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 Free Software Foundation, Inc.
16 Permission is granted to copy, distribute and/or modify this document
17 under the terms of the GNU Free Documentation License, Version 1.2 or
18 any later version published by the Free Software Foundation; with the
19 Invariant Sections being ``GNU General Public License'' and ``Funding
20 Free Software'', the Front-Cover texts being (a) (see below), and with
21 the Back-Cover Texts being (b) (see below). A copy of the license is
22 included in the gfdl(7) man page.
24 (a) The FSF's Front-Cover Text is:
28 (b) The FSF's Back-Cover Text is:
30 You have freedom to copy and modify this GNU Manual, like GNU
31 software. Copies published by the Free Software Foundation raise
32 funds for GNU development.
34 @c Set file name and title for the man page.
36 @settitle GNU project C and C++ compiler
38 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
39 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
40 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
41 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
42 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
43 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
44 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
46 Only the most useful options are listed here; see below for the
47 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
50 gpl(7), gfdl(7), fsf-funding(7),
51 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
52 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
53 @file{ld}, @file{binutils} and @file{gdb}.
56 For instructions on reporting bugs, see
57 @w{@uref{http://gcc.gnu.org/bugs.html}}.
60 See the Info entry for @command{gcc}, or
61 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
62 for contributors to GCC@.
67 @chapter GCC Command Options
68 @cindex GCC command options
69 @cindex command options
70 @cindex options, GCC command
72 @c man begin DESCRIPTION
73 When you invoke GCC, it normally does preprocessing, compilation,
74 assembly and linking. The ``overall options'' allow you to stop this
75 process at an intermediate stage. For example, the @option{-c} option
76 says not to run the linker. Then the output consists of object files
77 output by the assembler.
79 Other options are passed on to one stage of processing. Some options
80 control the preprocessor and others the compiler itself. Yet other
81 options control the assembler and linker; most of these are not
82 documented here, since you rarely need to use any of them.
84 @cindex C compilation options
85 Most of the command line options that you can use with GCC are useful
86 for C programs; when an option is only useful with another language
87 (usually C++), the explanation says so explicitly. If the description
88 for a particular option does not mention a source language, you can use
89 that option with all supported languages.
91 @cindex C++ compilation options
92 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
93 options for compiling C++ programs.
95 @cindex grouping options
96 @cindex options, grouping
97 The @command{gcc} program accepts options and file names as operands. Many
98 options have multi-letter names; therefore multiple single-letter options
99 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
102 @cindex order of options
103 @cindex options, order
104 You can mix options and other arguments. For the most part, the order
105 you use doesn't matter. Order does matter when you use several options
106 of the same kind; for example, if you specify @option{-L} more than once,
107 the directories are searched in the order specified.
109 Many options have long names starting with @samp{-f} or with
110 @samp{-W}---for example,
111 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
112 these have both positive and negative forms; the negative form of
113 @option{-ffoo} would be @option{-fno-foo}. This manual documents
114 only one of these two forms, whichever one is not the default.
118 @xref{Option Index}, for an index to GCC's options.
121 * Option Summary:: Brief list of all options, without explanations.
122 * Overall Options:: Controlling the kind of output:
123 an executable, object files, assembler files,
124 or preprocessed source.
125 * Invoking G++:: Compiling C++ programs.
126 * C Dialect Options:: Controlling the variant of C language compiled.
127 * C++ Dialect Options:: Variations on C++.
128 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
130 * Language Independent Options:: Controlling how diagnostics should be
132 * Warning Options:: How picky should the compiler be?
133 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
134 * Optimize Options:: How much optimization?
135 * Preprocessor Options:: Controlling header files and macro definitions.
136 Also, getting dependency information for Make.
137 * Assembler Options:: Passing options to the assembler.
138 * Link Options:: Specifying libraries and so on.
139 * Directory Options:: Where to find header files and libraries.
140 Where to find the compiler executable files.
141 * Spec Files:: How to pass switches to sub-processes.
142 * Target Options:: Running a cross-compiler, or an old version of GCC.
143 * Submodel Options:: Specifying minor hardware or convention variations,
144 such as 68010 vs 68020.
145 * Code Gen Options:: Specifying conventions for function calls, data layout
147 * Environment Variables:: Env vars that affect GCC.
148 * Precompiled Headers:: Compiling a header once, and using it many times.
149 * Running Protoize:: Automatically adding or removing function prototypes.
155 @section Option Summary
157 Here is a summary of all the options, grouped by type. Explanations are
158 in the following sections.
161 @item Overall Options
162 @xref{Overall Options,,Options Controlling the Kind of Output}.
163 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
164 -x @var{language} -v -### --help@r{[}=@var{class}@r{]} --target-help @gol
165 --version @@@var{file}}
167 @item C Language Options
168 @xref{C Dialect Options,,Options Controlling C Dialect}.
169 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
170 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
171 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
172 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
173 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
174 -fsigned-bitfields -fsigned-char @gol
175 -funsigned-bitfields -funsigned-char}
177 @item C++ Language Options
178 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
179 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
180 -fconserve-space -ffriend-injection @gol
181 -fno-elide-constructors @gol
182 -fno-enforce-eh-specs @gol
183 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
184 -fno-implicit-templates @gol
185 -fno-implicit-inline-templates @gol
186 -fno-implement-inlines -fms-extensions @gol
187 -fno-nonansi-builtins -fno-operator-names @gol
188 -fno-optional-diags -fpermissive @gol
189 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
190 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
191 -fno-default-inline -fvisibility-inlines-hidden @gol
192 -Wabi -Wctor-dtor-privacy @gol
193 -Wnon-virtual-dtor -Wreorder @gol
194 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
195 -Wno-non-template-friend -Wold-style-cast @gol
196 -Woverloaded-virtual -Wno-pmf-conversions @gol
199 @item Objective-C and Objective-C++ Language Options
200 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
201 Objective-C and Objective-C++ Dialects}.
202 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
203 -fgnu-runtime -fnext-runtime @gol
204 -fno-nil-receivers @gol
205 -fobjc-call-cxx-cdtors @gol
206 -fobjc-direct-dispatch @gol
207 -fobjc-exceptions @gol
209 -freplace-objc-classes @gol
212 -Wassign-intercept @gol
213 -Wno-protocol -Wselector @gol
214 -Wstrict-selector-match @gol
215 -Wundeclared-selector}
217 @item Language Independent Options
218 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
219 @gccoptlist{-fmessage-length=@var{n} @gol
220 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
221 -fdiagnostics-show-option}
223 @item Warning Options
224 @xref{Warning Options,,Options to Request or Suppress Warnings}.
225 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
226 -w -Wextra -Wall -Waggregate-return -Walways-true -Warray-bounds @gol
227 -Wno-attributes -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
228 -Wchar-subscripts -Wclobbered -Wcomment @gol
229 -Wconversion -Wcoverage-mismatch -Wno-deprecated-declarations @gol
230 -Wdisabled-optimization -Wno-div-by-zero @gol
231 -Wempty-body -Wno-endif-labels @gol
232 -Werror -Werror=* @gol
233 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
234 -Wno-format-extra-args -Wformat-nonliteral @gol
235 -Wformat-security -Wformat-y2k @gol
236 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
237 -Wimport -Wno-import -Winit-self -Winline @gol
238 -Wno-int-to-pointer-cast @gol
239 -Wno-invalid-offsetof -Winvalid-pch @gol
240 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
241 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
242 -Wmissing-format-attribute -Wmissing-include-dirs @gol
243 -Wmissing-noreturn @gol
244 -Wno-multichar -Wnonnull -Wno-overflow @gol
245 -Woverlength-strings -Wpacked -Wpadded @gol
246 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
247 -Wredundant-decls @gol
248 -Wreturn-type -Wsequence-point -Wshadow @gol
249 -Wsign-compare -Wstack-protector @gol
250 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
251 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
252 -Wstring-literal-comparison @gol
253 -Wswitch -Wswitch-default -Wswitch-enum @gol
254 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
255 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
256 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
257 -Wunused-value -Wunused-variable -Wvariadic-macros @gol
258 -Wvolatile-register-var -Wwrite-strings}
260 @item C-only Warning Options
261 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
262 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
263 -Wold-style-declaration -Wold-style-definition @gol
264 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
265 -Wdeclaration-after-statement -Wpointer-sign}
267 @item Debugging Options
268 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
269 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
270 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
271 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
272 -fdump-ipa-all -fdump-ipa-cgraph @gol
274 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
275 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
276 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
277 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
279 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
283 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
284 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-nrv -fdump-tree-vect @gol
288 -fdump-tree-sink @gol
289 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-salias @gol
291 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
293 -ftree-vectorizer-verbose=@var{n} @gol
294 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
295 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
296 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
297 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
298 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
299 -ftest-coverage -ftime-report -fvar-tracking @gol
300 -g -g@var{level} -gcoff -gdwarf-2 @gol
301 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
302 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
303 -print-multi-directory -print-multi-lib @gol
304 -print-prog-name=@var{program} -print-search-dirs -Q @gol
307 @item Optimization Options
308 @xref{Optimize Options,,Options that Control Optimization}.
309 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
310 -falign-labels=@var{n} -falign-loops=@var{n} @gol
311 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
312 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
313 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
314 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
315 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
316 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
317 -fexpensive-optimizations -ffast-math -ffloat-store @gol
318 -fforce-addr -fforward-propagate -ffunction-sections @gol
319 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
320 -fcrossjumping -fif-conversion -fif-conversion2 @gol
321 -finline-functions -finline-functions-called-once @gol
322 -finline-limit=@var{n} -fkeep-inline-functions @gol
323 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
324 -fmodulo-sched -fno-branch-count-reg @gol
325 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
326 -fno-function-cse -fno-guess-branch-probability @gol
327 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
328 -funsafe-math-optimizations -funsafe-loop-optimizations @gol
329 -ffinite-math-only -fno-signed-zeros @gol
330 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
331 -fomit-frame-pointer -foptimize-register-move @gol
332 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
333 -fprofile-generate -fprofile-use @gol
334 -fregmove -frename-registers @gol
335 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
336 -frerun-cse-after-loop @gol
337 -frounding-math -frtl-abstract-sequences @gol
338 -fschedule-insns -fschedule-insns2 @gol
339 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
340 -fsched-spec-load-dangerous @gol
341 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
342 -fsched2-use-superblocks @gol
343 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
344 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
345 -fno-split-wide-types -fstack-protector -fstack-protector-all @gol
346 -fstrict-aliasing -fstrict-overflow -ftracer -fthread-jumps @gol
347 -funroll-all-loops -funroll-loops -fpeel-loops @gol
348 -fsplit-ivs-in-unroller -funswitch-loops @gol
349 -fvariable-expansion-in-unroller @gol
350 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
351 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
352 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
353 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
354 -ftree-vect-loop-version -ftree-salias -fipa-pta -fweb @gol
355 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
356 --param @var{name}=@var{value}
357 -O -O0 -O1 -O2 -O3 -Os}
359 @item Preprocessor Options
360 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
361 @gccoptlist{-A@var{question}=@var{answer} @gol
362 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
363 -C -dD -dI -dM -dN @gol
364 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
365 -idirafter @var{dir} @gol
366 -include @var{file} -imacros @var{file} @gol
367 -iprefix @var{file} -iwithprefix @var{dir} @gol
368 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
369 -imultilib @var{dir} -isysroot @var{dir} @gol
370 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
371 -P -fworking-directory -remap @gol
372 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
373 -Xpreprocessor @var{option}}
375 @item Assembler Option
376 @xref{Assembler Options,,Passing Options to the Assembler}.
377 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
380 @xref{Link Options,,Options for Linking}.
381 @gccoptlist{@var{object-file-name} -l@var{library} @gol
382 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
383 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
384 -Wl,@var{option} -Xlinker @var{option} @gol
387 @item Directory Options
388 @xref{Directory Options,,Options for Directory Search}.
389 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
390 -specs=@var{file} -I- --sysroot=@var{dir}}
393 @c I wrote this xref this way to avoid overfull hbox. -- rms
394 @xref{Target Options}.
395 @gccoptlist{-V @var{version} -b @var{machine}}
397 @item Machine Dependent Options
398 @xref{Submodel Options,,Hardware Models and Configurations}.
399 @c This list is ordered alphanumerically by subsection name.
400 @c Try and put the significant identifier (CPU or system) first,
401 @c so users have a clue at guessing where the ones they want will be.
404 @gccoptlist{-EB -EL @gol
405 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
406 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
409 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
410 -mabi=@var{name} @gol
411 -mapcs-stack-check -mno-apcs-stack-check @gol
412 -mapcs-float -mno-apcs-float @gol
413 -mapcs-reentrant -mno-apcs-reentrant @gol
414 -msched-prolog -mno-sched-prolog @gol
415 -mlittle-endian -mbig-endian -mwords-little-endian @gol
416 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
417 -mthumb-interwork -mno-thumb-interwork @gol
418 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
419 -mstructure-size-boundary=@var{n} @gol
420 -mabort-on-noreturn @gol
421 -mlong-calls -mno-long-calls @gol
422 -msingle-pic-base -mno-single-pic-base @gol
423 -mpic-register=@var{reg} @gol
424 -mnop-fun-dllimport @gol
425 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
426 -mpoke-function-name @gol
428 -mtpcs-frame -mtpcs-leaf-frame @gol
429 -mcaller-super-interworking -mcallee-super-interworking @gol
433 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
434 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
436 @emph{Blackfin Options}
437 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
438 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
439 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
440 -mno-id-shared-library -mshared-library-id=@var{n} @gol
441 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
442 -msep-data -mno-sep-data -mlong-calls -mno-long-calls}
445 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
446 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
447 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
448 -mstack-align -mdata-align -mconst-align @gol
449 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
450 -melf -maout -melinux -mlinux -sim -sim2 @gol
451 -mmul-bug-workaround -mno-mul-bug-workaround}
454 @gccoptlist{-mmac -mpush-args}
456 @emph{Darwin Options}
457 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
458 -arch_only -bind_at_load -bundle -bundle_loader @gol
459 -client_name -compatibility_version -current_version @gol
461 -dependency-file -dylib_file -dylinker_install_name @gol
462 -dynamic -dynamiclib -exported_symbols_list @gol
463 -filelist -flat_namespace -force_cpusubtype_ALL @gol
464 -force_flat_namespace -headerpad_max_install_names @gol
465 -image_base -init -install_name -keep_private_externs @gol
466 -multi_module -multiply_defined -multiply_defined_unused @gol
467 -noall_load -no_dead_strip_inits_and_terms @gol
468 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
469 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
470 -private_bundle -read_only_relocs -sectalign @gol
471 -sectobjectsymbols -whyload -seg1addr @gol
472 -sectcreate -sectobjectsymbols -sectorder @gol
473 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
474 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
475 -segprot -segs_read_only_addr -segs_read_write_addr @gol
476 -single_module -static -sub_library -sub_umbrella @gol
477 -twolevel_namespace -umbrella -undefined @gol
478 -unexported_symbols_list -weak_reference_mismatches @gol
479 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
480 -mkernel -mone-byte-bool}
482 @emph{DEC Alpha Options}
483 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
484 -mieee -mieee-with-inexact -mieee-conformant @gol
485 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
486 -mtrap-precision=@var{mode} -mbuild-constants @gol
487 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
488 -mbwx -mmax -mfix -mcix @gol
489 -mfloat-vax -mfloat-ieee @gol
490 -mexplicit-relocs -msmall-data -mlarge-data @gol
491 -msmall-text -mlarge-text @gol
492 -mmemory-latency=@var{time}}
494 @emph{DEC Alpha/VMS Options}
495 @gccoptlist{-mvms-return-codes}
498 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
499 -mhard-float -msoft-float @gol
500 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
501 -mdouble -mno-double @gol
502 -mmedia -mno-media -mmuladd -mno-muladd @gol
503 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
504 -mlinked-fp -mlong-calls -malign-labels @gol
505 -mlibrary-pic -macc-4 -macc-8 @gol
506 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
507 -moptimize-membar -mno-optimize-membar @gol
508 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
509 -mvliw-branch -mno-vliw-branch @gol
510 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
511 -mno-nested-cond-exec -mtomcat-stats @gol
515 @emph{GNU/Linux Options}
516 @gccoptlist{-muclibc}
518 @emph{H8/300 Options}
519 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
522 @gccoptlist{-march=@var{architecture-type} @gol
523 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
524 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
525 -mfixed-range=@var{register-range} @gol
526 -mjump-in-delay -mlinker-opt -mlong-calls @gol
527 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
528 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
529 -mno-jump-in-delay -mno-long-load-store @gol
530 -mno-portable-runtime -mno-soft-float @gol
531 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
532 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
533 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
534 -munix=@var{unix-std} -nolibdld -static -threads}
536 @emph{i386 and x86-64 Options}
537 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
538 -mfpmath=@var{unit} @gol
539 -masm=@var{dialect} -mno-fancy-math-387 @gol
540 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
541 -mno-wide-multiply -mrtd -malign-double @gol
542 -mpreferred-stack-boundary=@var{num} @gol
543 -mmmx -msse -msse2 -msse3 -mssse3 -msse4a -m3dnow -mpopcnt -mabm @gol
544 -mthreads -mno-align-stringops -minline-all-stringops @gol
545 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
546 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
548 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
549 -mcmodel=@var{code-model} @gol
550 -m32 -m64 -mlarge-data-threshold=@var{num}}
553 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
554 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
555 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
556 -minline-float-divide-max-throughput @gol
557 -minline-int-divide-min-latency @gol
558 -minline-int-divide-max-throughput @gol
559 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
560 -mno-dwarf2-asm -mearly-stop-bits @gol
561 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
562 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
563 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
564 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
565 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
566 -mno-sched-prefer-non-data-spec-insns @gol
567 -mno-sched-prefer-non-control-spec-insns @gol
568 -mno-sched-count-spec-in-critical-path}
570 @emph{M32R/D Options}
571 @gccoptlist{-m32r2 -m32rx -m32r @gol
573 -malign-loops -mno-align-loops @gol
574 -missue-rate=@var{number} @gol
575 -mbranch-cost=@var{number} @gol
576 -mmodel=@var{code-size-model-type} @gol
577 -msdata=@var{sdata-type} @gol
578 -mno-flush-func -mflush-func=@var{name} @gol
579 -mno-flush-trap -mflush-trap=@var{number} @gol
583 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
585 @emph{M680x0 Options}
586 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
587 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
588 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
589 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
590 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
591 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
592 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
593 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
595 @emph{M68hc1x Options}
596 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
597 -mauto-incdec -minmax -mlong-calls -mshort @gol
598 -msoft-reg-count=@var{count}}
601 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
602 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
603 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
604 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
605 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
608 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
609 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
610 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
611 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
612 -mfp32 -mfp64 -mhard-float -msoft-float @gol
613 -msingle-float -mdouble-float -mdsp -mpaired-single -mips3d @gol
614 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
615 -G@var{num} -membedded-data -mno-embedded-data @gol
616 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
617 -msplit-addresses -mno-split-addresses @gol
618 -mexplicit-relocs -mno-explicit-relocs @gol
619 -mcheck-zero-division -mno-check-zero-division @gol
620 -mdivide-traps -mdivide-breaks @gol
621 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
622 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
623 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
624 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
625 -mfix-sb1 -mno-fix-sb1 @gol
626 -mflush-func=@var{func} -mno-flush-func @gol
627 -mbranch-likely -mno-branch-likely @gol
628 -mfp-exceptions -mno-fp-exceptions @gol
629 -mvr4130-align -mno-vr4130-align}
632 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
633 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
634 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
635 -mno-base-addresses -msingle-exit -mno-single-exit}
637 @emph{MN10300 Options}
638 @gccoptlist{-mmult-bug -mno-mult-bug @gol
639 -mam33 -mno-am33 @gol
640 -mam33-2 -mno-am33-2 @gol
641 -mreturn-pointer-on-d0 @gol
645 @gccoptlist{-mno-crt0 -mbacc -msim @gol
646 -march=@var{cpu-type} }
648 @emph{PDP-11 Options}
649 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
650 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
651 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
652 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
653 -mbranch-expensive -mbranch-cheap @gol
654 -msplit -mno-split -munix-asm -mdec-asm}
656 @emph{PowerPC Options}
657 See RS/6000 and PowerPC Options.
659 @emph{RS/6000 and PowerPC Options}
660 @gccoptlist{-mcpu=@var{cpu-type} @gol
661 -mtune=@var{cpu-type} @gol
662 -mpower -mno-power -mpower2 -mno-power2 @gol
663 -mpowerpc -mpowerpc64 -mno-powerpc @gol
664 -maltivec -mno-altivec @gol
665 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
666 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
667 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
668 -mmfpgpr -mno-mfpgpr @gol
669 -mnew-mnemonics -mold-mnemonics @gol
670 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
671 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
672 -malign-power -malign-natural @gol
673 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
674 -mstring -mno-string -mupdate -mno-update @gol
675 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
676 -mstrict-align -mno-strict-align -mrelocatable @gol
677 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
678 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
679 -mdynamic-no-pic -maltivec -mswdiv @gol
680 -mprioritize-restricted-insns=@var{priority} @gol
681 -msched-costly-dep=@var{dependence_type} @gol
682 -minsert-sched-nops=@var{scheme} @gol
683 -mcall-sysv -mcall-netbsd @gol
684 -maix-struct-return -msvr4-struct-return @gol
685 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
686 -misel -mno-isel @gol
687 -misel=yes -misel=no @gol
689 -mspe=yes -mspe=no @gol
690 -mvrsave -mno-vrsave @gol
691 -mmulhw -mno-mulhw @gol
692 -mdlmzb -mno-dlmzb @gol
693 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
694 -mprototype -mno-prototype @gol
695 -msim -mmvme -mads -myellowknife -memb -msdata @gol
696 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
698 @emph{S/390 and zSeries Options}
699 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
700 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
701 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
702 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
703 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
704 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
705 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
708 @gccoptlist{-mel -mel @gol
713 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
714 -m4-nofpu -m4-single-only -m4-single -m4 @gol
715 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
716 -m5-64media -m5-64media-nofpu @gol
717 -m5-32media -m5-32media-nofpu @gol
718 -m5-compact -m5-compact-nofpu @gol
719 -mb -ml -mdalign -mrelax @gol
720 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
721 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
722 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
723 -mdivsi3_libfunc=@var{name} @gol
724 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
728 @gccoptlist{-mcpu=@var{cpu-type} @gol
729 -mtune=@var{cpu-type} @gol
730 -mcmodel=@var{code-model} @gol
731 -m32 -m64 -mapp-regs -mno-app-regs @gol
732 -mfaster-structs -mno-faster-structs @gol
733 -mfpu -mno-fpu -mhard-float -msoft-float @gol
734 -mhard-quad-float -msoft-quad-float @gol
735 -mimpure-text -mno-impure-text -mlittle-endian @gol
736 -mstack-bias -mno-stack-bias @gol
737 -munaligned-doubles -mno-unaligned-doubles @gol
738 -mv8plus -mno-v8plus -mvis -mno-vis
739 -threads -pthreads -pthread}
742 @gccoptlist{-mwarn-reloc -merror-reloc @gol
743 -msafe-dma -munsafe-dma @gol
745 -msmall-mem -mlarge-mem -mstdmain @gol
746 -mfixed-range=@var{register-range}}
748 @emph{System V Options}
749 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
751 @emph{TMS320C3x/C4x Options}
752 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
753 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
754 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
755 -mparallel-insns -mparallel-mpy -mpreserve-float}
758 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
759 -mprolog-function -mno-prolog-function -mspace @gol
760 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
761 -mapp-regs -mno-app-regs @gol
762 -mdisable-callt -mno-disable-callt @gol
768 @gccoptlist{-mg -mgnu -munix}
770 @emph{x86-64 Options}
771 See i386 and x86-64 Options.
773 @emph{Xstormy16 Options}
776 @emph{Xtensa Options}
777 @gccoptlist{-mconst16 -mno-const16 @gol
778 -mfused-madd -mno-fused-madd @gol
779 -mtext-section-literals -mno-text-section-literals @gol
780 -mtarget-align -mno-target-align @gol
781 -mlongcalls -mno-longcalls}
783 @emph{zSeries Options}
784 See S/390 and zSeries Options.
786 @item Code Generation Options
787 @xref{Code Gen Options,,Options for Code Generation Conventions}.
788 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
789 -ffixed-@var{reg} -fexceptions @gol
790 -fnon-call-exceptions -funwind-tables @gol
791 -fasynchronous-unwind-tables @gol
792 -finhibit-size-directive -finstrument-functions @gol
793 -fno-common -fno-ident @gol
794 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
795 -fno-jump-tables @gol
796 -frecord-gcc-switches @gol
797 -freg-struct-return -fshort-enums @gol
798 -fshort-double -fshort-wchar @gol
799 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
800 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
801 -fargument-alias -fargument-noalias @gol
802 -fargument-noalias-global -fargument-noalias-anything
803 -fleading-underscore -ftls-model=@var{model} @gol
804 -ftrapv -fwrapv -fbounds-check @gol
809 * Overall Options:: Controlling the kind of output:
810 an executable, object files, assembler files,
811 or preprocessed source.
812 * C Dialect Options:: Controlling the variant of C language compiled.
813 * C++ Dialect Options:: Variations on C++.
814 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
816 * Language Independent Options:: Controlling how diagnostics should be
818 * Warning Options:: How picky should the compiler be?
819 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
820 * Optimize Options:: How much optimization?
821 * Preprocessor Options:: Controlling header files and macro definitions.
822 Also, getting dependency information for Make.
823 * Assembler Options:: Passing options to the assembler.
824 * Link Options:: Specifying libraries and so on.
825 * Directory Options:: Where to find header files and libraries.
826 Where to find the compiler executable files.
827 * Spec Files:: How to pass switches to sub-processes.
828 * Target Options:: Running a cross-compiler, or an old version of GCC.
831 @node Overall Options
832 @section Options Controlling the Kind of Output
834 Compilation can involve up to four stages: preprocessing, compilation
835 proper, assembly and linking, always in that order. GCC is capable of
836 preprocessing and compiling several files either into several
837 assembler input files, or into one assembler input file; then each
838 assembler input file produces an object file, and linking combines all
839 the object files (those newly compiled, and those specified as input)
840 into an executable file.
842 @cindex file name suffix
843 For any given input file, the file name suffix determines what kind of
848 C source code which must be preprocessed.
851 C source code which should not be preprocessed.
854 C++ source code which should not be preprocessed.
857 Objective-C source code. Note that you must link with the @file{libobjc}
858 library to make an Objective-C program work.
861 Objective-C source code which should not be preprocessed.
865 Objective-C++ source code. Note that you must link with the @file{libobjc}
866 library to make an Objective-C++ program work. Note that @samp{.M} refers
867 to a literal capital M@.
870 Objective-C++ source code which should not be preprocessed.
873 C, C++, Objective-C or Objective-C++ header file to be turned into a
878 @itemx @var{file}.cxx
879 @itemx @var{file}.cpp
880 @itemx @var{file}.CPP
881 @itemx @var{file}.c++
883 C++ source code which must be preprocessed. Note that in @samp{.cxx},
884 the last two letters must both be literally @samp{x}. Likewise,
885 @samp{.C} refers to a literal capital C@.
889 Objective-C++ source code which must be preprocessed.
892 Objective-C++ source code which should not be preprocessed.
896 C++ header file to be turned into a precompiled header.
899 @itemx @var{file}.for
900 @itemx @var{file}.FOR
901 Fixed form Fortran source code which should not be preprocessed.
904 @itemx @var{file}.fpp
905 @itemx @var{file}.FPP
906 Fixed form Fortran source code which must be preprocessed (with the traditional
910 @itemx @var{file}.f95
911 Free form Fortran source code which should not be preprocessed.
914 @itemx @var{file}.F95
915 Free form Fortran source code which must be preprocessed (with the
916 traditional preprocessor).
918 @c FIXME: Descriptions of Java file types.
925 Ada source code file which contains a library unit declaration (a
926 declaration of a package, subprogram, or generic, or a generic
927 instantiation), or a library unit renaming declaration (a package,
928 generic, or subprogram renaming declaration). Such files are also
931 @itemx @var{file}.adb
932 Ada source code file containing a library unit body (a subprogram or
933 package body). Such files are also called @dfn{bodies}.
935 @c GCC also knows about some suffixes for languages not yet included:
946 Assembler code which must be preprocessed.
949 An object file to be fed straight into linking.
950 Any file name with no recognized suffix is treated this way.
954 You can specify the input language explicitly with the @option{-x} option:
957 @item -x @var{language}
958 Specify explicitly the @var{language} for the following input files
959 (rather than letting the compiler choose a default based on the file
960 name suffix). This option applies to all following input files until
961 the next @option{-x} option. Possible values for @var{language} are:
963 c c-header c-cpp-output
964 c++ c++-header c++-cpp-output
965 objective-c objective-c-header objective-c-cpp-output
966 objective-c++ objective-c++-header objective-c++-cpp-output
967 assembler assembler-with-cpp
975 Turn off any specification of a language, so that subsequent files are
976 handled according to their file name suffixes (as they are if @option{-x}
977 has not been used at all).
979 @item -pass-exit-codes
980 @opindex pass-exit-codes
981 Normally the @command{gcc} program will exit with the code of 1 if any
982 phase of the compiler returns a non-success return code. If you specify
983 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
984 numerically highest error produced by any phase that returned an error
985 indication. The C, C++, and Fortran frontends return 4, if an internal
986 compiler error is encountered.
989 If you only want some of the stages of compilation, you can use
990 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
991 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
992 @command{gcc} is to stop. Note that some combinations (for example,
993 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
998 Compile or assemble the source files, but do not link. The linking
999 stage simply is not done. The ultimate output is in the form of an
1000 object file for each source file.
1002 By default, the object file name for a source file is made by replacing
1003 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1005 Unrecognized input files, not requiring compilation or assembly, are
1010 Stop after the stage of compilation proper; do not assemble. The output
1011 is in the form of an assembler code file for each non-assembler input
1014 By default, the assembler file name for a source file is made by
1015 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1017 Input files that don't require compilation are ignored.
1021 Stop after the preprocessing stage; do not run the compiler proper. The
1022 output is in the form of preprocessed source code, which is sent to the
1025 Input files which don't require preprocessing are ignored.
1027 @cindex output file option
1030 Place output in file @var{file}. This applies regardless to whatever
1031 sort of output is being produced, whether it be an executable file,
1032 an object file, an assembler file or preprocessed C code.
1034 If @option{-o} is not specified, the default is to put an executable
1035 file in @file{a.out}, the object file for
1036 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1037 assembler file in @file{@var{source}.s}, a precompiled header file in
1038 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1043 Print (on standard error output) the commands executed to run the stages
1044 of compilation. Also print the version number of the compiler driver
1045 program and of the preprocessor and the compiler proper.
1049 Like @option{-v} except the commands are not executed and all command
1050 arguments are quoted. This is useful for shell scripts to capture the
1051 driver-generated command lines.
1055 Use pipes rather than temporary files for communication between the
1056 various stages of compilation. This fails to work on some systems where
1057 the assembler is unable to read from a pipe; but the GNU assembler has
1062 If you are compiling multiple source files, this option tells the driver
1063 to pass all the source files to the compiler at once (for those
1064 languages for which the compiler can handle this). This will allow
1065 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1066 language for which this is supported is C@. If you pass source files for
1067 multiple languages to the driver, using this option, the driver will invoke
1068 the compiler(s) that support IMA once each, passing each compiler all the
1069 source files appropriate for it. For those languages that do not support
1070 IMA this option will be ignored, and the compiler will be invoked once for
1071 each source file in that language. If you use this option in conjunction
1072 with @option{-save-temps}, the compiler will generate multiple
1074 (one for each source file), but only one (combined) @file{.o} or
1079 Print (on the standard output) a description of the command line options
1080 understood by @command{gcc}. If the @option{-v} option is also specified
1081 then @option{--help} will also be passed on to the various processes
1082 invoked by @command{gcc}, so that they can display the command line options
1083 they accept. If the @option{-Wextra} option has also been specified
1084 (prior to the @option{--help} option), then command line options which
1085 have no documentation associated with them will also be displayed.
1088 @opindex target-help
1089 Print (on the standard output) a description of target-specific command
1090 line options for each tool.
1092 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1093 Print (on the standard output) a description of the command line
1094 options understood by the compiler that fit into a specific class.
1095 The class can be one of @var{optimizers}, @var{warnings}, @var{target}
1099 @item @var{optimizers}
1100 This will display all of the optimization options supported by the
1103 @item @var{warnings}
1104 This will display all of the options controlling warning messages
1105 produced by the compiler.
1108 This will display target-specific options. Unlike the
1109 @option{--target-help} option however, target-specific options of the
1110 linker and assembler will not be displayed. This is because those
1111 tools do not currently support the extended @option{--help=} syntax.
1114 This will display the values recognized by the @option{--param}
1118 It is possible to further refine the output of the @option{--help=}
1119 option by adding a comma separated list of qualifiers after the
1120 class. These can be any from the following list:
1124 Display only those options which are undocumented.
1127 Display options which take an argument that appears after an equal
1128 sign in the same continuous piece of text, such as:
1129 @samp{--help=target}.
1132 Display options which take an argument that appears as a separate word
1133 following the original option, such as: @samp{-o output-file}.
1136 Thus for example to display all the undocumented target-specific
1137 switches supported by the compiler the following can be used:
1140 --help=target,undocumented
1143 The sense of a qualifier can be inverted by prefixing it with the
1144 @var{^} character, so for example to display all binary warning
1145 options (i.e. ones that are either on or off and that do not take an
1146 argument), which have a description the following can be used:
1149 --help=warnings,^joined,^undocumented
1152 A class can also be used as a qualifier, although this usually
1153 restricts the output by so much that there is nothing to display. One
1154 case where it does work however is when one of the classes is
1155 @var{target}. So for example to display all the target-specific
1156 optimization options the following can be used:
1159 --help=target,optimizers
1162 The @option{--help=} option can be repeated on the command line. Each
1163 successive use will display its requested class of options, skipping
1164 those that have already been displayed.
1166 If the @option{-Q} option appears on the command line before the
1167 @option{--help=} option, then the descriptive text displayed by
1168 @option{--help=} is changed. Instead of describing the displayed
1169 options, an indication is given as to whether the option is enabled,
1170 disabled or set to a specific value (assuming that the compiler
1171 knows this at the point where the @option{--help=} option is used).
1173 Here is a truncated example from the ARM port of @command{gcc}:
1176 % gcc -Q -mabi=2 --help=target -c
1177 The following options are target specific:
1179 -mabort-on-noreturn [disabled]
1183 The output is sensitive to the effects of previous command line
1184 options, so for example it is possible to find out which optimizations
1185 are enabled at @option{-O2} by using:
1188 -O2 --help=optimizers
1191 Alternatively you can discover which binary optimizations are enabled
1192 by @option{-O3} by using:
1195 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1196 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1197 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1202 Display the version number and copyrights of the invoked GCC@.
1204 @include @value{srcdir}/../libiberty/at-file.texi
1208 @section Compiling C++ Programs
1210 @cindex suffixes for C++ source
1211 @cindex C++ source file suffixes
1212 C++ source files conventionally use one of the suffixes @samp{.C},
1213 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1214 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1215 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1216 files with these names and compiles them as C++ programs even if you
1217 call the compiler the same way as for compiling C programs (usually
1218 with the name @command{gcc}).
1222 However, the use of @command{gcc} does not add the C++ library.
1223 @command{g++} is a program that calls GCC and treats @samp{.c},
1224 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1225 files unless @option{-x} is used, and automatically specifies linking
1226 against the C++ library. This program is also useful when
1227 precompiling a C header file with a @samp{.h} extension for use in C++
1228 compilations. On many systems, @command{g++} is also installed with
1229 the name @command{c++}.
1231 @cindex invoking @command{g++}
1232 When you compile C++ programs, you may specify many of the same
1233 command-line options that you use for compiling programs in any
1234 language; or command-line options meaningful for C and related
1235 languages; or options that are meaningful only for C++ programs.
1236 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1237 explanations of options for languages related to C@.
1238 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1239 explanations of options that are meaningful only for C++ programs.
1241 @node C Dialect Options
1242 @section Options Controlling C Dialect
1243 @cindex dialect options
1244 @cindex language dialect options
1245 @cindex options, dialect
1247 The following options control the dialect of C (or languages derived
1248 from C, such as C++, Objective-C and Objective-C++) that the compiler
1252 @cindex ANSI support
1256 In C mode, support all ISO C90 programs. In C++ mode,
1257 remove GNU extensions that conflict with ISO C++.
1259 This turns off certain features of GCC that are incompatible with ISO
1260 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1261 such as the @code{asm} and @code{typeof} keywords, and
1262 predefined macros such as @code{unix} and @code{vax} that identify the
1263 type of system you are using. It also enables the undesirable and
1264 rarely used ISO trigraph feature. For the C compiler,
1265 it disables recognition of C++ style @samp{//} comments as well as
1266 the @code{inline} keyword.
1268 The alternate keywords @code{__asm__}, @code{__extension__},
1269 @code{__inline__} and @code{__typeof__} continue to work despite
1270 @option{-ansi}. You would not want to use them in an ISO C program, of
1271 course, but it is useful to put them in header files that might be included
1272 in compilations done with @option{-ansi}. Alternate predefined macros
1273 such as @code{__unix__} and @code{__vax__} are also available, with or
1274 without @option{-ansi}.
1276 The @option{-ansi} option does not cause non-ISO programs to be
1277 rejected gratuitously. For that, @option{-pedantic} is required in
1278 addition to @option{-ansi}. @xref{Warning Options}.
1280 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1281 option is used. Some header files may notice this macro and refrain
1282 from declaring certain functions or defining certain macros that the
1283 ISO standard doesn't call for; this is to avoid interfering with any
1284 programs that might use these names for other things.
1286 Functions which would normally be built in but do not have semantics
1287 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1288 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1289 built-in functions provided by GCC}, for details of the functions
1294 Determine the language standard. This option is currently only
1295 supported when compiling C or C++. A value for this option must be
1296 provided; possible values are
1301 ISO C90 (same as @option{-ansi}).
1303 @item iso9899:199409
1304 ISO C90 as modified in amendment 1.
1310 ISO C99. Note that this standard is not yet fully supported; see
1311 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1312 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1315 Default, ISO C90 plus GNU extensions (including some C99 features).
1319 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1320 this will become the default. The name @samp{gnu9x} is deprecated.
1323 The 1998 ISO C++ standard plus amendments.
1326 The same as @option{-std=c++98} plus GNU extensions. This is the
1327 default for C++ code.
1330 The working draft of the upcoming ISO C++0x standard. This option
1331 enables experimental features that are likely to be included in
1332 C++0x. The working draft is constantly changing, and any feature that is
1333 enabled by this flag may be removed from future versions of GCC if it is
1334 not part of the C++0x standard.
1337 The same as @option{-std=c++0x} plus GNU extensions. As with
1338 @option{-std=c++0x}, this option enables experimental features that may
1339 be removed in future versions of GCC.
1342 Even when this option is not specified, you can still use some of the
1343 features of newer standards in so far as they do not conflict with
1344 previous C standards. For example, you may use @code{__restrict__} even
1345 when @option{-std=c99} is not specified.
1347 The @option{-std} options specifying some version of ISO C have the same
1348 effects as @option{-ansi}, except that features that were not in ISO C90
1349 but are in the specified version (for example, @samp{//} comments and
1350 the @code{inline} keyword in ISO C99) are not disabled.
1352 @xref{Standards,,Language Standards Supported by GCC}, for details of
1353 these standard versions.
1355 @item -aux-info @var{filename}
1357 Output to the given filename prototyped declarations for all functions
1358 declared and/or defined in a translation unit, including those in header
1359 files. This option is silently ignored in any language other than C@.
1361 Besides declarations, the file indicates, in comments, the origin of
1362 each declaration (source file and line), whether the declaration was
1363 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1364 @samp{O} for old, respectively, in the first character after the line
1365 number and the colon), and whether it came from a declaration or a
1366 definition (@samp{C} or @samp{F}, respectively, in the following
1367 character). In the case of function definitions, a K&R-style list of
1368 arguments followed by their declarations is also provided, inside
1369 comments, after the declaration.
1373 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1374 keyword, so that code can use these words as identifiers. You can use
1375 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1376 instead. @option{-ansi} implies @option{-fno-asm}.
1378 In C++, this switch only affects the @code{typeof} keyword, since
1379 @code{asm} and @code{inline} are standard keywords. You may want to
1380 use the @option{-fno-gnu-keywords} flag instead, which has the same
1381 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1382 switch only affects the @code{asm} and @code{typeof} keywords, since
1383 @code{inline} is a standard keyword in ISO C99.
1386 @itemx -fno-builtin-@var{function}
1387 @opindex fno-builtin
1388 @cindex built-in functions
1389 Don't recognize built-in functions that do not begin with
1390 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1391 functions provided by GCC}, for details of the functions affected,
1392 including those which are not built-in functions when @option{-ansi} or
1393 @option{-std} options for strict ISO C conformance are used because they
1394 do not have an ISO standard meaning.
1396 GCC normally generates special code to handle certain built-in functions
1397 more efficiently; for instance, calls to @code{alloca} may become single
1398 instructions that adjust the stack directly, and calls to @code{memcpy}
1399 may become inline copy loops. The resulting code is often both smaller
1400 and faster, but since the function calls no longer appear as such, you
1401 cannot set a breakpoint on those calls, nor can you change the behavior
1402 of the functions by linking with a different library. In addition,
1403 when a function is recognized as a built-in function, GCC may use
1404 information about that function to warn about problems with calls to
1405 that function, or to generate more efficient code, even if the
1406 resulting code still contains calls to that function. For example,
1407 warnings are given with @option{-Wformat} for bad calls to
1408 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1409 known not to modify global memory.
1411 With the @option{-fno-builtin-@var{function}} option
1412 only the built-in function @var{function} is
1413 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1414 function is named this is not built-in in this version of GCC, this
1415 option is ignored. There is no corresponding
1416 @option{-fbuiltin-@var{function}} option; if you wish to enable
1417 built-in functions selectively when using @option{-fno-builtin} or
1418 @option{-ffreestanding}, you may define macros such as:
1421 #define abs(n) __builtin_abs ((n))
1422 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1427 @cindex hosted environment
1429 Assert that compilation takes place in a hosted environment. This implies
1430 @option{-fbuiltin}. A hosted environment is one in which the
1431 entire standard library is available, and in which @code{main} has a return
1432 type of @code{int}. Examples are nearly everything except a kernel.
1433 This is equivalent to @option{-fno-freestanding}.
1435 @item -ffreestanding
1436 @opindex ffreestanding
1437 @cindex hosted environment
1439 Assert that compilation takes place in a freestanding environment. This
1440 implies @option{-fno-builtin}. A freestanding environment
1441 is one in which the standard library may not exist, and program startup may
1442 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1443 This is equivalent to @option{-fno-hosted}.
1445 @xref{Standards,,Language Standards Supported by GCC}, for details of
1446 freestanding and hosted environments.
