An MQ
12370 register is included in processors supporting the POWER architecture.
12372 You use these options to specify which instructions are available on the
12373 processor you are using. The default value of these options is
12374 determined when configuring GCC@. Specifying the
12375 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12376 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12377 rather than the options listed above.
12379 The @option{-mpower} option allows GCC to generate instructions that
12380 are found only in the POWER architecture and to use the MQ register.
12381 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12382 to generate instructions that are present in the POWER2 architecture but
12383 not the original POWER architecture.
12385 The @option{-mpowerpc} option allows GCC to generate instructions that
12386 are found only in the 32-bit subset of the PowerPC architecture.
12387 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12388 GCC to use the optional PowerPC architecture instructions in the
12389 General Purpose group, including floating-point square root. Specifying
12390 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12391 use the optional PowerPC architecture instructions in the Graphics
12392 group, including floating-point select.
12394 The @option{-mmfcrf} option allows GCC to generate the move from
12395 condition register field instruction implemented on the POWER4
12396 processor and other processors that support the PowerPC V2.01
12398 The @option{-mpopcntb} option allows GCC to generate the popcount and
12399 double precision FP reciprocal estimate instruction implemented on the
12400 POWER5 processor and other processors that support the PowerPC V2.02
12402 The @option{-mfprnd} option allows GCC to generate the FP round to
12403 integer instructions implemented on the POWER5+ processor and other
12404 processors that support the PowerPC V2.03 architecture.
12405 The @option{-mcmpb} option allows GCC to generate the compare bytes
12406 instruction implemented on the POWER6 processor and other processors
12407 that support the PowerPC V2.05 architecture.
12408 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12409 general purpose register instructions implemented on the POWER6X
12410 processor and other processors that support the extended PowerPC V2.05
12412 The @option{-mdfp} option allows GCC to generate the decimal floating
12413 point instructions implemented on some POWER processors.
12415 The @option{-mpowerpc64} option allows GCC to generate the additional
12416 64-bit instructions that are found in the full PowerPC64 architecture
12417 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12418 @option{-mno-powerpc64}.
12420 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12421 will use only the instructions in the common subset of both
12422 architectures plus some special AIX common-mode calls, and will not use
12423 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12424 permits GCC to use any instruction from either architecture and to
12425 allow use of the MQ register; specify this for the Motorola MPC601.
12427 @item -mnew-mnemonics
12428 @itemx -mold-mnemonics
12429 @opindex mnew-mnemonics
12430 @opindex mold-mnemonics
12431 Select which mnemonics to use in the generated assembler code. With
12432 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12433 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12434 assembler mnemonics defined for the POWER architecture. Instructions
12435 defined in only one architecture have only one mnemonic; GCC uses that
12436 mnemonic irrespective of which of these options is specified.
12438 GCC defaults to the mnemonics appropriate for the architecture in
12439 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
12440 value of these option. Unless you are building a cross-compiler, you
12441 should normally not specify either @option{-mnew-mnemonics} or
12442 @option{-mold-mnemonics}, but should instead accept the default.
12444 @item -mcpu=@var{cpu_type}
12446 Set architecture type, register usage, choice of mnemonics, and
12447 instruction scheduling parameters for machine type @var{cpu_type}.
12448 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
12449 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
12450 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
12451 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
12452 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
12453 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
12454 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
12455 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
12456 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
12457 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
12459 @option{-mcpu=common} selects a completely generic processor. Code
12460 generated under this option will run on any POWER or PowerPC processor.
12461 GCC will use only the instructions in the common subset of both
12462 architectures, and will not use the MQ register. GCC assumes a generic
12463 processor model for scheduling purposes.
12465 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
12466 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
12467 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
12468 types, with an appropriate, generic processor model assumed for
12469 scheduling purposes.
12471 The other options specify a specific processor. Code generated under
12472 those options will run best on that processor, and may not run at all on
12475 The @option{-mcpu} options automatically enable or disable the
12478 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
12479 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
12480 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
12482 The particular options set for any particular CPU will vary between
12483 compiler versions, depending on what setting seems to produce optimal
12484 code for that CPU; it doesn't necessarily reflect the actual hardware's
12485 capabilities. If you wish to set an individual option to a particular
12486 value, you may specify it after the @option{-mcpu} option, like
12487 @samp{-mcpu=970 -mno-altivec}.
12489 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
12490 not enabled or disabled by the @option{-mcpu} option at present because
12491 AIX does not have full support for these options. You may still
12492 enable or disable them individually if you're sure it'll work in your
12495 @item -mtune=@var{cpu_type}
12497 Set the instruction scheduling parameters for machine type
12498 @var{cpu_type}, but do not set the architecture type, register usage, or
12499 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
12500 values for @var{cpu_type} are used for @option{-mtune} as for
12501 @option{-mcpu}. If both are specified, the code generated will use the
12502 architecture, registers, and mnemonics set by @option{-mcpu}, but the
12503 scheduling parameters set by @option{-mtune}.
12509 Generate code to compute division as reciprocal estimate and iterative
12510 refinement, creating opportunities for increased throughput. This
12511 feature requires: optional PowerPC Graphics instruction set for single
12512 precision and FRE instruction for double precision, assuming divides
12513 cannot generate user-visible traps, and the domain values not include
12514 Infinities, denormals or zero denominator.
12517 @itemx -mno-altivec
12519 @opindex mno-altivec
12520 Generate code that uses (does not use) AltiVec instructions, and also
12521 enable the use of built-in functions that allow more direct access to
12522 the AltiVec instruction set. You may also need to set
12523 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
12529 @opindex mno-vrsave
12530 Generate VRSAVE instructions when generating AltiVec code.
12533 @opindex msecure-plt
12534 Generate code that allows ld and ld.so to build executables and shared
12535 libraries with non-exec .plt and .got sections. This is a PowerPC
12536 32-bit SYSV ABI option.
12540 Generate code that uses a BSS .plt section that ld.so fills in, and
12541 requires .plt and .got sections that are both writable and executable.
12542 This is a PowerPC 32-bit SYSV ABI option.
12548 This switch enables or disables the generation of ISEL instructions.
12550 @item -misel=@var{yes/no}
12551 This switch has been deprecated. Use @option{-misel} and
12552 @option{-mno-isel} instead.
12558 This switch enables or disables the generation of SPE simd
12561 @item -mspe=@var{yes/no}
12562 This option has been deprecated. Use @option{-mspe} and
12563 @option{-mno-spe} instead.
12565 @item -mfloat-gprs=@var{yes/single/double/no}
12566 @itemx -mfloat-gprs
12567 @opindex mfloat-gprs
12568 This switch enables or disables the generation of floating point
12569 operations on the general purpose registers for architectures that
12572 The argument @var{yes} or @var{single} enables the use of
12573 single-precision floating point operations.
12575 The argument @var{double} enables the use of single and
12576 double-precision floating point operations.
12578 The argument @var{no} disables floating point operations on the
12579 general purpose registers.
12581 This option is currently only available on the MPC854x.
12587 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
12588 targets (including GNU/Linux). The 32-bit environment sets int, long
12589 and pointer to 32 bits and generates code that runs on any PowerPC
12590 variant. The 64-bit environment sets int to 32 bits and long and
12591 pointer to 64 bits, and generates code for PowerPC64, as for
12592 @option{-mpowerpc64}.
12595 @itemx -mno-fp-in-toc
12596 @itemx -mno-sum-in-toc
12597 @itemx -mminimal-toc
12599 @opindex mno-fp-in-toc
12600 @opindex mno-sum-in-toc
12601 @opindex mminimal-toc
12602 Modify generation of the TOC (Table Of Contents), which is created for
12603 every executable file. The @option{-mfull-toc} option is selected by
12604 default. In that case, GCC will allocate at least one TOC entry for
12605 each unique non-automatic variable reference in your program. GCC
12606 will also place floating-point constants in the TOC@. However, only
12607 16,384 entries are available in the TOC@.
12609 If you receive a linker error message that saying you have overflowed
12610 the available TOC space, you can reduce the amount of TOC space used
12611 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
12612 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
12613 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
12614 generate code to calculate the sum of an address and a constant at
12615 run-time instead of putting that sum into the TOC@. You may specify one
12616 or both of these options. Each causes GCC to produce very slightly
12617 slower and larger code at the expense of conserving TOC space.
12619 If you still run out of space in the TOC even when you specify both of
12620 these options, specify @option{-mminimal-toc} instead. This option causes
12621 GCC to make only one TOC entry for every file. When you specify this
12622 option, GCC will produce code that is slower and larger but which
12623 uses extremely little TOC space. You may wish to use this option
12624 only on files that contain less frequently executed code.
12630 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
12631 @code{long} type, and the infrastructure needed to support them.
12632 Specifying @option{-maix64} implies @option{-mpowerpc64} and
12633 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
12634 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
12637 @itemx -mno-xl-compat
12638 @opindex mxl-compat
12639 @opindex mno-xl-compat
12640 Produce code that conforms more closely to IBM XL compiler semantics
12641 when using AIX-compatible ABI. Pass floating-point arguments to
12642 prototyped functions beyond the register save area (RSA) on the stack
12643 in addition to argument FPRs. Do not assume that most significant
12644 double in 128-bit long double value is properly rounded when comparing
12645 values and converting to double. Use XL symbol names for long double
12648 The AIX calling convention was extended but not initially documented to
12649 handle an obscure K&R C case of calling a function that takes the
12650 address of its arguments with fewer arguments than declared. IBM XL
12651 compilers access floating point arguments which do not fit in the
12652 RSA from the stack when a subroutine is compiled without
12653 optimization. Because always storing floating-point arguments on the
12654 stack is inefficient and rarely needed, this option is not enabled by
12655 default and only is necessary when calling subroutines compiled by IBM
12656 XL compilers without optimization.
12660 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
12661 application written to use message passing with special startup code to
12662 enable the application to run. The system must have PE installed in the
12663 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
12664 must be overridden with the @option{-specs=} option to specify the
12665 appropriate directory location. The Parallel Environment does not
12666 support threads, so the @option{-mpe} option and the @option{-pthread}
12667 option are incompatible.
12669 @item -malign-natural
12670 @itemx -malign-power
12671 @opindex malign-natural
12672 @opindex malign-power
12673 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
12674 @option{-malign-natural} overrides the ABI-defined alignment of larger
12675 types, such as floating-point doubles, on their natural size-based boundary.
12676 The option @option{-malign-power} instructs GCC to follow the ABI-specified
12677 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
12679 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
12683 @itemx -mhard-float
12684 @opindex msoft-float
12685 @opindex mhard-float
12686 Generate code that does not use (uses) the floating-point register set.
12687 Software floating point emulation is provided if you use the
12688 @option{-msoft-float} option, and pass the option to GCC when linking.
12691 @itemx -mno-multiple
12693 @opindex mno-multiple
12694 Generate code that uses (does not use) the load multiple word
12695 instructions and the store multiple word instructions. These
12696 instructions are generated by default on POWER systems, and not
12697 generated on PowerPC systems. Do not use @option{-mmultiple} on little
12698 endian PowerPC systems, since those instructions do not work when the
12699 processor is in little endian mode. The exceptions are PPC740 and
12700 PPC750 which permit the instructions usage in little endian mode.
12705 @opindex mno-string
12706 Generate code that uses (does not use) the load string instructions
12707 and the store string word instructions to save multiple registers and
12708 do small block moves. These instructions are generated by default on
12709 POWER systems, and not generated on PowerPC systems. Do not use
12710 @option{-mstring} on little endian PowerPC systems, since those
12711 instructions do not work when the processor is in little endian mode.
12712 The exceptions are PPC740 and PPC750 which permit the instructions
12713 usage in little endian mode.
12718 @opindex mno-update
12719 Generate code that uses (does not use) the load or store instructions
12720 that update the base register to the address of the calculated memory
12721 location. These instructions are generated by default. If you use
12722 @option{-mno-update}, there is a small window between the time that the
12723 stack pointer is updated and the address of the previous frame is
12724 stored, which means code that walks the stack frame across interrupts or
12725 signals may get corrupted data.