1450 @cindex openmp parallel
1451 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1452 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1453 compiler generates parallel code according to the OpenMP Application
1454 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1456 @item -fms-extensions
1457 @opindex fms-extensions
1458 Accept some non-standard constructs used in Microsoft header files.
1460 Some cases of unnamed fields in structures and unions are only
1461 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1462 fields within structs/unions}, for details.
1466 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1467 options for strict ISO C conformance) implies @option{-trigraphs}.
1469 @item -no-integrated-cpp
1470 @opindex no-integrated-cpp
1471 Performs a compilation in two passes: preprocessing and compiling. This
1472 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1473 @option{-B} option. The user supplied compilation step can then add in
1474 an additional preprocessing step after normal preprocessing but before
1475 compiling. The default is to use the integrated cpp (internal cpp)
1477 The semantics of this option will change if "cc1", "cc1plus", and
1478 "cc1obj" are merged.
1480 @cindex traditional C language
1481 @cindex C language, traditional
1483 @itemx -traditional-cpp
1484 @opindex traditional-cpp
1485 @opindex traditional
1486 Formerly, these options caused GCC to attempt to emulate a pre-standard
1487 C compiler. They are now only supported with the @option{-E} switch.
1488 The preprocessor continues to support a pre-standard mode. See the GNU
1489 CPP manual for details.
1491 @item -fcond-mismatch
1492 @opindex fcond-mismatch
1493 Allow conditional expressions with mismatched types in the second and
1494 third arguments. The value of such an expression is void. This option
1495 is not supported for C++.
1497 @item -flax-vector-conversions
1498 @opindex flax-vector-conversions
1499 Allow implicit conversions between vectors with differing numbers of
1500 elements and/or incompatible element types. This option should not be
1503 @item -funsigned-char
1504 @opindex funsigned-char
1505 Let the type @code{char} be unsigned, like @code{unsigned char}.
1507 Each kind of machine has a default for what @code{char} should
1508 be. It is either like @code{unsigned char} by default or like
1509 @code{signed char} by default.
1511 Ideally, a portable program should always use @code{signed char} or
1512 @code{unsigned char} when it depends on the signedness of an object.
1513 But many programs have been written to use plain @code{char} and
1514 expect it to be signed, or expect it to be unsigned, depending on the
1515 machines they were written for. This option, and its inverse, let you
1516 make such a program work with the opposite default.
1518 The type @code{char} is always a distinct type from each of
1519 @code{signed char} or @code{unsigned char}, even though its behavior
1520 is always just like one of those two.
1523 @opindex fsigned-char
1524 Let the type @code{char} be signed, like @code{signed char}.
1526 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1527 the negative form of @option{-funsigned-char}. Likewise, the option
1528 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1530 @item -fsigned-bitfields
1531 @itemx -funsigned-bitfields
1532 @itemx -fno-signed-bitfields
1533 @itemx -fno-unsigned-bitfields
1534 @opindex fsigned-bitfields
1535 @opindex funsigned-bitfields
1536 @opindex fno-signed-bitfields
1537 @opindex fno-unsigned-bitfields
1538 These options control whether a bit-field is signed or unsigned, when the
1539 declaration does not use either @code{signed} or @code{unsigned}. By
1540 default, such a bit-field is signed, because this is consistent: the
1541 basic integer types such as @code{int} are signed types.
1544 @node C++ Dialect Options
1545 @section Options Controlling C++ Dialect
1547 @cindex compiler options, C++
1548 @cindex C++ options, command line
1549 @cindex options, C++
1550 This section describes the command-line options that are only meaningful
1551 for C++ programs; but you can also use most of the GNU compiler options
1552 regardless of what language your program is in. For example, you
1553 might compile a file @code{firstClass.C} like this:
1556 g++ -g -frepo -O -c firstClass.C
1560 In this example, only @option{-frepo} is an option meant
1561 only for C++ programs; you can use the other options with any
1562 language supported by GCC@.
1564 Here is a list of options that are @emph{only} for compiling C++ programs:
1568 @item -fabi-version=@var{n}
1569 @opindex fabi-version
1570 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1571 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1572 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1573 the version that conforms most closely to the C++ ABI specification.
1574 Therefore, the ABI obtained using version 0 will change as ABI bugs
1577 The default is version 2.
1579 @item -fno-access-control
1580 @opindex fno-access-control
1581 Turn off all access checking. This switch is mainly useful for working
1582 around bugs in the access control code.
1586 Check that the pointer returned by @code{operator new} is non-null
1587 before attempting to modify the storage allocated. This check is
1588 normally unnecessary because the C++ standard specifies that
1589 @code{operator new} will only return @code{0} if it is declared
1590 @samp{throw()}, in which case the compiler will always check the
1591 return value even without this option. In all other cases, when
1592 @code{operator new} has a non-empty exception specification, memory
1593 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1594 @samp{new (nothrow)}.
1596 @item -fconserve-space
1597 @opindex fconserve-space
1598 Put uninitialized or runtime-initialized global variables into the
1599 common segment, as C does. This saves space in the executable at the
1600 cost of not diagnosing duplicate definitions. If you compile with this
1601 flag and your program mysteriously crashes after @code{main()} has
1602 completed, you may have an object that is being destroyed twice because
1603 two definitions were merged.
1605 This option is no longer useful on most targets, now that support has
1606 been added for putting variables into BSS without making them common.
1608 @item -ffriend-injection
1609 @opindex ffriend-injection
1610 Inject friend functions into the enclosing namespace, so that they are
1611 visible outside the scope of the class in which they are declared.
1612 Friend functions were documented to work this way in the old Annotated
1613 C++ Reference Manual, and versions of G++ before 4.1 always worked
1614 that way. However, in ISO C++ a friend function which is not declared
1615 in an enclosing scope can only be found using argument dependent
1616 lookup. This option causes friends to be injected as they were in
1619 This option is for compatibility, and may be removed in a future
1622 @item -fno-elide-constructors
1623 @opindex fno-elide-constructors
1624 The C++ standard allows an implementation to omit creating a temporary
1625 which is only used to initialize another object of the same type.
1626 Specifying this option disables that optimization, and forces G++ to
1627 call the copy constructor in all cases.
1629 @item -fno-enforce-eh-specs
1630 @opindex fno-enforce-eh-specs
1631 Don't generate code to check for violation of exception specifications
1632 at runtime. This option violates the C++ standard, but may be useful
1633 for reducing code size in production builds, much like defining
1634 @samp{NDEBUG}. This does not give user code permission to throw
1635 exceptions in violation of the exception specifications; the compiler
1636 will still optimize based on the specifications, so throwing an
1637 unexpected exception will result in undefined behavior.
1640 @itemx -fno-for-scope
1642 @opindex fno-for-scope
1643 If @option{-ffor-scope} is specified, the scope of variables declared in
1644 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1645 as specified by the C++ standard.
1646 If @option{-fno-for-scope} is specified, the scope of variables declared in
1647 a @i{for-init-statement} extends to the end of the enclosing scope,
1648 as was the case in old versions of G++, and other (traditional)
1649 implementations of C++.
1651 The default if neither flag is given to follow the standard,
1652 but to allow and give a warning for old-style code that would
1653 otherwise be invalid, or have different behavior.
1655 @item -fno-gnu-keywords
1656 @opindex fno-gnu-keywords
1657 Do not recognize @code{typeof} as a keyword, so that code can use this
1658 word as an identifier. You can use the keyword @code{__typeof__} instead.
1659 @option{-ansi} implies @option{-fno-gnu-keywords}.
1661 @item -fno-implicit-templates
1662 @opindex fno-implicit-templates
1663 Never emit code for non-inline templates which are instantiated
1664 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1665 @xref{Template Instantiation}, for more information.
1667 @item -fno-implicit-inline-templates
1668 @opindex fno-implicit-inline-templates
1669 Don't emit code for implicit instantiations of inline templates, either.
1670 The default is to handle inlines differently so that compiles with and
1671 without optimization will need the same set of explicit instantiations.
1673 @item -fno-implement-inlines
1674 @opindex fno-implement-inlines
1675 To save space, do not emit out-of-line copies of inline functions
1676 controlled by @samp{#pragma implementation}. This will cause linker
1677 errors if these functions are not inlined everywhere they are called.
1679 @item -fms-extensions
1680 @opindex fms-extensions
1681 Disable pedantic warnings about constructs used in MFC, such as implicit
1682 int and getting a pointer to member function via non-standard syntax.
1684 @item -fno-nonansi-builtins
1685 @opindex fno-nonansi-builtins
1686 Disable built-in declarations of functions that are not mandated by
1687 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1688 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1690 @item -fno-operator-names
1691 @opindex fno-operator-names
1692 Do not treat the operator name keywords @code{and}, @code{bitand},
1693 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1694 synonyms as keywords.
1696 @item -fno-optional-diags
1697 @opindex fno-optional-diags
1698 Disable diagnostics that the standard says a compiler does not need to
1699 issue. Currently, the only such diagnostic issued by G++ is the one for
1700 a name having multiple meanings within a class.
1703 @opindex fpermissive
1704 Downgrade some diagnostics about nonconformant code from errors to
1705 warnings. Thus, using @option{-fpermissive} will allow some
1706 nonconforming code to compile.
1710 Enable automatic template instantiation at link time. This option also
1711 implies @option{-fno-implicit-templates}. @xref{Template
1712 Instantiation}, for more information.
1716 Disable generation of information about every class with virtual
1717 functions for use by the C++ runtime type identification features
1718 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1719 of the language, you can save some space by using this flag. Note that
1720 exception handling uses the same information, but it will generate it as
1721 needed. The @samp{dynamic_cast} operator can still be used for casts that
1722 do not require runtime type information, i.e. casts to @code{void *} or to
1723 unambiguous base classes.
1727 Emit statistics about front-end processing at the end of the compilation.
1728 This information is generally only useful to the G++ development team.
1730 @item -ftemplate-depth-@var{n}
1731 @opindex ftemplate-depth
1732 Set the maximum instantiation depth for template classes to @var{n}.
1733 A limit on the template instantiation depth is needed to detect
1734 endless recursions during template class instantiation. ANSI/ISO C++
1735 conforming programs must not rely on a maximum depth greater than 17.
1737 @item -fno-threadsafe-statics
1738 @opindex fno-threadsafe-statics
1739 Do not emit the extra code to use the routines specified in the C++
1740 ABI for thread-safe initialization of local statics. You can use this
1741 option to reduce code size slightly in code that doesn't need to be
1744 @item -fuse-cxa-atexit
1745 @opindex fuse-cxa-atexit
1746 Register destructors for objects with static storage duration with the
1747 @code{__cxa_atexit} function rather than the @code{atexit} function.
1748 This option is required for fully standards-compliant handling of static
1749 destructors, but will only work if your C library supports
1750 @code{__cxa_atexit}.
1752 @item -fno-use-cxa-get-exception-ptr
1753 @opindex fno-use-cxa-get-exception-ptr
1754 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1755 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1756 if the runtime routine is not available.
1758 @item -fvisibility-inlines-hidden
1759 @opindex fvisibility-inlines-hidden
1760 This switch declares that the user does not attempt to compare
1761 pointers to inline methods where the addresses of the two functions
1762 were taken in different shared objects.
1764 The effect of this is that GCC may, effectively, mark inline methods with
1765 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1766 appear in the export table of a DSO and do not require a PLT indirection
1767 when used within the DSO@. Enabling this option can have a dramatic effect
1768 on load and link times of a DSO as it massively reduces the size of the
1769 dynamic export table when the library makes heavy use of templates.
1771 The behavior of this switch is not quite the same as marking the
1772 methods as hidden directly, because it does not affect static variables
1773 local to the function or cause the compiler to deduce that
1774 the function is defined in only one shared object.
1776 You may mark a method as having a visibility explicitly to negate the
1777 effect of the switch for that method. For example, if you do want to
1778 compare pointers to a particular inline method, you might mark it as
1779 having default visibility. Marking the enclosing class with explicit
1780 visibility will have no effect.
1782 Explicitly instantiated inline methods are unaffected by this option
1783 as their linkage might otherwise cross a shared library boundary.
1784 @xref{Template Instantiation}.
1788 Do not use weak symbol support, even if it is provided by the linker.
1789 By default, G++ will use weak symbols if they are available. This
1790 option exists only for testing, and should not be used by end-users;
1791 it will result in inferior code and has no benefits. This option may
1792 be removed in a future release of G++.
1796 Do not search for header files in the standard directories specific to
1797 C++, but do still search the other standard directories. (This option
1798 is used when building the C++ library.)
1801 In addition, these optimization, warning, and code generation options
1802 have meanings only for C++ programs:
1805 @item -fno-default-inline
1806 @opindex fno-default-inline
1807 Do not assume @samp{inline} for functions defined inside a class scope.
1808 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1809 functions will have linkage like inline functions; they just won't be
1812 @item -Wabi @r{(C++ only)}
1814 Warn when G++ generates code that is probably not compatible with the
1815 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1816 all such cases, there are probably some cases that are not warned about,
1817 even though G++ is generating incompatible code. There may also be
1818 cases where warnings are emitted even though the code that is generated
1821 You should rewrite your code to avoid these warnings if you are
1822 concerned about the fact that code generated by G++ may not be binary
1823 compatible with code generated by other compilers.
1825 The known incompatibilities at this point include:
1830 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1831 pack data into the same byte as a base class. For example:
1834 struct A @{ virtual void f(); int f1 : 1; @};
1835 struct B : public A @{ int f2 : 1; @};
1839 In this case, G++ will place @code{B::f2} into the same byte
1840 as@code{A::f1}; other compilers will not. You can avoid this problem
1841 by explicitly padding @code{A} so that its size is a multiple of the
1842 byte size on your platform; that will cause G++ and other compilers to
1843 layout @code{B} identically.
1846 Incorrect handling of tail-padding for virtual bases. G++ does not use
1847 tail padding when laying out virtual bases. For example:
1850 struct A @{ virtual void f(); char c1; @};
1851 struct B @{ B(); char c2; @};
1852 struct C : public A, public virtual B @{@};
1856 In this case, G++ will not place @code{B} into the tail-padding for
1857 @code{A}; other compilers will. You can avoid this problem by
1858 explicitly padding @code{A} so that its size is a multiple of its
1859 alignment (ignoring virtual base classes); that will cause G++ and other
1860 compilers to layout @code{C} identically.
1863 Incorrect handling of bit-fields with declared widths greater than that
1864 of their underlying types, when the bit-fields appear in a union. For
1868 union U @{ int i : 4096; @};
1872 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1873 union too small by the number of bits in an @code{int}.
1876 Empty classes can be placed at incorrect offsets. For example:
1886 struct C : public B, public A @{@};
1890 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1891 it should be placed at offset zero. G++ mistakenly believes that the
1892 @code{A} data member of @code{B} is already at offset zero.
1895 Names of template functions whose types involve @code{typename} or
1896 template template parameters can be mangled incorrectly.
1899 template <typename Q>
1900 void f(typename Q::X) @{@}
1902 template <template <typename> class Q>
1903 void f(typename Q<int>::X) @{@}
1907 Instantiations of these templates may be mangled incorrectly.
1911 @item -Wctor-dtor-privacy @r{(C++ only)}
1912 @opindex Wctor-dtor-privacy
1913 Warn when a class seems unusable because all the constructors or
1914 destructors in that class are private, and it has neither friends nor
1915 public static member functions.
1917 @item -Wnon-virtual-dtor @r{(C++ only)}
1918 @opindex Wnon-virtual-dtor
1919 Warn when a class appears to be polymorphic, thereby requiring a virtual
1920 destructor, yet it declares a non-virtual one. This warning is also
1921 enabled if -Weffc++ is specified.
1923 @item -Wreorder @r{(C++ only)}
1925 @cindex reordering, warning
1926 @cindex warning for reordering of member initializers
1927 Warn when the order of member initializers given in the code does not
1928 match the order in which they must be executed. For instance:
1934 A(): j (0), i (1) @{ @}
1938 The compiler will rearrange the member initializers for @samp{i}
1939 and @samp{j} to match the declaration order of the members, emitting
1940 a warning to that effect. This warning is enabled by @option{-Wall}.
1943 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1946 @item -Weffc++ @r{(C++ only)}
1948 Warn about violations of the following style guidelines from Scott Meyers'
1949 @cite{Effective C++} book:
1953 Item 11: Define a copy constructor and an assignment operator for classes
1954 with dynamically allocated memory.
1957 Item 12: Prefer initialization to assignment in constructors.
1960 Item 14: Make destructors virtual in base classes.
1963 Item 15: Have @code{operator=} return a reference to @code{*this}.
1966 Item 23: Don't try to return a reference when you must return an object.
1970 Also warn about violations of the following style guidelines from
1971 Scott Meyers' @cite{More Effective C++} book:
1975 Item 6: Distinguish between prefix and postfix forms of increment and
1976 decrement operators.
1979 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1983 When selecting this option, be aware that the standard library
1984 headers do not obey all of these guidelines; use @samp{grep -v}
1985 to filter out those warnings.
1987 @item -Wno-deprecated @r{(C++ only)}
1988 @opindex Wno-deprecated
1989 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1991 @item -Wstrict-null-sentinel @r{(C++ only)}
1992 @opindex Wstrict-null-sentinel
1993 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1994 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1995 to @code{__null}. Although it is a null pointer constant not a null pointer,
1996 it is guaranteed to of the same size as a pointer. But this use is
1997 not portable across different compilers.
1999 @item -Wno-non-template-friend @r{(C++ only)}
2000 @opindex Wno-non-template-friend
2001 Disable warnings when non-templatized friend functions are declared
2002 within a template. Since the advent of explicit template specification
2003 support in G++, if the name of the friend is an unqualified-id (i.e.,
2004 @samp{friend foo(int)}), the C++ language specification demands that the
2005 friend declare or define an ordinary, nontemplate function. (Section
2006 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2007 could be interpreted as a particular specialization of a templatized
2008 function. Because this non-conforming behavior is no longer the default
2009 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2010 check existing code for potential trouble spots and is on by default.
2011 This new compiler behavior can be turned off with
2012 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2013 but disables the helpful warning.
2015 @item -Wold-style-cast @r{(C++ only)}
2016 @opindex Wold-style-cast
2017 Warn if an old-style (C-style) cast to a non-void type is used within
2018 a C++ program. The new-style casts (@samp{dynamic_cast},
2019 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2020 less vulnerable to unintended effects and much easier to search for.
2022 @item -Woverloaded-virtual @r{(C++ only)}
2023 @opindex Woverloaded-virtual
2024 @cindex overloaded virtual fn, warning
2025 @cindex warning for overloaded virtual fn
2026 Warn when a function declaration hides virtual functions from a
2027 base class. For example, in:
2034 struct B: public A @{
2039 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2047 will fail to compile.
2049 @item -Wno-pmf-conversions @r{(C++ only)}
2050 @opindex Wno-pmf-conversions
2051 Disable the diagnostic for converting a bound pointer to member function
2054 @item -Wsign-promo @r{(C++ only)}
2055 @opindex Wsign-promo
2056 Warn when overload resolution chooses a promotion from unsigned or
2057 enumerated type to a signed type, over a conversion to an unsigned type of
2058 the same size. Previous versions of G++ would try to preserve
2059 unsignedness, but the standard mandates the current behavior.
2064 A& operator = (int);
2074 In this example, G++ will synthesize a default @samp{A& operator =
2075 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2078 @node Objective-C and Objective-C++ Dialect Options
2079 @section Options Controlling Objective-C and Objective-C++ Dialects
2081 @cindex compiler options, Objective-C and Objective-C++
2082 @cindex Objective-C and Objective-C++ options, command line
2083 @cindex options, Objective-C and Objective-C++
2084 (NOTE: This manual does not describe the Objective-C and Objective-C++
2085 languages themselves. See @xref{Standards,,Language Standards
2086 Supported by GCC}, for references.)
2088 This section describes the command-line options that are only meaningful
2089 for Objective-C and Objective-C++ programs, but you can also use most of
2090 the language-independent GNU compiler options.
2091 For example, you might compile a file @code{some_class.m} like this:
2094 gcc -g -fgnu-runtime -O -c some_class.m
2098 In this example, @option{-fgnu-runtime} is an option meant only for
2099 Objective-C and Objective-C++ programs; you can use the other options with
2100 any language supported by GCC@.
2102 Note that since Objective-C is an extension of the C language, Objective-C
2103 compilations may also use options specific to the C front-end (e.g.,
2104 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2105 C++-specific options (e.g., @option{-Wabi}).
2107 Here is a list of options that are @emph{only} for compiling Objective-C
2108 and Objective-C++ programs:
2111 @item -fconstant-string-class=@var{class-name}
2112 @opindex fconstant-string-class
2113 Use @var{class-name} as the name of the class to instantiate for each
2114 literal string specified with the syntax @code{@@"@dots{}"}. The default
2115 class name is @code{NXConstantString} if the GNU runtime is being used, and
2116 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2117 @option{-fconstant-cfstrings} option, if also present, will override the
2118 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2119 to be laid out as constant CoreFoundation strings.
2122 @opindex fgnu-runtime
2123 Generate object code compatible with the standard GNU Objective-C
2124 runtime. This is the default for most types of systems.
2126 @item -fnext-runtime
2127 @opindex fnext-runtime
2128 Generate output compatible with the NeXT runtime. This is the default
2129 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2130 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2133 @item -fno-nil-receivers
2134 @opindex fno-nil-receivers
2135 Assume that all Objective-C message dispatches (e.g.,
2136 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2137 is not @code{nil}. This allows for more efficient entry points in the runtime
2138 to be used. Currently, this option is only available in conjunction with
2139 the NeXT runtime on Mac OS X 10.3 and later.
2141 @item -fobjc-call-cxx-cdtors
2142 @opindex fobjc-call-cxx-cdtors
2143 For each Objective-C class, check if any of its instance variables is a
2144 C++ object with a non-trivial default constructor. If so, synthesize a
2145 special @code{- (id) .cxx_construct} instance method that will run
2146 non-trivial default constructors on any such instance variables, in order,
2147 and then return @code{self}. Similarly, check if any instance variable
2148 is a C++ object with a non-trivial destructor, and if so, synthesize a
2149 special @code{- (void) .cxx_destruct} method that will run
2150 all such default destructors, in reverse order.
2152 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2153 thusly generated will only operate on instance variables declared in the
2154 current Objective-C class, and not those inherited from superclasses. It
2155 is the responsibility of the Objective-C runtime to invoke all such methods
2156 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2157 will be invoked by the runtime immediately after a new object
2158 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2159 be invoked immediately before the runtime deallocates an object instance.
2161 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2162 support for invoking the @code{- (id) .cxx_construct} and
2163 @code{- (void) .cxx_destruct} methods.
2165 @item -fobjc-direct-dispatch
2166 @opindex fobjc-direct-dispatch
2167 Allow fast jumps to the message dispatcher. On Darwin this is
2168 accomplished via the comm page.
2170 @item -fobjc-exceptions
2171 @opindex fobjc-exceptions
2172 Enable syntactic support for structured exception handling in Objective-C,
2173 similar to what is offered by C++ and Java. This option is
2174 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2183 @@catch (AnObjCClass *exc) @{
2190 @@catch (AnotherClass *exc) @{
2193 @@catch (id allOthers) @{
2203 The @code{@@throw} statement may appear anywhere in an Objective-C or
2204 Objective-C++ program; when used inside of a @code{@@catch} block, the
2205 @code{@@throw} may appear without an argument (as shown above), in which case
2206 the object caught by the @code{@@catch} will be rethrown.
2208 Note that only (pointers to) Objective-C objects may be thrown and
2209 caught using this scheme. When an object is thrown, it will be caught
2210 by the nearest @code{@@catch} clause capable of handling objects of that type,
2211 analogously to how @code{catch} blocks work in C++ and Java. A
2212 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2213 any and all Objective-C exceptions not caught by previous @code{@@catch}
2216 The @code{@@finally} clause, if present, will be executed upon exit from the
2217 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2218 regardless of whether any exceptions are thrown, caught or rethrown
2219 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2220 of the @code{finally} clause in Java.
2222 There are several caveats to using the new exception mechanism:
2226 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2227 idioms provided by the @code{NSException} class, the new
2228 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2229 systems, due to additional functionality needed in the (NeXT) Objective-C
2233 As mentioned above, the new exceptions do not support handling
2234 types other than Objective-C objects. Furthermore, when used from
2235 Objective-C++, the Objective-C exception model does not interoperate with C++
2236 exceptions at this time. This means you cannot @code{@@throw} an exception
2237 from Objective-C and @code{catch} it in C++, or vice versa
2238 (i.e., @code{throw @dots{} @@catch}).
2241 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2242 blocks for thread-safe execution:
2245 @@synchronized (ObjCClass *guard) @{
2250 Upon entering the @code{@@synchronized} block, a thread of execution shall
2251 first check whether a lock has been placed on the corresponding @code{guard}
2252 object by another thread. If it has, the current thread shall wait until
2253 the other thread relinquishes its lock. Once @code{guard} becomes available,
2254 the current thread will place its own lock on it, execute the code contained in
2255 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2256 making @code{guard} available to other threads).
2258 Unlike Java, Objective-C does not allow for entire methods to be marked
2259 @code{@@synchronized}. Note that throwing exceptions out of
2260 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2261 to be unlocked properly.
2265 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2267 @item -freplace-objc-classes
2268 @opindex freplace-objc-classes
2269 Emit a special marker instructing @command{ld(1)} not to statically link in
2270 the resulting object file, and allow @command{dyld(1)} to load it in at
2271 run time instead. This is used in conjunction with the Fix-and-Continue
2272 debugging mode, where the object file in question may be recompiled and
2273 dynamically reloaded in the course of program execution, without the need
2274 to restart the program itself. Currently, Fix-and-Continue functionality
2275 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2280 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2281 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2282 compile time) with static class references that get initialized at load time,
2283 which improves run-time performance. Specifying the @option{-fzero-link} flag
2284 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2285 to be retained. This is useful in Zero-Link debugging mode, since it allows
2286 for individual class implementations to be modified during program execution.
2290 Dump interface declarations for all classes seen in the source file to a
2291 file named @file{@var{sourcename}.decl}.
2293 @item -Wassign-intercept
2294 @opindex Wassign-intercept
2295 Warn whenever an Objective-C assignment is being intercepted by the
2299 @opindex Wno-protocol
2300 If a class is declared to implement a protocol, a warning is issued for
2301 every method in the protocol that is not implemented by the class. The
2302 default behavior is to issue a warning for every method not explicitly
2303 implemented in the class, even if a method implementation is inherited
2304 from the superclass. If you use the @option{-Wno-protocol} option, then
2305 methods inherited from the superclass are considered to be implemented,
2306 and no warning is issued for them.
2310 Warn if multiple methods of different types for the same selector are
2311 found during compilation. The check is performed on the list of methods
2312 in the final stage of compilation. Additionally, a check is performed
2313 for each selector appearing in a @code{@@selector(@dots{})}
2314 expression, and a corresponding method for that selector has been found
2315 during compilation. Because these checks scan the method table only at
2316 the end of compilation, these warnings are not produced if the final
2317 stage of compilation is not reached, for example because an error is
2318 found during compilation, or because the @option{-fsyntax-only} option is
2321 @item -Wstrict-selector-match
2322 @opindex Wstrict-selector-match
2323 Warn if multiple methods with differing argument and/or return types are
2324 found for a given selector when attempting to send a message using this
2325 selector to a receiver of type @code{id} or @code{Class}. When this flag
2326 is off (which is the default behavior), the compiler will omit such warnings
2327 if any differences found are confined to types which share the same size
2330 @item -Wundeclared-selector
2331 @opindex Wundeclared-selector
2332 Warn if a @code{@@selector(@dots{})} expression referring to an
2333 undeclared selector is found. A selector is considered undeclared if no
2334 method with that name has been declared before the
2335 @code{@@selector(@dots{})} expression, either explicitly in an
2336 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2337 an @code{@@implementation} section. This option always performs its
2338 checks as soon as a @code{@@selector(@dots{})} expression is found,
2339 while @option{-Wselector} only performs its checks in the final stage of
2340 compilation. This also enforces the coding style convention
2341 that methods and selectors must be declared before being used.
2343 @item -print-objc-runtime-info
2344 @opindex print-objc-runtime-info
2345 Generate C header describing the largest structure that is passed by
2350 @node Language Independent Options
2351 @section Options to Control Diagnostic Messages Formatting
2352 @cindex options to control diagnostics formatting
2353 @cindex diagnostic messages
2354 @cindex message formatting
2356 Traditionally, diagnostic messages have been formatted irrespective of
2357 the output device's aspect (e.g.@: its width, @dots{}). The options described
2358 below can be used to control the diagnostic messages formatting
2359 algorithm, e.g.@: how many characters per line, how often source location
2360 information should be reported. Right now, only the C++ front end can
2361 honor these options. However it is expected, in the near future, that
2362 the remaining front ends would be able to digest them correctly.
2365 @item -fmessage-length=@var{n}
2366 @opindex fmessage-length
2367 Try to format error messages so that they fit on lines of about @var{n}
2368 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2369 the front ends supported by GCC@. If @var{n} is zero, then no
2370 line-wrapping will be done; each error message will appear on a single
2373 @opindex fdiagnostics-show-location
2374 @item -fdiagnostics-show-location=once
2375 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2376 reporter to emit @emph{once} source location information; that is, in
2377 case the message is too long to fit on a single physical line and has to
2378 be wrapped, the source location won't be emitted (as prefix) again,
2379 over and over, in subsequent continuation lines. This is the default
2382 @item -fdiagnostics-show-location=every-line
2383 Only meaningful in line-wrapping mode. Instructs the diagnostic
2384 messages reporter to emit the same source location information (as
2385 prefix) for physical lines that result from the process of breaking
2386 a message which is too long to fit on a single line.
2388 @item -fdiagnostics-show-option
2389 @opindex fdiagnostics-show-option
2390 This option instructs the diagnostic machinery to add text to each
2391 diagnostic emitted, which indicates which command line option directly
2392 controls that diagnostic, when such an option is known to the
2393 diagnostic machinery.
2395 @item -Wcoverage-mismatch
2396 @opindex Wcoverage-mismatch
2397 Warn if feedback profiles do not match when using the
2398 @option{-fprofile-use} option.
2399 If a source file was changed between @option{-fprofile-gen} and
2400 @option{-fprofile-use}, the files with the profile feedback can fail
2401 to match the source file and GCC can not use the profile feedback
2402 information. By default, GCC emits an error message in this case.
2403 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2404 error. GCC does not use appropriate feedback profiles, so using this
2405 option can result in poorly optimized code. This option is useful
2406 only in the case of very minor changes such as bug fixes to an
2411 @node Warning Options
2412 @section Options to Request or Suppress Warnings
2413 @cindex options to control warnings
2414 @cindex warning messages
2415 @cindex messages, warning
2416 @cindex suppressing warnings
2418 Warnings are diagnostic messages that report constructions which
2419 are not inherently erroneous but which are risky or suggest there
2420 may have been an error.
2422 You can request many specific warnings with options beginning @samp{-W},
2423 for example @option{-Wimplicit} to request warnings on implicit
2424 declarations. Each of these specific warning options also has a
2425 negative form beginning @samp{-Wno-} to turn off warnings;
2426 for example, @option{-Wno-implicit}. This manual lists only one of the
2427 two forms, whichever is not the default.
2429 The following options control the amount and kinds of warnings produced
2430 by GCC; for further, language-specific options also refer to
2431 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2435 @cindex syntax checking
2437 @opindex fsyntax-only
2438 Check the code for syntax errors, but don't do anything beyond that.
2442 Issue all the warnings demanded by strict ISO C and ISO C++;
2443 reject all programs that use forbidden extensions, and some other
2444 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2445 version of the ISO C standard specified by any @option{-std} option used.
2447 Valid ISO C and ISO C++ programs should compile properly with or without
2448 this option (though a rare few will require @option{-ansi} or a
2449 @option{-std} option specifying the required version of ISO C)@. However,
2450 without this option, certain GNU extensions and traditional C and C++
2451 features are supported as well. With this option, they are rejected.
2453 @option{-pedantic} does not cause warning messages for use of the
2454 alternate keywords whose names begin and end with @samp{__}. Pedantic
2455 warnings are also disabled in the expression that follows
2456 @code{__extension__}. However, only system header files should use
2457 these escape routes; application programs should avoid them.
2458 @xref{Alternate Keywords}.
2460 Some users try to use @option{-pedantic} to check programs for strict ISO
2461 C conformance. They soon find that it does not do quite what they want:
2462 it finds some non-ISO practices, but not all---only those for which
2463 ISO C @emph{requires} a diagnostic, and some others for which
2464 diagnostics have been added.
2466 A feature to report any failure to conform to ISO C might be useful in
2467 some instances, but would require considerable additional work and would
2468 be quite different from @option{-pedantic}. We don't have plans to
2469 support such a feature in the near future.
2471 Where the standard specified with @option{-std} represents a GNU
2472 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2473 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2474 extended dialect is based. Warnings from @option{-pedantic} are given
2475 where they are required by the base standard. (It would not make sense
2476 for such warnings to be given only for features not in the specified GNU
2477 C dialect, since by definition the GNU dialects of C include all
2478 features the compiler supports with the given option, and there would be
2479 nothing to warn about.)
2481 @item -pedantic-errors
2482 @opindex pedantic-errors
2483 Like @option{-pedantic}, except that errors are produced rather than
2488 Inhibit all warning messages.
2492 Inhibit warning messages about the use of @samp{#import}.
2494 @item -Wchar-subscripts
2495 @opindex Wchar-subscripts
2496 Warn if an array subscript has type @code{char}. This is a common cause
2497 of error, as programmers often forget that this type is signed on some
2499 This warning is enabled by @option{-Wall}.
2503 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2504 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2505 This warning is enabled by @option{-Wall}.
2507 @item -Wfatal-errors
2508 @opindex Wfatal-errors
2509 This option causes the compiler to abort compilation on the first error
2510 occurred rather than trying to keep going and printing further error
2515 @opindex ffreestanding
2516 @opindex fno-builtin
2517 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2518 the arguments supplied have types appropriate to the format string
2519 specified, and that the conversions specified in the format string make
2520 sense. This includes standard functions, and others specified by format
2521 attributes (@pxref{Function Attributes}), in the @code{printf},
2522 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2523 not in the C standard) families (or other target-specific families).
2524 Which functions are checked without format attributes having been
2525 specified depends on the standard version selected, and such checks of
2526 functions without the attribute specified are disabled by
2527 @option{-ffreestanding} or @option{-fno-builtin}.
2529 The formats are checked against the format features supported by GNU
2530 libc version 2.2. These include all ISO C90 and C99 features, as well
2531 as features from the Single Unix Specification and some BSD and GNU
2532 extensions. Other library implementations may not support all these
2533 features; GCC does not support warning about features that go beyond a
2534 particular library's limitations. However, if @option{-pedantic} is used
2535 with @option{-Wformat}, warnings will be given about format features not
2536 in the selected standard version (but not for @code{strfmon} formats,
2537 since those are not in any version of the C standard). @xref{C Dialect
2538 Options,,Options Controlling C Dialect}.
2540 Since @option{-Wformat} also checks for null format arguments for
2541 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2543 @option{-Wformat} is included in @option{-Wall}. For more control over some
2544 aspects of format checking, the options @option{-Wformat-y2k},
2545 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2546 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2547 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2550 @opindex Wformat-y2k
2551 If @option{-Wformat} is specified, also warn about @code{strftime}
2552 formats which may yield only a two-digit year.
2554 @item -Wno-format-extra-args
2555 @opindex Wno-format-extra-args
2556 If @option{-Wformat} is specified, do not warn about excess arguments to a
2557 @code{printf} or @code{scanf} format function. The C standard specifies
2558 that such arguments are ignored.
2560 Where the unused arguments lie between used arguments that are
2561 specified with @samp{$} operand number specifications, normally
2562 warnings are still given, since the implementation could not know what
2563 type to pass to @code{va_arg} to skip the unused arguments. However,
2564 in the case of @code{scanf} formats, this option will suppress the
2565 warning if the unused arguments are all pointers, since the Single
2566 Unix Specification says that such unused arguments are allowed.
2568 @item -Wno-format-zero-length
2569 @opindex Wno-format-zero-length
2570 If @option{-Wformat} is specified, do not warn about zero-length formats.
2571 The C standard specifies that zero-length formats are allowed.
2573 @item -Wformat-nonliteral
2574 @opindex Wformat-nonliteral
2575 If @option{-Wformat} is specified, also warn if the format string is not a
2576 string literal and so cannot be checked, unless the format function
2577 takes its format arguments as a @code{va_list}.
2579 @item -Wformat-security
2580 @opindex Wformat-security
2581 If @option{-Wformat} is specified, also warn about uses of format
2582 functions that represent possible security problems. At present, this
2583 warns about calls to @code{printf} and @code{scanf} functions where the
2584 format string is not a string literal and there are no format arguments,
2585 as in @code{printf (foo);}. This may be a security hole if the format
2586 string came from untrusted input and contains @samp{%n}. (This is
2587 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2588 in future warnings may be added to @option{-Wformat-security} that are not
2589 included in @option{-Wformat-nonliteral}.)
2593 Enable @option{-Wformat} plus format checks not included in
2594 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2595 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2599 Warn about passing a null pointer for arguments marked as
2600 requiring a non-null value by the @code{nonnull} function attribute.
2602 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2603 can be disabled with the @option{-Wno-nonnull} option.
2605 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2607 Warn about uninitialized variables which are initialized with themselves.
2608 Note this option can only be used with the @option{-Wuninitialized} option,
2609 which in turn only works with @option{-O1} and above.
2611 For example, GCC will warn about @code{i} being uninitialized in the
2612 following snippet only when @option{-Winit-self} has been specified:
2623 @item -Wimplicit-int
2624 @opindex Wimplicit-int
2625 Warn when a declaration does not specify a type.
2626 This warning is enabled by @option{-Wall}.
2628 @item -Wimplicit-function-declaration
2629 @opindex Wimplicit-function-declaration
2630 @opindex Wno-implicit-function-declaration
2631 Give a warning whenever a function is used before being declared. In
2632 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2633 enabled by default and it is made into an error by
2634 @option{-pedantic-errors}. This warning is also enabled by
2639 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2640 This warning is enabled by @option{-Wall}.
2644 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2645 function with external linkage, returning int, taking either zero
2646 arguments, two, or three arguments of appropriate types.
2647 This warning is enabled by @option{-Wall}.
2649 @item -Wmissing-braces
2650 @opindex Wmissing-braces
2651 Warn if an aggregate or union initializer is not fully bracketed. In
2652 the following example, the initializer for @samp{a} is not fully
2653 bracketed, but that for @samp{b} is fully bracketed.
2656 int a[2][2] = @{ 0, 1, 2, 3 @};
2657 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2660 This warning is enabled by @option{-Wall}.
2662 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2663 @opindex Wmissing-include-dirs
2664 Warn if a user-supplied include directory does not exist.
2667 @opindex Wparentheses
2668 Warn if parentheses are omitted in certain contexts, such
2669 as when there is an assignment in a context where a truth value
2670 is expected, or when operators are nested whose precedence people
2671 often get confused about.
2673 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2674 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2675 interpretation from that of ordinary mathematical notation.
2677 Also warn about constructions where there may be confusion to which
2678 @code{if} statement an @code{else} branch belongs. Here is an example of
2693 In C/C++, every @code{else} branch belongs to the innermost possible
2694 @code{if} statement, which in this example is @code{if (b)}. This is
2695 often not what the programmer expected, as illustrated in the above
2696 example by indentation the programmer chose. When there is the
2697 potential for this confusion, GCC will issue a warning when this flag
2698 is specified. To eliminate the warning, add explicit braces around
2699 the innermost @code{if} statement so there is no way the @code{else}
2700 could belong to the enclosing @code{if}. The resulting code would
2717 This warning is enabled by @option{-Wall}.
2719 @item -Wsequence-point
2720 @opindex Wsequence-point
2721 Warn about code that may have undefined semantics because of violations
2722 of sequence point rules in the C and C++ standards.
2724 The C and C++ standards defines the order in which expressions in a C/C++
2725 program are evaluated in terms of @dfn{sequence points}, which represent
2726 a partial ordering between the execution of parts of the program: those
2727 executed before the sequence point, and those executed after it. These
2728 occur after the evaluation of a full expression (one which is not part
2729 of a larger expression), after the evaluation of the first operand of a
2730 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2731 function is called (but after the evaluation of its arguments and the
2732 expression denoting the called function), and in certain other places.
2733 Other than as expressed by the sequence point rules, the order of
2734 evaluation of subexpressions of an expression is not specified. All
2735 these rules describe only a partial order rather than a total order,
2736 since, for example, if two functions are called within one expression
2737 with no sequence point between them, the order in which the functions
2738 are called is not specified. However, the standards committee have
2739 ruled that function calls do not overlap.
2741 It is not specified when between sequence points modifications to the
2742 values of objects take effect. Programs whose behavior depends on this
2743 have undefined behavior; the C and C++ standards specify that ``Between
2744 the previous and next sequence point an object shall have its stored
2745 value modified at most once by the evaluation of an expression.
2746 Furthermore, the prior value shall be read only to determine the value
2747 to be stored.''. If a program breaks these rules, the results on any
2748 particular implementation are entirely unpredictable.
2750 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2751 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2752 diagnosed by this option, and it may give an occasional false positive
2753 result, but in general it has been found fairly effective at detecting
2754 this sort of problem in programs.
2756 The standard is worded confusingly, therefore there is some debate
2757 over the precise meaning of the sequence point rules in subtle cases.
2758 Links to discussions of the problem, including proposed formal
2759 definitions, may be found on the GCC readings page, at
2760 @w{@uref{http://gcc.gnu.org/readings.html}}.
2762 This warning is enabled by @option{-Wall} for C and C++.
2765 @opindex Wreturn-type
2766 Warn whenever a function is defined with a return-type that defaults to
2767 @code{int}. Also warn about any @code{return} statement with no
2768 return-value in a function whose return-type is not @code{void}.
2770 Also warn if the return type of a function has a type qualifier
2771 such as @code{const}. For ISO C such a type qualifier has no effect,
2772 since the value returned by a function is not an lvalue.
2773 For C++, the warning is only emitted for scalar types or @code{void}.
2774 ISO C prohibits qualified @code{void} return types on function
2775 definitions, so such return types always receive a warning
2776 even without this option.
2778 For C++, a function without return type always produces a diagnostic
2779 message, even when @option{-Wno-return-type} is specified. The only
2780 exceptions are @samp{main} and functions defined in system headers.
2782 This warning is enabled by @option{-Wall}.
2786 Warn whenever a @code{switch} statement has an index of enumerated type
2787 and lacks a @code{case} for one or more of the named codes of that
2788 enumeration. (The presence of a @code{default} label prevents this
2789 warning.) @code{case} labels outside the enumeration range also
2790 provoke warnings when this option is used.
2791 This warning is enabled by @option{-Wall}.
2793 @item -Wswitch-default
2794 @opindex Wswitch-switch
2795 Warn whenever a @code{switch} statement does not have a @code{default}
2799 @opindex Wswitch-enum
2800 Warn whenever a @code{switch} statement has an index of enumerated type
2801 and lacks a @code{case} for one or more of the named codes of that
2802 enumeration. @code{case} labels outside the enumeration range also
2803 provoke warnings when this option is used.
2807 Warn if any trigraphs are encountered that might change the meaning of
2808 the program (trigraphs within comments are not warned about).
2809 This warning is enabled by @option{-Wall}.
2811 @item -Wunused-function
2812 @opindex Wunused-function
2813 Warn whenever a static function is declared but not defined or a
2814 non-inline static function is unused.
2815 This warning is enabled by @option{-Wall}.
2817 @item -Wunused-label
2818 @opindex Wunused-label
2819 Warn whenever a label is declared but not used.
2820 This warning is enabled by @option{-Wall}.
2822 To suppress this warning use the @samp{unused} attribute
2823 (@pxref{Variable Attributes}).
2825 @item -Wunused-parameter
2826 @opindex Wunused-parameter
2827 Warn whenever a function parameter is unused aside from its declaration.
2829 To suppress this warning use the @samp{unused} attribute
2830 (@pxref{Variable Attributes}).
2832 @item -Wunused-variable
2833 @opindex Wunused-variable
2834 Warn whenever a local variable or non-constant static variable is unused
2835 aside from its declaration.
2836 This warning is enabled by @option{-Wall}.
2838 To suppress this warning use the @samp{unused} attribute
2839 (@pxref{Variable Attributes}).
2841 @item -Wunused-value
2842 @opindex Wunused-value
2843 Warn whenever a statement computes a result that is explicitly not
2844 used. To suppress this warning cast the unused expression to
2845 @samp{void}. This includes an expression-statement or the left-hand
2846 side of a comma expression that contains no side effects. For example,
2847 an expression such as @samp{x[i,j]} will cause a warning, while
2848 @samp{x[(void)i,j]} will not.
2850 This warning is enabled by @option{-Wall}.
2854 All the above @option{-Wunused} options combined.
2856 In order to get a warning about an unused function parameter, you must
2857 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2858 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2860 @item -Wuninitialized
2861 @opindex Wuninitialized
2862 Warn if an automatic variable is used without first being initialized or
2863 if a variable may be clobbered by a @code{setjmp} call.
2865 These warnings are possible only in optimizing compilation,
2866 because they require data flow information that is computed only
2867 when optimizing. If you do not specify @option{-O}, you will not get
2868 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2869 requiring @option{-O}.
2871 If you want to warn about code which uses the uninitialized value of the
2872 variable in its own initializer, use the @option{-Winit-self} option.
2874 These warnings occur for individual uninitialized or clobbered
2875 elements of structure, union or array variables as well as for
2876 variables which are uninitialized or clobbered as a whole. They do
2877 not occur for variables or elements declared @code{volatile}. Because
2878 these warnings depend on optimization, the exact variables or elements
2879 for which there are warnings will depend on the precise optimization
2880 options and version of GCC used.
2882 Note that there may be no warning about a variable that is used only
2883 to compute a value that itself is never used, because such
2884 computations may be deleted by data flow analysis before the warnings
2887 These warnings are made optional because GCC is not smart
2888 enough to see all the reasons why the code might be correct
2889 despite appearing to have an error. Here is one example of how
2910 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2911 always initialized, but GCC doesn't know this. Here is
2912 another common case:
2917 if (change_y) save_y = y, y = new_y;
2919 if (change_y) y = save_y;
2924 This has no bug because @code{save_y} is used only if it is set.
2926 @cindex @code{longjmp} warnings
2927 This option also warns when a non-volatile automatic variable might be
2928 changed by a call to @code{longjmp}. These warnings as well are possible
2929 only in optimizing compilation.
2931 The compiler sees only the calls to @code{setjmp}. It cannot know
2932 where @code{longjmp} will be called; in fact, a signal handler could
2933 call it at any point in the code. As a result, you may get a warning
2934 even when there is in fact no problem because @code{longjmp} cannot
2935 in fact be called at the place which would cause a problem.
2937 Some spurious warnings can be avoided if you declare all the functions
2938 you use that never return as @code{noreturn}. @xref{Function
2941 This warning is enabled by @option{-Wall}.
2943 @item -Wunknown-pragmas
2944 @opindex Wunknown-pragmas
2945 @cindex warning for unknown pragmas
2946 @cindex unknown pragmas, warning
2947 @cindex pragmas, warning of unknown
2948 Warn when a #pragma directive is encountered which is not understood by
2949 GCC@. If this command line option is used, warnings will even be issued
2950 for unknown pragmas in system header files. This is not the case if
2951 the warnings were only enabled by the @option{-Wall} command line option.
2954 @opindex Wno-pragmas
2956 Do not warn about misuses of pragmas, such as incorrect parameters,
2957 invalid syntax, or conflicts between pragmas. See also
2958 @samp{-Wunknown-pragmas}.
2960 @item -Wstrict-aliasing
2961 @opindex Wstrict-aliasing
2962 This option is only active when @option{-fstrict-aliasing} is active.
2963 It warns about code which might break the strict aliasing rules that the
2964 compiler is using for optimization. The warning does not catch all
2965 cases, but does attempt to catch the more common pitfalls. It is
2966 included in @option{-Wall}.
2968 @item -Wstrict-aliasing=2
2969 @opindex Wstrict-aliasing=2
2970 This option is only active when @option{-fstrict-aliasing} is active.
2971 It warns about code which might break the strict aliasing rules that the
2972 compiler is using for optimization. This warning catches more cases than
2973 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2974 cases that are safe.
2976 @item -Wstrict-overflow
2977 @item -Wstrict-overflow=@var{n}
2978 @opindex -Wstrict-overflow
2979 This option is only active when @option{-fstrict-overflow} is active.
2980 It warns about cases where the compiler optimizes based on the
2981 assumption that signed overflow does not occur. Note that it does not
2982 warn about all cases where the code might overflow: it only warns
2983 about cases where the compiler implements some optimization. Thus
2984 this warning depends on the optimization level.
2986 An optimization which assumes that signed overflow does not occur is
2987 perfectly safe if the values of the variables involved are such that
2988 overflow never does, in fact, occur. Therefore this warning can
2989 easily give a false positive: a warning about code which is not
2990 actually a problem. To help focus on important issues, several
2991 warning levels are defined.
2994 @item -Wstrict-overflow=1
2995 Warn about cases which are both questionable and easy to avoid. For
2996 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
2997 compiler will simplify this to @code{1}. @option{-Wstrict-overflow}
2998 (with no level) is the same as @option{-Wstrict-overflow=1}. This
2999 level of @option{-Wstrict-overflow} is enabled by @option{-Wall};
3000 higher levels are not, and must be explicitly requested.
3002 @item -Wstrict-overflow=2
3003 Also warn about other cases where a comparison is simplified to a
3004 constant. For example: @code{abs (x) >= 0}. This can only be
3005 simplified when @option{-fstrict-overflow} is in effect, because
3006 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3009 @item -Wstrict-overflow=3
3010 Also warn about other cases where a comparison is simplified. For
3011 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3013 @item -Wstrict-overflow=4
3014 Also warn about other simplifications not covered by the above cases.
3015 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3017 @item -Wstrict-overflow=5
3018 Also warn about cases where the compiler reduces the magnitude of a
3019 constant involved in a comparison. For example: @code{x + 2 > y} will
3020 be simplified to @code{x + 1 >= y}. This is reported only at the
3021 highest warning level because this simplification applies to many
3022 comparisons, so this warning level will give a very large number of
3026 @item -Warray-bounds
3027 @opindex Wno-array-bounds
3028 @opindex Warray-bounds
3029 This option is only active when @option{-ftree-vrp} is active
3030 (default for -O2 and above). It warns about subscripts to arrays
3031 that are always out of bounds. This warning is enabled by @option{-Wall}.
3035 All of the above @samp{-W} options combined. This enables all the
3036 warnings about constructions that some users consider questionable, and
3037 that are easy to avoid (or modify to prevent the warning), even in
3038 conjunction with macros. This also enables some language-specific
3039 warnings described in @ref{C++ Dialect Options} and
3040 @ref{Objective-C and Objective-C++ Dialect Options}.
3043 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
3044 Some of them warn about constructions that users generally do not
3045 consider questionable, but which occasionally you might wish to check
3046 for; others warn about constructions that are necessary or hard to avoid
3047 in some cases, and there is no simple way to modify the code to suppress
3054 (This option used to be called @option{-W}. The older name is still
3055 supported, but the newer name is more descriptive.) Print extra warning
3056 messages for these events:
3060 A function can return either with or without a value. (Falling
3061 off the end of the function body is considered returning without
3062 a value.) For example, this function would evoke such a
3076 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
3079 Storage-class specifiers like @code{static} are not the first things
3080 in a declaration. According to the C Standard, this usage is
3081 obsolescent. This warning can be independently controlled by
3082 @option{-Wold-style-declaration}.
3085 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
3089 A comparison between signed and unsigned values could produce an
3090 incorrect result when the signed value is converted to unsigned.
3091 (But don't warn if @option{-Wno-sign-compare} is also specified.)
3094 An aggregate has an initializer which does not initialize all members.
3095 This warning can be independently controlled by
3096 @option{-Wmissing-field-initializers}.
3099 An initialized field without side effects is overridden when using
3100 designated initializers (@pxref{Designated Inits, , Designated
3101 Initializers}). This warning can be independently controlled by
3102 @option{-Woverride-init}.
3105 A function parameter is declared without a type specifier in K&R-style
3106 functions. This warning can be independently controlled by
3107 @option{-Wmissing-parameter-type}.
3110 An empty body occurs in an @samp{if} or @samp{else} statement. This
3111 warning can be independently controlled by @option{-Wempty-body}.
3114 A pointer is compared against integer zero with @samp{<}, @samp{<=},
3115 @samp{>}, or @samp{>=}.
3118 A variable might be changed by @samp{longjmp} or @samp{vfork}.
3119 This warning can be independently controlled by @option{-Wclobbered}.
3121 @item @r{(C++ only)}
3122 An enumerator and a non-enumerator both appear in a conditional expression.
3124 @item @r{(C++ only)}
3125 A non-static reference or non-static @samp{const} member appears in a
3126 class without constructors.
3128 @item @r{(C++ only)}
3129 Ambiguous virtual bases.
3131 @item @r{(C++ only)}
3132 Subscripting an array which has been declared @samp{register}.
3134 @item @r{(C++ only)}
3135 Taking the address of a variable which has been declared @samp{register}.
3137 @item @r{(C++ only)}
3138 A base class is not initialized in a derived class' copy constructor.
3141 @item -Wno-div-by-zero
3142 @opindex Wno-div-by-zero
3143 @opindex Wdiv-by-zero
3144 Do not warn about compile-time integer division by zero. Floating point
3145 division by zero is not warned about, as it can be a legitimate way of
3146 obtaining infinities and NaNs.
3148 @item -Wsystem-headers
3149 @opindex Wsystem-headers
3150 @cindex warnings from system headers
3151 @cindex system headers, warnings from
3152 Print warning messages for constructs found in system header files.
3153 Warnings from system headers are normally suppressed, on the assumption
3154 that they usually do not indicate real problems and would only make the
3155 compiler output harder to read. Using this command line option tells
3156 GCC to emit warnings from system headers as if they occurred in user
3157 code. However, note that using @option{-Wall} in conjunction with this
3158 option will @emph{not} warn about unknown pragmas in system
3159 headers---for that, @option{-Wunknown-pragmas} must also be used.
3162 @opindex Wfloat-equal
3163 Warn if floating point values are used in equality comparisons.
3165 The idea behind this is that sometimes it is convenient (for the
3166 programmer) to consider floating-point values as approximations to
3167 infinitely precise real numbers. If you are doing this, then you need
3168 to compute (by analyzing the code, or in some other way) the maximum or
3169 likely maximum error that the computation introduces, and allow for it
3170 when performing comparisons (and when producing output, but that's a
3171 different problem). In particular, instead of testing for equality, you
3172 would check to see whether the two values have ranges that overlap; and
3173 this is done with the relational operators, so equality comparisons are
3176 @item -Wtraditional @r{(C only)}
3177 @opindex Wtraditional
3178 Warn about certain constructs that behave differently in traditional and
3179 ISO C@. Also warn about ISO C constructs that have no traditional C
3180 equivalent, and/or problematic constructs which should be avoided.
3184 Macro parameters that appear within string literals in the macro body.
3185 In traditional C macro replacement takes place within string literals,
3186 but does not in ISO C@.
3189 In traditional C, some preprocessor directives did not exist.
3190 Traditional preprocessors would only consider a line to be a directive
3191 if the @samp{#} appeared in column 1 on the line. Therefore
3192 @option{-Wtraditional} warns about directives that traditional C
3193 understands but would ignore because the @samp{#} does not appear as the
3194 first character on the line. It also suggests you hide directives like
3195 @samp{#pragma} not understood by traditional C by indenting them. Some
3196 traditional implementations would not recognize @samp{#elif}, so it
3197 suggests avoiding it altogether.
3200 A function-like macro that appears without arguments.
3203 The unary plus operator.
3206 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3207 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3208 constants.) Note, these suffixes appear in macros defined in the system
3209 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3210 Use of these macros in user code might normally lead to spurious
3211 warnings, however GCC's integrated preprocessor has enough context to
3212 avoid warning in these cases.
3215 A function declared external in one block and then used after the end of
3219 A @code{switch} statement has an operand of type @code{long}.
3222 A non-@code{static} function declaration follows a @code{static} one.
3223 This construct is not accepted by some traditional C compilers.
3226 The ISO type of an integer constant has a different width or
3227 signedness from its traditional type. This warning is only issued if
3228 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3229 typically represent bit patterns, are not warned about.
3232 Usage of ISO string concatenation is detected.
3235 Initialization of automatic aggregates.
3238 Identifier conflicts with labels. Traditional C lacks a separate
3239 namespace for labels.
3242 Initialization of unions. If the initializer is zero, the warning is
3243 omitted. This is done under the assumption that the zero initializer in
3244 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3245 initializer warnings and relies on default initialization to zero in the
3249 Conversions by prototypes between fixed/floating point values and vice
3250 versa. The absence of these prototypes when compiling with traditional
3251 C would cause serious problems. This is a subset of the possible
3252 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3255 Use of ISO C style function definitions. This warning intentionally is
3256 @emph{not} issued for prototype declarations or variadic functions
3257 because these ISO C features will appear in your code when using
3258 libiberty's traditional C compatibility macros, @code{PARAMS} and
3259 @code{VPARAMS}. This warning is also bypassed for nested functions
3260 because that feature is already a GCC extension and thus not relevant to
3261 traditional C compatibility.
3264 @item -Wtraditional-conversion @r{(C only)}
3265 @opindex Wtraditional-conversion
3266 Warn if a prototype causes a type conversion that is different from what
3267 would happen to the same argument in the absence of a prototype. This
3268 includes conversions of fixed point to floating and vice versa, and
3269 conversions changing the width or signedness of a fixed point argument
3270 except when the same as the default promotion.
3272 @item -Wdeclaration-after-statement @r{(C only)}
3273 @opindex Wdeclaration-after-statement
3274 Warn when a declaration is found after a statement in a block. This
3275 construct, known from C++, was introduced with ISO C99 and is by default
3276 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3277 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3281 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3283 @item -Wno-endif-labels
3284 @opindex Wno-endif-labels
3285 @opindex Wendif-labels
3286 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3290 Warn whenever a local variable shadows another local variable, parameter or
3291 global variable or whenever a built-in function is shadowed.
3293 @item -Wlarger-than-@var{len}
3294 @opindex Wlarger-than
3295 Warn whenever an object of larger than @var{len} bytes is defined.
3297 @item -Wunsafe-loop-optimizations
3298 @opindex Wunsafe-loop-optimizations
3299 Warn if the loop cannot be optimized because the compiler could not
3300 assume anything on the bounds of the loop indices. With
3301 @option{-funsafe-loop-optimizations} warn if the compiler made
3304 @item -Wpointer-arith
3305 @opindex Wpointer-arith
3306 Warn about anything that depends on the ``size of'' a function type or
3307 of @code{void}. GNU C assigns these types a size of 1, for
3308 convenience in calculations with @code{void *} pointers and pointers
3309 to functions. In C++, warn also when an arithmetic operation involves
3310 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3312 @item -Wbad-function-cast @r{(C only)}
3313 @opindex Wbad-function-cast
3314 Warn whenever a function call is cast to a non-matching type.
3315 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3318 Warn about ISO C constructs that are outside of the common subset of
3319 ISO C and ISO C++, e.g.@: request for implicit conversion from
3320 @code{void *} to a pointer to non-@code{void} type.
3322 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3323 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3324 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3325 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3329 Warn whenever a pointer is cast so as to remove a type qualifier from
3330 the target type. For example, warn if a @code{const char *} is cast
3331 to an ordinary @code{char *}.
3334 @opindex Wcast-align
3335 Warn whenever a pointer is cast such that the required alignment of the
3336 target is increased. For example, warn if a @code{char *} is cast to
3337 an @code{int *} on machines where integers can only be accessed at
3338 two- or four-byte boundaries.
3340 @item -Wwrite-strings
3341 @opindex Wwrite-strings
3342 When compiling C, give string constants the type @code{const
3343 char[@var{length}]} so that
3344 copying the address of one into a non-@code{const} @code{char *}
3345 pointer will get a warning; when compiling C++, warn about the
3346 deprecated conversion from string literals to @code{char *}. This
3347 warning, by default, is enabled for C++ programs.
3348 These warnings will help you find at
3349 compile time code that can try to write into a string constant, but
3350 only if you have been very careful about using @code{const} in
3351 declarations and prototypes. Otherwise, it will just be a nuisance;
3352 this is why we did not make @option{-Wall} request these warnings.
3356 Warn for variables that might be changed by @samp{longjmp} or
3357 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3360 @opindex Wconversion
3361 Warn for implicit conversions that may alter a value. This includes
3362 conversions between real and integer, like @code{abs (x)} when
3363 @code{x} is @code{double}; conversions between signed and unsigned,
3364 like @code{unsigned ui = -1}; and conversions to smaller types, like
3365 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3366 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3367 changed by the conversion like in @code{abs (2.0)}.
3370 @opindex Wempty-body
3371 An empty body occurs in an @samp{if} or @samp{else} statement.
3372 This warning is also enabled by @option{-Wextra}.
3374 @item -Wsign-compare
3375 @opindex Wsign-compare
3376 @cindex warning for comparison of signed and unsigned values
3377 @cindex comparison of signed and unsigned values, warning
3378 @cindex signed and unsigned values, comparison warning
3379 Warn when a comparison between signed and unsigned values could produce
3380 an incorrect result when the signed value is converted to unsigned.
3381 This warning is also enabled by @option{-Wextra}; to get the other warnings
3382 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3384 @item -Waggregate-return
3385 @opindex Waggregate-return
3386 Warn if any functions that return structures or unions are defined or
3387 called. (In languages where you can return an array, this also elicits
3391 @opindex Walways-true
3392 Warn about comparisons which are always true such as testing if
3393 unsigned values are greater than or equal to zero. This warning is
3394 enabled by @option{-Wall}.
3396 @item -Wno-attributes
3397 @opindex Wno-attributes
3398 @opindex Wattributes
3399 Do not warn if an unexpected @code{__attribute__} is used, such as
3400 unrecognized attributes, function attributes applied to variables,
3401 etc. This will not stop errors for incorrect use of supported
3404 @item -Wstrict-prototypes @r{(C only)}
3405 @opindex Wstrict-prototypes
3406 Warn if a function is declared or defined without specifying the
3407 argument types. (An old-style function definition is permitted without
3408 a warning if preceded by a declaration which specifies the argument
3411 @item -Wold-style-declaration @r{(C only)}
3412 @opindex Wold-style-declaration
3413 Warn for obsolescent usages, according to the C Standard, in a
3414 declaration. For example, warn if storage-class specifiers like
3415 @code{static} are not the first things in a declaration. This warning
3416 is also enabled by @option{-Wextra}.
3418 @item -Wold-style-definition @r{(C only)}
3419 @opindex Wold-style-definition
3420 Warn if an old-style function definition is used. A warning is given
3421 even if there is a previous prototype.
3423 @item -Wmissing-parameter-type @r{(C only)}
3424 @opindex Wmissing-parameter-type
3425 A function parameter is declared without a type specifier in K&R-style
3432 This warning is also enabled by @option{-Wextra}.
3434 @item -Wmissing-prototypes @r{(C only)}
3435 @opindex Wmissing-prototypes
3436 Warn if a global function is defined without a previous prototype
3437 declaration. This warning is issued even if the definition itself
3438 provides a prototype. The aim is to detect global functions that fail
3439 to be declared in header files.
3441 @item -Wmissing-declarations @r{(C and C++ only)}
3442 @opindex Wmissing-declarations
3443 Warn if a global function is defined without a previous declaration.
3444 Do so even if the definition itself provides a prototype.
3445 Use this option to detect global functions that are not declared in
3446 header files. In C++, no warnings are issued for function templates,
3447 or for inline functions, or for functions in anonymous namespaces.
3449 @item -Wmissing-field-initializers
3450 @opindex Wmissing-field-initializers
3453 Warn if a structure's initializer has some fields missing. For
3454 example, the following code would cause such a warning, because
3455 @code{x.h} is implicitly zero:
3458 struct s @{ int f, g, h; @};
3459 struct s x = @{ 3, 4 @};
3462 This option does not warn about designated initializers, so the following
3463 modification would not trigger a warning:
3466 struct s @{ int f, g, h; @};
3467 struct s x = @{ .f = 3, .g = 4 @};
3470 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3471 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3473 @item -Wmissing-noreturn
3474 @opindex Wmissing-noreturn
3475 Warn about functions which might be candidates for attribute @code{noreturn}.
3476 Note these are only possible candidates, not absolute ones. Care should
3477 be taken to manually verify functions actually do not ever return before
3478 adding the @code{noreturn} attribute, otherwise subtle code generation
3479 bugs could be introduced. You will not get a warning for @code{main} in
3480 hosted C environments.
3482 @item -Wmissing-format-attribute
3483 @opindex Wmissing-format-attribute
3485 Warn about function pointers which might be candidates for @code{format}
3486 attributes. Note these are only possible candidates, not absolute ones.
3487 GCC will guess that function pointers with @code{format} attributes that
3488 are used in assignment, initialization, parameter passing or return
3489 statements should have a corresponding @code{format} attribute in the
3490 resulting type. I.e.@: the left-hand side of the assignment or
3491 initialization, the type of the parameter variable, or the return type
3492 of the containing function respectively should also have a @code{format}
3493 attribute to avoid the warning.
3495 GCC will also warn about function definitions which might be
3496 candidates for @code{format} attributes. Again, these are only
3497 possible candidates. GCC will guess that @code{format} attributes
3498 might be appropriate for any function that calls a function like
3499 @code{vprintf} or @code{vscanf}, but this might not always be the
3500 case, and some functions for which @code{format} attributes are
3501 appropriate may not be detected.
3503 @item -Wno-multichar
3504 @opindex Wno-multichar
3506 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3507 Usually they indicate a typo in the user's code, as they have
3508 implementation-defined values, and should not be used in portable code.
3510 @item -Wnormalized=<none|id|nfc|nfkc>
3511 @opindex Wnormalized
3514 @cindex character set, input normalization
3515 In ISO C and ISO C++, two identifiers are different if they are
3516 different sequences of characters. However, sometimes when characters
3517 outside the basic ASCII character set are used, you can have two
3518 different character sequences that look the same. To avoid confusion,
3519 the ISO 10646 standard sets out some @dfn{normalization rules} which
3520 when applied ensure that two sequences that look the same are turned into
3521 the same sequence. GCC can warn you if you are using identifiers which
3522 have not been normalized; this option controls that warning.
3524 There are four levels of warning that GCC supports. The default is
3525 @option{-Wnormalized=nfc}, which warns about any identifier which is
3526 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3527 recommended form for most uses.
3529 Unfortunately, there are some characters which ISO C and ISO C++ allow
3530 in identifiers that when turned into NFC aren't allowable as
3531 identifiers. That is, there's no way to use these symbols in portable
3532 ISO C or C++ and have all your identifiers in NFC.
3533 @option{-Wnormalized=id} suppresses the warning for these characters.
3534 It is hoped that future versions of the standards involved will correct
3535 this, which is why this option is not the default.
3537 You can switch the warning off for all characters by writing
3538 @option{-Wnormalized=none}. You would only want to do this if you
3539 were using some other normalization scheme (like ``D''), because
3540 otherwise you can easily create bugs that are literally impossible to see.
3542 Some characters in ISO 10646 have distinct meanings but look identical
3543 in some fonts or display methodologies, especially once formatting has
3544 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3545 LETTER N'', will display just like a regular @code{n} which has been
3546 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3547 normalization scheme to convert all these into a standard form as
3548 well, and GCC will warn if your code is not in NFKC if you use
3549 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3550 about every identifier that contains the letter O because it might be
3551 confused with the digit 0, and so is not the default, but may be
3552 useful as a local coding convention if the programming environment is
3553 unable to be fixed to display these characters distinctly.
3555 @item -Wno-deprecated-declarations
3556 @opindex Wno-deprecated-declarations
3557 Do not warn about uses of functions (@pxref{Function Attributes}),
3558 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3559 Attributes}) marked as deprecated by using the @code{deprecated}
3563 @opindex Wno-overflow
3564 Do not warn about compile-time overflow in constant expressions.
3566 @item -Woverride-init
3567 @opindex Woverride-init
3570 Warn if an initialized field without side effects is overridden when
3571 using designated initializers (@pxref{Designated Inits, , Designated
3574 This warning is included in @option{-Wextra}. To get other
3575 @option{-Wextra} warnings without this one, use @samp{-Wextra
3576 -Wno-override-init}.
3580 Warn if a structure is given the packed attribute, but the packed
3581 attribute has no effect on the layout or size of the structure.
3582 Such structures may be mis-aligned for little benefit. For
3583 instance, in this code, the variable @code{f.x} in @code{struct bar}
3584 will be misaligned even though @code{struct bar} does not itself
3585 have the packed attribute:
3592 @} __attribute__((packed));
3602 Warn if padding is included in a structure, either to align an element
3603 of the structure or to align the whole structure. Sometimes when this
3604 happens it is possible to rearrange the fields of the structure to
3605 reduce the padding and so make the structure smaller.
3607 @item -Wredundant-decls
3608 @opindex Wredundant-decls
3609 Warn if anything is declared more than once in the same scope, even in
3610 cases where multiple declaration is valid and changes nothing.
3612 @item -Wnested-externs @r{(C only)}
3613 @opindex Wnested-externs
3614 Warn if an @code{extern} declaration is encountered within a function.
3616 @item -Wunreachable-code
3617 @opindex Wunreachable-code
3618 Warn if the compiler detects that code will never be executed.
3620 This option is intended to warn when the compiler detects that at
3621 least a whole line of source code will never be executed, because
3622 some condition is never satisfied or because it is after a
3623 procedure that never returns.
3625 It is possible for this option to produce a warning even though there
3626 are circumstances under which part of the affected line can be executed,
3627 so care should be taken when removing apparently-unreachable code.
3629 For instance, when a function is inlined, a warning may mean that the
3630 line is unreachable in only one inlined copy of the function.
3632 This option is not made part of @option{-Wall} because in a debugging
3633 version of a program there is often substantial code which checks
3634 correct functioning of the program and is, hopefully, unreachable
3635 because the program does work. Another common use of unreachable
3636 code is to provide behavior which is selectable at compile-time.
3640 Warn if a function can not be inlined and it was declared as inline.
3641 Even with this option, the compiler will not warn about failures to
3642 inline functions declared in system headers.
3644 The compiler uses a variety of heuristics to determine whether or not
3645 to inline a function. For example, the compiler takes into account
3646 the size of the function being inlined and the amount of inlining
3647 that has already been done in the current function. Therefore,
3648 seemingly insignificant changes in the source program can cause the
3649 warnings produced by @option{-Winline} to appear or disappear.
3651 @item -Wno-invalid-offsetof @r{(C++ only)}
3652 @opindex Wno-invalid-offsetof
3653 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3654 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3655 to a non-POD type is undefined. In existing C++ implementations,
3656 however, @samp{offsetof} typically gives meaningful results even when
3657 applied to certain kinds of non-POD types. (Such as a simple
3658 @samp{struct} that fails to be a POD type only by virtue of having a
3659 constructor.) This flag is for users who are aware that they are
3660 writing nonportable code and who have deliberately chosen to ignore the
3663 The restrictions on @samp{offsetof} may be relaxed in a future version
3664 of the C++ standard.
3666 @item -Wno-int-to-pointer-cast @r{(C only)}
3667 @opindex Wno-int-to-pointer-cast
3668 Suppress warnings from casts to pointer type of an integer of a
3671 @item -Wno-pointer-to-int-cast @r{(C only)}
3672 @opindex Wno-pointer-to-int-cast
3673 Suppress warnings from casts from a pointer to an integer type of a
3677 @opindex Winvalid-pch
3678 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3679 the search path but can't be used.
3683 @opindex Wno-long-long
3684 Warn if @samp{long long} type is used. This is default. To inhibit
3685 the warning messages, use @option{-Wno-long-long}. Flags
3686 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3687 only when @option{-pedantic} flag is used.
3689 @item -Wvariadic-macros
3690 @opindex Wvariadic-macros
3691 @opindex Wno-variadic-macros
3692 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3693 alternate syntax when in pedantic ISO C99 mode. This is default.
3694 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3696 @item -Wvolatile-register-var
3697 @opindex Wvolatile-register-var
3698 @opindex Wno-volatile-register-var
3699 Warn if a register variable is declared volatile. The volatile
3700 modifier does not inhibit all optimizations that may eliminate reads
3701 and/or writes to register variables.
3703 @item -Wdisabled-optimization
3704 @opindex Wdisabled-optimization
3705 Warn if a requested optimization pass is disabled. This warning does
3706 not generally indicate that there is anything wrong with your code; it
3707 merely indicates that GCC's optimizers were unable to handle the code
3708 effectively. Often, the problem is that your code is too big or too
3709 complex; GCC will refuse to optimize programs when the optimization
3710 itself is likely to take inordinate amounts of time.
3712 @item -Wpointer-sign
3713 @opindex Wpointer-sign
3714 @opindex Wno-pointer-sign
3715 Warn for pointer argument passing or assignment with different signedness.
3716 This option is only supported for C and Objective-C@. It is implied by
3717 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3718 @option{-Wno-pointer-sign}.
3722 Make all warnings into errors.
3726 Make the specified warning into an errors. The specifier for a
3727 warning is appended, for example @option{-Werror=switch} turns the
3728 warnings controlled by @option{-Wswitch} into errors. This switch
3729 takes a negative form, to be used to negate @option{-Werror} for
3730 specific warnings, for example @option{-Wno-error=switch} makes
3731 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3732 is in effect. You can use the @option{-fdiagnostics-show-option}
3733 option to have each controllable warning amended with the option which
3734 controls it, to determine what to use with this option.
3736 Note that specifying @option{-Werror=}@var{foo} automatically implies
3737 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3740 @item -Wstack-protector
3741 @opindex Wstack-protector
3742 This option is only active when @option{-fstack-protector} is active. It
3743 warns about functions that will not be protected against stack smashing.
3745 @item -Wstring-literal-comparison
3746 @opindex Wstring-literal-comparison
3747 Warn about suspicious comparisons to string literal constants. In C,
3748 direct comparisons against the memory address of a string literal, such
3749 as @code{if (x == "abc")}, typically indicate a programmer error, and
3750 even when intentional, result in unspecified behavior and are not portable.
3751 Usually these warnings alert that the programmer intended to use
3752 @code{strcmp}. This warning is enabled by @option{-Wall}.
3754 @item -Woverlength-strings
3755 @opindex Woverlength-strings
3756 Warn about string constants which are longer than the ``minimum
3757 maximum'' length specified in the C standard. Modern compilers
3758 generally allow string constants which are much longer than the
3759 standard's minimum limit, but very portable programs should avoid
3760 using longer strings.
3762 The limit applies @emph{after} string constant concatenation, and does
3763 not count the trailing NUL@. In C89, the limit was 509 characters; in
3764 C99, it was raised to 4095. C++98 does not specify a normative
3765 minimum maximum, so we do not diagnose overlength strings in C++@.
3767 This option is implied by @option{-pedantic}, and can be disabled with
3768 @option{-Wno-overlength-strings}.
3771 @node Debugging Options
3772 @section Options for Debugging Your Program or GCC
3773 @cindex options, debugging
3774 @cindex debugging information options
3776 GCC has various special options that are used for debugging
3777 either your program or GCC:
3782 Produce debugging information in the operating system's native format
3783 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3786 On most systems that use stabs format, @option{-g} enables use of extra
3787 debugging information that only GDB can use; this extra information
3788 makes debugging work better in GDB but will probably make other debuggers
3790 refuse to read the program. If you want to control for certain whether
3791 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3792 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3794 GCC allows you to use @option{-g} with
3795 @option{-O}. The shortcuts taken by optimized code may occasionally
3796 produce surprising results: some variables you declared may not exist
3797 at all; flow of control may briefly move where you did not expect it;
3798 some statements may not be executed because they compute constant
3799 results or their values were already at hand; some statements may
3800 execute in different places because they were moved out of loops.
3802 Nevertheless it proves possible to debug optimized output. This makes
3803 it reasonable to use the optimizer for programs that might have bugs.
3805 The following options are useful when GCC is generated with the
3806 capability for more than one debugging format.
3810 Produce debugging information for use by GDB@. This means to use the
3811 most expressive format available (DWARF 2, stabs, or the native format
3812 if neither of those are supported), including GDB extensions if at all
3817 Produce debugging information in stabs format (if that is supported),
3818 without GDB extensions. This is the format used by DBX on most BSD
3819 systems. On MIPS, Alpha and System V Release 4 systems this option
3820 produces stabs debugging output which is not understood by DBX or SDB@.
3821 On System V Release 4 systems this option requires the GNU assembler.
3823 @item -feliminate-unused-debug-symbols
3824 @opindex feliminate-unused-debug-symbols
3825 Produce debugging information in stabs format (if that is supported),
3826 for only symbols that are actually used.
3828 @item -femit-class-debug-always
3829 Instead of emitting debugging information for a C++ class in only one
3830 object file, emit it in all object files using the class. This option
3831 should be used only with debuggers that are unable to handle the way GCC
3832 normally emits debugging information for classes because using this
3833 option will increase the size of debugging information by as much as a
3838 Produce debugging information in stabs format (if that is supported),
3839 using GNU extensions understood only by the GNU debugger (GDB)@. The
3840 use of these extensions is likely to make other debuggers crash or
3841 refuse to read the program.
3845 Produce debugging information in COFF format (if that is supported).
3846 This is the format used by SDB on most System V systems prior to
3851 Produce debugging information in XCOFF format (if that is supported).
3852 This is the format used by the DBX debugger on IBM RS/6000 systems.
3856 Produce debugging information in XCOFF format (if that is supported),
3857 using GNU extensions understood only by the GNU debugger (GDB)@. The
3858 use of these extensions is likely to make other debuggers crash or
3859 refuse to read the program, and may cause assemblers other than the GNU
3860 assembler (GAS) to fail with an error.
3864 Produce debugging information in DWARF version 2 format (if that is
3865 supported). This is the format used by DBX on IRIX 6. With this
3866 option, GCC uses features of DWARF version 3 when they are useful;
3867 version 3 is upward compatible with version 2, but may still cause
3868 problems for older debuggers.
3872 Produce debugging information in VMS debug format (if that is
3873 supported). This is the format used by DEBUG on VMS systems.
3876 @itemx -ggdb@var{level}
3877 @itemx -gstabs@var{level}
3878 @itemx -gcoff@var{level}
3879 @itemx -gxcoff@var{level}
3880 @itemx -gvms@var{level}
3881 Request debugging information and also use @var{level} to specify how
3882 much information. The default level is 2.
3884 Level 0 produces no debug information at all. Thus, @option{-g0} negates
3887 Level 1 produces minimal information, enough for making backtraces in
3888 parts of the program that you don't plan to debug. This includes
3889 descriptions of functions and external variables, but no information
3890 about local variables and no line numbers.
3892 Level 3 includes extra information, such as all the macro definitions
3893 present in the program. Some debuggers support macro expansion when
3894 you use @option{-g3}.
3896 @option{-gdwarf-2} does not accept a concatenated debug level, because
3897 GCC used to support an option @option{-gdwarf} that meant to generate
3898 debug information in version 1 of the DWARF format (which is very
3899 different from version 2), and it would have been too confusing. That
3900 debug format is long obsolete, but the option cannot be changed now.
3901 Instead use an additional @option{-g@var{level}} option to change the
3902 debug level for DWARF2.
3904 @item -feliminate-dwarf2-dups
3905 @opindex feliminate-dwarf2-dups
3906 Compress DWARF2 debugging information by eliminating duplicated
3907 information about each symbol. This option only makes sense when
3908 generating DWARF2 debugging information with @option{-gdwarf-2}.
3910 @cindex @command{prof}
3913 Generate extra code to write profile information suitable for the
3914 analysis program @command{prof}. You must use this option when compiling
3915 the source files you want data about, and you must also use it when
3918 @cindex @command{gprof}
3921 Generate extra code to write profile information suitable for the
3922 analysis program @command{gprof}. You must use this option when compiling
3923 the source files you want data about, and you must also use it when
3928 Makes the compiler print out each function name as it is compiled, and
3929 print some statistics about each pass when it finishes.
3932 @opindex ftime-report
3933 Makes the compiler print some statistics about the time consumed by each
3934 pass when it finishes.
3937 @opindex fmem-report
3938 Makes the compiler print some statistics about permanent memory
3939 allocation when it finishes.
3941 @item -fpre-ipa-mem-report
3942 @opindex fpre-ipa-mem-report
3943 @item -fpost-ipa-mem-report
3944 @opindex fpost-ipa-mem-report
3945 Makes the compiler print some statistics about permanent memory
3946 allocation before or after interprocedural optimization.
3948 @item -fprofile-arcs
3949 @opindex fprofile-arcs
3950 Add code so that program flow @dfn{arcs} are instrumented. During
3951 execution the program records how many times each branch and call is
3952 executed and how many times it is taken or returns. When the compiled
3953 program exits it saves this data to a file called
3954 @file{@var{auxname}.gcda} for each source file. The data may be used for
3955 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3956 test coverage analysis (@option{-ftest-coverage}). Each object file's
3957 @var{auxname} is generated from the name of the output file, if
3958 explicitly specified and it is not the final executable, otherwise it is
3959 the basename of the source file. In both cases any suffix is removed
3960 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3961 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3962 @xref{Cross-profiling}.
3964 @cindex @command{gcov}
3968 This option is used to compile and link code instrumented for coverage
3969 analysis. The option is a synonym for @option{-fprofile-arcs}
3970 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3971 linking). See the documentation for those options for more details.
3976 Compile the source files with @option{-fprofile-arcs} plus optimization
3977 and code generation options. For test coverage analysis, use the
3978 additional @option{-ftest-coverage} option. You do not need to profile
3979 every source file in a program.
3982 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3983 (the latter implies the former).
3986 Run the program on a representative workload to generate the arc profile
3987 information. This may be repeated any number of times. You can run
3988 concurrent instances of your program, and provided that the file system
3989 supports locking, the data files will be correctly updated. Also
3990 @code{fork} calls are detected and correctly handled (double counting
3994 For profile-directed optimizations, compile the source files again with
3995 the same optimization and code generation options plus
3996 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3997 Control Optimization}).
4000 For test coverage analysis, use @command{gcov} to produce human readable
4001 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4002 @command{gcov} documentation for further information.
4006 With @option{-fprofile-arcs}, for each function of your program GCC
4007 creates a program flow graph, then finds a spanning tree for the graph.
4008 Only arcs that are not on the spanning tree have to be instrumented: the
4009 compiler adds code to count the number of times that these arcs are
4010 executed. When an arc is the only exit or only entrance to a block, the
4011 instrumentation code can be added to the block; otherwise, a new basic
4012 block must be created to hold the instrumentation code.
4015 @item -ftest-coverage
4016 @opindex ftest-coverage
4017 Produce a notes file that the @command{gcov} code-coverage utility
4018 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4019 show program coverage. Each source file's note file is called
4020 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4021 above for a description of @var{auxname} and instructions on how to
4022 generate test coverage data. Coverage data will match the source files
4023 more closely, if you do not optimize.
4025 @item -d@var{letters}
4026 @item -fdump-rtl-@var{pass}
4028 Says to make debugging dumps during compilation at times specified by
4029 @var{letters}. This is used for debugging the RTL-based passes of the
4030 compiler. The file names for most of the dumps are made by appending a
4031 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4032 from the name of the output file, if explicitly specified and it is not
4033 an executable, otherwise it is the basename of the source file.
4035 Most debug dumps can be enabled either passing a letter to the @option{-d}
4036 option, or with a long @option{-fdump-rtl} switch; here are the possible
4037 letters for use in @var{letters} and @var{pass}, and their meanings:
4042 Annotate the assembler output with miscellaneous debugging information.
4045 @itemx -fdump-rtl-bbro
4047 @opindex fdump-rtl-bbro
4048 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4051 @itemx -fdump-rtl-combine
4053 @opindex fdump-rtl-combine
4054 Dump after the RTL instruction combination pass, to the file
4055 @file{@var{file}.129r.combine}.
4058 @itemx -fdump-rtl-ce1
4059 @itemx -fdump-rtl-ce2
4061 @opindex fdump-rtl-ce1
4062 @opindex fdump-rtl-ce2
4063 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4064 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4065 and @option{-fdump-rtl-ce2} enable dumping after the second if
4066 conversion, to the file @file{@var{file}.130r.ce2}.
4069 @itemx -fdump-rtl-btl
4070 @itemx -fdump-rtl-dbr
4072 @opindex fdump-rtl-btl
4073 @opindex fdump-rtl-dbr
4074 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4075 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4076 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4077 scheduling, to @file{@var{file}.36.dbr}.
4081 Dump all macro definitions, at the end of preprocessing, in addition to
4085 @itemx -fdump-rtl-ce3
4087 @opindex fdump-rtl-ce3
4088 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4091 @itemx -fdump-rtl-cfg
4092 @itemx -fdump-rtl-life
4094 @opindex fdump-rtl-cfg
4095 @opindex fdump-rtl-life
4096 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4097 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4098 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4099 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4102 @itemx -fdump-rtl-greg
4104 @opindex fdump-rtl-greg
4105 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4108 @itemx -fdump-rtl-gcse
4109 @itemx -fdump-rtl-bypass
4111 @opindex fdump-rtl-gcse
4112 @opindex fdump-rtl-bypass
4113 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4114 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4115 enable dumping after jump bypassing and control flow optimizations, to
4116 @file{@var{file}.115r.bypass}.