12728 @itemx -mno-fused-madd
12729 @opindex mfused-madd
12730 @opindex mno-fused-madd
12731 Generate code that uses (does not use) the floating point multiply and
12732 accumulate instructions. These instructions are generated by default if
12733 hardware floating is used.
12739 Generate code that uses (does not use) the half-word multiply and
12740 multiply-accumulate instructions on the IBM 405 and 440 processors.
12741 These instructions are generated by default when targetting those
12748 Generate code that uses (does not use) the string-search @samp{dlmzb}
12749 instruction on the IBM 405 and 440 processors. This instruction is
12750 generated by default when targetting those processors.
12752 @item -mno-bit-align
12754 @opindex mno-bit-align
12755 @opindex mbit-align
12756 On System V.4 and embedded PowerPC systems do not (do) force structures
12757 and unions that contain bit-fields to be aligned to the base type of the
12760 For example, by default a structure containing nothing but 8
12761 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
12762 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
12763 the structure would be aligned to a 1 byte boundary and be one byte in
12766 @item -mno-strict-align
12767 @itemx -mstrict-align
12768 @opindex mno-strict-align
12769 @opindex mstrict-align
12770 On System V.4 and embedded PowerPC systems do not (do) assume that
12771 unaligned memory references will be handled by the system.
12773 @item -mrelocatable
12774 @itemx -mno-relocatable
12775 @opindex mrelocatable
12776 @opindex mno-relocatable
12777 On embedded PowerPC systems generate code that allows (does not allow)
12778 the program to be relocated to a different address at runtime. If you
12779 use @option{-mrelocatable} on any module, all objects linked together must
12780 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
12782 @item -mrelocatable-lib
12783 @itemx -mno-relocatable-lib
12784 @opindex mrelocatable-lib
12785 @opindex mno-relocatable-lib
12786 On embedded PowerPC systems generate code that allows (does not allow)
12787 the program to be relocated to a different address at runtime. Modules
12788 compiled with @option{-mrelocatable-lib} can be linked with either modules
12789 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
12790 with modules compiled with the @option{-mrelocatable} options.
12796 On System V.4 and embedded PowerPC systems do not (do) assume that
12797 register 2 contains a pointer to a global area pointing to the addresses
12798 used in the program.
12801 @itemx -mlittle-endian
12803 @opindex mlittle-endian
12804 On System V.4 and embedded PowerPC systems compile code for the
12805 processor in little endian mode. The @option{-mlittle-endian} option is
12806 the same as @option{-mlittle}.
12809 @itemx -mbig-endian
12811 @opindex mbig-endian
12812 On System V.4 and embedded PowerPC systems compile code for the
12813 processor in big endian mode. The @option{-mbig-endian} option is
12814 the same as @option{-mbig}.
12816 @item -mdynamic-no-pic
12817 @opindex mdynamic-no-pic
12818 On Darwin and Mac OS X systems, compile code so that it is not
12819 relocatable, but that its external references are relocatable. The
12820 resulting code is suitable for applications, but not shared
12823 @item -mprioritize-restricted-insns=@var{priority}
12824 @opindex mprioritize-restricted-insns
12825 This option controls the priority that is assigned to
12826 dispatch-slot restricted instructions during the second scheduling
12827 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
12828 @var{no/highest/second-highest} priority to dispatch slot restricted
12831 @item -msched-costly-dep=@var{dependence_type}
12832 @opindex msched-costly-dep
12833 This option controls which dependences are considered costly
12834 by the target during instruction scheduling. The argument
12835 @var{dependence_type} takes one of the following values:
12836 @var{no}: no dependence is costly,
12837 @var{all}: all dependences are costly,
12838 @var{true_store_to_load}: a true dependence from store to load is costly,
12839 @var{store_to_load}: any dependence from store to load is costly,
12840 @var{number}: any dependence which latency >= @var{number} is costly.
12842 @item -minsert-sched-nops=@var{scheme}
12843 @opindex minsert-sched-nops
12844 This option controls which nop insertion scheme will be used during
12845 the second scheduling pass. The argument @var{scheme} takes one of the
12847 @var{no}: Don't insert nops.
12848 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
12849 according to the scheduler's grouping.
12850 @var{regroup_exact}: Insert nops to force costly dependent insns into
12851 separate groups. Insert exactly as many nops as needed to force an insn
12852 to a new group, according to the estimated processor grouping.
12853 @var{number}: Insert nops to force costly dependent insns into
12854 separate groups. Insert @var{number} nops to force an insn to a new group.
12857 @opindex mcall-sysv
12858 On System V.4 and embedded PowerPC systems compile code using calling
12859 conventions that adheres to the March 1995 draft of the System V
12860 Application Binary Interface, PowerPC processor supplement. This is the
12861 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
12863 @item -mcall-sysv-eabi
12864 @opindex mcall-sysv-eabi
12865 Specify both @option{-mcall-sysv} and @option{-meabi} options.
12867 @item -mcall-sysv-noeabi
12868 @opindex mcall-sysv-noeabi
12869 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
12871 @item -mcall-solaris
12872 @opindex mcall-solaris
12873 On System V.4 and embedded PowerPC systems compile code for the Solaris
12877 @opindex mcall-linux
12878 On System V.4 and embedded PowerPC systems compile code for the
12879 Linux-based GNU system.
12883 On System V.4 and embedded PowerPC systems compile code for the
12884 Hurd-based GNU system.
12886 @item -mcall-netbsd
12887 @opindex mcall-netbsd
12888 On System V.4 and embedded PowerPC systems compile code for the
12889 NetBSD operating system.
12891 @item -maix-struct-return
12892 @opindex maix-struct-return
12893 Return all structures in memory (as specified by the AIX ABI)@.
12895 @item -msvr4-struct-return
12896 @opindex msvr4-struct-return
12897 Return structures smaller than 8 bytes in registers (as specified by the
12900 @item -mabi=@var{abi-type}
12902 Extend the current ABI with a particular extension, or remove such extension.
12903 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
12904 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
12908 Extend the current ABI with SPE ABI extensions. This does not change
12909 the default ABI, instead it adds the SPE ABI extensions to the current
12913 @opindex mabi=no-spe
12914 Disable Booke SPE ABI extensions for the current ABI@.
12916 @item -mabi=ibmlongdouble
12917 @opindex mabi=ibmlongdouble
12918 Change the current ABI to use IBM extended precision long double.
12919 This is a PowerPC 32-bit SYSV ABI option.
12921 @item -mabi=ieeelongdouble
12922 @opindex mabi=ieeelongdouble
12923 Change the current ABI to use IEEE extended precision long double.
12924 This is a PowerPC 32-bit Linux ABI option.
12927 @itemx -mno-prototype
12928 @opindex mprototype
12929 @opindex mno-prototype
12930 On System V.4 and embedded PowerPC systems assume that all calls to
12931 variable argument functions are properly prototyped. Otherwise, the
12932 compiler must insert an instruction before every non prototyped call to
12933 set or clear bit 6 of the condition code register (@var{CR}) to
12934 indicate whether floating point values were passed in the floating point
12935 registers in case the function takes a variable arguments. With
12936 @option{-mprototype}, only calls to prototyped variable argument functions
12937 will set or clear the bit.
12941 On embedded PowerPC systems, assume that the startup module is called
12942 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
12943 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
12948 On embedded PowerPC systems, assume that the startup module is called
12949 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
12954 On embedded PowerPC systems, assume that the startup module is called
12955 @file{crt0.o} and the standard C libraries are @file{libads.a} and
12958 @item -myellowknife
12959 @opindex myellowknife
12960 On embedded PowerPC systems, assume that the startup module is called
12961 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
12966 On System V.4 and embedded PowerPC systems, specify that you are
12967 compiling for a VxWorks system.
12971 Specify that you are compiling for the WindISS simulation environment.
12975 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
12976 header to indicate that @samp{eabi} extended relocations are used.
12982 On System V.4 and embedded PowerPC systems do (do not) adhere to the
12983 Embedded Applications Binary Interface (eabi) which is a set of
12984 modifications to the System V.4 specifications. Selecting @option{-meabi}
12985 means that the stack is aligned to an 8 byte boundary, a function
12986 @code{__eabi} is called to from @code{main} to set up the eabi
12987 environment, and the @option{-msdata} option can use both @code{r2} and
12988 @code{r13} to point to two separate small data areas. Selecting
12989 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
12990 do not call an initialization function from @code{main}, and the
12991 @option{-msdata} option will only use @code{r13} to point to a single
12992 small data area. The @option{-meabi} option is on by default if you
12993 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
12996 @opindex msdata=eabi
12997 On System V.4 and embedded PowerPC systems, put small initialized
12998 @code{const} global and static data in the @samp{.sdata2} section, which
12999 is pointed to by register @code{r2}. Put small initialized
13000 non-@code{const} global and static data in the @samp{.sdata} section,
13001 which is pointed to by register @code{r13}. Put small uninitialized
13002 global and static data in the @samp{.sbss} section, which is adjacent to
13003 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13004 incompatible with the @option{-mrelocatable} option. The
13005 @option{-msdata=eabi} option also sets the @option{-memb} option.
13008 @opindex msdata=sysv
13009 On System V.4 and embedded PowerPC systems, put small global and static
13010 data in the @samp{.sdata} section, which is pointed to by register
13011 @code{r13}. Put small uninitialized global and static data in the
13012 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13013 The @option{-msdata=sysv} option is incompatible with the
13014 @option{-mrelocatable} option.
13016 @item -msdata=default
13018 @opindex msdata=default
13020 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13021 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13022 same as @option{-msdata=sysv}.
13025 @opindex msdata-data
13026 On System V.4 and embedded PowerPC systems, put small global
13027 data in the @samp{.sdata} section. Put small uninitialized global
13028 data in the @samp{.sbss} section. Do not use register @code{r13}
13029 to address small data however. This is the default behavior unless
13030 other @option{-msdata} options are used.
13034 @opindex msdata=none
13036 On embedded PowerPC systems, put all initialized global and static data
13037 in the @samp{.data} section, and all uninitialized data in the
13038 @samp{.bss} section.
13042 @cindex smaller data references (PowerPC)
13043 @cindex .sdata/.sdata2 references (PowerPC)
13044 On embedded PowerPC systems, put global and static items less than or
13045 equal to @var{num} bytes into the small data or bss sections instead of
13046 the normal data or bss section. By default, @var{num} is 8. The
13047 @option{-G @var{num}} switch is also passed to the linker.
13048 All modules should be compiled with the same @option{-G @var{num}} value.
13051 @itemx -mno-regnames
13053 @opindex mno-regnames
13054 On System V.4 and embedded PowerPC systems do (do not) emit register
13055 names in the assembly language output using symbolic forms.
13058 @itemx -mno-longcall
13060 @opindex mno-longcall
13061 By default assume that all calls are far away so that a longer more
13062 expensive calling sequence is required. This is required for calls
13063 further than 32 megabytes (33,554,432 bytes) from the current location.
13064 A short call will be generated if the compiler knows
13065 the call cannot be that far away. This setting can be overridden by
13066 the @code{shortcall} function attribute, or by @code{#pragma
13069 Some linkers are capable of detecting out-of-range calls and generating
13070 glue code on the fly. On these systems, long calls are unnecessary and
13071 generate slower code. As of this writing, the AIX linker can do this,
13072 as can the GNU linker for PowerPC/64. It is planned to add this feature
13073 to the GNU linker for 32-bit PowerPC systems as well.
13075 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13076 callee, L42'', plus a ``branch island'' (glue code). The two target
13077 addresses represent the callee and the ``branch island''. The
13078 Darwin/PPC linker will prefer the first address and generate a ``bl
13079 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13080 otherwise, the linker will generate ``bl L42'' to call the ``branch
13081 island''. The ``branch island'' is appended to the body of the
13082 calling function; it computes the full 32-bit address of the callee
13085 On Mach-O (Darwin) systems, this option directs the compiler emit to
13086 the glue for every direct call, and the Darwin linker decides whether
13087 to use or discard it.
13089 In the future, we may cause GCC to ignore all longcall specifications
13090 when the linker is known to generate glue.
13094 Adds support for multithreading with the @dfn{pthreads} library.
13095 This option sets flags for both the preprocessor and linker.
13099 @node S/390 and zSeries Options
13100 @subsection S/390 and zSeries Options
13101 @cindex S/390 and zSeries Options
13103 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13107 @itemx -msoft-float
13108 @opindex mhard-float
13109 @opindex msoft-float
13110 Use (do not use) the hardware floating-point instructions and registers
13111 for floating-point operations. When @option{-msoft-float} is specified,
13112 functions in @file{libgcc.a} will be used to perform floating-point
13113 operations. When @option{-mhard-float} is specified, the compiler
13114 generates IEEE floating-point instructions. This is the default.
13116 @item -mlong-double-64
13117 @itemx -mlong-double-128
13118 @opindex mlong-double-64
13119 @opindex mlong-double-128
13120 These switches control the size of @code{long double} type. A size
13121 of 64bit makes the @code{long double} type equivalent to the @code{double}
13122 type. This is the default.
13125 @itemx -mno-backchain
13126 @opindex mbackchain
13127 @opindex mno-backchain
13128 Store (do not store) the address of the caller's frame as backchain pointer
13129 into the callee's stack frame.
13130 A backchain may be needed to allow debugging using tools that do not understand
13131 DWARF-2 call frame information.
13132 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13133 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13134 the backchain is placed into the topmost word of the 96/160 byte register
13137 In general, code compiled with @option{-mbackchain} is call-compatible with
13138 code compiled with @option{-mmo-backchain}; however, use of the backchain
13139 for debugging purposes usually requires that the whole binary is built with
13140 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13141 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13142 to build a linux kernel use @option{-msoft-float}.
13144 The default is to not maintain the backchain.
13146 @item -mpacked-stack
13147 @item -mno-packed-stack
13148 @opindex mpacked-stack
13149 @opindex mno-packed-stack
13150 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13151 specified, the compiler uses the all fields of the 96/160 byte register save
13152 area only for their default purpose; unused fields still take up stack space.
13153 When @option{-mpacked-stack} is specified, register save slots are densely
13154 packed at the top of the register save area; unused space is reused for other
13155 purposes, allowing for more efficient use of the available stack space.
13156 However, when @option{-mbackchain} is also in effect, the topmost word of
13157 the save area is always used to store the backchain, and the return address
13158 register is always saved two words below the backchain.
13160 As long as the stack frame backchain is not used, code generated with
13161 @option{-mpacked-stack} is call-compatible with code generated with
13162 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13163 S/390 or zSeries generated code that uses the stack frame backchain at run
13164 time, not just for debugging purposes. Such code is not call-compatible
13165 with code compiled with @option{-mpacked-stack}. Also, note that the
13166 combination of @option{-mbackchain},
13167 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13168 to build a linux kernel use @option{-msoft-float}.
13170 The default is to not use the packed stack layout.
13173 @itemx -mno-small-exec
13174 @opindex msmall-exec
13175 @opindex mno-small-exec
13176 Generate (or do not generate) code using the @code{bras} instruction
13177 to do subroutine calls.
13178 This only works reliably if the total executable size does not
13179 exceed 64k. The default is to use the @code{basr} instruction instead,
13180 which does not have this limitation.
13186 When @option{-m31} is specified, generate code compliant to the
13187 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13188 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13189 particular to generate 64-bit instructions. For the @samp{s390}
13190 targets, the default is @option{-m31}, while the @samp{s390x}
13191 targets default to @option{-m64}.
13197 When @option{-mzarch} is specified, generate code using the
13198 instructions available on z/Architecture.
13199 When @option{-mesa} is specified, generate code using the
13200 instructions available on ESA/390. Note that @option{-mesa} is
13201 not possible with @option{-m64}.
13202 When generating code compliant to the GNU/Linux for S/390 ABI,
13203 the default is @option{-mesa}. When generating code compliant
13204 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13210 Generate (or do not generate) code using the @code{mvcle} instruction
13211 to perform block moves. When @option{-mno-mvcle} is specified,
13212 use a @code{mvc} loop instead. This is the default unless optimizing for
13219 Print (or do not print) additional debug information when compiling.
13220 The default is to not print debug information.
13222 @item -march=@var{cpu-type}
13224 Generate code that will run on @var{cpu-type}, which is the name of a system
13225 representing a certain processor type. Possible values for
13226 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13227 When generating code using the instructions available on z/Architecture,
13228 the default is @option{-march=z900}. Otherwise, the default is
13229 @option{-march=g5}.
13231 @item -mtune=@var{cpu-type}
13233 Tune to @var{cpu-type} everything applicable about the generated code,
13234 except for the ABI and the set of available instructions.
13235 The list of @var{cpu-type} values is the same as for @option{-march}.
13236 The default is the value used for @option{-march}.
13239 @itemx -mno-tpf-trace
13240 @opindex mtpf-trace
13241 @opindex mno-tpf-trace
13242 Generate code that adds (does not add) in TPF OS specific branches to trace
13243 routines in the operating system. This option is off by default, even
13244 when compiling for the TPF OS@.
13247 @itemx -mno-fused-madd
13248 @opindex mfused-madd
13249 @opindex mno-fused-madd
13250 Generate code that uses (does not use) the floating point multiply and
13251 accumulate instructions. These instructions are generated by default if
13252 hardware floating point is used.
13254 @item -mwarn-framesize=@var{framesize}
13255 @opindex mwarn-framesize
13256 Emit a warning if the current function exceeds the given frame size. Because
13257 this is a compile time check it doesn't need to be a real problem when the program
13258 runs. It is intended to identify functions which most probably cause
13259 a stack overflow. It is useful to be used in an environment with limited stack
13260 size e.g.@: the linux kernel.
13262 @item -mwarn-dynamicstack
13263 @opindex mwarn-dynamicstack
13264 Emit a warning if the function calls alloca or uses dynamically
13265 sized arrays. This is generally a bad idea with a limited stack size.
13267 @item -mstack-guard=@var{stack-guard}
13268 @item -mstack-size=@var{stack-size}
13269 @opindex mstack-guard
13270 @opindex mstack-size
13271 If these options are provided the s390 back end emits additional instructions in
13272 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13273 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13274 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13275 the frame size of the compiled function is chosen.
13276 These options are intended to be used to help debugging stack overflow problems.
13277 The additionally emitted code causes only little overhead and hence can also be
13278 used in production like systems without greater performance degradation. The given
13279 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13280 @var{stack-guard} without exceeding 64k.
13281 In order to be efficient the extra code makes the assumption that the stack starts
13282 at an address aligned to the value given by @var{stack-size}.
13283 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13286 @node Score Options
13287 @subsection Score Options
13288 @cindex Score Options
13290 These options are defined for Score implementations:
13295 Compile code for big endian mode. This is the default.
13299 Compile code for little endian mode.
13303 Disable generate bcnz instruction.
13307 Enable generate unaligned load and store instruction.
13311 Enable the use of multiply-accumulate instructions. Disabled by default.
13315 Specify the SCORE5 as the target architecture.
13319 Specify the SCORE5U of the target architecture.
13323 Specify the SCORE7 as the target architecture. This is the default.
13327 Specify the SCORE7D as the target architecture.
13331 @subsection SH Options
13333 These @samp{-m} options are defined for the SH implementations:
13338 Generate code for the SH1.
13342 Generate code for the SH2.
13345 Generate code for the SH2e.
13349 Generate code for the SH3.
13353 Generate code for the SH3e.
13357 Generate code for the SH4 without a floating-point unit.
13359 @item -m4-single-only
13360 @opindex m4-single-only
13361 Generate code for the SH4 with a floating-point unit that only
13362 supports single-precision arithmetic.
13366 Generate code for the SH4 assuming the floating-point unit is in
13367 single-precision mode by default.
13371 Generate code for the SH4.
13375 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13376 floating-point unit is not used.
13378 @item -m4a-single-only
13379 @opindex m4a-single-only
13380 Generate code for the SH4a, in such a way that no double-precision
13381 floating point operations are used.
13384 @opindex m4a-single
13385 Generate code for the SH4a assuming the floating-point unit is in
13386 single-precision mode by default.
13390 Generate code for the SH4a.
13394 Same as @option{-m4a-nofpu}, except that it implicitly passes
13395 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13396 instructions at the moment.
13400 Compile code for the processor in big endian mode.
13404 Compile code for the processor in little endian mode.
13408 Align doubles at 64-bit boundaries. Note that this changes the calling
13409 conventions, and thus some functions from the standard C library will
13410 not work unless you recompile it first with @option{-mdalign}.
13414 Shorten some address references at link time, when possible; uses the
13415 linker option @option{-relax}.
13419 Use 32-bit offsets in @code{switch} tables. The default is to use
13424 Enable the use of the instruction @code{fmovd}.
13428 Comply with the calling conventions defined by Renesas.
13432 Comply with the calling conventions defined by Renesas.
13436 Comply with the calling conventions defined for GCC before the Renesas
13437 conventions were available. This option is the default for all
13438 targets of the SH toolchain except for @samp{sh-symbianelf}.
13441 @opindex mnomacsave
13442 Mark the @code{MAC} register as call-clobbered, even if
13443 @option{-mhitachi} is given.
13447 Increase IEEE-compliance of floating-point code.
13448 At the moment, this is equivalent to @option{-fno-finite-math-only}.
13449 When generating 16 bit SH opcodes, getting IEEE-conforming results for
13450 comparisons of NANs / infinities incurs extra overhead in every
13451 floating point comparison, therefore the default is set to
13452 @option{-ffinite-math-only}.
13454 @item -minline-ic_invalidate
13455 @opindex minline-ic_invalidate
13456 Inline code to invalidate instruction cache entries after setting up
13457 nested function trampolines.
13458 This option has no effect if -musermode is in effect and the selected
13459 code generation option (e.g. -m4) does not allow the use of the icbi
13461 If the selected code generation option does not allow the use of the icbi
13462 instruction, and -musermode is not in effect, the inlined code will
13463 manipulate the instruction cache address array directly with an associative
13464 write. This not only requires privileged mode, but it will also
13465 fail if the cache line had been mapped via the TLB and has become unmapped.
13469 Dump instruction size and location in the assembly code.
13472 @opindex mpadstruct
13473 This option is deprecated. It pads structures to multiple of 4 bytes,
13474 which is incompatible with the SH ABI@.
13478 Optimize for space instead of speed. Implied by @option{-Os}.
13481 @opindex mprefergot
13482 When generating position-independent code, emit function calls using
13483 the Global Offset Table instead of the Procedure Linkage Table.
13487 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
13488 if the inlined code would not work in user mode.
13489 This is the default when the target is @code{sh-*-linux*}.
13491 @item -multcost=@var{number}
13492 @opindex multcost=@var{number}
13493 Set the cost to assume for a multiply insn.
13495 @item -mdiv=@var{strategy}
13496 @opindex mdiv=@var{strategy}
13497 Set the division strategy to use for SHmedia code. @var{strategy} must be
13498 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
13499 inv:call2, inv:fp .
13500 "fp" performs the operation in floating point. This has a very high latency,
13501 but needs only a few instructions, so it might be a good choice if
13502 your code has enough easily exploitable ILP to allow the compiler to
13503 schedule the floating point instructions together with other instructions.
13504 Division by zero causes a floating point exception.
13505 "inv" uses integer operations to calculate the inverse of the divisor,
13506 and then multiplies the dividend with the inverse. This strategy allows
13507 cse and hoisting of the inverse calculation. Division by zero calculates
13508 an unspecified result, but does not trap.
13509 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
13510 have been found, or if the entire operation has been hoisted to the same
13511 place, the last stages of the inverse calculation are intertwined with the
13512 final multiply to reduce the overall latency, at the expense of using a few
13513 more instructions, and thus offering fewer scheduling opportunities with
13515 "call" calls a library function that usually implements the inv:minlat
13517 This gives high code density for m5-*media-nofpu compilations.
13518 "call2" uses a different entry point of the same library function, where it
13519 assumes that a pointer to a lookup table has already been set up, which
13520 exposes the pointer load to cse / code hoisting optimizations.
13521 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
13522 code generation, but if the code stays unoptimized, revert to the "call",
13523 "call2", or "fp" strategies, respectively. Note that the
13524 potentially-trapping side effect of division by zero is carried by a
13525 separate instruction, so it is possible that all the integer instructions
13526 are hoisted out, but the marker for the side effect stays where it is.