4119 @itemx -fdump-rtl-eh
4121 @opindex fdump-rtl-eh
4122 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4125 @itemx -fdump-rtl-sibling
4127 @opindex fdump-rtl-sibling
4128 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4131 @itemx -fdump-rtl-jump
4133 @opindex fdump-rtl-jump
4134 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4137 @itemx -fdump-rtl-stack
4139 @opindex fdump-rtl-stack
4140 Dump after conversion from GCC's "flat register file" registers to the
4141 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4144 @itemx -fdump-rtl-lreg
4146 @opindex fdump-rtl-lreg
4147 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4150 @itemx -fdump-rtl-loop2
4152 @opindex fdump-rtl-loop2
4153 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4154 loop optimization pass, to @file{@var{file}.119r.loop2},
4155 @file{@var{file}.120r.loop2_init},
4156 @file{@var{file}.121r.loop2_invariant}, and
4157 @file{@var{file}.125r.loop2_done}.
4160 @itemx -fdump-rtl-sms
4162 @opindex fdump-rtl-sms
4163 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4166 @itemx -fdump-rtl-mach
4168 @opindex fdump-rtl-mach
4169 Dump after performing the machine dependent reorganization pass, to
4170 @file{@var{file}.155r.mach}.
4173 @itemx -fdump-rtl-rnreg
4175 @opindex fdump-rtl-rnreg
4176 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4179 @itemx -fdump-rtl-regmove
4181 @opindex fdump-rtl-regmove
4182 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4185 @itemx -fdump-rtl-postreload
4187 @opindex fdump-rtl-postreload
4188 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4191 @itemx -fdump-rtl-expand
4193 @opindex fdump-rtl-expand
4194 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4197 @itemx -fdump-rtl-sched2
4199 @opindex fdump-rtl-sched2
4200 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4203 @itemx -fdump-rtl-cse
4205 @opindex fdump-rtl-cse
4206 Dump after CSE (including the jump optimization that sometimes follows
4207 CSE), to @file{@var{file}.113r.cse}.
4210 @itemx -fdump-rtl-sched1
4212 @opindex fdump-rtl-sched1
4213 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4216 @itemx -fdump-rtl-cse2
4218 @opindex fdump-rtl-cse2
4219 Dump after the second CSE pass (including the jump optimization that
4220 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4223 @itemx -fdump-rtl-tracer
4225 @opindex fdump-rtl-tracer
4226 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4229 @itemx -fdump-rtl-vpt
4230 @itemx -fdump-rtl-vartrack
4232 @opindex fdump-rtl-vpt
4233 @opindex fdump-rtl-vartrack
4234 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4235 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4236 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4237 to @file{@var{file}.154r.vartrack}.
4240 @itemx -fdump-rtl-flow2
4242 @opindex fdump-rtl-flow2
4243 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4246 @itemx -fdump-rtl-peephole2
4248 @opindex fdump-rtl-peephole2
4249 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4252 @itemx -fdump-rtl-web
4254 @opindex fdump-rtl-web
4255 Dump after live range splitting, to @file{@var{file}.126r.web}.
4258 @itemx -fdump-rtl-all
4260 @opindex fdump-rtl-all
4261 Produce all the dumps listed above.
4265 Produce a core dump whenever an error occurs.
4269 Print statistics on memory usage, at the end of the run, to
4274 Annotate the assembler output with a comment indicating which
4275 pattern and alternative was used. The length of each instruction is
4280 Dump the RTL in the assembler output as a comment before each instruction.
4281 Also turns on @option{-dp} annotation.
4285 For each of the other indicated dump files (either with @option{-d} or
4286 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4287 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4291 Just generate RTL for a function instead of compiling it. Usually used
4292 with @samp{r} (@option{-fdump-rtl-expand}).
4296 Dump debugging information during parsing, to standard error.
4300 @opindex fdump-noaddr
4301 When doing debugging dumps (see @option{-d} option above), suppress
4302 address output. This makes it more feasible to use diff on debugging
4303 dumps for compiler invocations with different compiler binaries and/or
4304 different text / bss / data / heap / stack / dso start locations.
4306 @item -fdump-unnumbered
4307 @opindex fdump-unnumbered
4308 When doing debugging dumps (see @option{-d} option above), suppress instruction
4309 numbers, line number note and address output. This makes it more feasible to
4310 use diff on debugging dumps for compiler invocations with different
4311 options, in particular with and without @option{-g}.
4313 @item -fdump-translation-unit @r{(C++ only)}
4314 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4315 @opindex fdump-translation-unit
4316 Dump a representation of the tree structure for the entire translation
4317 unit to a file. The file name is made by appending @file{.tu} to the
4318 source file name. If the @samp{-@var{options}} form is used, @var{options}
4319 controls the details of the dump as described for the
4320 @option{-fdump-tree} options.
4322 @item -fdump-class-hierarchy @r{(C++ only)}
4323 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4324 @opindex fdump-class-hierarchy
4325 Dump a representation of each class's hierarchy and virtual function
4326 table layout to a file. The file name is made by appending @file{.class}
4327 to the source file name. If the @samp{-@var{options}} form is used,
4328 @var{options} controls the details of the dump as described for the
4329 @option{-fdump-tree} options.
4331 @item -fdump-ipa-@var{switch}
4333 Control the dumping at various stages of inter-procedural analysis
4334 language tree to a file. The file name is generated by appending a switch
4335 specific suffix to the source file name. The following dumps are possible:
4339 Enables all inter-procedural analysis dumps; currently the only produced
4340 dump is the @samp{cgraph} dump.
4343 Dumps information about call-graph optimization, unused function removal,
4344 and inlining decisions.
4347 @item -fdump-tree-@var{switch}
4348 @itemx -fdump-tree-@var{switch}-@var{options}
4350 Control the dumping at various stages of processing the intermediate
4351 language tree to a file. The file name is generated by appending a switch
4352 specific suffix to the source file name. If the @samp{-@var{options}}
4353 form is used, @var{options} is a list of @samp{-} separated options that
4354 control the details of the dump. Not all options are applicable to all
4355 dumps, those which are not meaningful will be ignored. The following
4356 options are available
4360 Print the address of each node. Usually this is not meaningful as it
4361 changes according to the environment and source file. Its primary use
4362 is for tying up a dump file with a debug environment.
4364 Inhibit dumping of members of a scope or body of a function merely
4365 because that scope has been reached. Only dump such items when they
4366 are directly reachable by some other path. When dumping pretty-printed
4367 trees, this option inhibits dumping the bodies of control structures.
4369 Print a raw representation of the tree. By default, trees are
4370 pretty-printed into a C-like representation.
4372 Enable more detailed dumps (not honored by every dump option).
4374 Enable dumping various statistics about the pass (not honored by every dump
4377 Enable showing basic block boundaries (disabled in raw dumps).
4379 Enable showing virtual operands for every statement.
4381 Enable showing line numbers for statements.
4383 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4385 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4388 The following tree dumps are possible:
4392 Dump before any tree based optimization, to @file{@var{file}.original}.
4395 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4398 Dump after function inlining, to @file{@var{file}.inlined}.
4401 @opindex fdump-tree-gimple
4402 Dump each function before and after the gimplification pass to a file. The
4403 file name is made by appending @file{.gimple} to the source file name.
4406 @opindex fdump-tree-cfg
4407 Dump the control flow graph of each function to a file. The file name is
4408 made by appending @file{.cfg} to the source file name.
4411 @opindex fdump-tree-vcg
4412 Dump the control flow graph of each function to a file in VCG format. The
4413 file name is made by appending @file{.vcg} to the source file name. Note
4414 that if the file contains more than one function, the generated file cannot
4415 be used directly by VCG@. You will need to cut and paste each function's
4416 graph into its own separate file first.
4419 @opindex fdump-tree-ch
4420 Dump each function after copying loop headers. The file name is made by
4421 appending @file{.ch} to the source file name.
4424 @opindex fdump-tree-ssa
4425 Dump SSA related information to a file. The file name is made by appending
4426 @file{.ssa} to the source file name.
4429 @opindex fdump-tree-salias
4430 Dump structure aliasing variable information to a file. This file name
4431 is made by appending @file{.salias} to the source file name.
4434 @opindex fdump-tree-alias
4435 Dump aliasing information for each function. The file name is made by
4436 appending @file{.alias} to the source file name.
4439 @opindex fdump-tree-ccp
4440 Dump each function after CCP@. The file name is made by appending
4441 @file{.ccp} to the source file name.
4444 @opindex fdump-tree-storeccp
4445 Dump each function after STORE-CCP. The file name is made by appending
4446 @file{.storeccp} to the source file name.
4449 @opindex fdump-tree-pre
4450 Dump trees after partial redundancy elimination. The file name is made
4451 by appending @file{.pre} to the source file name.
4454 @opindex fdump-tree-fre
4455 Dump trees after full redundancy elimination. The file name is made
4456 by appending @file{.fre} to the source file name.
4459 @opindex fdump-tree-copyprop
4460 Dump trees after copy propagation. The file name is made
4461 by appending @file{.copyprop} to the source file name.
4463 @item store_copyprop
4464 @opindex fdump-tree-store_copyprop
4465 Dump trees after store copy-propagation. The file name is made
4466 by appending @file{.store_copyprop} to the source file name.
4469 @opindex fdump-tree-dce
4470 Dump each function after dead code elimination. The file name is made by
4471 appending @file{.dce} to the source file name.
4474 @opindex fdump-tree-mudflap
4475 Dump each function after adding mudflap instrumentation. The file name is
4476 made by appending @file{.mudflap} to the source file name.
4479 @opindex fdump-tree-sra
4480 Dump each function after performing scalar replacement of aggregates. The
4481 file name is made by appending @file{.sra} to the source file name.
4484 @opindex fdump-tree-sink
4485 Dump each function after performing code sinking. The file name is made
4486 by appending @file{.sink} to the source file name.
4489 @opindex fdump-tree-dom
4490 Dump each function after applying dominator tree optimizations. The file
4491 name is made by appending @file{.dom} to the source file name.
4494 @opindex fdump-tree-dse
4495 Dump each function after applying dead store elimination. The file
4496 name is made by appending @file{.dse} to the source file name.
4499 @opindex fdump-tree-phiopt
4500 Dump each function after optimizing PHI nodes into straightline code. The file
4501 name is made by appending @file{.phiopt} to the source file name.
4504 @opindex fdump-tree-forwprop
4505 Dump each function after forward propagating single use variables. The file
4506 name is made by appending @file{.forwprop} to the source file name.
4509 @opindex fdump-tree-copyrename
4510 Dump each function after applying the copy rename optimization. The file
4511 name is made by appending @file{.copyrename} to the source file name.
4514 @opindex fdump-tree-nrv
4515 Dump each function after applying the named return value optimization on
4516 generic trees. The file name is made by appending @file{.nrv} to the source
4520 @opindex fdump-tree-vect
4521 Dump each function after applying vectorization of loops. The file name is
4522 made by appending @file{.vect} to the source file name.
4525 @opindex fdump-tree-vrp
4526 Dump each function after Value Range Propagation (VRP). The file name
4527 is made by appending @file{.vrp} to the source file name.
4530 @opindex fdump-tree-all
4531 Enable all the available tree dumps with the flags provided in this option.
4534 @item -ftree-vectorizer-verbose=@var{n}
4535 @opindex ftree-vectorizer-verbose
4536 This option controls the amount of debugging output the vectorizer prints.
4537 This information is written to standard error, unless
4538 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4539 in which case it is output to the usual dump listing file, @file{.vect}.
4540 For @var{n}=0 no diagnostic information is reported.
4541 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4542 and the total number of loops that got vectorized.
4543 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4544 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4545 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4546 level that @option{-fdump-tree-vect-stats} uses.
4547 Higher verbosity levels mean either more information dumped for each
4548 reported loop, or same amount of information reported for more loops:
4549 If @var{n}=3, alignment related information is added to the reports.
4550 If @var{n}=4, data-references related information (e.g. memory dependences,
4551 memory access-patterns) is added to the reports.
4552 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4553 that did not pass the first analysis phase (i.e. may not be countable, or
4554 may have complicated control-flow).
4555 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4556 For @var{n}=7, all the information the vectorizer generates during its
4557 analysis and transformation is reported. This is the same verbosity level
4558 that @option{-fdump-tree-vect-details} uses.
4560 @item -frandom-seed=@var{string}
4561 @opindex frandom-string
4562 This option provides a seed that GCC uses when it would otherwise use
4563 random numbers. It is used to generate certain symbol names
4564 that have to be different in every compiled file. It is also used to
4565 place unique stamps in coverage data files and the object files that
4566 produce them. You can use the @option{-frandom-seed} option to produce
4567 reproducibly identical object files.
4569 The @var{string} should be different for every file you compile.
4571 @item -fsched-verbose=@var{n}
4572 @opindex fsched-verbose
4573 On targets that use instruction scheduling, this option controls the
4574 amount of debugging output the scheduler prints. This information is
4575 written to standard error, unless @option{-dS} or @option{-dR} is
4576 specified, in which case it is output to the usual dump
4577 listing file, @file{.sched} or @file{.sched2} respectively. However
4578 for @var{n} greater than nine, the output is always printed to standard
4581 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4582 same information as @option{-dRS}. For @var{n} greater than one, it
4583 also output basic block probabilities, detailed ready list information
4584 and unit/insn info. For @var{n} greater than two, it includes RTL
4585 at abort point, control-flow and regions info. And for @var{n} over
4586 four, @option{-fsched-verbose} also includes dependence info.
4590 Store the usual ``temporary'' intermediate files permanently; place them
4591 in the current directory and name them based on the source file. Thus,
4592 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4593 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4594 preprocessed @file{foo.i} output file even though the compiler now
4595 normally uses an integrated preprocessor.
4597 When used in combination with the @option{-x} command line option,
4598 @option{-save-temps} is sensible enough to avoid over writing an
4599 input source file with the same extension as an intermediate file.
4600 The corresponding intermediate file may be obtained by renaming the
4601 source file before using @option{-save-temps}.
4605 Report the CPU time taken by each subprocess in the compilation
4606 sequence. For C source files, this is the compiler proper and assembler
4607 (plus the linker if linking is done). The output looks like this:
4614 The first number on each line is the ``user time'', that is time spent
4615 executing the program itself. The second number is ``system time'',
4616 time spent executing operating system routines on behalf of the program.
4617 Both numbers are in seconds.
4619 @item -fvar-tracking
4620 @opindex fvar-tracking
4621 Run variable tracking pass. It computes where variables are stored at each
4622 position in code. Better debugging information is then generated
4623 (if the debugging information format supports this information).
4625 It is enabled by default when compiling with optimization (@option{-Os},
4626 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4627 the debug info format supports it.
4629 @item -print-file-name=@var{library}
4630 @opindex print-file-name
4631 Print the full absolute name of the library file @var{library} that
4632 would be used when linking---and don't do anything else. With this
4633 option, GCC does not compile or link anything; it just prints the
4636 @item -print-multi-directory
4637 @opindex print-multi-directory
4638 Print the directory name corresponding to the multilib selected by any
4639 other switches present in the command line. This directory is supposed
4640 to exist in @env{GCC_EXEC_PREFIX}.
4642 @item -print-multi-lib
4643 @opindex print-multi-lib
4644 Print the mapping from multilib directory names to compiler switches
4645 that enable them. The directory name is separated from the switches by
4646 @samp{;}, and each switch starts with an @samp{@@} instead of the
4647 @samp{-}, without spaces between multiple switches. This is supposed to
4648 ease shell-processing.
4650 @item -print-prog-name=@var{program}
4651 @opindex print-prog-name
4652 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4654 @item -print-libgcc-file-name
4655 @opindex print-libgcc-file-name
4656 Same as @option{-print-file-name=libgcc.a}.
4658 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4659 but you do want to link with @file{libgcc.a}. You can do
4662 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4665 @item -print-search-dirs
4666 @opindex print-search-dirs
4667 Print the name of the configured installation directory and a list of
4668 program and library directories @command{gcc} will search---and don't do anything else.
4670 This is useful when @command{gcc} prints the error message
4671 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4672 To resolve this you either need to put @file{cpp0} and the other compiler
4673 components where @command{gcc} expects to find them, or you can set the environment
4674 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4675 Don't forget the trailing @samp{/}.
4676 @xref{Environment Variables}.
4679 @opindex dumpmachine
4680 Print the compiler's target machine (for example,
4681 @samp{i686-pc-linux-gnu})---and don't do anything else.
4684 @opindex dumpversion
4685 Print the compiler version (for example, @samp{3.0})---and don't do
4690 Print the compiler's built-in specs---and don't do anything else. (This
4691 is used when GCC itself is being built.) @xref{Spec Files}.
4693 @item -feliminate-unused-debug-types
4694 @opindex feliminate-unused-debug-types
4695 Normally, when producing DWARF2 output, GCC will emit debugging
4696 information for all types declared in a compilation
4697 unit, regardless of whether or not they are actually used
4698 in that compilation unit. Sometimes this is useful, such as
4699 if, in the debugger, you want to cast a value to a type that is
4700 not actually used in your program (but is declared). More often,
4701 however, this results in a significant amount of wasted space.
4702 With this option, GCC will avoid producing debug symbol output
4703 for types that are nowhere used in the source file being compiled.
4706 @node Optimize Options
4707 @section Options That Control Optimization
4708 @cindex optimize options
4709 @cindex options, optimization
4711 These options control various sorts of optimizations.
4713 Without any optimization option, the compiler's goal is to reduce the
4714 cost of compilation and to make debugging produce the expected
4715 results. Statements are independent: if you stop the program with a
4716 breakpoint between statements, you can then assign a new value to any
4717 variable or change the program counter to any other statement in the
4718 function and get exactly the results you would expect from the source
4721 Turning on optimization flags makes the compiler attempt to improve
4722 the performance and/or code size at the expense of compilation time
4723 and possibly the ability to debug the program.
4725 The compiler performs optimization based on the knowledge it has of
4726 the program. Optimization levels @option{-O} and above, in
4727 particular, enable @emph{unit-at-a-time} mode, which allows the
4728 compiler to consider information gained from later functions in
4729 the file when compiling a function. Compiling multiple files at
4730 once to a single output file in @emph{unit-at-a-time} mode allows
4731 the compiler to use information gained from all of the files when
4732 compiling each of them.
4734 Not all optimizations are controlled directly by a flag. Only
4735 optimizations that have a flag are listed.
4742 Optimize. Optimizing compilation takes somewhat more time, and a lot
4743 more memory for a large function.
4745 With @option{-O}, the compiler tries to reduce code size and execution
4746 time, without performing any optimizations that take a great deal of
4749 @option{-O} turns on the following optimization flags:
4750 @gccoptlist{-fdefer-pop @gol
4751 -fdelayed-branch @gol
4752 -fguess-branch-probability @gol
4753 -fcprop-registers @gol
4754 -fif-conversion @gol
4755 -fif-conversion2 @gol
4756 -fsplit-wide-types @gol
4759 -ftree-dominator-opts @gol
4764 -ftree-copyrename @gol
4767 -funit-at-a-time @gol
4770 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4771 where doing so does not interfere with debugging.
4775 Optimize even more. GCC performs nearly all supported optimizations
4776 that do not involve a space-speed tradeoff. The compiler does not
4777 perform loop unrolling or function inlining when you specify @option{-O2}.
4778 As compared to @option{-O}, this option increases both compilation time
4779 and the performance of the generated code.
4781 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4782 also turns on the following optimization flags:
4783 @gccoptlist{-fthread-jumps @gol
4785 -foptimize-sibling-calls @gol
4786 -fcse-follow-jumps -fcse-skip-blocks @gol
4787 -fgcse -fgcse-lm @gol
4788 -fexpensive-optimizations @gol
4789 -frerun-cse-after-loop @gol
4792 -fschedule-insns -fschedule-insns2 @gol
4793 -fsched-interblock -fsched-spec @gol
4795 -fstrict-aliasing -fstrict-overflow @gol
4796 -fdelete-null-pointer-checks @gol
4797 -freorder-blocks -freorder-functions @gol
4798 -falign-functions -falign-jumps @gol
4799 -falign-loops -falign-labels @gol
4803 Please note the warning under @option{-fgcse} about
4804 invoking @option{-O2} on programs that use computed gotos.
4808 Optimize yet more. @option{-O3} turns on all optimizations specified by
4809 @option{-O2} and also turns on the @option{-finline-functions},
4810 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4814 Reduce compilation time and make debugging produce the expected
4815 results. This is the default.
4819 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4820 do not typically increase code size. It also performs further
4821 optimizations designed to reduce code size.
4823 @option{-Os} disables the following optimization flags:
4824 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4825 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4826 -fprefetch-loop-arrays -ftree-vect-loop-version}
4828 If you use multiple @option{-O} options, with or without level numbers,
4829 the last such option is the one that is effective.
4832 Options of the form @option{-f@var{flag}} specify machine-independent
4833 flags. Most flags have both positive and negative forms; the negative
4834 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4835 below, only one of the forms is listed---the one you typically will
4836 use. You can figure out the other form by either removing @samp{no-}
4839 The following options control specific optimizations. They are either
4840 activated by @option{-O} options or are related to ones that are. You
4841 can use the following flags in the rare cases when ``fine-tuning'' of
4842 optimizations to be performed is desired.
4845 @item -fno-default-inline
4846 @opindex fno-default-inline
4847 Do not make member functions inline by default merely because they are
4848 defined inside the class scope (C++ only). Otherwise, when you specify
4849 @w{@option{-O}}, member functions defined inside class scope are compiled
4850 inline by default; i.e., you don't need to add @samp{inline} in front of
4851 the member function name.
4853 @item -fno-defer-pop
4854 @opindex fno-defer-pop
4855 Always pop the arguments to each function call as soon as that function
4856 returns. For machines which must pop arguments after a function call,
4857 the compiler normally lets arguments accumulate on the stack for several
4858 function calls and pops them all at once.
4860 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4863 @opindex fforce-addr
4864 Force memory address constants to be copied into registers before
4865 doing arithmetic on them.
4867 @item -fforward-propagate
4868 @opindex fforward-propagate
4869 Perform a forward propagation pass on RTL. The pass tries to combine two
4870 instructions and checks if the result can be simplified. If loop unrolling
4871 is active, two passes are performed and the second is scheduled after
4874 This option is enabled by default at optimization levels @option{-O2},
4875 @option{-O3}, @option{-Os}.
4877 @item -fomit-frame-pointer
4878 @opindex fomit-frame-pointer
4879 Don't keep the frame pointer in a register for functions that
4880 don't need one. This avoids the instructions to save, set up and
4881 restore frame pointers; it also makes an extra register available
4882 in many functions. @strong{It also makes debugging impossible on
4885 On some machines, such as the VAX, this flag has no effect, because
4886 the standard calling sequence automatically handles the frame pointer
4887 and nothing is saved by pretending it doesn't exist. The
4888 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4889 whether a target machine supports this flag. @xref{Registers,,Register
4890 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4892 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4894 @item -foptimize-sibling-calls
4895 @opindex foptimize-sibling-calls
4896 Optimize sibling and tail recursive calls.
4898 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4902 Don't pay attention to the @code{inline} keyword. Normally this option
4903 is used to keep the compiler from expanding any functions inline.
4904 Note that if you are not optimizing, no functions can be expanded inline.
4906 @item -finline-functions
4907 @opindex finline-functions
4908 Integrate all simple functions into their callers. The compiler
4909 heuristically decides which functions are simple enough to be worth
4910 integrating in this way.
4912 If all calls to a given function are integrated, and the function is
4913 declared @code{static}, then the function is normally not output as
4914 assembler code in its own right.
4916 Enabled at level @option{-O3}.
4918 @item -finline-functions-called-once
4919 @opindex finline-functions-called-once
4920 Consider all @code{static} functions called once for inlining into their
4921 caller even if they are not marked @code{inline}. If a call to a given
4922 function is integrated, then the function is not output as assembler code
4925 Enabled if @option{-funit-at-a-time} is enabled.
4927 @item -fearly-inlining
4928 @opindex fearly-inlining
4929 Inline functions marked by @code{always_inline} and functions whose body seems
4930 smaller than the function call overhead early before doing
4931 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4932 makes profiling significantly cheaper and usually inlining faster on programs
4933 having large chains of nested wrapper functions.
4937 @item -finline-limit=@var{n}
4938 @opindex finline-limit
4939 By default, GCC limits the size of functions that can be inlined. This flag
4940 allows the control of this limit for functions that are explicitly marked as
4941 inline (i.e., marked with the inline keyword or defined within the class
4942 definition in c++). @var{n} is the size of functions that can be inlined in
4943 number of pseudo instructions (not counting parameter handling). The default
4944 value of @var{n} is 600.
4945 Increasing this value can result in more inlined code at
4946 the cost of compilation time and memory consumption. Decreasing usually makes
4947 the compilation faster and less code will be inlined (which presumably
4948 means slower programs). This option is particularly useful for programs that
4949 use inlining heavily such as those based on recursive templates with C++.
4951 Inlining is actually controlled by a number of parameters, which may be
4952 specified individually by using @option{--param @var{name}=@var{value}}.
4953 The @option{-finline-limit=@var{n}} option sets some of these parameters
4957 @item max-inline-insns-single
4958 is set to @var{n}/2.
4959 @item max-inline-insns-auto
4960 is set to @var{n}/2.
4961 @item min-inline-insns
4962 is set to 130 or @var{n}/4, whichever is smaller.
4963 @item max-inline-insns-rtl
4967 See below for a documentation of the individual
4968 parameters controlling inlining.
4970 @emph{Note:} pseudo instruction represents, in this particular context, an
4971 abstract measurement of function's size. In no way does it represent a count
4972 of assembly instructions and as such its exact meaning might change from one
4973 release to an another.
4975 @item -fkeep-inline-functions
4976 @opindex fkeep-inline-functions
4977 In C, emit @code{static} functions that are declared @code{inline}
4978 into the object file, even if the function has been inlined into all
4979 of its callers. This switch does not affect functions using the
4980 @code{extern inline} extension in GNU C89@. In C++, emit any and all
4981 inline functions into the object file.
4983 @item -fkeep-static-consts
4984 @opindex fkeep-static-consts
4985 Emit variables declared @code{static const} when optimization isn't turned
4986 on, even if the variables aren't referenced.
4988 GCC enables this option by default. If you want to force the compiler to
4989 check if the variable was referenced, regardless of whether or not
4990 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4992 @item -fmerge-constants
4993 Attempt to merge identical constants (string constants and floating point
4994 constants) across compilation units.
4996 This option is the default for optimized compilation if the assembler and
4997 linker support it. Use @option{-fno-merge-constants} to inhibit this
5000 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5002 @item -fmerge-all-constants
5003 Attempt to merge identical constants and identical variables.
5005 This option implies @option{-fmerge-constants}. In addition to
5006 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5007 arrays or initialized constant variables with integral or floating point
5008 types. Languages like C or C++ require each non-automatic variable to
5009 have distinct location, so using this option will result in non-conforming
5012 @item -fmodulo-sched
5013 @opindex fmodulo-sched
5014 Perform swing modulo scheduling immediately before the first scheduling
5015 pass. This pass looks at innermost loops and reorders their
5016 instructions by overlapping different iterations.
5018 @item -fno-branch-count-reg
5019 @opindex fno-branch-count-reg
5020 Do not use ``decrement and branch'' instructions on a count register,
5021 but instead generate a sequence of instructions that decrement a
5022 register, compare it against zero, then branch based upon the result.
5023 This option is only meaningful on architectures that support such
5024 instructions, which include x86, PowerPC, IA-64 and S/390.
5026 The default is @option{-fbranch-count-reg}.
5028 @item -fno-function-cse
5029 @opindex fno-function-cse
5030 Do not put function addresses in registers; make each instruction that
5031 calls a constant function contain the function's address explicitly.
5033 This option results in less efficient code, but some strange hacks
5034 that alter the assembler output may be confused by the optimizations
5035 performed when this option is not used.
5037 The default is @option{-ffunction-cse}
5039 @item -fno-zero-initialized-in-bss
5040 @opindex fno-zero-initialized-in-bss
5041 If the target supports a BSS section, GCC by default puts variables that
5042 are initialized to zero into BSS@. This can save space in the resulting
5045 This option turns off this behavior because some programs explicitly
5046 rely on variables going to the data section. E.g., so that the
5047 resulting executable can find the beginning of that section and/or make
5048 assumptions based on that.
5050 The default is @option{-fzero-initialized-in-bss}.
5052 @item -fbounds-check
5053 @opindex fbounds-check
5054 For front-ends that support it, generate additional code to check that
5055 indices used to access arrays are within the declared range. This is
5056 currently only supported by the Java and Fortran front-ends, where
5057 this option defaults to true and false respectively.
5059 @item -fmudflap -fmudflapth -fmudflapir
5063 @cindex bounds checking
5065 For front-ends that support it (C and C++), instrument all risky
5066 pointer/array dereferencing operations, some standard library
5067 string/heap functions, and some other associated constructs with
5068 range/validity tests. Modules so instrumented should be immune to
5069 buffer overflows, invalid heap use, and some other classes of C/C++
5070 programming errors. The instrumentation relies on a separate runtime
5071 library (@file{libmudflap}), which will be linked into a program if
5072 @option{-fmudflap} is given at link time. Run-time behavior of the
5073 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5074 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5077 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5078 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5079 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5080 instrumentation should ignore pointer reads. This produces less
5081 instrumentation (and therefore faster execution) and still provides
5082 some protection against outright memory corrupting writes, but allows
5083 erroneously read data to propagate within a program.
5085 @item -fthread-jumps
5086 @opindex fthread-jumps
5087 Perform optimizations where we check to see if a jump branches to a
5088 location where another comparison subsumed by the first is found. If
5089 so, the first branch is redirected to either the destination of the
5090 second branch or a point immediately following it, depending on whether
5091 the condition is known to be true or false.
5093 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5095 @item -fsplit-wide-types
5096 @opindex -fsplit-wide-types
5097 When using a type that occupies multiple registers, such as @code{long
5098 long} on a 32-bit system, split the registers apart and allocate them
5099 independently. This normally generates better code for those types,
5100 but may make debugging more difficult.
5102 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5105 @item -fcse-follow-jumps
5106 @opindex fcse-follow-jumps
5107 In common subexpression elimination, scan through jump instructions
5108 when the target of the jump is not reached by any other path. For
5109 example, when CSE encounters an @code{if} statement with an
5110 @code{else} clause, CSE will follow the jump when the condition
5113 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5115 @item -fcse-skip-blocks
5116 @opindex fcse-skip-blocks
5117 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5118 follow jumps which conditionally skip over blocks. When CSE
5119 encounters a simple @code{if} statement with no else clause,
5120 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5121 body of the @code{if}.
5123 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5125 @item -frerun-cse-after-loop
5126 @opindex frerun-cse-after-loop
5127 Re-run common subexpression elimination after loop optimizations has been
5130 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5134 Perform a global common subexpression elimination pass.
5135 This pass also performs global constant and copy propagation.
5137 @emph{Note:} When compiling a program using computed gotos, a GCC
5138 extension, you may get better runtime performance if you disable
5139 the global common subexpression elimination pass by adding
5140 @option{-fno-gcse} to the command line.
5142 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5146 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5147 attempt to move loads which are only killed by stores into themselves. This
5148 allows a loop containing a load/store sequence to be changed to a load outside
5149 the loop, and a copy/store within the loop.
5151 Enabled by default when gcse is enabled.
5155 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5156 global common subexpression elimination. This pass will attempt to move
5157 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5158 loops containing a load/store sequence can be changed to a load before
5159 the loop and a store after the loop.
5161 Not enabled at any optimization level.
5165 When @option{-fgcse-las} is enabled, the global common subexpression
5166 elimination pass eliminates redundant loads that come after stores to the
5167 same memory location (both partial and full redundancies).
5169 Not enabled at any optimization level.
5171 @item -fgcse-after-reload
5172 @opindex fgcse-after-reload
5173 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5174 pass is performed after reload. The purpose of this pass is to cleanup
5177 @item -funsafe-loop-optimizations
5178 @opindex funsafe-loop-optimizations
5179 If given, the loop optimizer will assume that loop indices do not
5180 overflow, and that the loops with nontrivial exit condition are not
5181 infinite. This enables a wider range of loop optimizations even if
5182 the loop optimizer itself cannot prove that these assumptions are valid.
5183 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5184 if it finds this kind of loop.
5186 @item -fcrossjumping
5187 @opindex crossjumping
5188 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5189 resulting code may or may not perform better than without cross-jumping.
5191 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5193 @item -fif-conversion
5194 @opindex if-conversion
5195 Attempt to transform conditional jumps into branch-less equivalents. This
5196 include use of conditional moves, min, max, set flags and abs instructions, and
5197 some tricks doable by standard arithmetics. The use of conditional execution
5198 on chips where it is available is controlled by @code{if-conversion2}.
5200 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5202 @item -fif-conversion2
5203 @opindex if-conversion2
5204 Use conditional execution (where available) to transform conditional jumps into
5205 branch-less equivalents.
5207 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5209 @item -fdelete-null-pointer-checks
5210 @opindex fdelete-null-pointer-checks
5211 Use global dataflow analysis to identify and eliminate useless checks
5212 for null pointers. The compiler assumes that dereferencing a null
5213 pointer would have halted the program. If a pointer is checked after
5214 it has already been dereferenced, it cannot be null.
5216 In some environments, this assumption is not true, and programs can
5217 safely dereference null pointers. Use
5218 @option{-fno-delete-null-pointer-checks} to disable this optimization
5219 for programs which depend on that behavior.
5221 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5223 @item -fexpensive-optimizations
5224 @opindex fexpensive-optimizations
5225 Perform a number of minor optimizations that are relatively expensive.
5227 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5229 @item -foptimize-register-move
5231 @opindex foptimize-register-move
5233 Attempt to reassign register numbers in move instructions and as
5234 operands of other simple instructions in order to maximize the amount of
5235 register tying. This is especially helpful on machines with two-operand
5238 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5241 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5243 @item -fdelayed-branch
5244 @opindex fdelayed-branch
5245 If supported for the target machine, attempt to reorder instructions
5246 to exploit instruction slots available after delayed branch
5249 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5251 @item -fschedule-insns
5252 @opindex fschedule-insns
5253 If supported for the target machine, attempt to reorder instructions to
5254 eliminate execution stalls due to required data being unavailable. This
5255 helps machines that have slow floating point or memory load instructions
5256 by allowing other instructions to be issued until the result of the load
5257 or floating point instruction is required.
5259 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5261 @item -fschedule-insns2
5262 @opindex fschedule-insns2
5263 Similar to @option{-fschedule-insns}, but requests an additional pass of
5264 instruction scheduling after register allocation has been done. This is
5265 especially useful on machines with a relatively small number of
5266 registers and where memory load instructions take more than one cycle.
5268 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5270 @item -fno-sched-interblock
5271 @opindex fno-sched-interblock
5272 Don't schedule instructions across basic blocks. This is normally
5273 enabled by default when scheduling before register allocation, i.e.@:
5274 with @option{-fschedule-insns} or at @option{-O2} or higher.
5276 @item -fno-sched-spec
5277 @opindex fno-sched-spec
5278 Don't allow speculative motion of non-load instructions. This is normally
5279 enabled by default when scheduling before register allocation, i.e.@:
5280 with @option{-fschedule-insns} or at @option{-O2} or higher.
5282 @item -fsched-spec-load
5283 @opindex fsched-spec-load
5284 Allow speculative motion of some load instructions. This only makes
5285 sense when scheduling before register allocation, i.e.@: with
5286 @option{-fschedule-insns} or at @option{-O2} or higher.
5288 @item -fsched-spec-load-dangerous
5289 @opindex fsched-spec-load-dangerous
5290 Allow speculative motion of more load instructions. This only makes
5291 sense when scheduling before register allocation, i.e.@: with
5292 @option{-fschedule-insns} or at @option{-O2} or higher.
5294 @item -fsched-stalled-insns=@var{n}
5295 @opindex fsched-stalled-insns
5296 Define how many insns (if any) can be moved prematurely from the queue
5297 of stalled insns into the ready list, during the second scheduling pass.
5299 @item -fsched-stalled-insns-dep=@var{n}
5300 @opindex fsched-stalled-insns-dep
5301 Define how many insn groups (cycles) will be examined for a dependency
5302 on a stalled insn that is candidate for premature removal from the queue
5303 of stalled insns. Has an effect only during the second scheduling pass,
5304 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5306 @item -fsched2-use-superblocks
5307 @opindex fsched2-use-superblocks
5308 When scheduling after register allocation, do use superblock scheduling
5309 algorithm. Superblock scheduling allows motion across basic block boundaries
5310 resulting on faster schedules. This option is experimental, as not all machine
5311 descriptions used by GCC model the CPU closely enough to avoid unreliable
5312 results from the algorithm.
5314 This only makes sense when scheduling after register allocation, i.e.@: with
5315 @option{-fschedule-insns2} or at @option{-O2} or higher.
5317 @item -fsched2-use-traces
5318 @opindex fsched2-use-traces
5319 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5320 allocation and additionally perform code duplication in order to increase the
5321 size of superblocks using tracer pass. See @option{-ftracer} for details on
5324 This mode should produce faster but significantly longer programs. Also
5325 without @option{-fbranch-probabilities} the traces constructed may not
5326 match the reality and hurt the performance. This only makes
5327 sense when scheduling after register allocation, i.e.@: with
5328 @option{-fschedule-insns2} or at @option{-O2} or higher.
5332 Eliminates redundant extension instructions and move the non redundant
5333 ones to optimal placement using LCM.
5335 @item -freschedule-modulo-scheduled-loops
5336 @opindex fscheduling-in-modulo-scheduled-loops
5337 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5338 we may want to prevent the later scheduling passes from changing its schedule, we use this
5339 option to control that.
5341 @item -fcaller-saves
5342 @opindex fcaller-saves
5343 Enable values to be allocated in registers that will be clobbered by
5344 function calls, by emitting extra instructions to save and restore the
5345 registers around such calls. Such allocation is done only when it
5346 seems to result in better code than would otherwise be produced.
5348 This option is always enabled by default on certain machines, usually
5349 those which have no call-preserved registers to use instead.
5351 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5354 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5355 enabled by default at @option{-O2} and @option{-O3}.
5358 Perform Full Redundancy Elimination (FRE) on trees. The difference
5359 between FRE and PRE is that FRE only considers expressions
5360 that are computed on all paths leading to the redundant computation.
5361 This analysis faster than PRE, though it exposes fewer redundancies.
5362 This flag is enabled by default at @option{-O} and higher.
5364 @item -ftree-copy-prop
5365 Perform copy propagation on trees. This pass eliminates unnecessary
5366 copy operations. This flag is enabled by default at @option{-O} and
5369 @item -ftree-store-copy-prop
5370 Perform copy propagation of memory loads and stores. This pass
5371 eliminates unnecessary copy operations in memory references
5372 (structures, global variables, arrays, etc). This flag is enabled by
5373 default at @option{-O2} and higher.
5376 Perform structural alias analysis on trees. This flag
5377 is enabled by default at @option{-O} and higher.
5380 Perform interprocedural pointer analysis.
5383 Perform forward store motion on trees. This flag is
5384 enabled by default at @option{-O} and higher.
5387 Perform sparse conditional constant propagation (CCP) on trees. This
5388 pass only operates on local scalar variables and is enabled by default
5389 at @option{-O} and higher.
5391 @item -ftree-store-ccp
5392 Perform sparse conditional constant propagation (CCP) on trees. This
5393 pass operates on both local scalar variables and memory stores and
5394 loads (global variables, structures, arrays, etc). This flag is
5395 enabled by default at @option{-O2} and higher.
5398 Perform dead code elimination (DCE) on trees. This flag is enabled by
5399 default at @option{-O} and higher.
5401 @item -ftree-dominator-opts
5402 Perform a variety of simple scalar cleanups (constant/copy
5403 propagation, redundancy elimination, range propagation and expression
5404 simplification) based on a dominator tree traversal. This also
5405 performs jump threading (to reduce jumps to jumps). This flag is
5406 enabled by default at @option{-O} and higher.
5409 Perform loop header copying on trees. This is beneficial since it increases
5410 effectiveness of code motion optimizations. It also saves one jump. This flag
5411 is enabled by default at @option{-O} and higher. It is not enabled
5412 for @option{-Os}, since it usually increases code size.
5414 @item -ftree-loop-optimize
5415 Perform loop optimizations on trees. This flag is enabled by default
5416 at @option{-O} and higher.
5418 @item -ftree-loop-linear
5419 Perform linear loop transformations on tree. This flag can improve cache
5420 performance and allow further loop optimizations to take place.
5422 @item -ftree-loop-im
5423 Perform loop invariant motion on trees. This pass moves only invariants that
5424 would be hard to handle at RTL level (function calls, operations that expand to
5425 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5426 operands of conditions that are invariant out of the loop, so that we can use
5427 just trivial invariantness analysis in loop unswitching. The pass also includes
5430 @item -ftree-loop-ivcanon
5431 Create a canonical counter for number of iterations in the loop for that
5432 determining number of iterations requires complicated analysis. Later
5433 optimizations then may determine the number easily. Useful especially
5434 in connection with unrolling.
5437 Perform induction variable optimizations (strength reduction, induction
5438 variable merging and induction variable elimination) on trees.
5441 Perform scalar replacement of aggregates. This pass replaces structure
5442 references with scalars to prevent committing structures to memory too
5443 early. This flag is enabled by default at @option{-O} and higher.
5445 @item -ftree-copyrename
5446 Perform copy renaming on trees. This pass attempts to rename compiler
5447 temporaries to other variables at copy locations, usually resulting in
5448 variable names which more closely resemble the original variables. This flag
5449 is enabled by default at @option{-O} and higher.
5452 Perform temporary expression replacement during the SSA->normal phase. Single
5453 use/single def temporaries are replaced at their use location with their
5454 defining expression. This results in non-GIMPLE code, but gives the expanders
5455 much more complex trees to work on resulting in better RTL generation. This is
5456 enabled by default at @option{-O} and higher.
5459 Perform live range splitting during the SSA->normal phase. Distinct live
5460 ranges of a variable are split into unique variables, allowing for better
5461 optimization later. This is enabled by default at @option{-O} and higher.
5463 @item -ftree-vectorize
5464 Perform loop vectorization on trees.
5466 @item -ftree-vect-loop-version
5467 @opindex ftree-vect-loop-version
5468 Perform loop versioning when doing loop vectorization on trees. When a loop
5469 appears to be vectorizable except that data alignment or data dependence cannot
5470 be determined at compile time then vectorized and non-vectorized versions of
5471 the loop are generated along with runtime checks for alignment or dependence
5472 to control which version is executed. This option is enabled by default
5473 except at level @option{-Os} where it is disabled.
5476 Perform Value Range Propagation on trees. This is similar to the
5477 constant propagation pass, but instead of values, ranges of values are
5478 propagated. This allows the optimizers to remove unnecessary range
5479 checks like array bound checks and null pointer checks. This is
5480 enabled by default at @option{-O2} and higher. Null pointer check
5481 elimination is only done if @option{-fdelete-null-pointer-checks} is
5486 Perform tail duplication to enlarge superblock size. This transformation
5487 simplifies the control flow of the function allowing other optimizations to do
5490 @item -funroll-loops
5491 @opindex funroll-loops
5492 Unroll loops whose number of iterations can be determined at compile
5493 time or upon entry to the loop. @option{-funroll-loops} implies
5494 @option{-frerun-cse-after-loop}. This option makes code larger,
5495 and may or may not make it run faster.