13527 A recombination to fp operations or a call is not possible in that case.
13528 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
13529 that the inverse calculation was nor separated from the multiply, they speed
13530 up division where the dividend fits into 20 bits (plus sign where applicable),
13531 by inserting a test to skip a number of operations in this case; this test
13532 slows down the case of larger dividends. inv20u assumes the case of a such
13533 a small dividend to be unlikely, and inv20l assumes it to be likely.
13535 @item -mdivsi3_libfunc=@var{name}
13536 @opindex mdivsi3_libfunc=@var{name}
13537 Set the name of the library function used for 32 bit signed division to
13538 @var{name}. This only affect the name used in the call and inv:call
13539 division strategies, and the compiler will still expect the same
13540 sets of input/output/clobbered registers as if this option was not present.
13542 @item -madjust-unroll
13543 @opindex madjust-unroll
13544 Throttle unrolling to avoid thrashing target registers.
13545 This option only has an effect if the gcc code base supports the
13546 TARGET_ADJUST_UNROLL_MAX target hook.
13548 @item -mindexed-addressing
13549 @opindex mindexed-addressing
13550 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
13551 This is only safe if the hardware and/or OS implement 32 bit wrap-around
13552 semantics for the indexed addressing mode. The architecture allows the
13553 implementation of processors with 64 bit MMU, which the OS could use to
13554 get 32 bit addressing, but since no current hardware implementation supports
13555 this or any other way to make the indexed addressing mode safe to use in
13556 the 32 bit ABI, the default is -mno-indexed-addressing.
13558 @item -mgettrcost=@var{number}
13559 @opindex mgettrcost=@var{number}
13560 Set the cost assumed for the gettr instruction to @var{number}.
13561 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
13565 Assume pt* instructions won't trap. This will generally generate better
13566 scheduled code, but is unsafe on current hardware. The current architecture
13567 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
13568 This has the unintentional effect of making it unsafe to schedule ptabs /
13569 ptrel before a branch, or hoist it out of a loop. For example,
13570 __do_global_ctors, a part of libgcc that runs constructors at program
13571 startup, calls functions in a list which is delimited by @minus{}1. With the
13572 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
13573 That means that all the constructors will be run a bit quicker, but when
13574 the loop comes to the end of the list, the program crashes because ptabs
13575 loads @minus{}1 into a target register. Since this option is unsafe for any
13576 hardware implementing the current architecture specification, the default
13577 is -mno-pt-fixed. Unless the user specifies a specific cost with
13578 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
13579 this deters register allocation using target registers for storing
13582 @item -minvalid-symbols
13583 @opindex minvalid-symbols
13584 Assume symbols might be invalid. Ordinary function symbols generated by
13585 the compiler will always be valid to load with movi/shori/ptabs or
13586 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
13587 to generate symbols that will cause ptabs / ptrel to trap.
13588 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
13589 It will then prevent cross-basic-block cse, hoisting and most scheduling
13590 of symbol loads. The default is @option{-mno-invalid-symbols}.
13593 @node SPARC Options
13594 @subsection SPARC Options
13595 @cindex SPARC options
13597 These @samp{-m} options are supported on the SPARC:
13600 @item -mno-app-regs
13602 @opindex mno-app-regs
13604 Specify @option{-mapp-regs} to generate output using the global registers
13605 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
13608 To be fully SVR4 ABI compliant at the cost of some performance loss,
13609 specify @option{-mno-app-regs}. You should compile libraries and system
13610 software with this option.
13613 @itemx -mhard-float
13615 @opindex mhard-float
13616 Generate output containing floating point instructions. This is the
13620 @itemx -msoft-float
13622 @opindex msoft-float
13623 Generate output containing library calls for floating point.
13624 @strong{Warning:} the requisite libraries are not available for all SPARC
13625 targets. Normally the facilities of the machine's usual C compiler are
13626 used, but this cannot be done directly in cross-compilation. You must make
13627 your own arrangements to provide suitable library functions for
13628 cross-compilation. The embedded targets @samp{sparc-*-aout} and
13629 @samp{sparclite-*-*} do provide software floating point support.
13631 @option{-msoft-float} changes the calling convention in the output file;
13632 therefore, it is only useful if you compile @emph{all} of a program with
13633 this option. In particular, you need to compile @file{libgcc.a}, the
13634 library that comes with GCC, with @option{-msoft-float} in order for
13637 @item -mhard-quad-float
13638 @opindex mhard-quad-float
13639 Generate output containing quad-word (long double) floating point
13642 @item -msoft-quad-float
13643 @opindex msoft-quad-float
13644 Generate output containing library calls for quad-word (long double)
13645 floating point instructions. The functions called are those specified
13646 in the SPARC ABI@. This is the default.
13648 As of this writing, there are no SPARC implementations that have hardware
13649 support for the quad-word floating point instructions. They all invoke
13650 a trap handler for one of these instructions, and then the trap handler
13651 emulates the effect of the instruction. Because of the trap handler overhead,
13652 this is much slower than calling the ABI library routines. Thus the
13653 @option{-msoft-quad-float} option is the default.
13655 @item -mno-unaligned-doubles
13656 @itemx -munaligned-doubles
13657 @opindex mno-unaligned-doubles
13658 @opindex munaligned-doubles
13659 Assume that doubles have 8 byte alignment. This is the default.
13661 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
13662 alignment only if they are contained in another type, or if they have an
13663 absolute address. Otherwise, it assumes they have 4 byte alignment.
13664 Specifying this option avoids some rare compatibility problems with code
13665 generated by other compilers. It is not the default because it results
13666 in a performance loss, especially for floating point code.
13668 @item -mno-faster-structs
13669 @itemx -mfaster-structs
13670 @opindex mno-faster-structs
13671 @opindex mfaster-structs
13672 With @option{-mfaster-structs}, the compiler assumes that structures
13673 should have 8 byte alignment. This enables the use of pairs of
13674 @code{ldd} and @code{std} instructions for copies in structure
13675 assignment, in place of twice as many @code{ld} and @code{st} pairs.
13676 However, the use of this changed alignment directly violates the SPARC
13677 ABI@. Thus, it's intended only for use on targets where the developer
13678 acknowledges that their resulting code will not be directly in line with
13679 the rules of the ABI@.
13681 @item -mimpure-text
13682 @opindex mimpure-text
13683 @option{-mimpure-text}, used in addition to @option{-shared}, tells
13684 the compiler to not pass @option{-z text} to the linker when linking a
13685 shared object. Using this option, you can link position-dependent
13686 code into a shared object.
13688 @option{-mimpure-text} suppresses the ``relocations remain against
13689 allocatable but non-writable sections'' linker error message.
13690 However, the necessary relocations will trigger copy-on-write, and the
13691 shared object is not actually shared across processes. Instead of
13692 using @option{-mimpure-text}, you should compile all source code with
13693 @option{-fpic} or @option{-fPIC}.
13695 This option is only available on SunOS and Solaris.
13697 @item -mcpu=@var{cpu_type}
13699 Set the instruction set, register set, and instruction scheduling parameters
13700 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
13701 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
13702 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
13703 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
13704 @samp{ultrasparc3}, and @samp{niagara}.
13706 Default instruction scheduling parameters are used for values that select
13707 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
13708 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
13710 Here is a list of each supported architecture and their supported
13715 v8: supersparc, hypersparc
13716 sparclite: f930, f934, sparclite86x
13718 v9: ultrasparc, ultrasparc3, niagara
13721 By default (unless configured otherwise), GCC generates code for the V7
13722 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
13723 additionally optimizes it for the Cypress CY7C602 chip, as used in the
13724 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
13725 SPARCStation 1, 2, IPX etc.
13727 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
13728 architecture. The only difference from V7 code is that the compiler emits
13729 the integer multiply and integer divide instructions which exist in SPARC-V8
13730 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
13731 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
13734 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
13735 the SPARC architecture. This adds the integer multiply, integer divide step
13736 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
13737 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
13738 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
13739 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
13740 MB86934 chip, which is the more recent SPARClite with FPU@.
13742 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
13743 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
13744 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
13745 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
13746 optimizes it for the TEMIC SPARClet chip.
13748 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
13749 architecture. This adds 64-bit integer and floating-point move instructions,
13750 3 additional floating-point condition code registers and conditional move
13751 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
13752 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
13753 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
13754 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
13755 @option{-mcpu=niagara}, the compiler additionally optimizes it for
13756 Sun UltraSPARC T1 chips.
13758 @item -mtune=@var{cpu_type}
13760 Set the instruction scheduling parameters for machine type
13761 @var{cpu_type}, but do not set the instruction set or register set that the
13762 option @option{-mcpu=@var{cpu_type}} would.
13764 The same values for @option{-mcpu=@var{cpu_type}} can be used for
13765 @option{-mtune=@var{cpu_type}}, but the only useful values are those
13766 that select a particular cpu implementation. Those are @samp{cypress},
13767 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
13768 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
13769 @samp{ultrasparc3}, and @samp{niagara}.
13774 @opindex mno-v8plus
13775 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
13776 difference from the V8 ABI is that the global and out registers are
13777 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
13778 mode for all SPARC-V9 processors.
13784 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
13785 Visual Instruction Set extensions. The default is @option{-mno-vis}.
13788 These @samp{-m} options are supported in addition to the above
13789 on SPARC-V9 processors in 64-bit environments:
13792 @item -mlittle-endian
13793 @opindex mlittle-endian
13794 Generate code for a processor running in little-endian mode. It is only
13795 available for a few configurations and most notably not on Solaris and Linux.
13801 Generate code for a 32-bit or 64-bit environment.
13802 The 32-bit environment sets int, long and pointer to 32 bits.
13803 The 64-bit environment sets int to 32 bits and long and pointer
13806 @item -mcmodel=medlow
13807 @opindex mcmodel=medlow
13808 Generate code for the Medium/Low code model: 64-bit addresses, programs
13809 must be linked in the low 32 bits of memory. Programs can be statically
13810 or dynamically linked.
13812 @item -mcmodel=medmid
13813 @opindex mcmodel=medmid
13814 Generate code for the Medium/Middle code model: 64-bit addresses, programs
13815 must be linked in the low 44 bits of memory, the text and data segments must
13816 be less than 2GB in size and the data segment must be located within 2GB of
13819 @item -mcmodel=medany
13820 @opindex mcmodel=medany
13821 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
13822 may be linked anywhere in memory, the text and data segments must be less
13823 than 2GB in size and the data segment must be located within 2GB of the
13826 @item -mcmodel=embmedany
13827 @opindex mcmodel=embmedany
13828 Generate code for the Medium/Anywhere code model for embedded systems:
13829 64-bit addresses, the text and data segments must be less than 2GB in
13830 size, both starting anywhere in memory (determined at link time). The
13831 global register %g4 points to the base of the data segment. Programs
13832 are statically linked and PIC is not supported.
13835 @itemx -mno-stack-bias
13836 @opindex mstack-bias
13837 @opindex mno-stack-bias
13838 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
13839 frame pointer if present, are offset by @minus{}2047 which must be added back
13840 when making stack frame references. This is the default in 64-bit mode.
13841 Otherwise, assume no such offset is present.
13844 These switches are supported in addition to the above on Solaris:
13849 Add support for multithreading using the Solaris threads library. This
13850 option sets flags for both the preprocessor and linker. This option does
13851 not affect the thread safety of object code produced by the compiler or
13852 that of libraries supplied with it.
13856 Add support for multithreading using the POSIX threads library. This
13857 option sets flags for both the preprocessor and linker. This option does
13858 not affect the thread safety of object code produced by the compiler or
13859 that of libraries supplied with it.
13863 This is a synonym for @option{-pthreads}.
13867 @subsection SPU Options
13868 @cindex SPU options
13870 These @samp{-m} options are supported on the SPU:
13874 @itemx -merror-reloc
13875 @opindex mwarn-reloc
13876 @opindex merror-reloc
13878 The loader for SPU does not handle dynamic relocations. By default, GCC
13879 will give an error when it generates code that requires a dynamic
13880 relocation. @option{-mno-error-reloc} disables the error,
13881 @option{-mwarn-reloc} will generate a warning instead.