5497 @item -funroll-all-loops
5498 @opindex funroll-all-loops
5499 Unroll all loops, even if their number of iterations is uncertain when
5500 the loop is entered. This usually makes programs run more slowly.
5501 @option{-funroll-all-loops} implies the same options as
5502 @option{-funroll-loops},
5504 @item -fsplit-ivs-in-unroller
5505 @opindex -fsplit-ivs-in-unroller
5506 Enables expressing of values of induction variables in later iterations
5507 of the unrolled loop using the value in the first iteration. This breaks
5508 long dependency chains, thus improving efficiency of the scheduling passes.
5510 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5511 same effect. However in cases the loop body is more complicated than
5512 a single basic block, this is not reliable. It also does not work at all
5513 on some of the architectures due to restrictions in the CSE pass.
5515 This optimization is enabled by default.
5517 @item -fvariable-expansion-in-unroller
5518 @opindex -fvariable-expansion-in-unroller
5519 With this option, the compiler will create multiple copies of some
5520 local variables when unrolling a loop which can result in superior code.
5522 @item -fprefetch-loop-arrays
5523 @opindex fprefetch-loop-arrays
5524 If supported by the target machine, generate instructions to prefetch
5525 memory to improve the performance of loops that access large arrays.
5527 This option may generate better or worse code; results are highly
5528 dependent on the structure of loops within the source code.
5530 Disabled at level @option{-Os}.
5533 @itemx -fno-peephole2
5534 @opindex fno-peephole
5535 @opindex fno-peephole2
5536 Disable any machine-specific peephole optimizations. The difference
5537 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5538 are implemented in the compiler; some targets use one, some use the
5539 other, a few use both.
5541 @option{-fpeephole} is enabled by default.
5542 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5544 @item -fno-guess-branch-probability
5545 @opindex fno-guess-branch-probability
5546 Do not guess branch probabilities using heuristics.
5548 GCC will use heuristics to guess branch probabilities if they are
5549 not provided by profiling feedback (@option{-fprofile-arcs}). These
5550 heuristics are based on the control flow graph. If some branch probabilities
5551 are specified by @samp{__builtin_expect}, then the heuristics will be
5552 used to guess branch probabilities for the rest of the control flow graph,
5553 taking the @samp{__builtin_expect} info into account. The interactions
5554 between the heuristics and @samp{__builtin_expect} can be complex, and in
5555 some cases, it may be useful to disable the heuristics so that the effects
5556 of @samp{__builtin_expect} are easier to understand.
5558 The default is @option{-fguess-branch-probability} at levels
5559 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5561 @item -freorder-blocks
5562 @opindex freorder-blocks
5563 Reorder basic blocks in the compiled function in order to reduce number of
5564 taken branches and improve code locality.
5566 Enabled at levels @option{-O2}, @option{-O3}.
5568 @item -freorder-blocks-and-partition
5569 @opindex freorder-blocks-and-partition
5570 In addition to reordering basic blocks in the compiled function, in order
5571 to reduce number of taken branches, partitions hot and cold basic blocks
5572 into separate sections of the assembly and .o files, to improve
5573 paging and cache locality performance.
5575 This optimization is automatically turned off in the presence of
5576 exception handling, for linkonce sections, for functions with a user-defined
5577 section attribute and on any architecture that does not support named
5580 @item -freorder-functions
5581 @opindex freorder-functions
5582 Reorder functions in the object file in order to
5583 improve code locality. This is implemented by using special
5584 subsections @code{.text.hot} for most frequently executed functions and
5585 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5586 the linker so object file format must support named sections and linker must
5587 place them in a reasonable way.
5589 Also profile feedback must be available in to make this option effective. See
5590 @option{-fprofile-arcs} for details.
5592 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5594 @item -fstrict-aliasing
5595 @opindex fstrict-aliasing
5596 Allows the compiler to assume the strictest aliasing rules applicable to
5597 the language being compiled. For C (and C++), this activates
5598 optimizations based on the type of expressions. In particular, an
5599 object of one type is assumed never to reside at the same address as an
5600 object of a different type, unless the types are almost the same. For
5601 example, an @code{unsigned int} can alias an @code{int}, but not a
5602 @code{void*} or a @code{double}. A character type may alias any other
5605 Pay special attention to code like this:
5618 The practice of reading from a different union member than the one most
5619 recently written to (called ``type-punning'') is common. Even with
5620 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5621 is accessed through the union type. So, the code above will work as
5622 expected. However, this code might not:
5633 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5635 @item -fstrict-overflow
5636 @opindex fstrict-overflow
5637 Allow the compiler to assume strict signed overflow rules, depending
5638 on the language being compiled. For C (and C++) this means that
5639 overflow when doing arithmetic with signed numbers is undefined, which
5640 means that the compiler may assume that it will not happen. This
5641 permits various optimizations. For example, the compiler will assume
5642 that an expression like @code{i + 10 > i} will always be true for
5643 signed @code{i}. This assumption is only valid if signed overflow is
5644 undefined, as the expression is false if @code{i + 10} overflows when
5645 using twos complement arithmetic. When this option is in effect any
5646 attempt to determine whether an operation on signed numbers will
5647 overflow must be written carefully to not actually involve overflow.
5649 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
5650 that signed overflow is fully defined: it wraps. When
5651 @option{-fwrapv} is used, there is no difference between
5652 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
5653 @option{-fwrapv} certain types of overflow are permitted. For
5654 example, if the compiler gets an overflow when doing arithmetic on
5655 constants, the overflowed value can still be used with
5656 @option{-fwrapv}, but not otherwise.
5658 The @option{-fstrict-overflow} option is enabled at levels
5659 @option{-O2}, @option{-O3}, @option{-Os}.
5661 @item -falign-functions
5662 @itemx -falign-functions=@var{n}
5663 @opindex falign-functions
5664 Align the start of functions to the next power-of-two greater than
5665 @var{n}, skipping up to @var{n} bytes. For instance,
5666 @option{-falign-functions=32} aligns functions to the next 32-byte
5667 boundary, but @option{-falign-functions=24} would align to the next
5668 32-byte boundary only if this can be done by skipping 23 bytes or less.
5670 @option{-fno-align-functions} and @option{-falign-functions=1} are
5671 equivalent and mean that functions will not be aligned.
5673 Some assemblers only support this flag when @var{n} is a power of two;
5674 in that case, it is rounded up.
5676 If @var{n} is not specified or is zero, use a machine-dependent default.
5678 Enabled at levels @option{-O2}, @option{-O3}.
5680 @item -falign-labels
5681 @itemx -falign-labels=@var{n}
5682 @opindex falign-labels
5683 Align all branch targets to a power-of-two boundary, skipping up to
5684 @var{n} bytes like @option{-falign-functions}. This option can easily
5685 make code slower, because it must insert dummy operations for when the
5686 branch target is reached in the usual flow of the code.
5688 @option{-fno-align-labels} and @option{-falign-labels=1} are
5689 equivalent and mean that labels will not be aligned.
5691 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5692 are greater than this value, then their values are used instead.
5694 If @var{n} is not specified or is zero, use a machine-dependent default
5695 which is very likely to be @samp{1}, meaning no alignment.
5697 Enabled at levels @option{-O2}, @option{-O3}.
5700 @itemx -falign-loops=@var{n}
5701 @opindex falign-loops
5702 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5703 like @option{-falign-functions}. The hope is that the loop will be
5704 executed many times, which will make up for any execution of the dummy
5707 @option{-fno-align-loops} and @option{-falign-loops=1} are
5708 equivalent and mean that loops will not be aligned.
5710 If @var{n} is not specified or is zero, use a machine-dependent default.
5712 Enabled at levels @option{-O2}, @option{-O3}.
5715 @itemx -falign-jumps=@var{n}
5716 @opindex falign-jumps
5717 Align branch targets to a power-of-two boundary, for branch targets
5718 where the targets can only be reached by jumping, skipping up to @var{n}
5719 bytes like @option{-falign-functions}. In this case, no dummy operations
5722 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5723 equivalent and mean that loops will not be aligned.
5725 If @var{n} is not specified or is zero, use a machine-dependent default.
5727 Enabled at levels @option{-O2}, @option{-O3}.
5729 @item -funit-at-a-time
5730 @opindex funit-at-a-time
5731 Parse the whole compilation unit before starting to produce code.
5732 This allows some extra optimizations to take place but consumes
5733 more memory (in general). There are some compatibility issues
5734 with @emph{unit-at-a-time} mode:
5737 enabling @emph{unit-at-a-time} mode may change the order
5738 in which functions, variables, and top-level @code{asm} statements
5739 are emitted, and will likely break code relying on some particular
5740 ordering. The majority of such top-level @code{asm} statements,
5741 though, can be replaced by @code{section} attributes. The
5742 @option{fno-toplevel-reorder} option may be used to keep the ordering
5743 used in the input file, at the cost of some optimizations.
5746 @emph{unit-at-a-time} mode removes unreferenced static variables
5747 and functions. This may result in undefined references
5748 when an @code{asm} statement refers directly to variables or functions
5749 that are otherwise unused. In that case either the variable/function
5750 shall be listed as an operand of the @code{asm} statement operand or,
5751 in the case of top-level @code{asm} statements the attribute @code{used}
5752 shall be used on the declaration.
5755 Static functions now can use non-standard passing conventions that
5756 may break @code{asm} statements calling functions directly. Again,
5757 attribute @code{used} will prevent this behavior.
5760 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5761 but this scheme may not be supported by future releases of GCC@.
5763 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5765 @item -fno-toplevel-reorder
5766 Do not reorder top-level functions, variables, and @code{asm}
5767 statements. Output them in the same order that they appear in the
5768 input file. When this option is used, unreferenced static variables
5769 will not be removed. This option is intended to support existing code
5770 which relies on a particular ordering. For new code, it is better to
5775 Constructs webs as commonly used for register allocation purposes and assign
5776 each web individual pseudo register. This allows the register allocation pass
5777 to operate on pseudos directly, but also strengthens several other optimization
5778 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5779 however, make debugging impossible, since variables will no longer stay in a
5782 Enabled by default with @option{-funroll-loops}.
5784 @item -fwhole-program
5785 @opindex fwhole-program
5786 Assume that the current compilation unit represents whole program being
5787 compiled. All public functions and variables with the exception of @code{main}
5788 and those merged by attribute @code{externally_visible} become static functions
5789 and in a affect gets more aggressively optimized by interprocedural optimizers.
5790 While this option is equivalent to proper use of @code{static} keyword for
5791 programs consisting of single file, in combination with option
5792 @option{--combine} this flag can be used to compile most of smaller scale C
5793 programs since the functions and variables become local for the whole combined
5794 compilation unit, not for the single source file itself.
5797 @item -fno-cprop-registers
5798 @opindex fno-cprop-registers
5799 After register allocation and post-register allocation instruction splitting,
5800 we perform a copy-propagation pass to try to reduce scheduling dependencies
5801 and occasionally eliminate the copy.
5803 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5805 @item -fprofile-generate
5806 @opindex fprofile-generate
5808 Enable options usually used for instrumenting application to produce
5809 profile useful for later recompilation with profile feedback based
5810 optimization. You must use @option{-fprofile-generate} both when
5811 compiling and when linking your program.
5813 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5816 @opindex fprofile-use
5817 Enable profile feedback directed optimizations, and optimizations
5818 generally profitable only with profile feedback available.
5820 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5821 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5823 By default, GCC emits an error message if the feedback profiles do not
5824 match the source code. This error can be turned into a warning by using
5825 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
5829 The following options control compiler behavior regarding floating
5830 point arithmetic. These options trade off between speed and
5831 correctness. All must be specifically enabled.
5835 @opindex ffloat-store
5836 Do not store floating point variables in registers, and inhibit other
5837 options that might change whether a floating point value is taken from a
5840 @cindex floating point precision
5841 This option prevents undesirable excess precision on machines such as
5842 the 68000 where the floating registers (of the 68881) keep more
5843 precision than a @code{double} is supposed to have. Similarly for the
5844 x86 architecture. For most programs, the excess precision does only
5845 good, but a few programs rely on the precise definition of IEEE floating
5846 point. Use @option{-ffloat-store} for such programs, after modifying
5847 them to store all pertinent intermediate computations into variables.
5851 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5852 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5853 @option{-fno-rounding-math}, @option{-fno-signaling-nans},
5854 @option{-fno-signed-zeros} and @option{fcx-limited-range}.
5856 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5858 This option should never be turned on by any @option{-O} option since
5859 it can result in incorrect output for programs which depend on
5860 an exact implementation of IEEE or ISO rules/specifications for
5863 @item -fno-math-errno
5864 @opindex fno-math-errno
5865 Do not set ERRNO after calling math functions that are executed
5866 with a single instruction, e.g., sqrt. A program that relies on
5867 IEEE exceptions for math error handling may want to use this flag
5868 for speed while maintaining IEEE arithmetic compatibility.
5870 This option should never be turned on by any @option{-O} option since
5871 it can result in incorrect output for programs which depend on
5872 an exact implementation of IEEE or ISO rules/specifications for
5875 The default is @option{-fmath-errno}.
5877 On Darwin systems, the math library never sets @code{errno}. There is
5878 therefore no reason for the compiler to consider the possibility that
5879 it might, and @option{-fno-math-errno} is the default.
5881 @item -funsafe-math-optimizations
5882 @opindex funsafe-math-optimizations
5883 Allow optimizations for floating-point arithmetic that (a) assume
5884 that arguments and results are valid and (b) may violate IEEE or
5885 ANSI standards. When used at link-time, it may include libraries
5886 or startup files that change the default FPU control word or other
5887 similar optimizations.
5889 This option should never be turned on by any @option{-O} option since
5890 it can result in incorrect output for programs which depend on
5891 an exact implementation of IEEE or ISO rules/specifications for
5894 The default is @option{-fno-unsafe-math-optimizations}.
5896 @item -ffinite-math-only
5897 @opindex ffinite-math-only
5898 Allow optimizations for floating-point arithmetic that assume
5899 that arguments and results are not NaNs or +-Infs.
5901 This option should never be turned on by any @option{-O} option since
5902 it can result in incorrect output for programs which depend on
5903 an exact implementation of IEEE or ISO rules/specifications.
5905 The default is @option{-fno-finite-math-only}.
5907 @item -fno-signed-zeros
5908 @opindex fno-signed-zeros
5909 Allow optimizations for floating point arithmetic that ignore the
5910 signedness of zero. IEEE arithmetic specifies the behavior of
5911 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
5912 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
5913 This option implies that the sign of a zero result isn't significant.
5915 The default is @option{-fsigned-zeros}.
5917 @item -fno-trapping-math
5918 @opindex fno-trapping-math
5919 Compile code assuming that floating-point operations cannot generate
5920 user-visible traps. These traps include division by zero, overflow,
5921 underflow, inexact result and invalid operation. This option implies
5922 @option{-fno-signaling-nans}. Setting this option may allow faster
5923 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5925 This option should never be turned on by any @option{-O} option since
5926 it can result in incorrect output for programs which depend on
5927 an exact implementation of IEEE or ISO rules/specifications for
5930 The default is @option{-ftrapping-math}.
5932 @item -frounding-math
5933 @opindex frounding-math
5934 Disable transformations and optimizations that assume default floating
5935 point rounding behavior. This is round-to-zero for all floating point
5936 to integer conversions, and round-to-nearest for all other arithmetic
5937 truncations. This option should be specified for programs that change
5938 the FP rounding mode dynamically, or that may be executed with a
5939 non-default rounding mode. This option disables constant folding of
5940 floating point expressions at compile-time (which may be affected by
5941 rounding mode) and arithmetic transformations that are unsafe in the
5942 presence of sign-dependent rounding modes.
5944 The default is @option{-fno-rounding-math}.
5946 This option is experimental and does not currently guarantee to
5947 disable all GCC optimizations that are affected by rounding mode.
5948 Future versions of GCC may provide finer control of this setting
5949 using C99's @code{FENV_ACCESS} pragma. This command line option
5950 will be used to specify the default state for @code{FENV_ACCESS}.
5952 @item -frtl-abstract-sequences
5953 @opindex frtl-abstract-sequences
5954 It is a size optimization method. This option is to find identical
5955 sequences of code, which can be turned into pseudo-procedures and
5956 then replace all occurrences with calls to the newly created
5957 subroutine. It is kind of an opposite of @option{-finline-functions}.
5958 This optimization runs at RTL level.
5960 @item -fsignaling-nans
5961 @opindex fsignaling-nans
5962 Compile code assuming that IEEE signaling NaNs may generate user-visible
5963 traps during floating-point operations. Setting this option disables
5964 optimizations that may change the number of exceptions visible with
5965 signaling NaNs. This option implies @option{-ftrapping-math}.
5967 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5970 The default is @option{-fno-signaling-nans}.
5972 This option is experimental and does not currently guarantee to
5973 disable all GCC optimizations that affect signaling NaN behavior.
5975 @item -fsingle-precision-constant
5976 @opindex fsingle-precision-constant
5977 Treat floating point constant as single precision constant instead of
5978 implicitly converting it to double precision constant.
5980 @item -fcx-limited-range
5981 @itemx -fno-cx-limited-range
5982 @opindex fcx-limited-range
5983 @opindex fno-cx-limited-range
5984 When enabled, this option states that a range reduction step is not
5985 needed when performing complex division. The default is
5986 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5988 This option controls the default setting of the ISO C99
5989 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5994 The following options control optimizations that may improve
5995 performance, but are not enabled by any @option{-O} options. This
5996 section includes experimental options that may produce broken code.
5999 @item -fbranch-probabilities
6000 @opindex fbranch-probabilities
6001 After running a program compiled with @option{-fprofile-arcs}
6002 (@pxref{Debugging Options,, Options for Debugging Your Program or
6003 @command{gcc}}), you can compile it a second time using
6004 @option{-fbranch-probabilities}, to improve optimizations based on
6005 the number of times each branch was taken. When the program
6006 compiled with @option{-fprofile-arcs} exits it saves arc execution
6007 counts to a file called @file{@var{sourcename}.gcda} for each source
6008 file. The information in this data file is very dependent on the
6009 structure of the generated code, so you must use the same source code
6010 and the same optimization options for both compilations.
6012 With @option{-fbranch-probabilities}, GCC puts a
6013 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6014 These can be used to improve optimization. Currently, they are only
6015 used in one place: in @file{reorg.c}, instead of guessing which path a
6016 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6017 exactly determine which path is taken more often.
6019 @item -fprofile-values
6020 @opindex fprofile-values
6021 If combined with @option{-fprofile-arcs}, it adds code so that some
6022 data about values of expressions in the program is gathered.
6024 With @option{-fbranch-probabilities}, it reads back the data gathered
6025 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6026 notes to instructions for their later usage in optimizations.
6028 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6032 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6033 a code to gather information about values of expressions.
6035 With @option{-fbranch-probabilities}, it reads back the data gathered
6036 and actually performs the optimizations based on them.
6037 Currently the optimizations include specialization of division operation
6038 using the knowledge about the value of the denominator.
6040 @item -frename-registers
6041 @opindex frename-registers
6042 Attempt to avoid false dependencies in scheduled code by making use
6043 of registers left over after register allocation. This optimization
6044 will most benefit processors with lots of registers. Depending on the
6045 debug information format adopted by the target, however, it can
6046 make debugging impossible, since variables will no longer stay in
6047 a ``home register''.
6049 Enabled by default with @option{-funroll-loops}.
6053 Perform tail duplication to enlarge superblock size. This transformation
6054 simplifies the control flow of the function allowing other optimizations to do
6057 Enabled with @option{-fprofile-use}.
6059 @item -funroll-loops
6060 @opindex funroll-loops
6061 Unroll loops whose number of iterations can be determined at compile time or
6062 upon entry to the loop. @option{-funroll-loops} implies
6063 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6064 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6065 small constant number of iterations). This option makes code larger, and may
6066 or may not make it run faster.
6068 Enabled with @option{-fprofile-use}.
6070 @item -funroll-all-loops
6071 @opindex funroll-all-loops
6072 Unroll all loops, even if their number of iterations is uncertain when
6073 the loop is entered. This usually makes programs run more slowly.
6074 @option{-funroll-all-loops} implies the same options as
6075 @option{-funroll-loops}.
6078 @opindex fpeel-loops
6079 Peels the loops for that there is enough information that they do not
6080 roll much (from profile feedback). It also turns on complete loop peeling
6081 (i.e.@: complete removal of loops with small constant number of iterations).
6083 Enabled with @option{-fprofile-use}.
6085 @item -fmove-loop-invariants
6086 @opindex fmove-loop-invariants
6087 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6088 at level @option{-O1}
6090 @item -funswitch-loops
6091 @opindex funswitch-loops
6092 Move branches with loop invariant conditions out of the loop, with duplicates
6093 of the loop on both branches (modified according to result of the condition).
6095 @item -ffunction-sections
6096 @itemx -fdata-sections
6097 @opindex ffunction-sections
6098 @opindex fdata-sections
6099 Place each function or data item into its own section in the output
6100 file if the target supports arbitrary sections. The name of the
6101 function or the name of the data item determines the section's name
6104 Use these options on systems where the linker can perform optimizations
6105 to improve locality of reference in the instruction space. Most systems
6106 using the ELF object format and SPARC processors running Solaris 2 have
6107 linkers with such optimizations. AIX may have these optimizations in
6110 Only use these options when there are significant benefits from doing
6111 so. When you specify these options, the assembler and linker will
6112 create larger object and executable files and will also be slower.
6113 You will not be able to use @code{gprof} on all systems if you
6114 specify this option and you may have problems with debugging if
6115 you specify both this option and @option{-g}.
6117 @item -fbranch-target-load-optimize
6118 @opindex fbranch-target-load-optimize
6119 Perform branch target register load optimization before prologue / epilogue
6121 The use of target registers can typically be exposed only during reload,
6122 thus hoisting loads out of loops and doing inter-block scheduling needs
6123 a separate optimization pass.
6125 @item -fbranch-target-load-optimize2
6126 @opindex fbranch-target-load-optimize2
6127 Perform branch target register load optimization after prologue / epilogue
6130 @item -fbtr-bb-exclusive
6131 @opindex fbtr-bb-exclusive
6132 When performing branch target register load optimization, don't reuse
6133 branch target registers in within any basic block.
6135 @item -fstack-protector
6136 Emit extra code to check for buffer overflows, such as stack smashing
6137 attacks. This is done by adding a guard variable to functions with
6138 vulnerable objects. This includes functions that call alloca, and
6139 functions with buffers larger than 8 bytes. The guards are initialized
6140 when a function is entered and then checked when the function exits.
6141 If a guard check fails, an error message is printed and the program exits.
6143 @item -fstack-protector-all
6144 Like @option{-fstack-protector} except that all functions are protected.
6146 @item -fsection-anchors
6147 @opindex fsection-anchors
6148 Try to reduce the number of symbolic address calculations by using
6149 shared ``anchor'' symbols to address nearby objects. This transformation
6150 can help to reduce the number of GOT entries and GOT accesses on some
6153 For example, the implementation of the following function @code{foo}:
6157 int foo (void) @{ return a + b + c; @}
6160 would usually calculate the addresses of all three variables, but if you
6161 compile it with @option{-fsection-anchors}, it will access the variables
6162 from a common anchor point instead. The effect is similar to the
6163 following pseudocode (which isn't valid C):
6168 register int *xr = &x;
6169 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6173 Not all targets support this option.
6175 @item --param @var{name}=@var{value}
6177 In some places, GCC uses various constants to control the amount of
6178 optimization that is done. For example, GCC will not inline functions
6179 that contain more that a certain number of instructions. You can
6180 control some of these constants on the command-line using the
6181 @option{--param} option.
6183 The names of specific parameters, and the meaning of the values, are
6184 tied to the internals of the compiler, and are subject to change
6185 without notice in future releases.
6187 In each case, the @var{value} is an integer. The allowable choices for
6188 @var{name} are given in the following table:
6191 @item salias-max-implicit-fields
6192 The maximum number of fields in a variable without direct
6193 structure accesses for which structure aliasing will consider trying
6194 to track each field. The default is 5
6196 @item salias-max-array-elements
6197 The maximum number of elements an array can have and its elements
6198 still be tracked individually by structure aliasing. The default is 4
6200 @item sra-max-structure-size
6201 The maximum structure size, in bytes, at which the scalar replacement
6202 of aggregates (SRA) optimization will perform block copies. The
6203 default value, 0, implies that GCC will select the most appropriate
6206 @item sra-field-structure-ratio
6207 The threshold ratio (as a percentage) between instantiated fields and
6208 the complete structure size. We say that if the ratio of the number
6209 of bytes in instantiated fields to the number of bytes in the complete
6210 structure exceeds this parameter, then block copies are not used. The
6213 @item max-crossjump-edges
6214 The maximum number of incoming edges to consider for crossjumping.
6215 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6216 the number of edges incoming to each block. Increasing values mean
6217 more aggressive optimization, making the compile time increase with
6218 probably small improvement in executable size.
6220 @item min-crossjump-insns
6221 The minimum number of instructions which must be matched at the end
6222 of two blocks before crossjumping will be performed on them. This
6223 value is ignored in the case where all instructions in the block being
6224 crossjumped from are matched. The default value is 5.
6226 @item max-grow-copy-bb-insns
6227 The maximum code size expansion factor when copying basic blocks
6228 instead of jumping. The expansion is relative to a jump instruction.
6229 The default value is 8.
6231 @item max-goto-duplication-insns
6232 The maximum number of instructions to duplicate to a block that jumps
6233 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6234 passes, GCC factors computed gotos early in the compilation process,
6235 and unfactors them as late as possible. Only computed jumps at the
6236 end of a basic blocks with no more than max-goto-duplication-insns are
6237 unfactored. The default value is 8.
6239 @item max-delay-slot-insn-search
6240 The maximum number of instructions to consider when looking for an
6241 instruction to fill a delay slot. If more than this arbitrary number of
6242 instructions is searched, the time savings from filling the delay slot
6243 will be minimal so stop searching. Increasing values mean more
6244 aggressive optimization, making the compile time increase with probably
6245 small improvement in executable run time.
6247 @item max-delay-slot-live-search
6248 When trying to fill delay slots, the maximum number of instructions to
6249 consider when searching for a block with valid live register
6250 information. Increasing this arbitrarily chosen value means more
6251 aggressive optimization, increasing the compile time. This parameter
6252 should be removed when the delay slot code is rewritten to maintain the
6255 @item max-gcse-memory
6256 The approximate maximum amount of memory that will be allocated in
6257 order to perform the global common subexpression elimination
6258 optimization. If more memory than specified is required, the
6259 optimization will not be done.
6261 @item max-gcse-passes
6262 The maximum number of passes of GCSE to run. The default is 1.
6264 @item max-pending-list-length
6265 The maximum number of pending dependencies scheduling will allow
6266 before flushing the current state and starting over. Large functions
6267 with few branches or calls can create excessively large lists which
6268 needlessly consume memory and resources.
6270 @item max-inline-insns-single
6271 Several parameters control the tree inliner used in gcc.
6272 This number sets the maximum number of instructions (counted in GCC's
6273 internal representation) in a single function that the tree inliner
6274 will consider for inlining. This only affects functions declared
6275 inline and methods implemented in a class declaration (C++).
6276 The default value is 450.
6278 @item max-inline-insns-auto
6279 When you use @option{-finline-functions} (included in @option{-O3}),
6280 a lot of functions that would otherwise not be considered for inlining
6281 by the compiler will be investigated. To those functions, a different
6282 (more restrictive) limit compared to functions declared inline can
6284 The default value is 90.
6286 @item large-function-insns
6287 The limit specifying really large functions. For functions larger than this
6288 limit after inlining inlining is constrained by
6289 @option{--param large-function-growth}. This parameter is useful primarily
6290 to avoid extreme compilation time caused by non-linear algorithms used by the
6292 This parameter is ignored when @option{-funit-at-a-time} is not used.
6293 The default value is 2700.
6295 @item large-function-growth
6296 Specifies maximal growth of large function caused by inlining in percents.
6297 This parameter is ignored when @option{-funit-at-a-time} is not used.
6298 The default value is 100 which limits large function growth to 2.0 times
6301 @item large-unit-insns
6302 The limit specifying large translation unit. Growth caused by inlining of
6303 units larger than this limit is limited by @option{--param inline-unit-growth}.
6304 For small units this might be too tight (consider unit consisting of function A
6305 that is inline and B that just calls A three time. If B is small relative to
6306 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6307 large units consisting of small inlininable functions however the overall unit
6308 growth limit is needed to avoid exponential explosion of code size. Thus for
6309 smaller units, the size is increased to @option{--param large-unit-insns}
6310 before applying @option{--param inline-unit-growth}. The default is 10000
6312 @item inline-unit-growth
6313 Specifies maximal overall growth of the compilation unit caused by inlining.
6314 This parameter is ignored when @option{-funit-at-a-time} is not used.
6315 The default value is 30 which limits unit growth to 1.3 times the original
6318 @item large-stack-frame
6319 The limit specifying large stack frames. While inlining the algorithm is trying
6320 to not grow past this limit too much. Default value is 256 bytes.
6322 @item large-stack-frame-growth
6323 Specifies maximal growth of large stack frames caused by inlining in percents.
6324 The default value is 1000 which limits large stack frame growth to 11 times
6327 @item max-inline-insns-recursive
6328 @itemx max-inline-insns-recursive-auto
6329 Specifies maximum number of instructions out-of-line copy of self recursive inline
6330 function can grow into by performing recursive inlining.
6332 For functions declared inline @option{--param max-inline-insns-recursive} is
6333 taken into account. For function not declared inline, recursive inlining
6334 happens only when @option{-finline-functions} (included in @option{-O3}) is
6335 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6336 default value is 450.
6338 @item max-inline-recursive-depth
6339 @itemx max-inline-recursive-depth-auto
6340 Specifies maximum recursion depth used by the recursive inlining.
6342 For functions declared inline @option{--param max-inline-recursive-depth} is
6343 taken into account. For function not declared inline, recursive inlining
6344 happens only when @option{-finline-functions} (included in @option{-O3}) is
6345 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6346 default value is 450.
6348 @item min-inline-recursive-probability
6349 Recursive inlining is profitable only for function having deep recursion
6350 in average and can hurt for function having little recursion depth by
6351 increasing the prologue size or complexity of function body to other
6354 When profile feedback is available (see @option{-fprofile-generate}) the actual
6355 recursion depth can be guessed from probability that function will recurse via
6356 given call expression. This parameter limits inlining only to call expression
6357 whose probability exceeds given threshold (in percents). The default value is
6360 @item inline-call-cost
6361 Specify cost of call instruction relative to simple arithmetics operations
6362 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6363 functions and at the same time increases size of leaf function that is believed to
6364 reduce function size by being inlined. In effect it increases amount of
6365 inlining for code having large abstraction penalty (many functions that just
6366 pass the arguments to other functions) and decrease inlining for code with low
6367 abstraction penalty. The default value is 16.
6369 @item min-vect-loop-bound
6370 The minimum number of iterations under which a loop will not get vectorized
6371 when @option{-ftree-vectorize} is used. The number of iterations after
6372 vectorization needs to be greater than the value specified by this option
6373 to allow vectorization. The default value is 0.
6375 @item max-unrolled-insns
6376 The maximum number of instructions that a loop should have if that loop
6377 is unrolled, and if the loop is unrolled, it determines how many times
6378 the loop code is unrolled.
6380 @item max-average-unrolled-insns
6381 The maximum number of instructions biased by probabilities of their execution
6382 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6383 it determines how many times the loop code is unrolled.
6385 @item max-unroll-times
6386 The maximum number of unrollings of a single loop.
6388 @item max-peeled-insns
6389 The maximum number of instructions that a loop should have if that loop
6390 is peeled, and if the loop is peeled, it determines how many times
6391 the loop code is peeled.
6393 @item max-peel-times
6394 The maximum number of peelings of a single loop.
6396 @item max-completely-peeled-insns
6397 The maximum number of insns of a completely peeled loop.
6399 @item max-completely-peel-times
6400 The maximum number of iterations of a loop to be suitable for complete peeling.
6402 @item max-unswitch-insns
6403 The maximum number of insns of an unswitched loop.
6405 @item max-unswitch-level
6406 The maximum number of branches unswitched in a single loop.
6409 The minimum cost of an expensive expression in the loop invariant motion.
6411 @item iv-consider-all-candidates-bound
6412 Bound on number of candidates for induction variables below that
6413 all candidates are considered for each use in induction variable
6414 optimizations. Only the most relevant candidates are considered
6415 if there are more candidates, to avoid quadratic time complexity.
6417 @item iv-max-considered-uses
6418 The induction variable optimizations give up on loops that contain more
6419 induction variable uses.
6421 @item iv-always-prune-cand-set-bound
6422 If number of candidates in the set is smaller than this value,
6423 we always try to remove unnecessary ivs from the set during its
6424 optimization when a new iv is added to the set.
6426 @item scev-max-expr-size
6427 Bound on size of expressions used in the scalar evolutions analyzer.
6428 Large expressions slow the analyzer.
6430 @item vect-max-version-checks
6431 The maximum number of runtime checks that can be performed when doing
6432 loop versioning in the vectorizer. See option ftree-vect-loop-version
6433 for more information.
6435 @item max-iterations-to-track
6437 The maximum number of iterations of a loop the brute force algorithm
6438 for analysis of # of iterations of the loop tries to evaluate.
6440 @item hot-bb-count-fraction
6441 Select fraction of the maximal count of repetitions of basic block in program
6442 given basic block needs to have to be considered hot.
6444 @item hot-bb-frequency-fraction
6445 Select fraction of the maximal frequency of executions of basic block in
6446 function given basic block needs to have to be considered hot
6448 @item max-predicted-iterations
6449 The maximum number of loop iterations we predict statically. This is useful
6450 in cases where function contain single loop with known bound and other loop
6451 with unknown. We predict the known number of iterations correctly, while
6452 the unknown number of iterations average to roughly 10. This means that the
6453 loop without bounds would appear artificially cold relative to the other one.
6455 @item tracer-dynamic-coverage
6456 @itemx tracer-dynamic-coverage-feedback
6458 This value is used to limit superblock formation once the given percentage of
6459 executed instructions is covered. This limits unnecessary code size
6462 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6463 feedback is available. The real profiles (as opposed to statically estimated
6464 ones) are much less balanced allowing the threshold to be larger value.
6466 @item tracer-max-code-growth
6467 Stop tail duplication once code growth has reached given percentage. This is
6468 rather hokey argument, as most of the duplicates will be eliminated later in
6469 cross jumping, so it may be set to much higher values than is the desired code
6472 @item tracer-min-branch-ratio
6474 Stop reverse growth when the reverse probability of best edge is less than this
6475 threshold (in percent).
6477 @item tracer-min-branch-ratio
6478 @itemx tracer-min-branch-ratio-feedback
6480 Stop forward growth if the best edge do have probability lower than this
6483 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6484 compilation for profile feedback and one for compilation without. The value
6485 for compilation with profile feedback needs to be more conservative (higher) in
6486 order to make tracer effective.
6488 @item max-cse-path-length
6490 Maximum number of basic blocks on path that cse considers. The default is 10.
6493 The maximum instructions CSE process before flushing. The default is 1000.
6495 @item max-aliased-vops
6497 Maximum number of virtual operands per statement allowed to represent
6498 aliases before triggering the alias grouping heuristic. Alias
6499 grouping reduces compile times and memory consumption needed for
6500 aliasing at the expense of precision loss in alias information.
6502 @item ggc-min-expand
6504 GCC uses a garbage collector to manage its own memory allocation. This
6505 parameter specifies the minimum percentage by which the garbage
6506 collector's heap should be allowed to expand between collections.
6507 Tuning this may improve compilation speed; it has no effect on code
6510 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6511 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6512 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6513 GCC is not able to calculate RAM on a particular platform, the lower
6514 bound of 30% is used. Setting this parameter and
6515 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6516 every opportunity. This is extremely slow, but can be useful for
6519 @item ggc-min-heapsize
6521 Minimum size of the garbage collector's heap before it begins bothering
6522 to collect garbage. The first collection occurs after the heap expands
6523 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6524 tuning this may improve compilation speed, and has no effect on code
6527 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6528 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6529 with a lower bound of 4096 (four megabytes) and an upper bound of
6530 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6531 particular platform, the lower bound is used. Setting this parameter
6532 very large effectively disables garbage collection. Setting this
6533 parameter and @option{ggc-min-expand} to zero causes a full collection
6534 to occur at every opportunity.
6536 @item max-reload-search-insns
6537 The maximum number of instruction reload should look backward for equivalent
6538 register. Increasing values mean more aggressive optimization, making the
6539 compile time increase with probably slightly better performance. The default
6542 @item max-cselib-memory-locations
6543 The maximum number of memory locations cselib should take into account.
6544 Increasing values mean more aggressive optimization, making the compile time
6545 increase with probably slightly better performance. The default value is 500.
6547 @item max-flow-memory-locations
6548 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6549 The default value is 100.
6551 @item reorder-blocks-duplicate
6552 @itemx reorder-blocks-duplicate-feedback
6554 Used by basic block reordering pass to decide whether to use unconditional
6555 branch or duplicate the code on its destination. Code is duplicated when its
6556 estimated size is smaller than this value multiplied by the estimated size of
6557 unconditional jump in the hot spots of the program.
6559 The @option{reorder-block-duplicate-feedback} is used only when profile
6560 feedback is available and may be set to higher values than
6561 @option{reorder-block-duplicate} since information about the hot spots is more
6564 @item max-sched-ready-insns
6565 The maximum number of instructions ready to be issued the scheduler should
6566 consider at any given time during the first scheduling pass. Increasing
6567 values mean more thorough searches, making the compilation time increase
6568 with probably little benefit. The default value is 100.
6570 @item max-sched-region-blocks
6571 The maximum number of blocks in a region to be considered for
6572 interblock scheduling. The default value is 10.
6574 @item max-sched-region-insns
6575 The maximum number of insns in a region to be considered for
6576 interblock scheduling. The default value is 100.
6579 The minimum probability (in percents) of reaching a source block
6580 for interblock speculative scheduling. The default value is 40.
6582 @item max-sched-extend-regions-iters
6583 The maximum number of iterations through CFG to extend regions.
6584 0 - disable region extension,
6585 N - do at most N iterations.
6586 The default value is 0.
6588 @item max-sched-insn-conflict-delay
6589 The maximum conflict delay for an insn to be considered for speculative motion.
6590 The default value is 3.
6592 @item sched-spec-prob-cutoff
6593 The minimal probability of speculation success (in percents), so that
6594 speculative insn will be scheduled.
6595 The default value is 40.
6597 @item max-last-value-rtl
6599 The maximum size measured as number of RTLs that can be recorded in an expression
6600 in combiner for a pseudo register as last known value of that register. The default
6603 @item integer-share-limit
6604 Small integer constants can use a shared data structure, reducing the
6605 compiler's memory usage and increasing its speed. This sets the maximum
6606 value of a shared integer constant's. The default value is 256.
6608 @item min-virtual-mappings
6609 Specifies the minimum number of virtual mappings in the incremental
6610 SSA updater that should be registered to trigger the virtual mappings
6611 heuristic defined by virtual-mappings-ratio. The default value is
6614 @item virtual-mappings-ratio
6615 If the number of virtual mappings is virtual-mappings-ratio bigger
6616 than the number of virtual symbols to be updated, then the incremental
6617 SSA updater switches to a full update for those symbols. The default
6620 @item ssp-buffer-size
6621 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6622 protection when @option{-fstack-protection} is used.
6624 @item max-jump-thread-duplication-stmts
6625 Maximum number of statements allowed in a block that needs to be
6626 duplicated when threading jumps.
6628 @item max-fields-for-field-sensitive
6629 Maximum number of fields in a structure we will treat in
6630 a field sensitive manner during pointer analysis.
6632 @item prefetch-latency
6633 Estimate on average number of instructions that are executed before
6634 prefetch finishes. The distance we prefetch ahead is proportional
6635 to this constant. Increasing this number may also lead to less
6636 streams being prefetched (see @option{simultaneous-prefetches}).
6638 @item simultaneous-prefetches
6639 Maximum number of prefetches that can run at the same time.
6641 @item l1-cache-line-size
6642 The size of cache line in L1 cache, in bytes.
6645 The number of cache lines in L1 cache.
6647 @item verify-canonical-types
6648 Whether the compiler should verify the ``canonical'' types used for
6649 type equality comparisons within the C++ and Objective-C++ front
6650 ends. Set to 1 (the default when GCC is configured with
6651 --enable-checking) to enable verification, 0 to disable verification
6652 (the default when GCC is configured with --disable-checking).
6657 @node Preprocessor Options
6658 @section Options Controlling the Preprocessor
6659 @cindex preprocessor options
6660 @cindex options, preprocessor
6662 These options control the C preprocessor, which is run on each C source
6663 file before actual compilation.
6665 If you use the @option{-E} option, nothing is done except preprocessing.
6666 Some of these options make sense only together with @option{-E} because
6667 they cause the preprocessor output to be unsuitable for actual
6672 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6673 and pass @var{option} directly through to the preprocessor. If
6674 @var{option} contains commas, it is split into multiple options at the
6675 commas. However, many options are modified, translated or interpreted
6676 by the compiler driver before being passed to the preprocessor, and
6677 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6678 interface is undocumented and subject to change, so whenever possible
6679 you should avoid using @option{-Wp} and let the driver handle the
6682 @item -Xpreprocessor @var{option}
6683 @opindex preprocessor
6684 Pass @var{option} as an option to the preprocessor. You can use this to
6685 supply system-specific preprocessor options which GCC does not know how to
6688 If you want to pass an option that takes an argument, you must use
6689 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6692 @include cppopts.texi
6694 @node Assembler Options
6695 @section Passing Options to the Assembler
6697 @c prevent bad page break with this line
6698 You can pass options to the assembler.
6701 @item -Wa,@var{option}
6703 Pass @var{option} as an option to the assembler. If @var{option}
6704 contains commas, it is split into multiple options at the commas.
6706 @item -Xassembler @var{option}
6708 Pass @var{option} as an option to the assembler. You can use this to
6709 supply system-specific assembler options which GCC does not know how to
6712 If you want to pass an option that takes an argument, you must use
6713 @option{-Xassembler} twice, once for the option and once for the argument.
6718 @section Options for Linking
6719 @cindex link options
6720 @cindex options, linking
6722 These options come into play when the compiler links object files into
6723 an executable output file. They are meaningless if the compiler is
6724 not doing a link step.
6728 @item @var{object-file-name}
6729 A file name that does not end in a special recognized suffix is
6730 considered to name an object file or library. (Object files are
6731 distinguished from libraries by the linker according to the file
6732 contents.) If linking is done, these object files are used as input
6741 If any of these options is used, then the linker is not run, and
6742 object file names should not be used as arguments. @xref{Overall
6746 @item -l@var{library}
6747 @itemx -l @var{library}
6749 Search the library named @var{library} when linking. (The second
6750 alternative with the library as a separate argument is only for
6751 POSIX compliance and is not recommended.)
6753 It makes a difference where in the command you write this option; the
6754 linker searches and processes libraries and object files in the order they
6755 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6756 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6757 to functions in @samp{z}, those functions may not be loaded.