13884 @itemx -munsafe-dma
13886 @opindex munsafe-dma
13888 Instructions which initiate or test completion of DMA must not be
13889 reordered with respect to loads and stores of the memory which is being
13890 accessed. Users typically address this problem using the volatile
13891 keyword, but that can lead to inefficient code in places where the
13892 memory is known to not change. Rather than mark the memory as volatile
13893 we treat the DMA instructions as potentially effecting all memory. With
13894 @option{-munsafe-dma} users must use the volatile keyword to protect
13897 @item -mbranch-hints
13898 @opindex mbranch-hints
13900 By default, GCC will generate a branch hint instruction to avoid
13901 pipeline stalls for always taken or probably taken branches. A hint
13902 will not be generated closer than 8 instructions away from its branch.
13903 There is little reason to disable them, except for debugging purposes,
13904 or to make an object a little bit smaller.
13908 @opindex msmall-mem
13909 @opindex mlarge-mem
13911 By default, GCC generates code assuming that addresses are never larger
13912 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
13913 a full 32 bit address.
13918 By default, GCC links against startup code that assumes the SPU-style
13919 main function interface (which has an unconventional parameter list).
13920 With @option{-mstdmain}, GCC will link your program against startup
13921 code that assumes a C99-style interface to @code{main}, including a
13922 local copy of @code{argv} strings.
13924 @item -mfixed-range=@var{register-range}
13925 @opindex mfixed-range
13926 Generate code treating the given register range as fixed registers.
13927 A fixed register is one that the register allocator can not use. This is
13928 useful when compiling kernel code. A register range is specified as
13929 two registers separated by a dash. Multiple register ranges can be
13930 specified separated by a comma.
13934 @node System V Options
13935 @subsection Options for System V
13937 These additional options are available on System V Release 4 for
13938 compatibility with other compilers on those systems:
13943 Create a shared object.
13944 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
13948 Identify the versions of each tool used by the compiler, in a
13949 @code{.ident} assembler directive in the output.
13953 Refrain from adding @code{.ident} directives to the output file (this is
13956 @item -YP,@var{dirs}
13958 Search the directories @var{dirs}, and no others, for libraries
13959 specified with @option{-l}.
13961 @item -Ym,@var{dir}
13963 Look in the directory @var{dir} to find the M4 preprocessor.
13964 The assembler uses this option.
13965 @c This is supposed to go with a -Yd for predefined M4 macro files, but
13966 @c the generic assembler that comes with Solaris takes just -Ym.
13969 @node TMS320C3x/C4x Options
13970 @subsection TMS320C3x/C4x Options
13971 @cindex TMS320C3x/C4x Options
13973 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
13977 @item -mcpu=@var{cpu_type}
13979 Set the instruction set, register set, and instruction scheduling
13980 parameters for machine type @var{cpu_type}. Supported values for
13981 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
13982 @samp{c44}. The default is @samp{c40} to generate code for the
13987 @itemx -msmall-memory
13989 @opindex mbig-memory
13991 @opindex msmall-memory
13993 Generates code for the big or small memory model. The small memory
13994 model assumed that all data fits into one 64K word page. At run-time
13995 the data page (DP) register must be set to point to the 64K page
13996 containing the .bss and .data program sections. The big memory model is
13997 the default and requires reloading of the DP register for every direct
14004 Allow (disallow) allocation of general integer operands into the block
14005 count register BK@.
14011 Enable (disable) generation of code using decrement and branch,
14012 DBcond(D), instructions. This is enabled by default for the C4x. To be
14013 on the safe side, this is disabled for the C3x, since the maximum
14014 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
14015 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
14016 that it can utilize the decrement and branch instruction, but will give
14017 up if there is more than one memory reference in the loop. Thus a loop
14018 where the loop counter is decremented can generate slightly more
14019 efficient code, in cases where the RPTB instruction cannot be utilized.
14021 @item -mdp-isr-reload
14023 @opindex mdp-isr-reload
14025 Force the DP register to be saved on entry to an interrupt service
14026 routine (ISR), reloaded to point to the data section, and restored on
14027 exit from the ISR@. This should not be required unless someone has
14028 violated the small memory model by modifying the DP register, say within
14035 For the C3x use the 24-bit MPYI instruction for integer multiplies
14036 instead of a library call to guarantee 32-bit results. Note that if one
14037 of the operands is a constant, then the multiplication will be performed
14038 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
14039 then squaring operations are performed inline instead of a library call.
14042 @itemx -mno-fast-fix
14044 @opindex mno-fast-fix
14045 The C3x/C4x FIX instruction to convert a floating point value to an
14046 integer value chooses the nearest integer less than or equal to the
14047 floating point value rather than to the nearest integer. Thus if the
14048 floating point number is negative, the result will be incorrectly
14049 truncated an additional code is necessary to detect and correct this
14050 case. This option can be used to disable generation of the additional
14051 code required to correct the result.
14057 Enable (disable) generation of repeat block sequences using the RPTB
14058 instruction for zero overhead looping. The RPTB construct is only used
14059 for innermost loops that do not call functions or jump across the loop
14060 boundaries. There is no advantage having nested RPTB loops due to the
14061 overhead required to save and restore the RC, RS, and RE registers.
14062 This is enabled by default with @option{-O2}.
14064 @item -mrpts=@var{count}
14068 Enable (disable) the use of the single instruction repeat instruction
14069 RPTS@. If a repeat block contains a single instruction, and the loop
14070 count can be guaranteed to be less than the value @var{count}, GCC will
14071 emit a RPTS instruction instead of a RPTB@. If no value is specified,
14072 then a RPTS will be emitted even if the loop count cannot be determined
14073 at compile time. Note that the repeated instruction following RPTS does
14074 not have to be reloaded from memory each iteration, thus freeing up the
14075 CPU buses for operands. However, since interrupts are blocked by this
14076 instruction, it is disabled by default.
14078 @item -mloop-unsigned
14079 @itemx -mno-loop-unsigned
14080 @opindex mloop-unsigned
14081 @opindex mno-loop-unsigned
14082 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
14083 is @math{2^{31} + 1} since these instructions test if the iteration count is
14084 negative to terminate the loop. If the iteration count is unsigned
14085 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
14086 exceeded. This switch allows an unsigned iteration count.
14090 Try to emit an assembler syntax that the TI assembler (asm30) is happy
14091 with. This also enforces compatibility with the API employed by the TI
14092 C3x C compiler. For example, long doubles are passed as structures
14093 rather than in floating point registers.
14099 Generate code that uses registers (stack) for passing arguments to functions.
14100 By default, arguments are passed in registers where possible rather
14101 than by pushing arguments on to the stack.
14103 @item -mparallel-insns
14104 @itemx -mno-parallel-insns
14105 @opindex mparallel-insns
14106 @opindex mno-parallel-insns
14107 Allow the generation of parallel instructions. This is enabled by
14108 default with @option{-O2}.
14110 @item -mparallel-mpy
14111 @itemx -mno-parallel-mpy
14112 @opindex mparallel-mpy
14113 @opindex mno-parallel-mpy
14114 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
14115 provided @option{-mparallel-insns} is also specified. These instructions have
14116 tight register constraints which can pessimize the code generation
14117 of large functions.
14122 @subsection V850 Options
14123 @cindex V850 Options
14125 These @samp{-m} options are defined for V850 implementations:
14129 @itemx -mno-long-calls
14130 @opindex mlong-calls
14131 @opindex mno-long-calls
14132 Treat all calls as being far away (near). If calls are assumed to be
14133 far away, the compiler will always load the functions address up into a
14134 register, and call indirect through the pointer.
14140 Do not optimize (do optimize) basic blocks that use the same index
14141 pointer 4 or more times to copy pointer into the @code{ep} register, and
14142 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14143 option is on by default if you optimize.
14145 @item -mno-prolog-function
14146 @itemx -mprolog-function
14147 @opindex mno-prolog-function
14148 @opindex mprolog-function
14149 Do not use (do use) external functions to save and restore registers
14150 at the prologue and epilogue of a function. The external functions
14151 are slower, but use less code space if more than one function saves
14152 the same number of registers. The @option{-mprolog-function} option
14153 is on by default if you optimize.
14157 Try to make the code as small as possible. At present, this just turns
14158 on the @option{-mep} and @option{-mprolog-function} options.
14160 @item -mtda=@var{n}
14162 Put static or global variables whose size is @var{n} bytes or less into
14163 the tiny data area that register @code{ep} points to. The tiny data
14164 area can hold up to 256 bytes in total (128 bytes for byte references).
14166 @item -msda=@var{n}
14168 Put static or global variables whose size is @var{n} bytes or less into
14169 the small data area that register @code{gp} points to. The small data
14170 area can hold up to 64 kilobytes.
14172 @item -mzda=@var{n}
14174 Put static or global variables whose size is @var{n} bytes or less into
14175 the first 32 kilobytes of memory.
14179 Specify that the target processor is the V850.
14182 @opindex mbig-switch
14183 Generate code suitable for big switch tables. Use this option only if
14184 the assembler/linker complain about out of range branches within a switch
14189 This option will cause r2 and r5 to be used in the code generated by
14190 the compiler. This setting is the default.
14192 @item -mno-app-regs
14193 @opindex mno-app-regs
14194 This option will cause r2 and r5 to be treated as fixed registers.
14198 Specify that the target processor is the V850E1. The preprocessor
14199 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14200 this option is used.
14204 Specify that the target processor is the V850E@. The preprocessor
14205 constant @samp{__v850e__} will be defined if this option is used.
14207 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14208 are defined then a default target processor will be chosen and the
14209 relevant @samp{__v850*__} preprocessor constant will be defined.
14211 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14212 defined, regardless of which processor variant is the target.
14214 @item -mdisable-callt
14215 @opindex mdisable-callt
14216 This option will suppress generation of the CALLT instruction for the
14217 v850e and v850e1 flavors of the v850 architecture. The default is
14218 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14223 @subsection VAX Options
14224 @cindex VAX options
14226 These @samp{-m} options are defined for the VAX:
14231 Do not output certain jump instructions (@code{aobleq} and so on)
14232 that the Unix assembler for the VAX cannot handle across long
14237 Do output those jump instructions, on the assumption that you
14238 will assemble with the GNU assembler.
14242 Output code for g-format floating point numbers instead of d-format.
14245 @node VxWorks Options
14246 @subsection VxWorks Options
14247 @cindex VxWorks Options
14249 The options in this section are defined for all VxWorks targets.
14250 Options specific to the target hardware are listed with the other
14251 options for that target.
14256 GCC can generate code for both VxWorks kernels and real time processes
14257 (RTPs). This option switches from the former to the latter. It also
14258 defines the preprocessor macro @code{__RTP__}.
14261 @opindex non-static
14262 Link an RTP executable against shared libraries rather than static
14263 libraries. The options @option{-static} and @option{-shared} can
14264 also be used for RTPs (@pxref{Link Options}); @option{-static}
14271 These options are passed down to the linker. They are defined for
14272 compatibility with Diab.
14275 @opindex Xbind-lazy
14276 Enable lazy binding of function calls. This option is equivalent to
14277 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14281 Disable lazy binding of function calls. This option is the default and
14282 is defined for compatibility with Diab.
14285 @node x86-64 Options
14286 @subsection x86-64 Options
14287 @cindex x86-64 options
14289 These are listed under @xref{i386 and x86-64 Options}.
14291 @node Xstormy16 Options
14292 @subsection Xstormy16 Options
14293 @cindex Xstormy16 Options
14295 These options are defined for Xstormy16:
14300 Choose startup files and linker script suitable for the simulator.
14303 @node Xtensa Options
14304 @subsection Xtensa Options
14305 @cindex Xtensa Options
14307 These options are supported for Xtensa targets:
14311 @itemx -mno-const16
14313 @opindex mno-const16
14314 Enable or disable use of @code{CONST16} instructions for loading
14315 constant values. The @code{CONST16} instruction is currently not a
14316 standard option from Tensilica. When enabled, @code{CONST16}
14317 instructions are always used in place of the standard @code{L32R}
14318 instructions. The use of @code{CONST16} is enabled by default only if
14319 the @code{L32R} instruction is not available.
14322 @itemx -mno-fused-madd
14323 @opindex mfused-madd
14324 @opindex mno-fused-madd
14325 Enable or disable use of fused multiply/add and multiply/subtract
14326 instructions in the floating-point option. This has no effect if the
14327 floating-point option is not also enabled. Disabling fused multiply/add
14328 and multiply/subtract instructions forces the compiler to use separate
14329 instructions for the multiply and add/subtract operations. This may be
14330 desirable in some cases where strict IEEE 754-compliant results are
14331 required: the fused multiply add/subtract instructions do not round the
14332 intermediate result, thereby producing results with @emph{more} bits of
14333 precision than specified by the IEEE standard. Disabling fused multiply
14334 add/subtract instructions also ensures that the program output is not
14335 sensitive to the compiler's ability to combine multiply and add/subtract
14338 @item -mtext-section-literals
14339 @itemx -mno-text-section-literals
14340 @opindex mtext-section-literals
14341 @opindex mno-text-section-literals
14342 Control the treatment of literal pools. The default is
14343 @option{-mno-text-section-literals}, which places literals in a separate
14344 section in the output file. This allows the literal pool to be placed
14345 in a data RAM/ROM, and it also allows the linker to combine literal
14346 pools from separate object files to remove redundant literals and
14347 improve code size. With @option{-mtext-section-literals}, the literals
14348 are interspersed in the text section in order to keep them as close as
14349 possible to their references. This may be necessary for large assembly
14352 @item -mtarget-align
14353 @itemx -mno-target-align
14354 @opindex mtarget-align
14355 @opindex mno-target-align
14356 When this option is enabled, GCC instructs the assembler to
14357 automatically align instructions to reduce branch penalties at the
14358 expense of some code density. The assembler attempts to widen density
14359 instructions to align branch targets and the instructions following call
14360 instructions. If there are not enough preceding safe density
14361 instructions to align a target, no widening will be performed. The
14362 default is @option{-mtarget-align}. These options do not affect the
14363 treatment of auto-aligned instructions like @code{LOOP}, which the
14364 assembler will always align, either by widening density instructions or
14365 by inserting no-op instructions.
14368 @itemx -mno-longcalls
14369 @opindex mlongcalls
14370 @opindex mno-longcalls
14371 When this option is enabled, GCC instructs the assembler to translate
14372 direct calls to indirect calls unless it can determine that the target
14373 of a direct call is in the range allowed by the call instruction. This
14374 translation typically occurs for calls to functions in other source
14375 files. Specifically, the assembler translates a direct @code{CALL}
14376 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14377 The default is @option{-mno-longcalls}. This option should be used in
14378 programs where the call target can potentially be out of range. This
14379 option is implemented in the assembler, not the compiler, so the
14380 assembly code generated by GCC will still show direct call
14381 instructions---look at the disassembled object code to see the actual
14382 instructions. Note that the assembler will use an indirect call for
14383 every cross-file call, not just those that really will be out of range.
14386 @node zSeries Options
14387 @subsection zSeries Options
14388 @cindex zSeries options
14390 These are listed under @xref{S/390 and zSeries Options}.
14392 @node Code Gen Options
14393 @section Options for Code Generation Conventions
14394 @cindex code generation conventions
14395 @cindex options, code generation
14396 @cindex run-time options
14398 These machine-independent options control the interface conventions
14399 used in code generation.
14401 Most of them have both positive and negative forms; the negative form
14402 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14403 one of the forms is listed---the one which is not the default. You
14404 can figure out the other form by either removing @samp{no-} or adding
14408 @item -fbounds-check
14409 @opindex fbounds-check
14410 For front-ends that support it, generate additional code to check that
14411 indices used to access arrays are within the declared range. This is
14412 currently only supported by the Java and Fortran front-ends, where
14413 this option defaults to true and false respectively.
14417 This option generates traps for signed overflow on addition, subtraction,
14418 multiplication operations.
14422 This option instructs the compiler to assume that signed arithmetic
14423 overflow of addition, subtraction and multiplication wraps around
14424 using twos-complement representation. This flag enables some optimizations
14425 and disables others. This option is enabled by default for the Java
14426 front-end, as required by the Java language specification.
14429 @opindex fexceptions
14430 Enable exception handling. Generates extra code needed to propagate
14431 exceptions. For some targets, this implies GCC will generate frame
14432 unwind information for all functions, which can produce significant data
14433 size overhead, although it does not affect execution. If you do not
14434 specify this option, GCC will enable it by default for languages like
14435 C++ which normally require exception handling, and disable it for
14436 languages like C that do not normally require it. However, you may need
14437 to enable this option when compiling C code that needs to interoperate
14438 properly with exception handlers written in C++. You may also wish to
14439 disable this option if you are compiling older C++ programs that don't
14440 use exception handling.
14442 @item -fnon-call-exceptions
14443 @opindex fnon-call-exceptions
14444 Generate code that allows trapping instructions to throw exceptions.
14445 Note that this requires platform-specific runtime support that does
14446 not exist everywhere. Moreover, it only allows @emph{trapping}
14447 instructions to throw exceptions, i.e.@: memory references or floating
14448 point instructions. It does not allow exceptions to be thrown from
14449 arbitrary signal handlers such as @code{SIGALRM}.
14451 @item -funwind-tables
14452 @opindex funwind-tables
14453 Similar to @option{-fexceptions}, except that it will just generate any needed
14454 static data, but will not affect the generated code in any other way.
14455 You will normally not enable this option; instead, a language processor
14456 that needs this handling would enable it on your behalf.
14458 @item -fasynchronous-unwind-tables
14459 @opindex fasynchronous-unwind-tables
14460 Generate unwind table in dwarf2 format, if supported by target machine. The
14461 table is exact at each instruction boundary, so it can be used for stack
14462 unwinding from asynchronous events (such as debugger or garbage collector).
14464 @item -fpcc-struct-return
14465 @opindex fpcc-struct-return
14466 Return ``short'' @code{struct} and @code{union} values in memory like
14467 longer ones, rather than in registers. This convention is less
14468 efficient, but it has the advantage of allowing intercallability between
14469 GCC-compiled files and files compiled with other compilers, particularly
14470 the Portable C Compiler (pcc).
14472 The precise convention for returning structures in memory depends
14473 on the target configuration macros.
14475 Short structures and unions are those whose size and alignment match
14476 that of some integer type.
14478 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14479 switch is not binary compatible with code compiled with the
14480 @option{-freg-struct-return} switch.
14481 Use it to conform to a non-default application binary interface.
14483 @item -freg-struct-return
14484 @opindex freg-struct-return
14485 Return @code{struct} and @code{union} values in registers when possible.
14486 This is more efficient for small structures than
14487 @option{-fpcc-struct-return}.
14489 If you specify neither @option{-fpcc-struct-return} nor
14490 @option{-freg-struct-return}, GCC defaults to whichever convention is
14491 standard for the target. If there is no standard convention, GCC
14492 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14493 the principal compiler. In those cases, we can choose the standard, and
14494 we chose the more efficient register return alternative.
14496 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14497 switch is not binary compatible with code compiled with the
14498 @option{-fpcc-struct-return} switch.
14499 Use it to conform to a non-default application binary interface.
14501 @item -fshort-enums
14502 @opindex fshort-enums
14503 Allocate to an @code{enum} type only as many bytes as it needs for the
14504 declared range of possible values. Specifically, the @code{enum} type
14505 will be equivalent to the smallest integer type which has enough room.
14507 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14508 code that is not binary compatible with code generated without that switch.
14509 Use it to conform to a non-default application binary interface.
14511 @item -fshort-double
14512 @opindex fshort-double
14513 Use the same size for @code{double} as for @code{float}.
14515 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14516 code that is not binary compatible with code generated without that switch.
14517 Use it to conform to a non-default application binary interface.
14519 @item -fshort-wchar
14520 @opindex fshort-wchar
14521 Override the underlying type for @samp{wchar_t} to be @samp{short
14522 unsigned int} instead of the default for the target. This option is
14523 useful for building programs to run under WINE@.
14525 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14526 code that is not binary compatible with code generated without that switch.
14527 Use it to conform to a non-default application binary interface.
14530 @opindex fno-common
14531 In C, allocate even uninitialized global variables in the data section of the
14532 object file, rather than generating them as common blocks. This has the
14533 effect that if the same variable is declared (without @code{extern}) in
14534 two different compilations, you will get an error when you link them.
14535 The only reason this might be useful is if you wish to verify that the
14536 program will work on other systems which always work this way.
14540 Ignore the @samp{#ident} directive.
14542 @item -finhibit-size-directive
14543 @opindex finhibit-size-directive
14544 Don't output a @code{.size} assembler directive, or anything else that
14545 would cause trouble if the function is split in the middle, and the
14546 two halves are placed at locations far apart in memory. This option is
14547 used when compiling @file{crtstuff.c}; you should not need to use it
14550 @item -fverbose-asm
14551 @opindex fverbose-asm
14552 Put extra commentary information in the generated assembly code to
14553 make it more readable. This option is generally only of use to those
14554 who actually need to read the generated assembly code (perhaps while
14555 debugging the compiler itself).
14557 @option{-fno-verbose-asm}, the default, causes the
14558 extra information to be omitted and is useful when comparing two assembler
14561 @item -frecord-gcc-switches
14562 @opindex frecord-gcc-switches
14563 This switch causes the command line that was used to invoke the
14564 compiler to be recorded into the object file that is being created.
14565 This switch is only implemented on some targets and the exact format
14566 of the recording is target and binary file format dependent, but it
14567 usually takes the form of a section containing ASCII text. This
14568 switch is related to the @option{-fverbose-asm} switch, but that
14569 switch only records information in the assembler output file as
14570 comments, so it never reaches the object file.
14574 @cindex global offset table
14576 Generate position-independent code (PIC) suitable for use in a shared
14577 library, if supported for the target machine. Such code accesses all
14578 constant addresses through a global offset table (GOT)@. The dynamic
14579 loader resolves the GOT entries when the program starts (the dynamic
14580 loader is not part of GCC; it is part of the operating system). If
14581 the GOT size for the linked executable exceeds a machine-specific
14582 maximum size, you get an error message from the linker indicating that
14583 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14584 instead. (These maximums are 8k on the SPARC and 32k
14585 on the m68k and RS/6000. The 386 has no such limit.)
14587 Position-independent code requires special support, and therefore works
14588 only on certain machines. For the 386, GCC supports PIC for System V
14589 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14590 position-independent.
14592 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14597 If supported for the target machine, emit position-independent code,
14598 suitable for dynamic linking and avoiding any limit on the size of the
14599 global offset table. This option makes a difference on the m68k,
14600 PowerPC and SPARC@.
14602 Position-independent code requires special support, and therefore works
14603 only on certain machines.
14605 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14612 These options are similar to @option{-fpic} and @option{-fPIC}, but
14613 generated position independent code can be only linked into executables.
14614 Usually these options are used when @option{-pie} GCC option will be
14615 used during linking.
14617 @option{-fpie} and @option{-fPIE} both define the macros
14618 @code{__pie__} and @code{__PIE__}. The macros have the value 1
14619 for @option{-fpie} and 2 for @option{-fPIE}.
14621 @item -fno-jump-tables
14622 @opindex fno-jump-tables
14623 Do not use jump tables for switch statements even where it would be
14624 more efficient than other code generation strategies. This option is
14625 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14626 building code which forms part of a dynamic linker and cannot
14627 reference the address of a jump table. On some targets, jump tables
14628 do not require a GOT and this option is not needed.
14630 @item -ffixed-@var{reg}
14632 Treat the register named @var{reg} as a fixed register; generated code
14633 should never refer to it (except perhaps as a stack pointer, frame
14634 pointer or in some other fixed role).
14636 @var{reg} must be the name of a register. The register names accepted
14637 are machine-specific and are defined in the @code{REGISTER_NAMES}
14638 macro in the machine description macro file.
14640 This flag does not have a negative form, because it specifies a
14643 @item -fcall-used-@var{reg}
14644 @opindex fcall-used
14645 Treat the register named @var{reg} as an allocable register that is
14646 clobbered by function calls. It may be allocated for temporaries or
14647 variables that do not live across a call. Functions compiled this way
14648 will not save and restore the register @var{reg}.