6759 The linker searches a standard list of directories for the library,
6760 which is actually a file named @file{lib@var{library}.a}. The linker
6761 then uses this file as if it had been specified precisely by name.
6763 The directories searched include several standard system directories
6764 plus any that you specify with @option{-L}.
6766 Normally the files found this way are library files---archive files
6767 whose members are object files. The linker handles an archive file by
6768 scanning through it for members which define symbols that have so far
6769 been referenced but not defined. But if the file that is found is an
6770 ordinary object file, it is linked in the usual fashion. The only
6771 difference between using an @option{-l} option and specifying a file name
6772 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6773 and searches several directories.
6777 You need this special case of the @option{-l} option in order to
6778 link an Objective-C or Objective-C++ program.
6781 @opindex nostartfiles
6782 Do not use the standard system startup files when linking.
6783 The standard system libraries are used normally, unless @option{-nostdlib}
6784 or @option{-nodefaultlibs} is used.
6786 @item -nodefaultlibs
6787 @opindex nodefaultlibs
6788 Do not use the standard system libraries when linking.
6789 Only the libraries you specify will be passed to the linker.
6790 The standard startup files are used normally, unless @option{-nostartfiles}
6791 is used. The compiler may generate calls to @code{memcmp},
6792 @code{memset}, @code{memcpy} and @code{memmove}.
6793 These entries are usually resolved by entries in
6794 libc. These entry points should be supplied through some other
6795 mechanism when this option is specified.
6799 Do not use the standard system startup files or libraries when linking.
6800 No startup files and only the libraries you specify will be passed to
6801 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6802 @code{memcpy} and @code{memmove}.
6803 These entries are usually resolved by entries in
6804 libc. These entry points should be supplied through some other
6805 mechanism when this option is specified.
6807 @cindex @option{-lgcc}, use with @option{-nostdlib}
6808 @cindex @option{-nostdlib} and unresolved references
6809 @cindex unresolved references and @option{-nostdlib}
6810 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6811 @cindex @option{-nodefaultlibs} and unresolved references
6812 @cindex unresolved references and @option{-nodefaultlibs}
6813 One of the standard libraries bypassed by @option{-nostdlib} and
6814 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6815 that GCC uses to overcome shortcomings of particular machines, or special
6816 needs for some languages.
6817 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6818 Collection (GCC) Internals},
6819 for more discussion of @file{libgcc.a}.)
6820 In most cases, you need @file{libgcc.a} even when you want to avoid
6821 other standard libraries. In other words, when you specify @option{-nostdlib}
6822 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6823 This ensures that you have no unresolved references to internal GCC
6824 library subroutines. (For example, @samp{__main}, used to ensure C++
6825 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6826 GNU Compiler Collection (GCC) Internals}.)
6830 Produce a position independent executable on targets which support it.
6831 For predictable results, you must also specify the same set of options
6832 that were used to generate code (@option{-fpie}, @option{-fPIE},
6833 or model suboptions) when you specify this option.
6837 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6838 that support it. This instructs the linker to add all symbols, not
6839 only used ones, to the dynamic symbol table. This option is needed
6840 for some uses of @code{dlopen} or to allow obtaining backtraces
6841 from within a program.
6845 Remove all symbol table and relocation information from the executable.
6849 On systems that support dynamic linking, this prevents linking with the shared
6850 libraries. On other systems, this option has no effect.
6854 Produce a shared object which can then be linked with other objects to
6855 form an executable. Not all systems support this option. For predictable
6856 results, you must also specify the same set of options that were used to
6857 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6858 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6859 needs to build supplementary stub code for constructors to work. On
6860 multi-libbed systems, @samp{gcc -shared} must select the correct support
6861 libraries to link against. Failing to supply the correct flags may lead
6862 to subtle defects. Supplying them in cases where they are not necessary
6865 @item -shared-libgcc
6866 @itemx -static-libgcc
6867 @opindex shared-libgcc
6868 @opindex static-libgcc
6869 On systems that provide @file{libgcc} as a shared library, these options
6870 force the use of either the shared or static version respectively.
6871 If no shared version of @file{libgcc} was built when the compiler was
6872 configured, these options have no effect.
6874 There are several situations in which an application should use the
6875 shared @file{libgcc} instead of the static version. The most common
6876 of these is when the application wishes to throw and catch exceptions
6877 across different shared libraries. In that case, each of the libraries
6878 as well as the application itself should use the shared @file{libgcc}.
6880 Therefore, the G++ and GCJ drivers automatically add
6881 @option{-shared-libgcc} whenever you build a shared library or a main
6882 executable, because C++ and Java programs typically use exceptions, so
6883 this is the right thing to do.
6885 If, instead, you use the GCC driver to create shared libraries, you may
6886 find that they will not always be linked with the shared @file{libgcc}.
6887 If GCC finds, at its configuration time, that you have a non-GNU linker
6888 or a GNU linker that does not support option @option{--eh-frame-hdr},
6889 it will link the shared version of @file{libgcc} into shared libraries
6890 by default. Otherwise, it will take advantage of the linker and optimize
6891 away the linking with the shared version of @file{libgcc}, linking with
6892 the static version of libgcc by default. This allows exceptions to
6893 propagate through such shared libraries, without incurring relocation
6894 costs at library load time.
6896 However, if a library or main executable is supposed to throw or catch
6897 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6898 for the languages used in the program, or using the option
6899 @option{-shared-libgcc}, such that it is linked with the shared
6904 Bind references to global symbols when building a shared object. Warn
6905 about any unresolved references (unless overridden by the link editor
6906 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6909 @item -Xlinker @var{option}
6911 Pass @var{option} as an option to the linker. You can use this to
6912 supply system-specific linker options which GCC does not know how to
6915 If you want to pass an option that takes an argument, you must use
6916 @option{-Xlinker} twice, once for the option and once for the argument.
6917 For example, to pass @option{-assert definitions}, you must write
6918 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6919 @option{-Xlinker "-assert definitions"}, because this passes the entire
6920 string as a single argument, which is not what the linker expects.
6922 @item -Wl,@var{option}
6924 Pass @var{option} as an option to the linker. If @var{option} contains
6925 commas, it is split into multiple options at the commas.
6927 @item -u @var{symbol}
6929 Pretend the symbol @var{symbol} is undefined, to force linking of
6930 library modules to define it. You can use @option{-u} multiple times with
6931 different symbols to force loading of additional library modules.
6934 @node Directory Options
6935 @section Options for Directory Search
6936 @cindex directory options
6937 @cindex options, directory search
6940 These options specify directories to search for header files, for
6941 libraries and for parts of the compiler:
6946 Add the directory @var{dir} to the head of the list of directories to be
6947 searched for header files. This can be used to override a system header
6948 file, substituting your own version, since these directories are
6949 searched before the system header file directories. However, you should
6950 not use this option to add directories that contain vendor-supplied
6951 system header files (use @option{-isystem} for that). If you use more than
6952 one @option{-I} option, the directories are scanned in left-to-right
6953 order; the standard system directories come after.
6955 If a standard system include directory, or a directory specified with
6956 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6957 option will be ignored. The directory will still be searched but as a
6958 system directory at its normal position in the system include chain.
6959 This is to ensure that GCC's procedure to fix buggy system headers and
6960 the ordering for the include_next directive are not inadvertently changed.
6961 If you really need to change the search order for system directories,
6962 use the @option{-nostdinc} and/or @option{-isystem} options.
6964 @item -iquote@var{dir}
6966 Add the directory @var{dir} to the head of the list of directories to
6967 be searched for header files only for the case of @samp{#include
6968 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6969 otherwise just like @option{-I}.
6973 Add directory @var{dir} to the list of directories to be searched
6976 @item -B@var{prefix}
6978 This option specifies where to find the executables, libraries,
6979 include files, and data files of the compiler itself.
6981 The compiler driver program runs one or more of the subprograms
6982 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6983 @var{prefix} as a prefix for each program it tries to run, both with and
6984 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6986 For each subprogram to be run, the compiler driver first tries the
6987 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6988 was not specified, the driver tries two standard prefixes, which are
6989 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6990 those results in a file name that is found, the unmodified program
6991 name is searched for using the directories specified in your
6992 @env{PATH} environment variable.
6994 The compiler will check to see if the path provided by the @option{-B}
6995 refers to a directory, and if necessary it will add a directory
6996 separator character at the end of the path.
6998 @option{-B} prefixes that effectively specify directory names also apply
6999 to libraries in the linker, because the compiler translates these
7000 options into @option{-L} options for the linker. They also apply to
7001 includes files in the preprocessor, because the compiler translates these
7002 options into @option{-isystem} options for the preprocessor. In this case,
7003 the compiler appends @samp{include} to the prefix.
7005 The run-time support file @file{libgcc.a} can also be searched for using
7006 the @option{-B} prefix, if needed. If it is not found there, the two
7007 standard prefixes above are tried, and that is all. The file is left
7008 out of the link if it is not found by those means.
7010 Another way to specify a prefix much like the @option{-B} prefix is to use
7011 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7014 As a special kludge, if the path provided by @option{-B} is
7015 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7016 9, then it will be replaced by @file{[dir/]include}. This is to help
7017 with boot-strapping the compiler.
7019 @item -specs=@var{file}
7021 Process @var{file} after the compiler reads in the standard @file{specs}
7022 file, in order to override the defaults that the @file{gcc} driver
7023 program uses when determining what switches to pass to @file{cc1},
7024 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7025 @option{-specs=@var{file}} can be specified on the command line, and they
7026 are processed in order, from left to right.
7028 @item --sysroot=@var{dir}
7030 Use @var{dir} as the logical root directory for headers and libraries.
7031 For example, if the compiler would normally search for headers in
7032 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7033 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7035 If you use both this option and the @option{-isysroot} option, then
7036 the @option{--sysroot} option will apply to libraries, but the
7037 @option{-isysroot} option will apply to header files.
7039 The GNU linker (beginning with version 2.16) has the necessary support
7040 for this option. If your linker does not support this option, the
7041 header file aspect of @option{--sysroot} will still work, but the
7042 library aspect will not.
7046 This option has been deprecated. Please use @option{-iquote} instead for
7047 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7048 Any directories you specify with @option{-I} options before the @option{-I-}
7049 option are searched only for the case of @samp{#include "@var{file}"};
7050 they are not searched for @samp{#include <@var{file}>}.
7052 If additional directories are specified with @option{-I} options after
7053 the @option{-I-}, these directories are searched for all @samp{#include}
7054 directives. (Ordinarily @emph{all} @option{-I} directories are used
7057 In addition, the @option{-I-} option inhibits the use of the current
7058 directory (where the current input file came from) as the first search
7059 directory for @samp{#include "@var{file}"}. There is no way to
7060 override this effect of @option{-I-}. With @option{-I.} you can specify
7061 searching the directory which was current when the compiler was
7062 invoked. That is not exactly the same as what the preprocessor does
7063 by default, but it is often satisfactory.
7065 @option{-I-} does not inhibit the use of the standard system directories
7066 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7073 @section Specifying subprocesses and the switches to pass to them
7076 @command{gcc} is a driver program. It performs its job by invoking a
7077 sequence of other programs to do the work of compiling, assembling and
7078 linking. GCC interprets its command-line parameters and uses these to
7079 deduce which programs it should invoke, and which command-line options
7080 it ought to place on their command lines. This behavior is controlled
7081 by @dfn{spec strings}. In most cases there is one spec string for each
7082 program that GCC can invoke, but a few programs have multiple spec
7083 strings to control their behavior. The spec strings built into GCC can
7084 be overridden by using the @option{-specs=} command-line switch to specify
7087 @dfn{Spec files} are plaintext files that are used to construct spec
7088 strings. They consist of a sequence of directives separated by blank
7089 lines. The type of directive is determined by the first non-whitespace
7090 character on the line and it can be one of the following:
7093 @item %@var{command}
7094 Issues a @var{command} to the spec file processor. The commands that can
7098 @item %include <@var{file}>
7100 Search for @var{file} and insert its text at the current point in the
7103 @item %include_noerr <@var{file}>
7104 @cindex %include_noerr
7105 Just like @samp{%include}, but do not generate an error message if the include
7106 file cannot be found.
7108 @item %rename @var{old_name} @var{new_name}
7110 Rename the spec string @var{old_name} to @var{new_name}.
7114 @item *[@var{spec_name}]:
7115 This tells the compiler to create, override or delete the named spec
7116 string. All lines after this directive up to the next directive or
7117 blank line are considered to be the text for the spec string. If this
7118 results in an empty string then the spec will be deleted. (Or, if the
7119 spec did not exist, then nothing will happened.) Otherwise, if the spec
7120 does not currently exist a new spec will be created. If the spec does
7121 exist then its contents will be overridden by the text of this
7122 directive, unless the first character of that text is the @samp{+}
7123 character, in which case the text will be appended to the spec.
7125 @item [@var{suffix}]:
7126 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7127 and up to the next directive or blank line are considered to make up the
7128 spec string for the indicated suffix. When the compiler encounters an
7129 input file with the named suffix, it will processes the spec string in
7130 order to work out how to compile that file. For example:
7137 This says that any input file whose name ends in @samp{.ZZ} should be
7138 passed to the program @samp{z-compile}, which should be invoked with the
7139 command-line switch @option{-input} and with the result of performing the
7140 @samp{%i} substitution. (See below.)
7142 As an alternative to providing a spec string, the text that follows a
7143 suffix directive can be one of the following:
7146 @item @@@var{language}
7147 This says that the suffix is an alias for a known @var{language}. This is
7148 similar to using the @option{-x} command-line switch to GCC to specify a
7149 language explicitly. For example:
7156 Says that .ZZ files are, in fact, C++ source files.
7159 This causes an error messages saying:
7162 @var{name} compiler not installed on this system.
7166 GCC already has an extensive list of suffixes built into it.
7167 This directive will add an entry to the end of the list of suffixes, but
7168 since the list is searched from the end backwards, it is effectively
7169 possible to override earlier entries using this technique.
7173 GCC has the following spec strings built into it. Spec files can
7174 override these strings or create their own. Note that individual
7175 targets can also add their own spec strings to this list.
7178 asm Options to pass to the assembler
7179 asm_final Options to pass to the assembler post-processor
7180 cpp Options to pass to the C preprocessor
7181 cc1 Options to pass to the C compiler
7182 cc1plus Options to pass to the C++ compiler
7183 endfile Object files to include at the end of the link
7184 link Options to pass to the linker
7185 lib Libraries to include on the command line to the linker
7186 libgcc Decides which GCC support library to pass to the linker
7187 linker Sets the name of the linker
7188 predefines Defines to be passed to the C preprocessor
7189 signed_char Defines to pass to CPP to say whether @code{char} is signed
7191 startfile Object files to include at the start of the link
7194 Here is a small example of a spec file:
7200 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7203 This example renames the spec called @samp{lib} to @samp{old_lib} and
7204 then overrides the previous definition of @samp{lib} with a new one.
7205 The new definition adds in some extra command-line options before
7206 including the text of the old definition.
7208 @dfn{Spec strings} are a list of command-line options to be passed to their
7209 corresponding program. In addition, the spec strings can contain
7210 @samp{%}-prefixed sequences to substitute variable text or to
7211 conditionally insert text into the command line. Using these constructs
7212 it is possible to generate quite complex command lines.
7214 Here is a table of all defined @samp{%}-sequences for spec
7215 strings. Note that spaces are not generated automatically around the
7216 results of expanding these sequences. Therefore you can concatenate them
7217 together or combine them with constant text in a single argument.
7221 Substitute one @samp{%} into the program name or argument.
7224 Substitute the name of the input file being processed.
7227 Substitute the basename of the input file being processed.
7228 This is the substring up to (and not including) the last period
7229 and not including the directory.
7232 This is the same as @samp{%b}, but include the file suffix (text after
7236 Marks the argument containing or following the @samp{%d} as a
7237 temporary file name, so that that file will be deleted if GCC exits
7238 successfully. Unlike @samp{%g}, this contributes no text to the
7241 @item %g@var{suffix}
7242 Substitute a file name that has suffix @var{suffix} and is chosen
7243 once per compilation, and mark the argument in the same way as
7244 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7245 name is now chosen in a way that is hard to predict even when previously
7246 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7247 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7248 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7249 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7250 was simply substituted with a file name chosen once per compilation,
7251 without regard to any appended suffix (which was therefore treated
7252 just like ordinary text), making such attacks more likely to succeed.
7254 @item %u@var{suffix}
7255 Like @samp{%g}, but generates a new temporary file name even if
7256 @samp{%u@var{suffix}} was already seen.
7258 @item %U@var{suffix}
7259 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7260 new one if there is no such last file name. In the absence of any
7261 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7262 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7263 would involve the generation of two distinct file names, one
7264 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7265 simply substituted with a file name chosen for the previous @samp{%u},
7266 without regard to any appended suffix.
7268 @item %j@var{suffix}
7269 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7270 writable, and if save-temps is off; otherwise, substitute the name
7271 of a temporary file, just like @samp{%u}. This temporary file is not
7272 meant for communication between processes, but rather as a junk
7275 @item %|@var{suffix}
7276 @itemx %m@var{suffix}
7277 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7278 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7279 all. These are the two most common ways to instruct a program that it
7280 should read from standard input or write to standard output. If you
7281 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7282 construct: see for example @file{f/lang-specs.h}.
7284 @item %.@var{SUFFIX}
7285 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7286 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7287 terminated by the next space or %.
7290 Marks the argument containing or following the @samp{%w} as the
7291 designated output file of this compilation. This puts the argument
7292 into the sequence of arguments that @samp{%o} will substitute later.
7295 Substitutes the names of all the output files, with spaces
7296 automatically placed around them. You should write spaces
7297 around the @samp{%o} as well or the results are undefined.
7298 @samp{%o} is for use in the specs for running the linker.
7299 Input files whose names have no recognized suffix are not compiled
7300 at all, but they are included among the output files, so they will
7304 Substitutes the suffix for object files. Note that this is
7305 handled specially when it immediately follows @samp{%g, %u, or %U},
7306 because of the need for those to form complete file names. The
7307 handling is such that @samp{%O} is treated exactly as if it had already
7308 been substituted, except that @samp{%g, %u, and %U} do not currently
7309 support additional @var{suffix} characters following @samp{%O} as they would
7310 following, for example, @samp{.o}.
7313 Substitutes the standard macro predefinitions for the
7314 current target machine. Use this when running @code{cpp}.
7317 Like @samp{%p}, but puts @samp{__} before and after the name of each
7318 predefined macro, except for macros that start with @samp{__} or with
7319 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7323 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7324 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7325 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7326 and @option{-imultilib} as necessary.
7329 Current argument is the name of a library or startup file of some sort.
7330 Search for that file in a standard list of directories and substitute
7331 the full name found.
7334 Print @var{str} as an error message. @var{str} is terminated by a newline.
7335 Use this when inconsistent options are detected.
7338 Substitute the contents of spec string @var{name} at this point.
7341 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7343 @item %x@{@var{option}@}
7344 Accumulate an option for @samp{%X}.
7347 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7351 Output the accumulated assembler options specified by @option{-Wa}.
7354 Output the accumulated preprocessor options specified by @option{-Wp}.
7357 Process the @code{asm} spec. This is used to compute the
7358 switches to be passed to the assembler.
7361 Process the @code{asm_final} spec. This is a spec string for
7362 passing switches to an assembler post-processor, if such a program is
7366 Process the @code{link} spec. This is the spec for computing the
7367 command line passed to the linker. Typically it will make use of the
7368 @samp{%L %G %S %D and %E} sequences.
7371 Dump out a @option{-L} option for each directory that GCC believes might
7372 contain startup files. If the target supports multilibs then the
7373 current multilib directory will be prepended to each of these paths.
7376 Process the @code{lib} spec. This is a spec string for deciding which
7377 libraries should be included on the command line to the linker.
7380 Process the @code{libgcc} spec. This is a spec string for deciding
7381 which GCC support library should be included on the command line to the linker.
7384 Process the @code{startfile} spec. This is a spec for deciding which
7385 object files should be the first ones passed to the linker. Typically
7386 this might be a file named @file{crt0.o}.
7389 Process the @code{endfile} spec. This is a spec string that specifies
7390 the last object files that will be passed to the linker.
7393 Process the @code{cpp} spec. This is used to construct the arguments
7394 to be passed to the C preprocessor.
7397 Process the @code{cc1} spec. This is used to construct the options to be
7398 passed to the actual C compiler (@samp{cc1}).
7401 Process the @code{cc1plus} spec. This is used to construct the options to be
7402 passed to the actual C++ compiler (@samp{cc1plus}).
7405 Substitute the variable part of a matched option. See below.
7406 Note that each comma in the substituted string is replaced by
7410 Remove all occurrences of @code{-S} from the command line. Note---this
7411 command is position dependent. @samp{%} commands in the spec string
7412 before this one will see @code{-S}, @samp{%} commands in the spec string
7413 after this one will not.
7415 @item %:@var{function}(@var{args})
7416 Call the named function @var{function}, passing it @var{args}.
7417 @var{args} is first processed as a nested spec string, then split
7418 into an argument vector in the usual fashion. The function returns
7419 a string which is processed as if it had appeared literally as part
7420 of the current spec.
7422 The following built-in spec functions are provided:
7425 @item @code{if-exists}
7426 The @code{if-exists} spec function takes one argument, an absolute
7427 pathname to a file. If the file exists, @code{if-exists} returns the
7428 pathname. Here is a small example of its usage:
7432 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7435 @item @code{if-exists-else}
7436 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7437 spec function, except that it takes two arguments. The first argument is
7438 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7439 returns the pathname. If it does not exist, it returns the second argument.
7440 This way, @code{if-exists-else} can be used to select one file or another,
7441 based on the existence of the first. Here is a small example of its usage:
7445 crt0%O%s %:if-exists(crti%O%s) \
7446 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7449 @item @code{replace-outfile}
7450 The @code{replace-outfile} spec function takes two arguments. It looks for the
7451 first argument in the outfiles array and replaces it with the second argument. Here
7452 is a small example of its usage:
7455 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7461 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7462 If that switch was not specified, this substitutes nothing. Note that
7463 the leading dash is omitted when specifying this option, and it is
7464 automatically inserted if the substitution is performed. Thus the spec
7465 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7466 and would output the command line option @option{-foo}.
7468 @item %W@{@code{S}@}
7469 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7472 @item %@{@code{S}*@}
7473 Substitutes all the switches specified to GCC whose names start
7474 with @code{-S}, but which also take an argument. This is used for
7475 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7476 GCC considers @option{-o foo} as being
7477 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7478 text, including the space. Thus two arguments would be generated.
7480 @item %@{@code{S}*&@code{T}*@}
7481 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7482 (the order of @code{S} and @code{T} in the spec is not significant).
7483 There can be any number of ampersand-separated variables; for each the
7484 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7486 @item %@{@code{S}:@code{X}@}
7487 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7489 @item %@{!@code{S}:@code{X}@}
7490 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7492 @item %@{@code{S}*:@code{X}@}
7493 Substitutes @code{X} if one or more switches whose names start with
7494 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7495 once, no matter how many such switches appeared. However, if @code{%*}
7496 appears somewhere in @code{X}, then @code{X} will be substituted once
7497 for each matching switch, with the @code{%*} replaced by the part of
7498 that switch that matched the @code{*}.
7500 @item %@{.@code{S}:@code{X}@}
7501 Substitutes @code{X}, if processing a file with suffix @code{S}.
7503 @item %@{!.@code{S}:@code{X}@}
7504 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7506 @item %@{@code{S}|@code{P}:@code{X}@}
7507 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7508 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7509 although they have a stronger binding than the @samp{|}. If @code{%*}
7510 appears in @code{X}, all of the alternatives must be starred, and only
7511 the first matching alternative is substituted.
7513 For example, a spec string like this:
7516 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7519 will output the following command-line options from the following input
7520 command-line options:
7525 -d fred.c -foo -baz -boggle
7526 -d jim.d -bar -baz -boggle
7529 @item %@{S:X; T:Y; :D@}
7531 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7532 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7533 be as many clauses as you need. This may be combined with @code{.},
7534 @code{!}, @code{|}, and @code{*} as needed.
7539 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7540 construct may contain other nested @samp{%} constructs or spaces, or
7541 even newlines. They are processed as usual, as described above.
7542 Trailing white space in @code{X} is ignored. White space may also
7543 appear anywhere on the left side of the colon in these constructs,
7544 except between @code{.} or @code{*} and the corresponding word.
7546 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7547 handled specifically in these constructs. If another value of
7548 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7549 @option{-W} switch is found later in the command line, the earlier
7550 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7551 just one letter, which passes all matching options.
7553 The character @samp{|} at the beginning of the predicate text is used to
7554 indicate that a command should be piped to the following command, but
7555 only if @option{-pipe} is specified.
7557 It is built into GCC which switches take arguments and which do not.
7558 (You might think it would be useful to generalize this to allow each
7559 compiler's spec to say which switches take arguments. But this cannot
7560 be done in a consistent fashion. GCC cannot even decide which input
7561 files have been specified without knowing which switches take arguments,
7562 and it must know which input files to compile in order to tell which
7565 GCC also knows implicitly that arguments starting in @option{-l} are to be
7566 treated as compiler output files, and passed to the linker in their
7567 proper position among the other output files.
7569 @c man begin OPTIONS
7571 @node Target Options
7572 @section Specifying Target Machine and Compiler Version
7573 @cindex target options
7574 @cindex cross compiling
7575 @cindex specifying machine version
7576 @cindex specifying compiler version and target machine
7577 @cindex compiler version, specifying
7578 @cindex target machine, specifying
7580 The usual way to run GCC is to run the executable called @file{gcc}, or
7581 @file{<machine>-gcc} when cross-compiling, or
7582 @file{<machine>-gcc-<version>} to run a version other than the one that
7583 was installed last. Sometimes this is inconvenient, so GCC provides
7584 options that will switch to another cross-compiler or version.
7587 @item -b @var{machine}
7589 The argument @var{machine} specifies the target machine for compilation.
7591 The value to use for @var{machine} is the same as was specified as the
7592 machine type when configuring GCC as a cross-compiler. For
7593 example, if a cross-compiler was configured with @samp{configure
7594 arm-elf}, meaning to compile for an arm processor with elf binaries,
7595 then you would specify @option{-b arm-elf} to run that cross compiler.
7596 Because there are other options beginning with @option{-b}, the
7597 configuration must contain a hyphen.
7599 @item -V @var{version}
7601 The argument @var{version} specifies which version of GCC to run.
7602 This is useful when multiple versions are installed. For example,
7603 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7606 The @option{-V} and @option{-b} options work by running the
7607 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7608 use them if you can just run that directly.
7610 @node Submodel Options
7611 @section Hardware Models and Configurations
7612 @cindex submodel options
7613 @cindex specifying hardware config
7614 @cindex hardware models and configurations, specifying
7615 @cindex machine dependent options
7617 Earlier we discussed the standard option @option{-b} which chooses among
7618 different installed compilers for completely different target
7619 machines, such as VAX vs.@: 68000 vs.@: 80386.
7621 In addition, each of these target machine types can have its own
7622 special options, starting with @samp{-m}, to choose among various
7623 hardware models or configurations---for example, 68010 vs 68020,
7624 floating coprocessor or none. A single installed version of the
7625 compiler can compile for any model or configuration, according to the
7628 Some configurations of the compiler also support additional special
7629 options, usually for compatibility with other compilers on the same
7632 @c This list is ordered alphanumerically by subsection name.
7633 @c It should be the same order and spelling as these options are listed
7634 @c in Machine Dependent Options
7640 * Blackfin Options::
7644 * DEC Alpha Options::
7645 * DEC Alpha/VMS Options::
7647 * GNU/Linux Options::
7650 * i386 and x86-64 Options::
7663 * RS/6000 and PowerPC Options::
7664 * S/390 and zSeries Options::
7669 * System V Options::
7670 * TMS320C3x/C4x Options::
7674 * Xstormy16 Options::
7680 @subsection ARC Options
7683 These options are defined for ARC implementations:
7688 Compile code for little endian mode. This is the default.
7692 Compile code for big endian mode.
7695 @opindex mmangle-cpu
7696 Prepend the name of the cpu to all public symbol names.
7697 In multiple-processor systems, there are many ARC variants with different
7698 instruction and register set characteristics. This flag prevents code
7699 compiled for one cpu to be linked with code compiled for another.
7700 No facility exists for handling variants that are ``almost identical''.
7701 This is an all or nothing option.
7703 @item -mcpu=@var{cpu}
7705 Compile code for ARC variant @var{cpu}.
7706 Which variants are supported depend on the configuration.
7707 All variants support @option{-mcpu=base}, this is the default.
7709 @item -mtext=@var{text-section}
7710 @itemx -mdata=@var{data-section}
7711 @itemx -mrodata=@var{readonly-data-section}
7715 Put functions, data, and readonly data in @var{text-section},
7716 @var{data-section}, and @var{readonly-data-section} respectively
7717 by default. This can be overridden with the @code{section} attribute.
7718 @xref{Variable Attributes}.
7723 @subsection ARM Options
7726 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7730 @item -mabi=@var{name}
7732 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7733 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7736 @opindex mapcs-frame
7737 Generate a stack frame that is compliant with the ARM Procedure Call
7738 Standard for all functions, even if this is not strictly necessary for
7739 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7740 with this option will cause the stack frames not to be generated for
7741 leaf functions. The default is @option{-mno-apcs-frame}.
7745 This is a synonym for @option{-mapcs-frame}.
7748 @c not currently implemented
7749 @item -mapcs-stack-check
7750 @opindex mapcs-stack-check
7751 Generate code to check the amount of stack space available upon entry to
7752 every function (that actually uses some stack space). If there is
7753 insufficient space available then either the function
7754 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7755 called, depending upon the amount of stack space required. The run time
7756 system is required to provide these functions. The default is
7757 @option{-mno-apcs-stack-check}, since this produces smaller code.
7759 @c not currently implemented
7761 @opindex mapcs-float
7762 Pass floating point arguments using the float point registers. This is
7763 one of the variants of the APCS@. This option is recommended if the
7764 target hardware has a floating point unit or if a lot of floating point
7765 arithmetic is going to be performed by the code. The default is
7766 @option{-mno-apcs-float}, since integer only code is slightly increased in
7767 size if @option{-mapcs-float} is used.
7769 @c not currently implemented
7770 @item -mapcs-reentrant
7771 @opindex mapcs-reentrant
7772 Generate reentrant, position independent code. The default is
7773 @option{-mno-apcs-reentrant}.
7776 @item -mthumb-interwork
7777 @opindex mthumb-interwork
7778 Generate code which supports calling between the ARM and Thumb
7779 instruction sets. Without this option the two instruction sets cannot
7780 be reliably used inside one program. The default is
7781 @option{-mno-thumb-interwork}, since slightly larger code is generated
7782 when @option{-mthumb-interwork} is specified.
7784 @item -mno-sched-prolog
7785 @opindex mno-sched-prolog
7786 Prevent the reordering of instructions in the function prolog, or the
7787 merging of those instruction with the instructions in the function's
7788 body. This means that all functions will start with a recognizable set
7789 of instructions (or in fact one of a choice from a small set of
7790 different function prologues), and this information can be used to
7791 locate the start if functions inside an executable piece of code. The
7792 default is @option{-msched-prolog}.
7795 @opindex mhard-float
7796 Generate output containing floating point instructions. This is the
7800 @opindex msoft-float
7801 Generate output containing library calls for floating point.
7802 @strong{Warning:} the requisite libraries are not available for all ARM
7803 targets. Normally the facilities of the machine's usual C compiler are
7804 used, but this cannot be done directly in cross-compilation. You must make
7805 your own arrangements to provide suitable library functions for
7808 @option{-msoft-float} changes the calling convention in the output file;
7809 therefore, it is only useful if you compile @emph{all} of a program with
7810 this option. In particular, you need to compile @file{libgcc.a}, the
7811 library that comes with GCC, with @option{-msoft-float} in order for
7814 @item -mfloat-abi=@var{name}
7816 Specifies which ABI to use for floating point values. Permissible values
7817 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7819 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7820 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7821 of floating point instructions, but still uses the soft-float calling
7824 @item -mlittle-endian
7825 @opindex mlittle-endian
7826 Generate code for a processor running in little-endian mode. This is
7827 the default for all standard configurations.
7830 @opindex mbig-endian
7831 Generate code for a processor running in big-endian mode; the default is
7832 to compile code for a little-endian processor.
7834 @item -mwords-little-endian
7835 @opindex mwords-little-endian
7836 This option only applies when generating code for big-endian processors.
7837 Generate code for a little-endian word order but a big-endian byte
7838 order. That is, a byte order of the form @samp{32107654}. Note: this
7839 option should only be used if you require compatibility with code for
7840 big-endian ARM processors generated by versions of the compiler prior to
7843 @item -mcpu=@var{name}
7845 This specifies the name of the target ARM processor. GCC uses this name
7846 to determine what kind of instructions it can emit when generating
7847 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7848 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7849 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7850 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7851 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7852 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7853 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7854 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7855 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7856 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7857 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7858 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7859 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7860 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
7861 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
7862 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
7864 @itemx -mtune=@var{name}
7866 This option is very similar to the @option{-mcpu=} option, except that
7867 instead of specifying the actual target processor type, and hence
7868 restricting which instructions can be used, it specifies that GCC should
7869 tune the performance of the code as if the target were of the type
7870 specified in this option, but still choosing the instructions that it
7871 will generate based on the cpu specified by a @option{-mcpu=} option.
7872 For some ARM implementations better performance can be obtained by using
7875 @item -march=@var{name}
7877 This specifies the name of the target ARM architecture. GCC uses this
7878 name to determine what kind of instructions it can emit when generating
7879 assembly code. This option can be used in conjunction with or instead
7880 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7881 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7882 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7883 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
7884 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
7886 @item -mfpu=@var{name}
7887 @itemx -mfpe=@var{number}
7888 @itemx -mfp=@var{number}
7892 This specifies what floating point hardware (or hardware emulation) is
7893 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7894 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7895 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7896 with older versions of GCC@.
7898 If @option{-msoft-float} is specified this specifies the format of
7899 floating point values.
7901 @item -mstructure-size-boundary=@var{n}
7902 @opindex mstructure-size-boundary
7903 The size of all structures and unions will be rounded up to a multiple
7904 of the number of bits set by this option. Permissible values are 8, 32
7905 and 64. The default value varies for different toolchains. For the COFF
7906 targeted toolchain the default value is 8. A value of 64 is only allowed
7907 if the underlying ABI supports it.
7909 Specifying the larger number can produce faster, more efficient code, but
7910 can also increase the size of the program. Different values are potentially
7911 incompatible. Code compiled with one value cannot necessarily expect to
7912 work with code or libraries compiled with another value, if they exchange
7913 information using structures or unions.
7915 @item -mabort-on-noreturn
7916 @opindex mabort-on-noreturn
7917 Generate a call to the function @code{abort} at the end of a
7918 @code{noreturn} function. It will be executed if the function tries to
7922 @itemx -mno-long-calls
7923 @opindex mlong-calls
7924 @opindex mno-long-calls
7925 Tells the compiler to perform function calls by first loading the
7926 address of the function into a register and then performing a subroutine
7927 call on this register. This switch is needed if the target function
7928 will lie outside of the 64 megabyte addressing range of the offset based
7929 version of subroutine call instruction.
7931 Even if this switch is enabled, not all function calls will be turned
7932 into long calls. The heuristic is that static functions, functions
7933 which have the @samp{short-call} attribute, functions that are inside
7934 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7935 definitions have already been compiled within the current compilation
7936 unit, will not be turned into long calls. The exception to this rule is
7937 that weak function definitions, functions with the @samp{long-call}
7938 attribute or the @samp{section} attribute, and functions that are within
7939 the scope of a @samp{#pragma long_calls} directive, will always be
7940 turned into long calls.
7942 This feature is not enabled by default. Specifying
7943 @option{-mno-long-calls} will restore the default behavior, as will
7944 placing the function calls within the scope of a @samp{#pragma
7945 long_calls_off} directive. Note these switches have no effect on how
7946 the compiler generates code to handle function calls via function
7949 @item -mnop-fun-dllimport
7950 @opindex mnop-fun-dllimport
7951 Disable support for the @code{dllimport} attribute.
7953 @item -msingle-pic-base
7954 @opindex msingle-pic-base
7955 Treat the register used for PIC addressing as read-only, rather than
7956 loading it in the prologue for each function. The run-time system is
7957 responsible for initializing this register with an appropriate value
7958 before execution begins.
7960 @item -mpic-register=@var{reg}
7961 @opindex mpic-register
7962 Specify the register to be used for PIC addressing. The default is R10
7963 unless stack-checking is enabled, when R9 is used.
7965 @item -mcirrus-fix-invalid-insns
7966 @opindex mcirrus-fix-invalid-insns
7967 @opindex mno-cirrus-fix-invalid-insns
7968 Insert NOPs into the instruction stream to in order to work around
7969 problems with invalid Maverick instruction combinations. This option
7970 is only valid if the @option{-mcpu=ep9312} option has been used to
7971 enable generation of instructions for the Cirrus Maverick floating
7972 point co-processor. This option is not enabled by default, since the
7973 problem is only present in older Maverick implementations. The default
7974 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7977 @item -mpoke-function-name
7978 @opindex mpoke-function-name
7979 Write the name of each function into the text section, directly
7980 preceding the function prologue. The generated code is similar to this:
7984 .ascii "arm_poke_function_name", 0
7987 .word 0xff000000 + (t1 - t0)
7988 arm_poke_function_name
7990 stmfd sp!, @{fp, ip, lr, pc@}
7994 When performing a stack backtrace, code can inspect the value of
7995 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7996 location @code{pc - 12} and the top 8 bits are set, then we know that
7997 there is a function name embedded immediately preceding this location
7998 and has length @code{((pc[-3]) & 0xff000000)}.
8002 Generate code for the Thumb instruction set. The default is to
8003 use the 32-bit ARM instruction set.
8004 This option automatically enables either 16-bit Thumb-1 or
8005 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8006 and @option{-march=@var{name}} options.
8009 @opindex mtpcs-frame
8010 Generate a stack frame that is compliant with the Thumb Procedure Call
8011 Standard for all non-leaf functions. (A leaf function is one that does
8012 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8014 @item -mtpcs-leaf-frame
8015 @opindex mtpcs-leaf-frame
8016 Generate a stack frame that is compliant with the Thumb Procedure Call
8017 Standard for all leaf functions. (A leaf function is one that does
8018 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8020 @item -mcallee-super-interworking
8021 @opindex mcallee-super-interworking
8022 Gives all externally visible functions in the file being compiled an ARM
8023 instruction set header which switches to Thumb mode before executing the
8024 rest of the function. This allows these functions to be called from
8025 non-interworking code.
8027 @item -mcaller-super-interworking
8028 @opindex mcaller-super-interworking
8029 Allows calls via function pointers (including virtual functions) to
8030 execute correctly regardless of whether the target code has been
8031 compiled for interworking or not. There is a small overhead in the cost
8032 of executing a function pointer if this option is enabled.
8034 @item -mtp=@var{name}
8036 Specify the access model for the thread local storage pointer. The valid
8037 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8038 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8039 (supported in the arm6k architecture), and @option{auto}, which uses the
8040 best available method for the selected processor. The default setting is
8046 @subsection AVR Options
8049 These options are defined for AVR implementations:
8052 @item -mmcu=@var{mcu}
8054 Specify ATMEL AVR instruction set or MCU type.
8056 Instruction set avr1 is for the minimal AVR core, not supported by the C
8057 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8058 attiny11, attiny12, attiny15, attiny28).
8060 Instruction set avr2 (default) is for the classic AVR core with up to
8061 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8062 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8063 at90c8534, at90s8535).
8065 Instruction set avr3 is for the classic AVR core with up to 128K program
8066 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8068 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8069 memory space (MCU types: atmega8, atmega83, atmega85).
8071 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8072 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8073 atmega64, atmega128, at43usb355, at94k).
8077 Output instruction sizes to the asm file.
8079 @item -minit-stack=@var{N}
8080 @opindex minit-stack
8081 Specify the initial stack address, which may be a symbol or numeric value,
8082 @samp{__stack} is the default.
8084 @item -mno-interrupts
8085 @opindex mno-interrupts
8086 Generated code is not compatible with hardware interrupts.
8087 Code size will be smaller.
8089 @item -mcall-prologues
8090 @opindex mcall-prologues
8091 Functions prologues/epilogues expanded as call to appropriate
8092 subroutines. Code size will be smaller.
8094 @item -mno-tablejump
8095 @opindex mno-tablejump
8096 Do not generate tablejump insns which sometimes increase code size.
8099 @opindex mtiny-stack
8100 Change only the low 8 bits of the stack pointer.
8104 Assume int to be 8 bit integer. This affects the sizes of all types: A
8105 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8106 and long long will be 4 bytes. Please note that this option does not
8107 comply to the C standards, but it will provide you with smaller code
8111 @node Blackfin Options
8112 @subsection Blackfin Options
8113 @cindex Blackfin Options
8116 @item -momit-leaf-frame-pointer
8117 @opindex momit-leaf-frame-pointer
8118 Don't keep the frame pointer in a register for leaf functions. This
8119 avoids the instructions to save, set up and restore frame pointers and
8120 makes an extra register available in leaf functions. The option
8121 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8122 which might make debugging harder.
8124 @item -mspecld-anomaly
8125 @opindex mspecld-anomaly
8126 When enabled, the compiler will ensure that the generated code does not
8127 contain speculative loads after jump instructions. This option is enabled
8130 @item -mno-specld-anomaly
8131 @opindex mno-specld-anomaly
8132 Don't generate extra code to prevent speculative loads from occurring.
8134 @item -mcsync-anomaly
8135 @opindex mcsync-anomaly
8136 When enabled, the compiler will ensure that the generated code does not
8137 contain CSYNC or SSYNC instructions too soon after conditional branches.
8138 This option is enabled by default.
8140 @item -mno-csync-anomaly
8141 @opindex mno-csync-anomaly
8142 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8143 occurring too soon after a conditional branch.
8147 When enabled, the compiler is free to take advantage of the knowledge that
8148 the entire program fits into the low 64k of memory.
8151 @opindex mno-low-64k
8152 Assume that the program is arbitrarily large. This is the default.
8154 @item -mstack-check-l1
8155 @opindex mstack-check-l1
8156 Do stack checking using information placed into L1 scratchpad memory by the
8159 @item -mid-shared-library
8160 @opindex mid-shared-library
8161 Generate code that supports shared libraries via the library ID method.
8162 This allows for execute in place and shared libraries in an environment
8163 without virtual memory management. This option implies @option{-fPIC}.
8165 @item -mno-id-shared-library
8166 @opindex mno-id-shared-library
8167 Generate code that doesn't assume ID based shared libraries are being used.
8168 This is the default.
8170 @item -mleaf-id-shared-library
8171 @opindex mleaf-id-shared-library
8172 Generate code that supports shared libraries via the library ID method,
8173 but assumes that this library or executable won't link against any other
8174 ID shared libraries. That allows the compiler to use faster code for jumps
8177 @item -mno-leaf-id-shared-library
8178 @opindex mno-leaf-id-shared-library
8179 Do not assume that the code being compiled won't link against any ID shared
8180 libraries. Slower code will be generated for jump and call insns.