14650 It is an error to used this flag with the frame pointer or stack pointer.
14651 Use of this flag for other registers that have fixed pervasive roles in
14652 the machine's execution model will produce disastrous results.
14654 This flag does not have a negative form, because it specifies a
14657 @item -fcall-saved-@var{reg}
14658 @opindex fcall-saved
14659 Treat the register named @var{reg} as an allocable register saved by
14660 functions. It may be allocated even for temporaries or variables that
14661 live across a call. Functions compiled this way will save and restore
14662 the register @var{reg} if they use it.
14664 It is an error to used this flag with the frame pointer or stack pointer.
14665 Use of this flag for other registers that have fixed pervasive roles in
14666 the machine's execution model will produce disastrous results.
14668 A different sort of disaster will result from the use of this flag for
14669 a register in which function values may be returned.
14671 This flag does not have a negative form, because it specifies a
14674 @item -fpack-struct[=@var{n}]
14675 @opindex fpack-struct
14676 Without a value specified, pack all structure members together without
14677 holes. When a value is specified (which must be a small power of two), pack
14678 structure members according to this value, representing the maximum
14679 alignment (that is, objects with default alignment requirements larger than
14680 this will be output potentially unaligned at the next fitting location.
14682 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14683 code that is not binary compatible with code generated without that switch.
14684 Additionally, it makes the code suboptimal.
14685 Use it to conform to a non-default application binary interface.
14687 @item -finstrument-functions
14688 @opindex finstrument-functions
14689 Generate instrumentation calls for entry and exit to functions. Just
14690 after function entry and just before function exit, the following
14691 profiling functions will be called with the address of the current
14692 function and its call site. (On some platforms,
14693 @code{__builtin_return_address} does not work beyond the current
14694 function, so the call site information may not be available to the
14695 profiling functions otherwise.)
14698 void __cyg_profile_func_enter (void *this_fn,
14700 void __cyg_profile_func_exit (void *this_fn,
14704 The first argument is the address of the start of the current function,
14705 which may be looked up exactly in the symbol table.
14707 This instrumentation is also done for functions expanded inline in other
14708 functions. The profiling calls will indicate where, conceptually, the
14709 inline function is entered and exited. This means that addressable
14710 versions of such functions must be available. If all your uses of a
14711 function are expanded inline, this may mean an additional expansion of
14712 code size. If you use @samp{extern inline} in your C code, an
14713 addressable version of such functions must be provided. (This is
14714 normally the case anyways, but if you get lucky and the optimizer always
14715 expands the functions inline, you might have gotten away without
14716 providing static copies.)
14718 A function may be given the attribute @code{no_instrument_function}, in
14719 which case this instrumentation will not be done. This can be used, for
14720 example, for the profiling functions listed above, high-priority
14721 interrupt routines, and any functions from which the profiling functions
14722 cannot safely be called (perhaps signal handlers, if the profiling
14723 routines generate output or allocate memory).
14725 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
14726 @opindex finstrument-functions-exclude-file-list
14728 Set the list of functions that are excluded from instrumentation (see
14729 the description of @code{-finstrument-functions}). If the file that
14730 contains a function definition matches with one of @var{file}, then
14731 that function is not instrumented. The match is done on substrings:
14732 if the @var{file} parameter is a substring of the file name, it is
14733 considered to be a match.
14736 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
14737 will exclude any inline function defined in files whose pathnames
14738 contain @code{/bits/stl} or @code{include/sys}.
14740 If, for some reason, you want to include letter @code{','} in one of
14741 @var{sym}, write @code{'\,'}. For example,
14742 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
14743 (note the single quote surrounding the option).
14745 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
14746 @opindex finstrument-functions-exclude-function-list
14748 This is similar to @code{-finstrument-functions-exclude-file-list},
14749 but this option sets the list of function names to be excluded from
14750 instrumentation. The function name to be matched is its user-visible
14751 name, such as @code{vector<int> blah(const vector<int> &)}, not the
14752 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
14753 match is done on substrings: if the @var{sym} parameter is a substring
14754 of the function name, it is considered to be a match.
14756 @item -fstack-check
14757 @opindex fstack-check
14758 Generate code to verify that you do not go beyond the boundary of the
14759 stack. You should specify this flag if you are running in an
14760 environment with multiple threads, but only rarely need to specify it in
14761 a single-threaded environment since stack overflow is automatically
14762 detected on nearly all systems if there is only one stack.
14764 Note that this switch does not actually cause checking to be done; the
14765 operating system must do that. The switch causes generation of code
14766 to ensure that the operating system sees the stack being extended.
14768 @item -fstack-limit-register=@var{reg}
14769 @itemx -fstack-limit-symbol=@var{sym}
14770 @itemx -fno-stack-limit
14771 @opindex fstack-limit-register
14772 @opindex fstack-limit-symbol
14773 @opindex fno-stack-limit
14774 Generate code to ensure that the stack does not grow beyond a certain value,
14775 either the value of a register or the address of a symbol. If the stack
14776 would grow beyond the value, a signal is raised. For most targets,
14777 the signal is raised before the stack overruns the boundary, so
14778 it is possible to catch the signal without taking special precautions.
14780 For instance, if the stack starts at absolute address @samp{0x80000000}
14781 and grows downwards, you can use the flags
14782 @option{-fstack-limit-symbol=__stack_limit} and
14783 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
14784 of 128KB@. Note that this may only work with the GNU linker.
14786 @cindex aliasing of parameters
14787 @cindex parameters, aliased
14788 @item -fargument-alias
14789 @itemx -fargument-noalias
14790 @itemx -fargument-noalias-global
14791 @itemx -fargument-noalias-anything
14792 @opindex fargument-alias
14793 @opindex fargument-noalias
14794 @opindex fargument-noalias-global
14795 @opindex fargument-noalias-anything
14796 Specify the possible relationships among parameters and between
14797 parameters and global data.
14799 @option{-fargument-alias} specifies that arguments (parameters) may
14800 alias each other and may alias global storage.@*
14801 @option{-fargument-noalias} specifies that arguments do not alias
14802 each other, but may alias global storage.@*
14803 @option{-fargument-noalias-global} specifies that arguments do not
14804 alias each other and do not alias global storage.
14805 @option{-fargument-noalias-anything} specifies that arguments do not
14806 alias any other storage.
14808 Each language will automatically use whatever option is required by
14809 the language standard. You should not need to use these options yourself.
14811 @item -fleading-underscore
14812 @opindex fleading-underscore
14813 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
14814 change the way C symbols are represented in the object file. One use
14815 is to help link with legacy assembly code.
14817 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
14818 generate code that is not binary compatible with code generated without that
14819 switch. Use it to conform to a non-default application binary interface.
14820 Not all targets provide complete support for this switch.
14822 @item -ftls-model=@var{model}
14823 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
14824 The @var{model} argument should be one of @code{global-dynamic},
14825 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
14827 The default without @option{-fpic} is @code{initial-exec}; with
14828 @option{-fpic} the default is @code{global-dynamic}.
14830 @item -fvisibility=@var{default|internal|hidden|protected}
14831 @opindex fvisibility
14832 Set the default ELF image symbol visibility to the specified option---all
14833 symbols will be marked with this unless overridden within the code.
14834 Using this feature can very substantially improve linking and
14835 load times of shared object libraries, produce more optimized
14836 code, provide near-perfect API export and prevent symbol clashes.
14837 It is @strong{strongly} recommended that you use this in any shared objects
14840 Despite the nomenclature, @code{default} always means public ie;
14841 available to be linked against from outside the shared object.
14842 @code{protected} and @code{internal} are pretty useless in real-world
14843 usage so the only other commonly used option will be @code{hidden}.
14844 The default if @option{-fvisibility} isn't specified is
14845 @code{default}, i.e., make every
14846 symbol public---this causes the same behavior as previous versions of
14849 A good explanation of the benefits offered by ensuring ELF
14850 symbols have the correct visibility is given by ``How To Write
14851 Shared Libraries'' by Ulrich Drepper (which can be found at
14852 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
14853 solution made possible by this option to marking things hidden when
14854 the default is public is to make the default hidden and mark things
14855 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
14856 and @code{__attribute__ ((visibility("default")))} instead of
14857 @code{__declspec(dllexport)} you get almost identical semantics with
14858 identical syntax. This is a great boon to those working with
14859 cross-platform projects.
14861 For those adding visibility support to existing code, you may find
14862 @samp{#pragma GCC visibility} of use. This works by you enclosing
14863 the declarations you wish to set visibility for with (for example)
14864 @samp{#pragma GCC visibility push(hidden)} and
14865 @samp{#pragma GCC visibility pop}.
14866 Bear in mind that symbol visibility should be viewed @strong{as
14867 part of the API interface contract} and thus all new code should
14868 always specify visibility when it is not the default ie; declarations
14869 only for use within the local DSO should @strong{always} be marked explicitly
14870 as hidden as so to avoid PLT indirection overheads---making this
14871 abundantly clear also aids readability and self-documentation of the code.
14872 Note that due to ISO C++ specification requirements, operator new and
14873 operator delete must always be of default visibility.
14875 Be aware that headers from outside your project, in particular system
14876 headers and headers from any other library you use, may not be
14877 expecting to be compiled with visibility other than the default. You
14878 may need to explicitly say @samp{#pragma GCC visibility push(default)}
14879 before including any such headers.
14881 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
14882 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
14883 no modifications. However, this means that calls to @samp{extern}
14884 functions with no explicit visibility will use the PLT, so it is more
14885 effective to use @samp{__attribute ((visibility))} and/or
14886 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
14887 declarations should be treated as hidden.
14889 Note that @samp{-fvisibility} does affect C++ vague linkage
14890 entities. This means that, for instance, an exception class that will
14891 be thrown between DSOs must be explicitly marked with default
14892 visibility so that the @samp{type_info} nodes will be unified between
14895 An overview of these techniques, their benefits and how to use them
14896 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
14902 @node Environment Variables
14903 @section Environment Variables Affecting GCC
14904 @cindex environment variables
14906 @c man begin ENVIRONMENT
14907 This section describes several environment variables that affect how GCC
14908 operates. Some of them work by specifying directories or prefixes to use
14909 when searching for various kinds of files. Some are used to specify other
14910 aspects of the compilation environment.
14912 Note that you can also specify places to search using options such as
14913 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
14914 take precedence over places specified using environment variables, which
14915 in turn take precedence over those specified by the configuration of GCC@.
14916 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
14917 GNU Compiler Collection (GCC) Internals}.
14922 @c @itemx LC_COLLATE
14924 @c @itemx LC_MONETARY
14925 @c @itemx LC_NUMERIC
14930 @c @findex LC_COLLATE
14931 @findex LC_MESSAGES
14932 @c @findex LC_MONETARY
14933 @c @findex LC_NUMERIC
14937 These environment variables control the way that GCC uses
14938 localization information that allow GCC to work with different
14939 national conventions. GCC inspects the locale categories
14940 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
14941 so. These locale categories can be set to any value supported by your
14942 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
14943 Kingdom encoded in UTF-8.
14945 The @env{LC_CTYPE} environment variable specifies character
14946 classification. GCC uses it to determine the character boundaries in
14947 a string; this is needed for some multibyte encodings that contain quote
14948 and escape characters that would otherwise be interpreted as a string
14951 The @env{LC_MESSAGES} environment variable specifies the language to
14952 use in diagnostic messages.
14954 If the @env{LC_ALL} environment variable is set, it overrides the value
14955 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
14956 and @env{LC_MESSAGES} default to the value of the @env{LANG}
14957 environment variable. If none of these variables are set, GCC
14958 defaults to traditional C English behavior.
14962 If @env{TMPDIR} is set, it specifies the directory to use for temporary
14963 files. GCC uses temporary files to hold the output of one stage of
14964 compilation which is to be used as input to the next stage: for example,
14965 the output of the preprocessor, which is the input to the compiler
14968 @item GCC_EXEC_PREFIX
14969 @findex GCC_EXEC_PREFIX
14970 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
14971 names of the subprograms executed by the compiler. No slash is added
14972 when this prefix is combined with the name of a subprogram, but you can
14973 specify a prefix that ends with a slash if you wish.