8182 @item -mshared-library-id=n
8183 @opindex mshared-library-id
8184 Specified the identification number of the ID based shared library being
8185 compiled. Specifying a value of 0 will generate more compact code, specifying
8186 other values will force the allocation of that number to the current
8187 library but is no more space or time efficient than omitting this option.
8191 Generate code that allows the data segment to be located in a different
8192 area of memory from the text segment. This allows for execute in place in
8193 an environment without virtual memory management by eliminating relocations
8194 against the text section.
8197 @opindex mno-sep-data
8198 Generate code that assumes that the data segment follows the text segment.
8199 This is the default.
8202 @itemx -mno-long-calls
8203 @opindex mlong-calls
8204 @opindex mno-long-calls
8205 Tells the compiler to perform function calls by first loading the
8206 address of the function into a register and then performing a subroutine
8207 call on this register. This switch is needed if the target function
8208 will lie outside of the 24 bit addressing range of the offset based
8209 version of subroutine call instruction.
8211 This feature is not enabled by default. Specifying
8212 @option{-mno-long-calls} will restore the default behavior. Note these
8213 switches have no effect on how the compiler generates code to handle
8214 function calls via function pointers.
8218 @subsection CRIS Options
8219 @cindex CRIS Options
8221 These options are defined specifically for the CRIS ports.
8224 @item -march=@var{architecture-type}
8225 @itemx -mcpu=@var{architecture-type}
8228 Generate code for the specified architecture. The choices for
8229 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8230 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8231 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8234 @item -mtune=@var{architecture-type}
8236 Tune to @var{architecture-type} everything applicable about the generated
8237 code, except for the ABI and the set of available instructions. The
8238 choices for @var{architecture-type} are the same as for
8239 @option{-march=@var{architecture-type}}.
8241 @item -mmax-stack-frame=@var{n}
8242 @opindex mmax-stack-frame
8243 Warn when the stack frame of a function exceeds @var{n} bytes.
8245 @item -melinux-stacksize=@var{n}
8246 @opindex melinux-stacksize
8247 Only available with the @samp{cris-axis-aout} target. Arranges for
8248 indications in the program to the kernel loader that the stack of the
8249 program should be set to @var{n} bytes.
8255 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8256 @option{-march=v3} and @option{-march=v8} respectively.
8258 @item -mmul-bug-workaround
8259 @itemx -mno-mul-bug-workaround
8260 @opindex mmul-bug-workaround
8261 @opindex mno-mul-bug-workaround
8262 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8263 models where it applies. This option is active by default.
8267 Enable CRIS-specific verbose debug-related information in the assembly
8268 code. This option also has the effect to turn off the @samp{#NO_APP}
8269 formatted-code indicator to the assembler at the beginning of the
8274 Do not use condition-code results from previous instruction; always emit
8275 compare and test instructions before use of condition codes.
8277 @item -mno-side-effects
8278 @opindex mno-side-effects
8279 Do not emit instructions with side-effects in addressing modes other than
8283 @itemx -mno-stack-align
8285 @itemx -mno-data-align
8286 @itemx -mconst-align
8287 @itemx -mno-const-align
8288 @opindex mstack-align
8289 @opindex mno-stack-align
8290 @opindex mdata-align
8291 @opindex mno-data-align
8292 @opindex mconst-align
8293 @opindex mno-const-align
8294 These options (no-options) arranges (eliminate arrangements) for the
8295 stack-frame, individual data and constants to be aligned for the maximum
8296 single data access size for the chosen CPU model. The default is to
8297 arrange for 32-bit alignment. ABI details such as structure layout are
8298 not affected by these options.
8306 Similar to the stack- data- and const-align options above, these options
8307 arrange for stack-frame, writable data and constants to all be 32-bit,
8308 16-bit or 8-bit aligned. The default is 32-bit alignment.
8310 @item -mno-prologue-epilogue
8311 @itemx -mprologue-epilogue
8312 @opindex mno-prologue-epilogue
8313 @opindex mprologue-epilogue
8314 With @option{-mno-prologue-epilogue}, the normal function prologue and
8315 epilogue that sets up the stack-frame are omitted and no return
8316 instructions or return sequences are generated in the code. Use this
8317 option only together with visual inspection of the compiled code: no
8318 warnings or errors are generated when call-saved registers must be saved,
8319 or storage for local variable needs to be allocated.
8325 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8326 instruction sequences that load addresses for functions from the PLT part
8327 of the GOT rather than (traditional on other architectures) calls to the
8328 PLT@. The default is @option{-mgotplt}.
8332 Legacy no-op option only recognized with the cris-axis-aout target.
8336 Legacy no-op option only recognized with the cris-axis-elf and
8337 cris-axis-linux-gnu targets.
8341 Only recognized with the cris-axis-aout target, where it selects a
8342 GNU/linux-like multilib, include files and instruction set for
8347 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8351 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8352 to link with input-output functions from a simulator library. Code,
8353 initialized data and zero-initialized data are allocated consecutively.
8357 Like @option{-sim}, but pass linker options to locate initialized data at
8358 0x40000000 and zero-initialized data at 0x80000000.
8362 @subsection CRX Options
8365 These options are defined specifically for the CRX ports.
8371 Enable the use of multiply-accumulate instructions. Disabled by default.
8375 Push instructions will be used to pass outgoing arguments when functions
8376 are called. Enabled by default.
8379 @node Darwin Options
8380 @subsection Darwin Options
8381 @cindex Darwin options
8383 These options are defined for all architectures running the Darwin operating
8386 FSF GCC on Darwin does not create ``fat'' object files; it will create
8387 an object file for the single architecture that it was built to
8388 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8389 @option{-arch} options are used; it does so by running the compiler or
8390 linker multiple times and joining the results together with
8393 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8394 @samp{i686}) is determined by the flags that specify the ISA
8395 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8396 @option{-force_cpusubtype_ALL} option can be used to override this.
8398 The Darwin tools vary in their behavior when presented with an ISA
8399 mismatch. The assembler, @file{as}, will only permit instructions to
8400 be used that are valid for the subtype of the file it is generating,
8401 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8402 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8403 and print an error if asked to create a shared library with a less
8404 restrictive subtype than its input files (for instance, trying to put
8405 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8406 for executables, @file{ld}, will quietly give the executable the most
8407 restrictive subtype of any of its input files.
8412 Add the framework directory @var{dir} to the head of the list of
8413 directories to be searched for header files. These directories are
8414 interleaved with those specified by @option{-I} options and are
8415 scanned in a left-to-right order.
8417 A framework directory is a directory with frameworks in it. A
8418 framework is a directory with a @samp{"Headers"} and/or
8419 @samp{"PrivateHeaders"} directory contained directly in it that ends
8420 in @samp{".framework"}. The name of a framework is the name of this
8421 directory excluding the @samp{".framework"}. Headers associated with
8422 the framework are found in one of those two directories, with
8423 @samp{"Headers"} being searched first. A subframework is a framework
8424 directory that is in a framework's @samp{"Frameworks"} directory.
8425 Includes of subframework headers can only appear in a header of a
8426 framework that contains the subframework, or in a sibling subframework
8427 header. Two subframeworks are siblings if they occur in the same
8428 framework. A subframework should not have the same name as a
8429 framework, a warning will be issued if this is violated. Currently a
8430 subframework cannot have subframeworks, in the future, the mechanism
8431 may be extended to support this. The standard frameworks can be found
8432 in @samp{"/System/Library/Frameworks"} and
8433 @samp{"/Library/Frameworks"}. An example include looks like
8434 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8435 the name of the framework and header.h is found in the
8436 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8440 Emit debugging information for symbols that are used. For STABS
8441 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8442 This is by default ON@.
8446 Emit debugging information for all symbols and types.
8448 @item -mmacosx-version-min=@var{version}
8449 The earliest version of MacOS X that this executable will run on
8450 is @var{version}. Typical values of @var{version} include @code{10.1},
8451 @code{10.2}, and @code{10.3.9}.
8453 The default for this option is to make choices that seem to be most
8458 Enable kernel development mode. The @option{-mkernel} option sets
8459 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8460 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8461 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8462 applicable. This mode also sets @option{-mno-altivec},
8463 @option{-msoft-float}, @option{-fno-builtin} and
8464 @option{-mlong-branch} for PowerPC targets.
8466 @item -mone-byte-bool
8467 @opindex -mone-byte-bool
8468 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8469 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8470 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8471 option has no effect on x86.
8473 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8474 to generate code that is not binary compatible with code generated
8475 without that switch. Using this switch may require recompiling all
8476 other modules in a program, including system libraries. Use this
8477 switch to conform to a non-default data model.
8479 @item -mfix-and-continue
8480 @itemx -ffix-and-continue
8481 @itemx -findirect-data
8482 @opindex mfix-and-continue
8483 @opindex ffix-and-continue
8484 @opindex findirect-data
8485 Generate code suitable for fast turn around development. Needed to
8486 enable gdb to dynamically load @code{.o} files into already running
8487 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8488 are provided for backwards compatibility.
8492 Loads all members of static archive libraries.
8493 See man ld(1) for more information.
8495 @item -arch_errors_fatal
8496 @opindex arch_errors_fatal
8497 Cause the errors having to do with files that have the wrong architecture
8501 @opindex bind_at_load
8502 Causes the output file to be marked such that the dynamic linker will
8503 bind all undefined references when the file is loaded or launched.
8507 Produce a Mach-o bundle format file.
8508 See man ld(1) for more information.
8510 @item -bundle_loader @var{executable}
8511 @opindex bundle_loader
8512 This option specifies the @var{executable} that will be loading the build
8513 output file being linked. See man ld(1) for more information.
8516 @opindex -dynamiclib
8517 When passed this option, GCC will produce a dynamic library instead of
8518 an executable when linking, using the Darwin @file{libtool} command.
8520 @item -force_cpusubtype_ALL
8521 @opindex -force_cpusubtype_ALL
8522 This causes GCC's output file to have the @var{ALL} subtype, instead of
8523 one controlled by the @option{-mcpu} or @option{-march} option.
8525 @item -allowable_client @var{client_name}
8527 @itemx -compatibility_version
8528 @itemx -current_version
8530 @itemx -dependency-file
8532 @itemx -dylinker_install_name
8534 @itemx -exported_symbols_list
8536 @itemx -flat_namespace
8537 @itemx -force_flat_namespace
8538 @itemx -headerpad_max_install_names
8541 @itemx -install_name
8542 @itemx -keep_private_externs
8543 @itemx -multi_module
8544 @itemx -multiply_defined
8545 @itemx -multiply_defined_unused
8547 @itemx -no_dead_strip_inits_and_terms
8548 @itemx -nofixprebinding
8551 @itemx -noseglinkedit
8552 @itemx -pagezero_size
8554 @itemx -prebind_all_twolevel_modules
8555 @itemx -private_bundle
8556 @itemx -read_only_relocs
8558 @itemx -sectobjectsymbols
8562 @itemx -sectobjectsymbols
8565 @itemx -segs_read_only_addr
8566 @itemx -segs_read_write_addr
8567 @itemx -seg_addr_table
8568 @itemx -seg_addr_table_filename
8571 @itemx -segs_read_only_addr
8572 @itemx -segs_read_write_addr
8573 @itemx -single_module
8576 @itemx -sub_umbrella
8577 @itemx -twolevel_namespace
8580 @itemx -unexported_symbols_list
8581 @itemx -weak_reference_mismatches
8584 @opindex allowable_client
8585 @opindex client_name
8586 @opindex compatibility_version
8587 @opindex current_version
8589 @opindex dependency-file
8591 @opindex dylinker_install_name
8593 @opindex exported_symbols_list
8595 @opindex flat_namespace
8596 @opindex force_flat_namespace
8597 @opindex headerpad_max_install_names
8600 @opindex install_name
8601 @opindex keep_private_externs
8602 @opindex multi_module
8603 @opindex multiply_defined
8604 @opindex multiply_defined_unused
8606 @opindex no_dead_strip_inits_and_terms
8607 @opindex nofixprebinding
8608 @opindex nomultidefs
8610 @opindex noseglinkedit
8611 @opindex pagezero_size
8613 @opindex prebind_all_twolevel_modules
8614 @opindex private_bundle
8615 @opindex read_only_relocs
8617 @opindex sectobjectsymbols
8621 @opindex sectobjectsymbols
8624 @opindex segs_read_only_addr
8625 @opindex segs_read_write_addr
8626 @opindex seg_addr_table
8627 @opindex seg_addr_table_filename
8628 @opindex seglinkedit
8630 @opindex segs_read_only_addr
8631 @opindex segs_read_write_addr
8632 @opindex single_module
8634 @opindex sub_library
8635 @opindex sub_umbrella
8636 @opindex twolevel_namespace
8639 @opindex unexported_symbols_list
8640 @opindex weak_reference_mismatches
8641 @opindex whatsloaded
8643 These options are passed to the Darwin linker. The Darwin linker man page
8644 describes them in detail.
8647 @node DEC Alpha Options
8648 @subsection DEC Alpha Options
8650 These @samp{-m} options are defined for the DEC Alpha implementations:
8653 @item -mno-soft-float
8655 @opindex mno-soft-float
8656 @opindex msoft-float
8657 Use (do not use) the hardware floating-point instructions for
8658 floating-point operations. When @option{-msoft-float} is specified,
8659 functions in @file{libgcc.a} will be used to perform floating-point
8660 operations. Unless they are replaced by routines that emulate the
8661 floating-point operations, or compiled in such a way as to call such
8662 emulations routines, these routines will issue floating-point
8663 operations. If you are compiling for an Alpha without floating-point
8664 operations, you must ensure that the library is built so as not to call
8667 Note that Alpha implementations without floating-point operations are
8668 required to have floating-point registers.
8673 @opindex mno-fp-regs
8674 Generate code that uses (does not use) the floating-point register set.
8675 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8676 register set is not used, floating point operands are passed in integer
8677 registers as if they were integers and floating-point results are passed
8678 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8679 so any function with a floating-point argument or return value called by code
8680 compiled with @option{-mno-fp-regs} must also be compiled with that
8683 A typical use of this option is building a kernel that does not use,
8684 and hence need not save and restore, any floating-point registers.
8688 The Alpha architecture implements floating-point hardware optimized for
8689 maximum performance. It is mostly compliant with the IEEE floating
8690 point standard. However, for full compliance, software assistance is
8691 required. This option generates code fully IEEE compliant code
8692 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8693 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8694 defined during compilation. The resulting code is less efficient but is
8695 able to correctly support denormalized numbers and exceptional IEEE
8696 values such as not-a-number and plus/minus infinity. Other Alpha
8697 compilers call this option @option{-ieee_with_no_inexact}.
8699 @item -mieee-with-inexact
8700 @opindex mieee-with-inexact
8701 This is like @option{-mieee} except the generated code also maintains
8702 the IEEE @var{inexact-flag}. Turning on this option causes the
8703 generated code to implement fully-compliant IEEE math. In addition to
8704 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8705 macro. On some Alpha implementations the resulting code may execute
8706 significantly slower than the code generated by default. Since there is
8707 very little code that depends on the @var{inexact-flag}, you should
8708 normally not specify this option. Other Alpha compilers call this
8709 option @option{-ieee_with_inexact}.
8711 @item -mfp-trap-mode=@var{trap-mode}
8712 @opindex mfp-trap-mode
8713 This option controls what floating-point related traps are enabled.
8714 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8715 The trap mode can be set to one of four values:
8719 This is the default (normal) setting. The only traps that are enabled
8720 are the ones that cannot be disabled in software (e.g., division by zero
8724 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8728 Like @samp{u}, but the instructions are marked to be safe for software
8729 completion (see Alpha architecture manual for details).
8732 Like @samp{su}, but inexact traps are enabled as well.
8735 @item -mfp-rounding-mode=@var{rounding-mode}
8736 @opindex mfp-rounding-mode
8737 Selects the IEEE rounding mode. Other Alpha compilers call this option
8738 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8743 Normal IEEE rounding mode. Floating point numbers are rounded towards
8744 the nearest machine number or towards the even machine number in case
8748 Round towards minus infinity.
8751 Chopped rounding mode. Floating point numbers are rounded towards zero.
8754 Dynamic rounding mode. A field in the floating point control register
8755 (@var{fpcr}, see Alpha architecture reference manual) controls the
8756 rounding mode in effect. The C library initializes this register for
8757 rounding towards plus infinity. Thus, unless your program modifies the
8758 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8761 @item -mtrap-precision=@var{trap-precision}
8762 @opindex mtrap-precision
8763 In the Alpha architecture, floating point traps are imprecise. This
8764 means without software assistance it is impossible to recover from a
8765 floating trap and program execution normally needs to be terminated.
8766 GCC can generate code that can assist operating system trap handlers
8767 in determining the exact location that caused a floating point trap.
8768 Depending on the requirements of an application, different levels of
8769 precisions can be selected:
8773 Program precision. This option is the default and means a trap handler
8774 can only identify which program caused a floating point exception.
8777 Function precision. The trap handler can determine the function that
8778 caused a floating point exception.
8781 Instruction precision. The trap handler can determine the exact
8782 instruction that caused a floating point exception.
8785 Other Alpha compilers provide the equivalent options called
8786 @option{-scope_safe} and @option{-resumption_safe}.
8788 @item -mieee-conformant
8789 @opindex mieee-conformant
8790 This option marks the generated code as IEEE conformant. You must not
8791 use this option unless you also specify @option{-mtrap-precision=i} and either
8792 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8793 is to emit the line @samp{.eflag 48} in the function prologue of the
8794 generated assembly file. Under DEC Unix, this has the effect that
8795 IEEE-conformant math library routines will be linked in.
8797 @item -mbuild-constants
8798 @opindex mbuild-constants
8799 Normally GCC examines a 32- or 64-bit integer constant to
8800 see if it can construct it from smaller constants in two or three
8801 instructions. If it cannot, it will output the constant as a literal and
8802 generate code to load it from the data segment at runtime.
8804 Use this option to require GCC to construct @emph{all} integer constants
8805 using code, even if it takes more instructions (the maximum is six).
8807 You would typically use this option to build a shared library dynamic
8808 loader. Itself a shared library, it must relocate itself in memory
8809 before it can find the variables and constants in its own data segment.
8815 Select whether to generate code to be assembled by the vendor-supplied
8816 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8834 Indicate whether GCC should generate code to use the optional BWX,
8835 CIX, FIX and MAX instruction sets. The default is to use the instruction
8836 sets supported by the CPU type specified via @option{-mcpu=} option or that
8837 of the CPU on which GCC was built if none was specified.
8842 @opindex mfloat-ieee
8843 Generate code that uses (does not use) VAX F and G floating point
8844 arithmetic instead of IEEE single and double precision.
8846 @item -mexplicit-relocs
8847 @itemx -mno-explicit-relocs
8848 @opindex mexplicit-relocs
8849 @opindex mno-explicit-relocs
8850 Older Alpha assemblers provided no way to generate symbol relocations
8851 except via assembler macros. Use of these macros does not allow
8852 optimal instruction scheduling. GNU binutils as of version 2.12
8853 supports a new syntax that allows the compiler to explicitly mark
8854 which relocations should apply to which instructions. This option
8855 is mostly useful for debugging, as GCC detects the capabilities of
8856 the assembler when it is built and sets the default accordingly.
8860 @opindex msmall-data
8861 @opindex mlarge-data
8862 When @option{-mexplicit-relocs} is in effect, static data is
8863 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8864 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8865 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8866 16-bit relocations off of the @code{$gp} register. This limits the
8867 size of the small data area to 64KB, but allows the variables to be
8868 directly accessed via a single instruction.
8870 The default is @option{-mlarge-data}. With this option the data area
8871 is limited to just below 2GB@. Programs that require more than 2GB of
8872 data must use @code{malloc} or @code{mmap} to allocate the data in the
8873 heap instead of in the program's data segment.
8875 When generating code for shared libraries, @option{-fpic} implies
8876 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8880 @opindex msmall-text
8881 @opindex mlarge-text
8882 When @option{-msmall-text} is used, the compiler assumes that the
8883 code of the entire program (or shared library) fits in 4MB, and is
8884 thus reachable with a branch instruction. When @option{-msmall-data}
8885 is used, the compiler can assume that all local symbols share the
8886 same @code{$gp} value, and thus reduce the number of instructions
8887 required for a function call from 4 to 1.
8889 The default is @option{-mlarge-text}.
8891 @item -mcpu=@var{cpu_type}
8893 Set the instruction set and instruction scheduling parameters for
8894 machine type @var{cpu_type}. You can specify either the @samp{EV}
8895 style name or the corresponding chip number. GCC supports scheduling
8896 parameters for the EV4, EV5 and EV6 family of processors and will
8897 choose the default values for the instruction set from the processor
8898 you specify. If you do not specify a processor type, GCC will default
8899 to the processor on which the compiler was built.
8901 Supported values for @var{cpu_type} are
8907 Schedules as an EV4 and has no instruction set extensions.
8911 Schedules as an EV5 and has no instruction set extensions.
8915 Schedules as an EV5 and supports the BWX extension.
8920 Schedules as an EV5 and supports the BWX and MAX extensions.
8924 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8928 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8931 @item -mtune=@var{cpu_type}
8933 Set only the instruction scheduling parameters for machine type
8934 @var{cpu_type}. The instruction set is not changed.
8936 @item -mmemory-latency=@var{time}
8937 @opindex mmemory-latency
8938 Sets the latency the scheduler should assume for typical memory
8939 references as seen by the application. This number is highly
8940 dependent on the memory access patterns used by the application
8941 and the size of the external cache on the machine.
8943 Valid options for @var{time} are
8947 A decimal number representing clock cycles.
8953 The compiler contains estimates of the number of clock cycles for
8954 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8955 (also called Dcache, Scache, and Bcache), as well as to main memory.
8956 Note that L3 is only valid for EV5.
8961 @node DEC Alpha/VMS Options
8962 @subsection DEC Alpha/VMS Options
8964 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8967 @item -mvms-return-codes
8968 @opindex mvms-return-codes
8969 Return VMS condition codes from main. The default is to return POSIX
8970 style condition (e.g.@ error) codes.
8974 @subsection FRV Options
8981 Only use the first 32 general purpose registers.
8986 Use all 64 general purpose registers.
8991 Use only the first 32 floating point registers.
8996 Use all 64 floating point registers
8999 @opindex mhard-float
9001 Use hardware instructions for floating point operations.
9004 @opindex msoft-float
9006 Use library routines for floating point operations.
9011 Dynamically allocate condition code registers.
9016 Do not try to dynamically allocate condition code registers, only
9017 use @code{icc0} and @code{fcc0}.
9022 Change ABI to use double word insns.
9027 Do not use double word instructions.
9032 Use floating point double instructions.
9037 Do not use floating point double instructions.
9042 Use media instructions.
9047 Do not use media instructions.
9052 Use multiply and add/subtract instructions.
9057 Do not use multiply and add/subtract instructions.
9062 Select the FDPIC ABI, that uses function descriptors to represent
9063 pointers to functions. Without any PIC/PIE-related options, it
9064 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9065 assumes GOT entries and small data are within a 12-bit range from the
9066 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9067 are computed with 32 bits.
9070 @opindex minline-plt
9072 Enable inlining of PLT entries in function calls to functions that are
9073 not known to bind locally. It has no effect without @option{-mfdpic}.
9074 It's enabled by default if optimizing for speed and compiling for
9075 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9076 optimization option such as @option{-O3} or above is present in the
9082 Assume a large TLS segment when generating thread-local code.
9087 Do not assume a large TLS segment when generating thread-local code.
9092 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9093 that is known to be in read-only sections. It's enabled by default,
9094 except for @option{-fpic} or @option{-fpie}: even though it may help
9095 make the global offset table smaller, it trades 1 instruction for 4.
9096 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9097 one of which may be shared by multiple symbols, and it avoids the need
9098 for a GOT entry for the referenced symbol, so it's more likely to be a
9099 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9101 @item -multilib-library-pic
9102 @opindex multilib-library-pic
9104 Link with the (library, not FD) pic libraries. It's implied by
9105 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9106 @option{-fpic} without @option{-mfdpic}. You should never have to use
9112 Follow the EABI requirement of always creating a frame pointer whenever
9113 a stack frame is allocated. This option is enabled by default and can
9114 be disabled with @option{-mno-linked-fp}.
9117 @opindex mlong-calls
9119 Use indirect addressing to call functions outside the current
9120 compilation unit. This allows the functions to be placed anywhere
9121 within the 32-bit address space.
9123 @item -malign-labels
9124 @opindex malign-labels
9126 Try to align labels to an 8-byte boundary by inserting nops into the
9127 previous packet. This option only has an effect when VLIW packing
9128 is enabled. It doesn't create new packets; it merely adds nops to
9132 @opindex mlibrary-pic
9134 Generate position-independent EABI code.
9139 Use only the first four media accumulator registers.
9144 Use all eight media accumulator registers.
9149 Pack VLIW instructions.
9154 Do not pack VLIW instructions.
9159 Do not mark ABI switches in e_flags.
9164 Enable the use of conditional-move instructions (default).
9166 This switch is mainly for debugging the compiler and will likely be removed
9167 in a future version.
9169 @item -mno-cond-move
9170 @opindex mno-cond-move
9172 Disable the use of conditional-move instructions.
9174 This switch is mainly for debugging the compiler and will likely be removed
9175 in a future version.
9180 Enable the use of conditional set instructions (default).
9182 This switch is mainly for debugging the compiler and will likely be removed
9183 in a future version.
9188 Disable the use of conditional set instructions.
9190 This switch is mainly for debugging the compiler and will likely be removed
9191 in a future version.
9196 Enable the use of conditional execution (default).
9198 This switch is mainly for debugging the compiler and will likely be removed
9199 in a future version.
9201 @item -mno-cond-exec
9202 @opindex mno-cond-exec
9204 Disable the use of conditional execution.
9206 This switch is mainly for debugging the compiler and will likely be removed
9207 in a future version.
9210 @opindex mvliw-branch
9212 Run a pass to pack branches into VLIW instructions (default).
9214 This switch is mainly for debugging the compiler and will likely be removed
9215 in a future version.
9217 @item -mno-vliw-branch
9218 @opindex mno-vliw-branch
9220 Do not run a pass to pack branches into VLIW instructions.
9222 This switch is mainly for debugging the compiler and will likely be removed
9223 in a future version.
9225 @item -mmulti-cond-exec
9226 @opindex mmulti-cond-exec
9228 Enable optimization of @code{&&} and @code{||} in conditional execution
9231 This switch is mainly for debugging the compiler and will likely be removed
9232 in a future version.
9234 @item -mno-multi-cond-exec
9235 @opindex mno-multi-cond-exec
9237 Disable optimization of @code{&&} and @code{||} in conditional execution.
9239 This switch is mainly for debugging the compiler and will likely be removed
9240 in a future version.
9242 @item -mnested-cond-exec
9243 @opindex mnested-cond-exec
9245 Enable nested conditional execution optimizations (default).
9247 This switch is mainly for debugging the compiler and will likely be removed
9248 in a future version.
9250 @item -mno-nested-cond-exec
9251 @opindex mno-nested-cond-exec
9253 Disable nested conditional execution optimizations.
9255 This switch is mainly for debugging the compiler and will likely be removed
9256 in a future version.
9258 @item -moptimize-membar
9259 @opindex moptimize-membar
9261 This switch removes redundant @code{membar} instructions from the
9262 compiler generated code. It is enabled by default.
9264 @item -mno-optimize-membar
9265 @opindex mno-optimize-membar
9267 This switch disables the automatic removal of redundant @code{membar}
9268 instructions from the generated code.
9270 @item -mtomcat-stats
9271 @opindex mtomcat-stats
9273 Cause gas to print out tomcat statistics.
9275 @item -mcpu=@var{cpu}
9278 Select the processor type for which to generate code. Possible values are
9279 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9280 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9284 @node GNU/Linux Options
9285 @subsection GNU/Linux Options
9287 These @samp{-m} options are defined for GNU/Linux targets:
9292 Use the GNU C library instead of uClibc. This is the default except
9293 on @samp{*-*-linux-*uclibc*} targets.
9297 Use uClibc instead of the GNU C library. This is the default on
9298 @samp{*-*-linux-*uclibc*} targets.
9301 @node H8/300 Options
9302 @subsection H8/300 Options
9304 These @samp{-m} options are defined for the H8/300 implementations:
9309 Shorten some address references at link time, when possible; uses the
9310 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9311 ld, Using ld}, for a fuller description.
9315 Generate code for the H8/300H@.
9319 Generate code for the H8S@.
9323 Generate code for the H8S and H8/300H in the normal mode. This switch
9324 must be used either with @option{-mh} or @option{-ms}.
9328 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9332 Make @code{int} data 32 bits by default.
9336 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9337 The default for the H8/300H and H8S is to align longs and floats on 4
9339 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9340 This option has no effect on the H8/300.
9344 @subsection HPPA Options
9345 @cindex HPPA Options
9347 These @samp{-m} options are defined for the HPPA family of computers:
9350 @item -march=@var{architecture-type}
9352 Generate code for the specified architecture. The choices for
9353 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9354 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9355 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9356 architecture option for your machine. Code compiled for lower numbered
9357 architectures will run on higher numbered architectures, but not the
9361 @itemx -mpa-risc-1-1
9362 @itemx -mpa-risc-2-0
9363 @opindex mpa-risc-1-0
9364 @opindex mpa-risc-1-1
9365 @opindex mpa-risc-2-0
9366 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9369 @opindex mbig-switch
9370 Generate code suitable for big switch tables. Use this option only if
9371 the assembler/linker complain about out of range branches within a switch
9374 @item -mjump-in-delay
9375 @opindex mjump-in-delay
9376 Fill delay slots of function calls with unconditional jump instructions
9377 by modifying the return pointer for the function call to be the target
9378 of the conditional jump.
9380 @item -mdisable-fpregs
9381 @opindex mdisable-fpregs
9382 Prevent floating point registers from being used in any manner. This is
9383 necessary for compiling kernels which perform lazy context switching of
9384 floating point registers. If you use this option and attempt to perform
9385 floating point operations, the compiler will abort.
9387 @item -mdisable-indexing
9388 @opindex mdisable-indexing
9389 Prevent the compiler from using indexing address modes. This avoids some
9390 rather obscure problems when compiling MIG generated code under MACH@.
9392 @item -mno-space-regs
9393 @opindex mno-space-regs
9394 Generate code that assumes the target has no space registers. This allows
9395 GCC to generate faster indirect calls and use unscaled index address modes.
9397 Such code is suitable for level 0 PA systems and kernels.
9399 @item -mfast-indirect-calls
9400 @opindex mfast-indirect-calls
9401 Generate code that assumes calls never cross space boundaries. This
9402 allows GCC to emit code which performs faster indirect calls.
9404 This option will not work in the presence of shared libraries or nested
9407 @item -mfixed-range=@var{register-range}
9408 @opindex mfixed-range
9409 Generate code treating the given register range as fixed registers.
9410 A fixed register is one that the register allocator can not use. This is
9411 useful when compiling kernel code. A register range is specified as
9412 two registers separated by a dash. Multiple register ranges can be
9413 specified separated by a comma.
9415 @item -mlong-load-store
9416 @opindex mlong-load-store
9417 Generate 3-instruction load and store sequences as sometimes required by
9418 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9421 @item -mportable-runtime
9422 @opindex mportable-runtime
9423 Use the portable calling conventions proposed by HP for ELF systems.
9427 Enable the use of assembler directives only GAS understands.
9429 @item -mschedule=@var{cpu-type}
9431 Schedule code according to the constraints for the machine type
9432 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9433 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9434 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9435 proper scheduling option for your machine. The default scheduling is
9439 @opindex mlinker-opt
9440 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9441 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9442 linkers in which they give bogus error messages when linking some programs.
9445 @opindex msoft-float
9446 Generate output containing library calls for floating point.
9447 @strong{Warning:} the requisite libraries are not available for all HPPA
9448 targets. Normally the facilities of the machine's usual C compiler are
9449 used, but this cannot be done directly in cross-compilation. You must make
9450 your own arrangements to provide suitable library functions for
9451 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9452 does provide software floating point support.
9454 @option{-msoft-float} changes the calling convention in the output file;
9455 therefore, it is only useful if you compile @emph{all} of a program with
9456 this option. In particular, you need to compile @file{libgcc.a}, the
9457 library that comes with GCC, with @option{-msoft-float} in order for
9462 Generate the predefine, @code{_SIO}, for server IO@. The default is
9463 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9464 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9465 options are available under HP-UX and HI-UX@.
9469 Use GNU ld specific options. This passes @option{-shared} to ld when
9470 building a shared library. It is the default when GCC is configured,
9471 explicitly or implicitly, with the GNU linker. This option does not
9472 have any affect on which ld is called, it only changes what parameters
9473 are passed to that ld. The ld that is called is determined by the
9474 @option{--with-ld} configure option, GCC's program search path, and
9475 finally by the user's @env{PATH}. The linker used by GCC can be printed
9476 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9477 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9481 Use HP ld specific options. This passes @option{-b} to ld when building
9482 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9483 links. It is the default when GCC is configured, explicitly or
9484 implicitly, with the HP linker. This option does not have any affect on
9485 which ld is called, it only changes what parameters are passed to that
9486 ld. The ld that is called is determined by the @option{--with-ld}
9487 configure option, GCC's program search path, and finally by the user's
9488 @env{PATH}. The linker used by GCC can be printed using @samp{which
9489 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9490 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9493 @opindex mno-long-calls
9494 Generate code that uses long call sequences. This ensures that a call
9495 is always able to reach linker generated stubs. The default is to generate
9496 long calls only when the distance from the call site to the beginning
9497 of the function or translation unit, as the case may be, exceeds a
9498 predefined limit set by the branch type being used. The limits for
9499 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9500 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9503 Distances are measured from the beginning of functions when using the
9504 @option{-ffunction-sections} option, or when using the @option{-mgas}
9505 and @option{-mno-portable-runtime} options together under HP-UX with
9508 It is normally not desirable to use this option as it will degrade
9509 performance. However, it may be useful in large applications,
9510 particularly when partial linking is used to build the application.
9512 The types of long calls used depends on the capabilities of the
9513 assembler and linker, and the type of code being generated. The
9514 impact on systems that support long absolute calls, and long pic
9515 symbol-difference or pc-relative calls should be relatively small.
9516 However, an indirect call is used on 32-bit ELF systems in pic code
9517 and it is quite long.
9519 @item -munix=@var{unix-std}
9521 Generate compiler predefines and select a startfile for the specified
9522 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9523 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9524 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9525 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9526 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9529 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9530 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9531 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9532 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9533 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9534 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9536 It is @emph{important} to note that this option changes the interfaces
9537 for various library routines. It also affects the operational behavior
9538 of the C library. Thus, @emph{extreme} care is needed in using this
9541 Library code that is intended to operate with more than one UNIX
9542 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9543 as appropriate. Most GNU software doesn't provide this capability.
9547 Suppress the generation of link options to search libdld.sl when the
9548 @option{-static} option is specified on HP-UX 10 and later.
9552 The HP-UX implementation of setlocale in libc has a dependency on
9553 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9554 when the @option{-static} option is specified, special link options
9555 are needed to resolve this dependency.
9557 On HP-UX 10 and later, the GCC driver adds the necessary options to
9558 link with libdld.sl when the @option{-static} option is specified.
9559 This causes the resulting binary to be dynamic. On the 64-bit port,
9560 the linkers generate dynamic binaries by default in any case. The
9561 @option{-nolibdld} option can be used to prevent the GCC driver from
9562 adding these link options.
9566 Add support for multithreading with the @dfn{dce thread} library
9567 under HP-UX@. This option sets flags for both the preprocessor and
9571 @node i386 and x86-64 Options
9572 @subsection Intel 386 and AMD x86-64 Options
9573 @cindex i386 Options
9574 @cindex x86-64 Options
9575 @cindex Intel 386 Options
9576 @cindex AMD x86-64 Options
9578 These @samp{-m} options are defined for the i386 and x86-64 family of
9582 @item -mtune=@var{cpu-type}
9584 Tune to @var{cpu-type} everything applicable about the generated code, except
9585 for the ABI and the set of available instructions. The choices for
9589 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9590 If you know the CPU on which your code will run, then you should use
9591 the corresponding @option{-mtune} option instead of
9592 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9593 of your application will have, then you should use this option.
9595 As new processors are deployed in the marketplace, the behavior of this
9596 option will change. Therefore, if you upgrade to a newer version of
9597 GCC, the code generated option will change to reflect the processors
9598 that were most common when that version of GCC was released.
9600 There is no @option{-march=generic} option because @option{-march}
9601 indicates the instruction set the compiler can use, and there is no
9602 generic instruction set applicable to all processors. In contrast,
9603 @option{-mtune} indicates the processor (or, in this case, collection of
9604 processors) for which the code is optimized.
9606 This selects the CPU to tune for at compilation time by determining
9607 the processor type of the compiling machine. Using @option{-mtune=native}
9608 will produce code optimized for the local machine under the constraints
9609 of the selected instruction set. Using @option{-march=native} will
9610 enable all instruction subsets supported by the local machine (hence
9611 the result might not run on different machines).
9613 Original Intel's i386 CPU@.
9615 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9617 Intel Pentium CPU with no MMX support.
9619 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9621 Intel PentiumPro CPU@.
9623 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9624 instruction set will be used, so the code will run on all i686 family chips.
9626 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9627 @item pentium3, pentium3m
9628 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9631 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9632 support. Used by Centrino notebooks.
9633 @item pentium4, pentium4m
9634 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9636 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9639 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9640 SSE2 and SSE3 instruction set support.
9642 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
9643 instruction set support.
9645 AMD K6 CPU with MMX instruction set support.
9647 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9648 @item athlon, athlon-tbird
9649 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9651 @item athlon-4, athlon-xp, athlon-mp
9652 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9653 instruction set support.
9654 @item k8, opteron, athlon64, athlon-fx
9655 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9656 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9658 AMD Family 10 core based CPUs with x86-64 instruction set support. (This
9659 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
9660 instruction set extensions.)
9662 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9665 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9666 instruction set support.
9668 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9669 implemented for this chip.)
9671 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9672 implemented for this chip.)
9674 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
9677 While picking a specific @var{cpu-type} will schedule things appropriately
9678 for that particular chip, the compiler will not generate any code that
9679 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9682 @item -march=@var{cpu-type}
9684 Generate instructions for the machine type @var{cpu-type}. The choices
9685 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9686 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9688 @item -mcpu=@var{cpu-type}
9690 A deprecated synonym for @option{-mtune}.
9699 @opindex mpentiumpro
9700 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9701 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9702 These synonyms are deprecated.
9704 @item -mfpmath=@var{unit}
9706 Generate floating point arithmetics for selected unit @var{unit}. The choices
9711 Use the standard 387 floating point coprocessor present majority of chips and
9712 emulated otherwise. Code compiled with this option will run almost everywhere.
9713 The temporary results are computed in 80bit precision instead of precision
9714 specified by the type resulting in slightly different results compared to most
9715 of other chips. See @option{-ffloat-store} for more detailed description.
9717 This is the default choice for i386 compiler.
9720 Use scalar floating point instructions present in the SSE instruction set.
9721 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9722 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9723 instruction set supports only single precision arithmetics, thus the double and
9724 extended precision arithmetics is still done using 387. Later version, present
9725 only in Pentium4 and the future AMD x86-64 chips supports double precision
9728 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9729 or @option{-msse2} switches to enable SSE extensions and make this option
9730 effective. For the x86-64 compiler, these extensions are enabled by default.
9732 The resulting code should be considerably faster in the majority of cases and avoid
9733 the numerical instability problems of 387 code, but may break some existing
9734 code that expects temporaries to be 80bit.
9736 This is the default choice for the x86-64 compiler.
9739 Attempt to utilize both instruction sets at once. This effectively double the
9740 amount of available registers and on chips with separate execution units for
9741 387 and SSE the execution resources too. Use this option with care, as it is
9742 still experimental, because the GCC register allocator does not model separate
9743 functional units well resulting in instable performance.
9746 @item -masm=@var{dialect}
9747 @opindex masm=@var{dialect}
9748 Output asm instructions using selected @var{dialect}. Supported
9749 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9750 not support @samp{intel}.
9755 @opindex mno-ieee-fp
9756 Control whether or not the compiler uses IEEE floating point
9757 comparisons. These handle correctly the case where the result of a
9758 comparison is unordered.
9761 @opindex msoft-float
9762 Generate output containing library calls for floating point.
9763 @strong{Warning:} the requisite libraries are not part of GCC@.
9764 Normally the facilities of the machine's usual C compiler are used, but
9765 this can't be done directly in cross-compilation. You must make your
9766 own arrangements to provide suitable library functions for
9769 On machines where a function returns floating point results in the 80387
9770 register stack, some floating point opcodes may be emitted even if
9771 @option{-msoft-float} is used.
9773 @item -mno-fp-ret-in-387
9774 @opindex mno-fp-ret-in-387
9775 Do not use the FPU registers for return values of functions.
9777 The usual calling convention has functions return values of types
9778 @code{float} and @code{double} in an FPU register, even if there
9779 is no FPU@. The idea is that the operating system should emulate
9782 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9783 in ordinary CPU registers instead.
9785 @item -mno-fancy-math-387
9786 @opindex mno-fancy-math-387
9787 Some 387 emulators do not support the @code{sin}, @code{cos} and
9788 @code{sqrt} instructions for the 387. Specify this option to avoid
9789 generating those instructions. This option is the default on FreeBSD,
9790 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9791 indicates that the target cpu will always have an FPU and so the
9792 instruction will not need emulation. As of revision 2.6.1, these
9793 instructions are not generated unless you also use the
9794 @option{-funsafe-math-optimizations} switch.
9796 @item -malign-double
9797 @itemx -mno-align-double
9798 @opindex malign-double
9799 @opindex mno-align-double
9800 Control whether GCC aligns @code{double}, @code{long double}, and
9801 @code{long long} variables on a two word boundary or a one word
9802 boundary. Aligning @code{double} variables on a two word boundary will
9803 produce code that runs somewhat faster on a @samp{Pentium} at the
9804 expense of more memory.
9806 On x86-64, @option{-malign-double} is enabled by default.
9808 @strong{Warning:} if you use the @option{-malign-double} switch,
9809 structures containing the above types will be aligned differently than
9810 the published application binary interface specifications for the 386
9811 and will not be binary compatible with structures in code compiled
9812 without that switch.
9814 @item -m96bit-long-double
9815 @itemx -m128bit-long-double
9816 @opindex m96bit-long-double
9817 @opindex m128bit-long-double
9818 These switches control the size of @code{long double} type. The i386
9819 application binary interface specifies the size to be 96 bits,
9820 so @option{-m96bit-long-double} is the default in 32 bit mode.
9822 Modern architectures (Pentium and newer) would prefer @code{long double}
9823 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9824 conforming to the ABI, this would not be possible. So specifying a
9825 @option{-m128bit-long-double} will align @code{long double}
9826 to a 16 byte boundary by padding the @code{long double} with an additional
9829 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9830 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9832 Notice that neither of these options enable any extra precision over the x87
9833 standard of 80 bits for a @code{long double}.