14975 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
14976 an appropriate prefix to use based on the pathname it was invoked with.
14978 If GCC cannot find the subprogram using the specified prefix, it
14979 tries looking in the usual places for the subprogram.
14981 The default value of @env{GCC_EXEC_PREFIX} is
14982 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
14983 the installed compiler. In many cases @var{prefix} is the value
14984 of @code{prefix} when you ran the @file{configure} script.
14986 Other prefixes specified with @option{-B} take precedence over this prefix.
14988 This prefix is also used for finding files such as @file{crt0.o} that are
14991 In addition, the prefix is used in an unusual way in finding the
14992 directories to search for header files. For each of the standard
14993 directories whose name normally begins with @samp{/usr/local/lib/gcc}
14994 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
14995 replacing that beginning with the specified prefix to produce an
14996 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
14997 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
14998 These alternate directories are searched first; the standard directories
14999 come next. If a standard directory begins with the configured
15000 @var{prefix} then the value of @var{prefix} is replaced by
15001 @env{GCC_EXEC_PREFIX} when looking for header files.
15003 @item COMPILER_PATH
15004 @findex COMPILER_PATH
15005 The value of @env{COMPILER_PATH} is a colon-separated list of
15006 directories, much like @env{PATH}. GCC tries the directories thus
15007 specified when searching for subprograms, if it can't find the
15008 subprograms using @env{GCC_EXEC_PREFIX}.
15011 @findex LIBRARY_PATH
15012 The value of @env{LIBRARY_PATH} is a colon-separated list of
15013 directories, much like @env{PATH}. When configured as a native compiler,
15014 GCC tries the directories thus specified when searching for special
15015 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15016 using GCC also uses these directories when searching for ordinary
15017 libraries for the @option{-l} option (but directories specified with
15018 @option{-L} come first).
15022 @cindex locale definition
15023 This variable is used to pass locale information to the compiler. One way in
15024 which this information is used is to determine the character set to be used
15025 when character literals, string literals and comments are parsed in C and C++.
15026 When the compiler is configured to allow multibyte characters,
15027 the following values for @env{LANG} are recognized:
15031 Recognize JIS characters.
15033 Recognize SJIS characters.
15035 Recognize EUCJP characters.
15038 If @env{LANG} is not defined, or if it has some other value, then the
15039 compiler will use mblen and mbtowc as defined by the default locale to
15040 recognize and translate multibyte characters.
15044 Some additional environments variables affect the behavior of the
15047 @include cppenv.texi
15051 @node Precompiled Headers
15052 @section Using Precompiled Headers
15053 @cindex precompiled headers
15054 @cindex speed of compilation
15056 Often large projects have many header files that are included in every
15057 source file. The time the compiler takes to process these header files
15058 over and over again can account for nearly all of the time required to
15059 build the project. To make builds faster, GCC allows users to
15060 `precompile' a header file; then, if builds can use the precompiled
15061 header file they will be much faster.
15063 To create a precompiled header file, simply compile it as you would any
15064 other file, if necessary using the @option{-x} option to make the driver
15065 treat it as a C or C++ header file. You will probably want to use a
15066 tool like @command{make} to keep the precompiled header up-to-date when
15067 the headers it contains change.
15069 A precompiled header file will be searched for when @code{#include} is
15070 seen in the compilation. As it searches for the included file
15071 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15072 compiler looks for a precompiled header in each directory just before it
15073 looks for the include file in that directory. The name searched for is
15074 the name specified in the @code{#include} with @samp{.gch} appended. If
15075 the precompiled header file can't be used, it is ignored.
15077 For instance, if you have @code{#include "all.h"}, and you have
15078 @file{all.h.gch} in the same directory as @file{all.h}, then the
15079 precompiled header file will be used if possible, and the original
15080 header will be used otherwise.
15082 Alternatively, you might decide to put the precompiled header file in a
15083 directory and use @option{-I} to ensure that directory is searched
15084 before (or instead of) the directory containing the original header.
15085 Then, if you want to check that the precompiled header file is always
15086 used, you can put a file of the same name as the original header in this
15087 directory containing an @code{#error} command.
15089 This also works with @option{-include}. So yet another way to use
15090 precompiled headers, good for projects not designed with precompiled
15091 header files in mind, is to simply take most of the header files used by
15092 a project, include them from another header file, precompile that header
15093 file, and @option{-include} the precompiled header. If the header files
15094 have guards against multiple inclusion, they will be skipped because
15095 they've already been included (in the precompiled header).
15097 If you need to precompile the same header file for different
15098 languages, targets, or compiler options, you can instead make a
15099 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15100 header in the directory, perhaps using @option{-o}. It doesn't matter
15101 what you call the files in the directory, every precompiled header in
15102 the directory will be considered. The first precompiled header
15103 encountered in the directory that is valid for this compilation will
15104 be used; they're searched in no particular order.
15106 There are many other possibilities, limited only by your imagination,
15107 good sense, and the constraints of your build system.
15109 A precompiled header file can be used only when these conditions apply:
15113 Only one precompiled header can be used in a particular compilation.
15116 A precompiled header can't be used once the first C token is seen. You
15117 can have preprocessor directives before a precompiled header; you can
15118 even include a precompiled header from inside another header, so long as
15119 there are no C tokens before the @code{#include}.
15122 The precompiled header file must be produced for the same language as
15123 the current compilation. You can't use a C precompiled header for a C++
15127 The precompiled header file must have been produced by the same compiler
15128 binary as the current compilation is using.
15131 Any macros defined before the precompiled header is included must
15132 either be defined in the same way as when the precompiled header was
15133 generated, or must not affect the precompiled header, which usually
15134 means that they don't appear in the precompiled header at all.
15136 The @option{-D} option is one way to define a macro before a
15137 precompiled header is included; using a @code{#define} can also do it.
15138 There are also some options that define macros implicitly, like
15139 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15142 @item If debugging information is output when using the precompiled
15143 header, using @option{-g} or similar, the same kind of debugging information
15144 must have been output when building the precompiled header. However,
15145 a precompiled header built using @option{-g} can be used in a compilation
15146 when no debugging information is being output.
15148 @item The same @option{-m} options must generally be used when building
15149 and using the precompiled header. @xref{Submodel Options},
15150 for any cases where this rule is relaxed.
15152 @item Each of the following options must be the same when building and using
15153 the precompiled header:
15155 @gccoptlist{-fexceptions -funit-at-a-time}
15158 Some other command-line options starting with @option{-f},
15159 @option{-p}, or @option{-O} must be defined in the same way as when
15160 the precompiled header was generated. At present, it's not clear
15161 which options are safe to change and which are not; the safest choice
15162 is to use exactly the same options when generating and using the
15163 precompiled header. The following are known to be safe:
15165 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15166 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15167 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15172 For all of these except the last, the compiler will automatically
15173 ignore the precompiled header if the conditions aren't met. If you
15174 find an option combination that doesn't work and doesn't cause the
15175 precompiled header to be ignored, please consider filing a bug report,
15178 If you do use differing options when generating and using the
15179 precompiled header, the actual behavior will be a mixture of the
15180 behavior for the options. For instance, if you use @option{-g} to
15181 generate the precompiled header but not when using it, you may or may
15182 not get debugging information for routines in the precompiled header.
15184 @node Running Protoize
15185 @section Running Protoize
15187 The program @code{protoize} is an optional part of GCC@. You can use
15188 it to add prototypes to a program, thus converting the program to ISO
15189 C in one respect. The companion program @code{unprotoize} does the
15190 reverse: it removes argument types from any prototypes that are found.
15192 When you run these programs, you must specify a set of source files as
15193 command line arguments. The conversion programs start out by compiling
15194 these files to see what functions they define. The information gathered
15195 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15197 After scanning comes actual conversion. The specified files are all
15198 eligible to be converted; any files they include (whether sources or
15199 just headers) are eligible as well.
15201 But not all the eligible files are converted. By default,
15202 @code{protoize} and @code{unprotoize} convert only source and header
15203 files in the current directory. You can specify additional directories
15204 whose files should be converted with the @option{-d @var{directory}}
15205 option. You can also specify particular files to exclude with the
15206 @option{-x @var{file}} option. A file is converted if it is eligible, its
15207 directory name matches one of the specified directory names, and its
15208 name within the directory has not been excluded.
15210 Basic conversion with @code{protoize} consists of rewriting most
15211 function definitions and function declarations to specify the types of
15212 the arguments. The only ones not rewritten are those for varargs
15215 @code{protoize} optionally inserts prototype declarations at the
15216 beginning of the source file, to make them available for any calls that
15217 precede the function's definition. Or it can insert prototype
15218 declarations with block scope in the blocks where undeclared functions
15221 Basic conversion with @code{unprotoize} consists of rewriting most
15222 function declarations to remove any argument types, and rewriting
15223 function definitions to the old-style pre-ISO form.
15225 Both conversion programs print a warning for any function declaration or
15226 definition that they can't convert. You can suppress these warnings
15229 The output from @code{protoize} or @code{unprotoize} replaces the
15230 original source file. The original file is renamed to a name ending
15231 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15232 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15233 for DOS) file already exists, then the source file is simply discarded.
15235 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15236 scan the program and collect information about the functions it uses.
15237 So neither of these programs will work until GCC is installed.
15239 Here is a table of the options you can use with @code{protoize} and
15240 @code{unprotoize}. Each option works with both programs unless
15244 @item -B @var{directory}
15245 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15246 usual directory (normally @file{/usr/local/lib}). This file contains
15247 prototype information about standard system functions. This option
15248 applies only to @code{protoize}.
15250 @item -c @var{compilation-options}
15251 Use @var{compilation-options} as the options when running @command{gcc} to
15252 produce the @samp{.X} files. The special option @option{-aux-info} is
15253 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15255 Note that the compilation options must be given as a single argument to
15256 @code{protoize} or @code{unprotoize}. If you want to specify several
15257 @command{gcc} options, you must quote the entire set of compilation options
15258 to make them a single word in the shell.
15260 There are certain @command{gcc} arguments that you cannot use, because they
15261 would produce the wrong kind of output. These include @option{-g},
15262 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15263 the @var{compilation-options}, they are ignored.
15266 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15267 systems) instead of @samp{.c}. This is convenient if you are converting
15268 a C program to C++. This option applies only to @code{protoize}.
15271 Add explicit global declarations. This means inserting explicit
15272 declarations at the beginning of each source file for each function
15273 that is called in the file and was not declared. These declarations
15274 precede the first function definition that contains a call to an
15275 undeclared function. This option applies only to @code{protoize}.
15277 @item -i @var{string}
15278 Indent old-style parameter declarations with the string @var{string}.
15279 This option applies only to @code{protoize}.
15281 @code{unprotoize} converts prototyped function definitions to old-style
15282 function definitions, where the arguments are declared between the
15283 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15284 uses five spaces as the indentation. If you want to indent with just
15285 one space instead, use @option{-i " "}.
15288 Keep the @samp{.X} files. Normally, they are deleted after conversion
15292 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15293 a prototype declaration for each function in each block which calls the
15294 function without any declaration. This option applies only to
15298 Make no real changes. This mode just prints information about the conversions
15299 that would have been done without @option{-n}.
15302 Make no @samp{.save} files. The original files are simply deleted.
15303 Use this option with caution.
15305 @item -p @var{program}
15306 Use the program @var{program} as the compiler. Normally, the name
15307 @file{gcc} is used.
15310 Work quietly. Most warnings are suppressed.
15313 Print the version number, just like @option{-v} for @command{gcc}.
15316 If you need special compiler options to compile one of your program's
15317 source files, then you should generate that file's @samp{.X} file
15318 specially, by running @command{gcc} on that source file with the
15319 appropriate options and the option @option{-aux-info}. Then run
15320 @code{protoize} on the entire set of files. @code{protoize} will use
15321 the existing @samp{.X} file because it is newer than the source file.
15325 gcc -Dfoo=bar file1.c -aux-info file1.X
15330 You need to include the special files along with the rest in the
15331 @code{protoize} command, even though their @samp{.X} files already
15332 exist, because otherwise they won't get converted.
15334 @xref{Protoize Caveats}, for more information on how to use
15335 @code{protoize} successfully.