9835 @strong{Warning:} if you override the default value for your target ABI, the
9836 structures and arrays containing @code{long double} variables will change
9837 their size as well as function calling convention for function taking
9838 @code{long double} will be modified. Hence they will not be binary
9839 compatible with arrays or structures in code compiled without that switch.
9841 @item -mmlarge-data-threshold=@var{number}
9842 @opindex mlarge-data-threshold=@var{number}
9843 When @option{-mcmodel=medium} is specified, the data greater than
9844 @var{threshold} are placed in large data section. This value must be the
9845 same across all object linked into the binary and defaults to 65535.
9848 @itemx -mno-svr3-shlib
9849 @opindex msvr3-shlib
9850 @opindex mno-svr3-shlib
9851 Control whether GCC places uninitialized local variables into the
9852 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9853 into @code{bss}. These options are meaningful only on System V Release 3.
9857 Use a different function-calling convention, in which functions that
9858 take a fixed number of arguments return with the @code{ret} @var{num}
9859 instruction, which pops their arguments while returning. This saves one
9860 instruction in the caller since there is no need to pop the arguments
9863 You can specify that an individual function is called with this calling
9864 sequence with the function attribute @samp{stdcall}. You can also
9865 override the @option{-mrtd} option by using the function attribute
9866 @samp{cdecl}. @xref{Function Attributes}.
9868 @strong{Warning:} this calling convention is incompatible with the one
9869 normally used on Unix, so you cannot use it if you need to call
9870 libraries compiled with the Unix compiler.
9872 Also, you must provide function prototypes for all functions that
9873 take variable numbers of arguments (including @code{printf});
9874 otherwise incorrect code will be generated for calls to those
9877 In addition, seriously incorrect code will result if you call a
9878 function with too many arguments. (Normally, extra arguments are
9879 harmlessly ignored.)
9881 @item -mregparm=@var{num}
9883 Control how many registers are used to pass integer arguments. By
9884 default, no registers are used to pass arguments, and at most 3
9885 registers can be used. You can control this behavior for a specific
9886 function by using the function attribute @samp{regparm}.
9887 @xref{Function Attributes}.
9889 @strong{Warning:} if you use this switch, and
9890 @var{num} is nonzero, then you must build all modules with the same
9891 value, including any libraries. This includes the system libraries and
9895 @opindex msseregparm
9896 Use SSE register passing conventions for float and double arguments
9897 and return values. You can control this behavior for a specific
9898 function by using the function attribute @samp{sseregparm}.
9899 @xref{Function Attributes}.
9901 @strong{Warning:} if you use this switch then you must build all
9902 modules with the same value, including any libraries. This includes
9903 the system libraries and startup modules.
9905 @item -mstackrealign
9906 @opindex mstackrealign
9907 Realign the stack at entry. On the Intel x86, the
9908 @option{-mstackrealign} option will generate an alternate prologue and
9909 epilogue that realigns the runtime stack. This supports mixing legacy
9910 codes that keep a 4-byte aligned stack with modern codes that keep a
9911 16-byte stack for SSE compatibility. The alternate prologue and
9912 epilogue are slower and bigger than the regular ones, and the
9913 alternate prologue requires an extra scratch register; this lowers the
9914 number of registers available if used in conjunction with the
9915 @code{regparm} attribute. The @option{-mstackrealign} option is
9916 incompatible with the nested function prologue; this is considered a
9917 hard error. See also the attribute @code{force_align_arg_pointer},
9918 applicable to individual functions.
9920 @item -mpreferred-stack-boundary=@var{num}
9921 @opindex mpreferred-stack-boundary
9922 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9923 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9924 the default is 4 (16 bytes or 128 bits).
9926 On Pentium and PentiumPro, @code{double} and @code{long double} values
9927 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9928 suffer significant run time performance penalties. On Pentium III, the
9929 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
9930 properly if it is not 16 byte aligned.
9932 To ensure proper alignment of this values on the stack, the stack boundary
9933 must be as aligned as that required by any value stored on the stack.
9934 Further, every function must be generated such that it keeps the stack
9935 aligned. Thus calling a function compiled with a higher preferred
9936 stack boundary from a function compiled with a lower preferred stack
9937 boundary will most likely misalign the stack. It is recommended that
9938 libraries that use callbacks always use the default setting.
9940 This extra alignment does consume extra stack space, and generally
9941 increases code size. Code that is sensitive to stack space usage, such
9942 as embedded systems and operating system kernels, may want to reduce the
9943 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9969 These switches enable or disable the use of instructions in the MMX,
9970 SSE, SSE2, SSE3, SSSE3, SSE4A, ABM or 3DNow! extended instruction sets.
9971 These extensions are also available as built-in functions: see
9972 @ref{X86 Built-in Functions}, for details of the functions enabled and
9973 disabled by these switches.
9975 To have SSE/SSE2 instructions generated automatically from floating-point
9976 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9978 These options will enable GCC to use these extended instructions in
9979 generated code, even without @option{-mfpmath=sse}. Applications which
9980 perform runtime CPU detection must compile separate files for each
9981 supported architecture, using the appropriate flags. In particular,
9982 the file containing the CPU detection code should be compiled without
9986 @itemx -mno-push-args
9988 @opindex mno-push-args
9989 Use PUSH operations to store outgoing parameters. This method is shorter
9990 and usually equally fast as method using SUB/MOV operations and is enabled
9991 by default. In some cases disabling it may improve performance because of
9992 improved scheduling and reduced dependencies.
9994 @item -maccumulate-outgoing-args
9995 @opindex maccumulate-outgoing-args
9996 If enabled, the maximum amount of space required for outgoing arguments will be
9997 computed in the function prologue. This is faster on most modern CPUs
9998 because of reduced dependencies, improved scheduling and reduced stack usage
9999 when preferred stack boundary is not equal to 2. The drawback is a notable
10000 increase in code size. This switch implies @option{-mno-push-args}.
10004 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10005 on thread-safe exception handling must compile and link all code with the
10006 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10007 @option{-D_MT}; when linking, it links in a special thread helper library
10008 @option{-lmingwthrd} which cleans up per thread exception handling data.
10010 @item -mno-align-stringops
10011 @opindex mno-align-stringops
10012 Do not align destination of inlined string operations. This switch reduces
10013 code size and improves performance in case the destination is already aligned,
10014 but GCC doesn't know about it.
10016 @item -minline-all-stringops
10017 @opindex minline-all-stringops
10018 By default GCC inlines string operations only when destination is known to be
10019 aligned at least to 4 byte boundary. This enables more inlining, increase code
10020 size, but may improve performance of code that depends on fast memcpy, strlen
10021 and memset for short lengths.
10023 @item -minline-stringops-dynamically
10024 @opindex minline-stringops-dynamically
10025 For string operation of unknown size, inline runtime checks so for small
10026 blocks inline code is used, while for large blocks library call is used.
10028 @item -mstringop-strategy=@var{alg}
10029 @opindex mstringop-strategy=@var{alg}
10030 Overwrite internal decision heuristic about particular algorithm to inline
10031 string operation with. The allowed values are @code{rep_byte},
10032 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10033 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10034 expanding inline loop, @code{libcall} for always expanding library call.
10036 @item -momit-leaf-frame-pointer
10037 @opindex momit-leaf-frame-pointer
10038 Don't keep the frame pointer in a register for leaf functions. This
10039 avoids the instructions to save, set up and restore frame pointers and
10040 makes an extra register available in leaf functions. The option
10041 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10042 which might make debugging harder.
10044 @item -mtls-direct-seg-refs
10045 @itemx -mno-tls-direct-seg-refs
10046 @opindex mtls-direct-seg-refs
10047 Controls whether TLS variables may be accessed with offsets from the
10048 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10049 or whether the thread base pointer must be added. Whether or not this
10050 is legal depends on the operating system, and whether it maps the
10051 segment to cover the entire TLS area.
10053 For systems that use GNU libc, the default is on.
10056 These @samp{-m} switches are supported in addition to the above
10057 on AMD x86-64 processors in 64-bit environments.
10064 Generate code for a 32-bit or 64-bit environment.
10065 The 32-bit environment sets int, long and pointer to 32 bits and
10066 generates code that runs on any i386 system.
10067 The 64-bit environment sets int to 32 bits and long and pointer
10068 to 64 bits and generates code for AMD's x86-64 architecture.
10070 @item -mno-red-zone
10071 @opindex no-red-zone
10072 Do not use a so called red zone for x86-64 code. The red zone is mandated
10073 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10074 stack pointer that will not be modified by signal or interrupt handlers
10075 and therefore can be used for temporary data without adjusting the stack
10076 pointer. The flag @option{-mno-red-zone} disables this red zone.
10078 @item -mcmodel=small
10079 @opindex mcmodel=small
10080 Generate code for the small code model: the program and its symbols must
10081 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10082 Programs can be statically or dynamically linked. This is the default
10085 @item -mcmodel=kernel
10086 @opindex mcmodel=kernel
10087 Generate code for the kernel code model. The kernel runs in the
10088 negative 2 GB of the address space.
10089 This model has to be used for Linux kernel code.
10091 @item -mcmodel=medium
10092 @opindex mcmodel=medium
10093 Generate code for the medium model: The program is linked in the lower 2
10094 GB of the address space but symbols can be located anywhere in the
10095 address space. Programs can be statically or dynamically linked, but
10096 building of shared libraries are not supported with the medium model.
10098 @item -mcmodel=large
10099 @opindex mcmodel=large
10100 Generate code for the large model: This model makes no assumptions
10101 about addresses and sizes of sections. Currently GCC does not implement
10105 @node IA-64 Options
10106 @subsection IA-64 Options
10107 @cindex IA-64 Options
10109 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10113 @opindex mbig-endian
10114 Generate code for a big endian target. This is the default for HP-UX@.
10116 @item -mlittle-endian
10117 @opindex mlittle-endian
10118 Generate code for a little endian target. This is the default for AIX5
10124 @opindex mno-gnu-as
10125 Generate (or don't) code for the GNU assembler. This is the default.
10126 @c Also, this is the default if the configure option @option{--with-gnu-as}
10132 @opindex mno-gnu-ld
10133 Generate (or don't) code for the GNU linker. This is the default.
10134 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10139 Generate code that does not use a global pointer register. The result
10140 is not position independent code, and violates the IA-64 ABI@.
10142 @item -mvolatile-asm-stop
10143 @itemx -mno-volatile-asm-stop
10144 @opindex mvolatile-asm-stop
10145 @opindex mno-volatile-asm-stop
10146 Generate (or don't) a stop bit immediately before and after volatile asm
10149 @item -mregister-names
10150 @itemx -mno-register-names
10151 @opindex mregister-names
10152 @opindex mno-register-names
10153 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10154 the stacked registers. This may make assembler output more readable.
10160 Disable (or enable) optimizations that use the small data section. This may
10161 be useful for working around optimizer bugs.
10163 @item -mconstant-gp
10164 @opindex mconstant-gp
10165 Generate code that uses a single constant global pointer value. This is
10166 useful when compiling kernel code.
10170 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10171 This is useful when compiling firmware code.
10173 @item -minline-float-divide-min-latency
10174 @opindex minline-float-divide-min-latency
10175 Generate code for inline divides of floating point values
10176 using the minimum latency algorithm.
10178 @item -minline-float-divide-max-throughput
10179 @opindex minline-float-divide-max-throughput
10180 Generate code for inline divides of floating point values
10181 using the maximum throughput algorithm.
10183 @item -minline-int-divide-min-latency
10184 @opindex minline-int-divide-min-latency
10185 Generate code for inline divides of integer values
10186 using the minimum latency algorithm.
10188 @item -minline-int-divide-max-throughput
10189 @opindex minline-int-divide-max-throughput
10190 Generate code for inline divides of integer values
10191 using the maximum throughput algorithm.
10193 @item -minline-sqrt-min-latency
10194 @opindex minline-sqrt-min-latency
10195 Generate code for inline square roots
10196 using the minimum latency algorithm.
10198 @item -minline-sqrt-max-throughput
10199 @opindex minline-sqrt-max-throughput
10200 Generate code for inline square roots
10201 using the maximum throughput algorithm.
10203 @item -mno-dwarf2-asm
10204 @itemx -mdwarf2-asm
10205 @opindex mno-dwarf2-asm
10206 @opindex mdwarf2-asm
10207 Don't (or do) generate assembler code for the DWARF2 line number debugging
10208 info. This may be useful when not using the GNU assembler.
10210 @item -mearly-stop-bits
10211 @itemx -mno-early-stop-bits
10212 @opindex mearly-stop-bits
10213 @opindex mno-early-stop-bits
10214 Allow stop bits to be placed earlier than immediately preceding the
10215 instruction that triggered the stop bit. This can improve instruction
10216 scheduling, but does not always do so.
10218 @item -mfixed-range=@var{register-range}
10219 @opindex mfixed-range
10220 Generate code treating the given register range as fixed registers.
10221 A fixed register is one that the register allocator can not use. This is
10222 useful when compiling kernel code. A register range is specified as
10223 two registers separated by a dash. Multiple register ranges can be
10224 specified separated by a comma.
10226 @item -mtls-size=@var{tls-size}
10228 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
10231 @item -mtune=@var{cpu-type}
10233 Tune the instruction scheduling for a particular CPU, Valid values are
10234 itanium, itanium1, merced, itanium2, and mckinley.
10240 Add support for multithreading using the POSIX threads library. This
10241 option sets flags for both the preprocessor and linker. It does
10242 not affect the thread safety of object code produced by the compiler or
10243 that of libraries supplied with it. These are HP-UX specific flags.
10249 Generate code for a 32-bit or 64-bit environment.
10250 The 32-bit environment sets int, long and pointer to 32 bits.
10251 The 64-bit environment sets int to 32 bits and long and pointer
10252 to 64 bits. These are HP-UX specific flags.
10254 @item -mno-sched-br-data-spec
10255 @itemx -msched-br-data-spec
10256 @opindex -mno-sched-br-data-spec
10257 @opindex -msched-br-data-spec
10258 (Dis/En)able data speculative scheduling before reload.
10259 This will result in generation of the ld.a instructions and
10260 the corresponding check instructions (ld.c / chk.a).
10261 The default is 'disable'.
10263 @item -msched-ar-data-spec
10264 @itemx -mno-sched-ar-data-spec
10265 @opindex -msched-ar-data-spec
10266 @opindex -mno-sched-ar-data-spec
10267 (En/Dis)able data speculative scheduling after reload.
10268 This will result in generation of the ld.a instructions and
10269 the corresponding check instructions (ld.c / chk.a).
10270 The default is 'enable'.
10272 @item -mno-sched-control-spec
10273 @itemx -msched-control-spec
10274 @opindex -mno-sched-control-spec
10275 @opindex -msched-control-spec
10276 (Dis/En)able control speculative scheduling. This feature is
10277 available only during region scheduling (i.e. before reload).
10278 This will result in generation of the ld.s instructions and
10279 the corresponding check instructions chk.s .
10280 The default is 'disable'.
10282 @item -msched-br-in-data-spec
10283 @itemx -mno-sched-br-in-data-spec
10284 @opindex -msched-br-in-data-spec
10285 @opindex -mno-sched-br-in-data-spec
10286 (En/Dis)able speculative scheduling of the instructions that
10287 are dependent on the data speculative loads before reload.
10288 This is effective only with @option{-msched-br-data-spec} enabled.
10289 The default is 'enable'.
10291 @item -msched-ar-in-data-spec
10292 @itemx -mno-sched-ar-in-data-spec
10293 @opindex -msched-ar-in-data-spec
10294 @opindex -mno-sched-ar-in-data-spec
10295 (En/Dis)able speculative scheduling of the instructions that
10296 are dependent on the data speculative loads after reload.
10297 This is effective only with @option{-msched-ar-data-spec} enabled.
10298 The default is 'enable'.
10300 @item -msched-in-control-spec
10301 @itemx -mno-sched-in-control-spec
10302 @opindex -msched-in-control-spec
10303 @opindex -mno-sched-in-control-spec
10304 (En/Dis)able speculative scheduling of the instructions that
10305 are dependent on the control speculative loads.
10306 This is effective only with @option{-msched-control-spec} enabled.
10307 The default is 'enable'.
10310 @itemx -mno-sched-ldc
10311 @opindex -msched-ldc
10312 @opindex -mno-sched-ldc
10313 (En/Dis)able use of simple data speculation checks ld.c .
10314 If disabled, only chk.a instructions will be emitted to check
10315 data speculative loads.
10316 The default is 'enable'.
10318 @item -mno-sched-control-ldc
10319 @itemx -msched-control-ldc
10320 @opindex -mno-sched-control-ldc
10321 @opindex -msched-control-ldc
10322 (Dis/En)able use of ld.c instructions to check control speculative loads.
10323 If enabled, in case of control speculative load with no speculatively
10324 scheduled dependent instructions this load will be emitted as ld.sa and
10325 ld.c will be used to check it.
10326 The default is 'disable'.
10328 @item -mno-sched-spec-verbose
10329 @itemx -msched-spec-verbose
10330 @opindex -mno-sched-spec-verbose
10331 @opindex -msched-spec-verbose
10332 (Dis/En)able printing of the information about speculative motions.
10334 @item -mno-sched-prefer-non-data-spec-insns
10335 @itemx -msched-prefer-non-data-spec-insns
10336 @opindex -mno-sched-prefer-non-data-spec-insns
10337 @opindex -msched-prefer-non-data-spec-insns
10338 If enabled, data speculative instructions will be chosen for schedule
10339 only if there are no other choices at the moment. This will make
10340 the use of the data speculation much more conservative.
10341 The default is 'disable'.
10343 @item -mno-sched-prefer-non-control-spec-insns
10344 @itemx -msched-prefer-non-control-spec-insns
10345 @opindex -mno-sched-prefer-non-control-spec-insns
10346 @opindex -msched-prefer-non-control-spec-insns
10347 If enabled, control speculative instructions will be chosen for schedule
10348 only if there are no other choices at the moment. This will make
10349 the use of the control speculation much more conservative.
10350 The default is 'disable'.
10352 @item -mno-sched-count-spec-in-critical-path
10353 @itemx -msched-count-spec-in-critical-path
10354 @opindex -mno-sched-count-spec-in-critical-path
10355 @opindex -msched-count-spec-in-critical-path
10356 If enabled, speculative dependencies will be considered during
10357 computation of the instructions priorities. This will make the use of the
10358 speculation a bit more conservative.
10359 The default is 'disable'.
10364 @subsection M32C Options
10365 @cindex M32C options
10368 @item -mcpu=@var{name}
10370 Select the CPU for which code is generated. @var{name} may be one of
10371 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
10372 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
10373 the M32C/80 series.
10377 Specifies that the program will be run on the simulator. This causes
10378 an alternate runtime library to be linked in which supports, for
10379 example, file I/O. You must not use this option when generating
10380 programs that will run on real hardware; you must provide your own
10381 runtime library for whatever I/O functions are needed.
10383 @item -memregs=@var{number}
10385 Specifies the number of memory-based pseudo-registers GCC will use
10386 during code generation. These pseudo-registers will be used like real
10387 registers, so there is a tradeoff between GCC's ability to fit the
10388 code into available registers, and the performance penalty of using
10389 memory instead of registers. Note that all modules in a program must
10390 be compiled with the same value for this option. Because of that, you
10391 must not use this option with the default runtime libraries gcc
10396 @node M32R/D Options
10397 @subsection M32R/D Options
10398 @cindex M32R/D options
10400 These @option{-m} options are defined for Renesas M32R/D architectures:
10405 Generate code for the M32R/2@.
10409 Generate code for the M32R/X@.
10413 Generate code for the M32R@. This is the default.
10415 @item -mmodel=small
10416 @opindex mmodel=small
10417 Assume all objects live in the lower 16MB of memory (so that their addresses
10418 can be loaded with the @code{ld24} instruction), and assume all subroutines
10419 are reachable with the @code{bl} instruction.
10420 This is the default.
10422 The addressability of a particular object can be set with the
10423 @code{model} attribute.
10425 @item -mmodel=medium
10426 @opindex mmodel=medium
10427 Assume objects may be anywhere in the 32-bit address space (the compiler
10428 will generate @code{seth/add3} instructions to load their addresses), and
10429 assume all subroutines are reachable with the @code{bl} instruction.
10431 @item -mmodel=large
10432 @opindex mmodel=large
10433 Assume objects may be anywhere in the 32-bit address space (the compiler
10434 will generate @code{seth/add3} instructions to load their addresses), and
10435 assume subroutines may not be reachable with the @code{bl} instruction
10436 (the compiler will generate the much slower @code{seth/add3/jl}
10437 instruction sequence).
10440 @opindex msdata=none
10441 Disable use of the small data area. Variables will be put into
10442 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10443 @code{section} attribute has been specified).
10444 This is the default.
10446 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10447 Objects may be explicitly put in the small data area with the
10448 @code{section} attribute using one of these sections.
10450 @item -msdata=sdata
10451 @opindex msdata=sdata
10452 Put small global and static data in the small data area, but do not
10453 generate special code to reference them.
10456 @opindex msdata=use
10457 Put small global and static data in the small data area, and generate
10458 special instructions to reference them.
10462 @cindex smaller data references
10463 Put global and static objects less than or equal to @var{num} bytes
10464 into the small data or bss sections instead of the normal data or bss
10465 sections. The default value of @var{num} is 8.
10466 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10467 for this option to have any effect.
10469 All modules should be compiled with the same @option{-G @var{num}} value.
10470 Compiling with different values of @var{num} may or may not work; if it
10471 doesn't the linker will give an error message---incorrect code will not be
10476 Makes the M32R specific code in the compiler display some statistics
10477 that might help in debugging programs.
10479 @item -malign-loops
10480 @opindex malign-loops
10481 Align all loops to a 32-byte boundary.
10483 @item -mno-align-loops
10484 @opindex mno-align-loops
10485 Do not enforce a 32-byte alignment for loops. This is the default.
10487 @item -missue-rate=@var{number}
10488 @opindex missue-rate=@var{number}
10489 Issue @var{number} instructions per cycle. @var{number} can only be 1
10492 @item -mbranch-cost=@var{number}
10493 @opindex mbranch-cost=@var{number}
10494 @var{number} can only be 1 or 2. If it is 1 then branches will be
10495 preferred over conditional code, if it is 2, then the opposite will
10498 @item -mflush-trap=@var{number}
10499 @opindex mflush-trap=@var{number}
10500 Specifies the trap number to use to flush the cache. The default is
10501 12. Valid numbers are between 0 and 15 inclusive.
10503 @item -mno-flush-trap
10504 @opindex mno-flush-trap
10505 Specifies that the cache cannot be flushed by using a trap.
10507 @item -mflush-func=@var{name}
10508 @opindex mflush-func=@var{name}
10509 Specifies the name of the operating system function to call to flush
10510 the cache. The default is @emph{_flush_cache}, but a function call
10511 will only be used if a trap is not available.
10513 @item -mno-flush-func
10514 @opindex mno-flush-func
10515 Indicates that there is no OS function for flushing the cache.
10519 @node M680x0 Options
10520 @subsection M680x0 Options
10521 @cindex M680x0 options
10523 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
10524 The default settings depend on which architecture was selected when
10525 the compiler was configured; the defaults for the most common choices
10529 @item -march=@var{arch}
10531 Generate code for a specific M680x0 or ColdFire instruction set
10532 architecture. Permissible values of @var{arch} for M680x0
10533 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
10534 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
10535 architectures are selected according to Freescale's ISA classification
10536 and the permissible values are: @samp{isaa}, @samp{isaaplus},
10537 @samp{isab} and @samp{isac}.
10539 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
10540 code for a ColdFire target. The @var{arch} in this macro is one of the
10541 @option{-march} arguments given above.
10543 When used together, @option{-march} and @option{-mtune} select code
10544 that runs on a family of similar processors but that is optimized
10545 for a particular microarchitecture.
10547 @item -mcpu=@var{cpu}
10549 Generate code for a specific M680x0 or ColdFire processor.
10550 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
10551 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
10552 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
10553 below, which also classifies the CPUs into families:
10555 @multitable @columnfractions 0.20 0.80
10556 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
10557 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
10558 @item @samp{5206e} @tab @samp{5206e}
10559 @item @samp{5208} @tab @samp{5207} @samp{5208}
10560 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
10561 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
10562 @item @samp{5216} @tab @samp{5214} @samp{5216}
10563 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
10564 @item @samp{5225} @tab @samp{5224} @samp{5225}
10565 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
10566 @item @samp{5249} @tab @samp{5249}
10567 @item @samp{5250} @tab @samp{5250}
10568 @item @samp{5271} @tab @samp{5270} @samp{5271}
10569 @item @samp{5272} @tab @samp{5272}
10570 @item @samp{5275} @tab @samp{5274} @samp{5275}
10571 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
10572 @item @samp{5307} @tab @samp{5307}
10573 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
10574 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
10575 @item @samp{5407} @tab @samp{5407}
10576 @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}
10579 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
10580 @var{arch} is compatible with @var{cpu}. Other combinations of
10581 @option{-mcpu} and @option{-march} are rejected.
10583 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
10584 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
10585 where the value of @var{family} is given by the table above.
10587 @item -mtune=@var{tune}
10589 Tune the code for a particular microarchitecture, within the
10590 constraints set by @option{-march} and @option{-mcpu}.
10591 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
10592 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
10593 and @samp{cpu32}. The ColdFire microarchitectures
10594 are: @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
10596 You can also use @option{-mtune=68020-40} for code that needs
10597 to run relatively well on 68020, 68030 and 68040 targets.
10598 @option{-mtune=68020-60} is similar but includes 68060 targets
10599 as well. These two options select the same tuning decisions as
10600 @option{-m68020-40} and @option{-m68020-60} respectively.
10602 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
10603 when tuning for 680x0 architecture @var{arch}. It also defines
10604 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
10605 option is used. If gcc is tuning for a range of architectures,
10606 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
10607 it defines the macros for every architecture in the range.
10609 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
10610 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
10611 of the arguments given above.
10617 Generate output for a 68000. This is the default
10618 when the compiler is configured for 68000-based systems.
10619 It is equivalent to @option{-march=68000}.
10621 Use this option for microcontrollers with a 68000 or EC000 core,
10622 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10626 Generate output for a 68010. This is the default
10627 when the compiler is configured for 68010-based systems.
10628 It is equivalent to @option{-march=68010}.
10634 Generate output for a 68020. This is the default
10635 when the compiler is configured for 68020-based systems.
10636 It is equivalent to @option{-march=68020}.
10640 Generate output for a 68030. This is the default when the compiler is
10641 configured for 68030-based systems. It is equivalent to
10642 @option{-march=68030}.
10646 Generate output for a 68040. This is the default when the compiler is
10647 configured for 68040-based systems. It is equivalent to
10648 @option{-march=68040}.
10650 This option inhibits the use of 68881/68882 instructions that have to be
10651 emulated by software on the 68040. Use this option if your 68040 does not
10652 have code to emulate those instructions.
10656 Generate output for a 68060. This is the default when the compiler is
10657 configured for 68060-based systems. It is equivalent to
10658 @option{-march=68060}.
10660 This option inhibits the use of 68020 and 68881/68882 instructions that
10661 have to be emulated by software on the 68060. Use this option if your 68060
10662 does not have code to emulate those instructions.
10666 Generate output for a CPU32. This is the default
10667 when the compiler is configured for CPU32-based systems.
10668 It is equivalent to @option{-march=cpu32}.
10670 Use this option for microcontrollers with a
10671 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10672 68336, 68340, 68341, 68349 and 68360.
10676 Generate output for a 520X ColdFire CPU. This is the default
10677 when the compiler is configured for 520X-based systems.
10678 It is equivalent to @option{-mcpu=5206}, and is now deprecated
10679 in favor of that option.
10681 Use this option for microcontroller with a 5200 core, including
10682 the MCF5202, MCF5203, MCF5204 and MCF5206.
10686 Generate output for a 5206e ColdFire CPU. The option is now
10687 deprecated in favor of the equivalent @option{-mcpu=5206e}.
10691 Generate output for a member of the ColdFire 528X family.
10692 The option is now deprecated in favor of the equivalent
10693 @option{-mcpu=528x}.
10697 Generate output for a ColdFire 5307 CPU. The option is now deprecated
10698 in favor of the equivalent @option{-mcpu=5307}.
10702 Generate output for a ColdFire 5407 CPU. The option is now deprecated
10703 in favor of the equivalent @option{-mcpu=5407}.
10707 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
10708 This includes use of hardware floating point instructions.
10709 The option is equivalent to @option{-mcpu=547x}, and is now
10710 deprecated in favor of that option.
10714 Generate output for a 68040, without using any of the new instructions.
10715 This results in code which can run relatively efficiently on either a
10716 68020/68881 or a 68030 or a 68040. The generated code does use the
10717 68881 instructions that are emulated on the 68040.
10719 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
10723 Generate output for a 68060, without using any of the new instructions.
10724 This results in code which can run relatively efficiently on either a
10725 68020/68881 or a 68030 or a 68040. The generated code does use the
10726 68881 instructions that are emulated on the 68060.
10728 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
10732 @opindex mhard-float
10734 Generate floating-point instructions. This is the default for 68020
10735 and above, and for ColdFire devices that have an FPU. It defines the
10736 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
10737 on ColdFire targets.
10740 @opindex msoft-float
10741 Do not generate floating-point instructions; use library calls instead.
10742 This is the default for 68000, 68010, and 68832 targets. It is also
10743 the default for ColdFire devices that have no FPU.
10749 Generate (do not generate) ColdFire hardware divide and remainder
10750 instructions. If @option{-march} is used without @option{-mcpu},
10751 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
10752 architectures. Otherwise, the default is taken from the target CPU
10753 (either the default CPU, or the one specified by @option{-mcpu}). For
10754 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
10755 @option{-mcpu=5206e}.
10757 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
10761 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10762 Additionally, parameters passed on the stack are also aligned to a
10763 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10766 @opindex -mno-short
10767 Do not consider type @code{int} to be 16 bits wide. This is the default.
10770 @itemx -mno-bitfield
10771 @opindex mnobitfield
10772 @opindex mno-bitfield
10773 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10774 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10778 Do use the bit-field instructions. The @option{-m68020} option implies
10779 @option{-mbitfield}. This is the default if you use a configuration
10780 designed for a 68020.
10784 Use a different function-calling convention, in which functions
10785 that take a fixed number of arguments return with the @code{rtd}
10786 instruction, which pops their arguments while returning. This
10787 saves one instruction in the caller since there is no need to pop
10788 the arguments there.
10790 This calling convention is incompatible with the one normally
10791 used on Unix, so you cannot use it if you need to call libraries
10792 compiled with the Unix compiler.
10794 Also, you must provide function prototypes for all functions that
10795 take variable numbers of arguments (including @code{printf});
10796 otherwise incorrect code will be generated for calls to those
10799 In addition, seriously incorrect code will result if you call a
10800 function with too many arguments. (Normally, extra arguments are
10801 harmlessly ignored.)
10803 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10804 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10808 Do not use the calling conventions selected by @option{-mrtd}.
10809 This is the default.
10812 @itemx -mno-align-int
10813 @opindex malign-int
10814 @opindex mno-align-int
10815 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10816 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10817 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10818 Aligning variables on 32-bit boundaries produces code that runs somewhat
10819 faster on processors with 32-bit busses at the expense of more memory.
10821 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10822 align structures containing the above types differently than
10823 most published application binary interface specifications for the m68k.
10827 Use the pc-relative addressing mode of the 68000 directly, instead of
10828 using a global offset table. At present, this option implies @option{-fpic},
10829 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10830 not presently supported with @option{-mpcrel}, though this could be supported for
10831 68020 and higher processors.
10833 @item -mno-strict-align
10834 @itemx -mstrict-align
10835 @opindex mno-strict-align
10836 @opindex mstrict-align
10837 Do not (do) assume that unaligned memory references will be handled by
10841 Generate code that allows the data segment to be located in a different
10842 area of memory from the text segment. This allows for execute in place in
10843 an environment without virtual memory management. This option implies
10846 @item -mno-sep-data
10847 Generate code that assumes that the data segment follows the text segment.
10848 This is the default.
10850 @item -mid-shared-library
10851 Generate code that supports shared libraries via the library ID method.
10852 This allows for execute in place and shared libraries in an environment
10853 without virtual memory management. This option implies @option{-fPIC}.
10855 @item -mno-id-shared-library
10856 Generate code that doesn't assume ID based shared libraries are being used.
10857 This is the default.
10859 @item -mshared-library-id=n
10860 Specified the identification number of the ID based shared library being
10861 compiled. Specifying a value of 0 will generate more compact code, specifying
10862 other values will force the allocation of that number to the current
10863 library but is no more space or time efficient than omitting this option.
10867 @node M68hc1x Options
10868 @subsection M68hc1x Options
10869 @cindex M68hc1x options
10871 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10872 microcontrollers. The default values for these options depends on
10873 which style of microcontroller was selected when the compiler was configured;
10874 the defaults for the most common choices are given below.
10881 Generate output for a 68HC11. This is the default
10882 when the compiler is configured for 68HC11-based systems.
10888 Generate output for a 68HC12. This is the default
10889 when the compiler is configured for 68HC12-based systems.
10895 Generate output for a 68HCS12.
10897 @item -mauto-incdec
10898 @opindex mauto-incdec
10899 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10906 Enable the use of 68HC12 min and max instructions.
10909 @itemx -mno-long-calls
10910 @opindex mlong-calls
10911 @opindex mno-long-calls
10912 Treat all calls as being far away (near). If calls are assumed to be
10913 far away, the compiler will use the @code{call} instruction to
10914 call a function and the @code{rtc} instruction for returning.
10918 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10920 @item -msoft-reg-count=@var{count}
10921 @opindex msoft-reg-count
10922 Specify the number of pseudo-soft registers which are used for the
10923 code generation. The maximum number is 32. Using more pseudo-soft
10924 register may or may not result in better code depending on the program.
10925 The default is 4 for 68HC11 and 2 for 68HC12.
10929 @node MCore Options
10930 @subsection MCore Options
10931 @cindex MCore options
10933 These are the @samp{-m} options defined for the Motorola M*Core
10939 @itemx -mno-hardlit
10941 @opindex mno-hardlit
10942 Inline constants into the code stream if it can be done in two
10943 instructions or less.
10949 Use the divide instruction. (Enabled by default).
10951 @item -mrelax-immediate
10952 @itemx -mno-relax-immediate
10953 @opindex mrelax-immediate
10954 @opindex mno-relax-immediate
10955 Allow arbitrary sized immediates in bit operations.
10957 @item -mwide-bitfields
10958 @itemx -mno-wide-bitfields
10959 @opindex mwide-bitfields
10960 @opindex mno-wide-bitfields
10961 Always treat bit-fields as int-sized.
10963 @item -m4byte-functions
10964 @itemx -mno-4byte-functions
10965 @opindex m4byte-functions
10966 @opindex mno-4byte-functions
10967 Force all functions to be aligned to a four byte boundary.
10969 @item -mcallgraph-data
10970 @itemx -mno-callgraph-data
10971 @opindex mcallgraph-data
10972 @opindex mno-callgraph-data
10973 Emit callgraph information.
10976 @itemx -mno-slow-bytes
10977 @opindex mslow-bytes
10978 @opindex mno-slow-bytes
10979 Prefer word access when reading byte quantities.
10981 @item -mlittle-endian
10982 @itemx -mbig-endian
10983 @opindex mlittle-endian
10984 @opindex mbig-endian
10985 Generate code for a little endian target.
10991 Generate code for the 210 processor.
10995 @subsection MIPS Options
10996 @cindex MIPS options
11002 Generate big-endian code.
11006 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11009 @item -march=@var{arch}
11011 Generate code that will run on @var{arch}, which can be the name of a
11012 generic MIPS ISA, or the name of a particular processor.
11014 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11015 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11016 The processor names are:
11017 @samp{4kc}, @samp{4km}, @samp{4kp},
11018 @samp{4kec}, @samp{4kem}, @samp{4kep},
11019 @samp{5kc}, @samp{5kf},
11021 @samp{24kc}, @samp{24kf}, @samp{24kx},
11022 @samp{24kec}, @samp{24kef}, @samp{24kex},
11023 @samp{34kc}, @samp{34kf}, @samp{34kx},
11026 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11027 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11028 @samp{rm7000}, @samp{rm9000},
11031 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11032 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11033 The special value @samp{from-abi} selects the
11034 most compatible architecture for the selected ABI (that is,
11035 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11037 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11038 (for example, @samp{-march=r2k}). Prefixes are optional, and
11039 @samp{vr} may be written @samp{r}.
11041 GCC defines two macros based on the value of this option. The first
11042 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11043 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11044 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11045 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11046 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11048 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11049 above. In other words, it will have the full prefix and will not
11050 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11051 the macro names the resolved architecture (either @samp{"mips1"} or
11052 @samp{"mips3"}). It names the default architecture when no
11053 @option{-march} option is given.
11055 @item -mtune=@var{arch}
11057 Optimize for @var{arch}. Among other things, this option controls
11058 the way instructions are scheduled, and the perceived cost of arithmetic
11059 operations. The list of @var{arch} values is the same as for
11062 When this option is not used, GCC will optimize for the processor
11063 specified by @option{-march}. By using @option{-march} and
11064 @option{-mtune} together, it is possible to generate code that will
11065 run on a family of processors, but optimize the code for one
11066 particular member of that family.
11068 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11069 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11070 @samp{-march} ones described above.
11074 Equivalent to @samp{-march=mips1}.
11078 Equivalent to @samp{-march=mips2}.
11082 Equivalent to @samp{-march=mips3}.
11086 Equivalent to @samp{-march=mips4}.
11090 Equivalent to @samp{-march=mips32}.
11094 Equivalent to @samp{-march=mips32r2}.
11098 Equivalent to @samp{-march=mips64}.
11103 @opindex mno-mips16
11104 Generate (do not generate) MIPS16 code. If GCC is targetting a
11105 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11117 Generate code for the given ABI@.
11119 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11120 generates 64-bit code when you select a 64-bit architecture, but you
11121 can use @option{-mgp32} to get 32-bit code instead.
11123 For information about the O64 ABI, see
11124 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11126 GCC supports a variant of the o32 ABI in which floating-point registers
11127 are 64 rather than 32 bits wide. You can select this combination with
11128 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11129 and @samp{mfhc1} instructions and is therefore only supported for
11130 MIPS32R2 processors.
11132 The register assignments for arguments and return values remain the
11133 same, but each scalar value is passed in a single 64-bit register
11134 rather than a pair of 32-bit registers. For example, scalar
11135 floating-point values are returned in @samp{$f0} only, not a
11136 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
11137 remains the same, but all 64 bits are saved.
11140 @itemx -mno-abicalls
11142 @opindex mno-abicalls
11143 Generate (do not generate) code that is suitable for SVR4-style
11144 dynamic objects. @option{-mabicalls} is the default for SVR4-based
11149 Generate (do not generate) code that is fully position-independent,
11150 and that can therefore be linked into shared libraries. This option
11151 only affects @option{-mabicalls}.
11153 All @option{-mabicalls} code has traditionally been position-independent,
11154 regardless of options like @option{-fPIC} and @option{-fpic}. However,
11155 as an extension, the GNU toolchain allows executables to use absolute
11156 accesses for locally-binding symbols. It can also use shorter GP
11157 initialization sequences and generate direct calls to locally-defined
11158 functions. This mode is selected by @option{-mno-shared}.
11160 @option{-mno-shared} depends on binutils 2.16 or higher and generates
11161 objects that can only be linked by the GNU linker. However, the option
11162 does not affect the ABI of the final executable; it only affects the ABI
11163 of relocatable objects. Using @option{-mno-shared} will generally make
11164 executables both smaller and quicker.
11166 @option{-mshared} is the default.
11172 Lift (do not lift) the usual restrictions on the size of the global
11175 GCC normally uses a single instruction to load values from the GOT@.
11176 While this is relatively efficient, it will only work if the GOT
11177 is smaller than about 64k. Anything larger will cause the linker
11178 to report an error such as:
11180 @cindex relocation truncated to fit (MIPS)
11182 relocation truncated to fit: R_MIPS_GOT16 foobar
11185 If this happens, you should recompile your code with @option{-mxgot}.
11186 It should then work with very large GOTs, although it will also be
11187 less efficient, since it will take three instructions to fetch the
11188 value of a global symbol.
11190 Note that some linkers can create multiple GOTs. If you have such a
11191 linker, you should only need to use @option{-mxgot} when a single object
11192 file accesses more than 64k's worth of GOT entries. Very few do.
11194 These options have no effect unless GCC is generating position
11199 Assume that general-purpose registers are 32 bits wide.
11203 Assume that general-purpose registers are 64 bits wide.
11207 Assume that floating-point registers are 32 bits wide.
11211 Assume that floating-point registers are 64 bits wide.
11214 @opindex mhard-float
11215 Use floating-point coprocessor instructions.
11218 @opindex msoft-float
11219 Do not use floating-point coprocessor instructions. Implement
11220 floating-point calculations using library calls instead.
11222 @item -msingle-float
11223 @opindex msingle-float
11224 Assume that the floating-point coprocessor only supports single-precision
11227 @itemx -mdouble-float
11228 @opindex mdouble-float
11229 Assume that the floating-point coprocessor supports double-precision
11230 operations. This is the default.
11236 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
11238 @itemx -mpaired-single
11239 @itemx -mno-paired-single
11240 @opindex mpaired-single
11241 @opindex mno-paired-single
11242 Use (do not use) paired-single floating-point instructions.
11243 @xref{MIPS Paired-Single Support}. This option can only be used
11244 when generating 64-bit code and requires hardware floating-point
11245 support to be enabled.
11250 @opindex mno-mips3d
11251 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
11252 The option @option{-mips3d} implies @option{-mpaired-single}.
11256 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
11257 an explanation of the default and the way that the pointer size is
11262 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
11264 The default size of @code{int}s, @code{long}s and pointers depends on
11265 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
11266 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
11267 32-bit @code{long}s. Pointers are the same size as @code{long}s,
11268 or the same size as integer registers, whichever is smaller.
11274 Assume (do not assume) that all symbols have 32-bit values, regardless
11275 of the selected ABI@. This option is useful in combination with
11276 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
11277 to generate shorter and faster references to symbolic addresses.
11281 @cindex smaller data references (MIPS)
11282 @cindex gp-relative references (MIPS)
11283 Put global and static items less than or equal to @var{num} bytes into
11284 the small data or bss section instead of the normal data or bss section.
11285 This allows the data to be accessed using a single instruction.
11287 All modules should be compiled with the same @option{-G @var{num}}
11290 @item -membedded-data
11291 @itemx -mno-embedded-data
11292 @opindex membedded-data
11293 @opindex mno-embedded-data
11294 Allocate variables to the read-only data section first if possible, then
11295 next in the small data section if possible, otherwise in data. This gives
11296 slightly slower code than the default, but reduces the amount of RAM required
11297 when executing, and thus may be preferred for some embedded systems.
11299 @item -muninit-const-in-rodata
11300 @itemx -mno-uninit-const-in-rodata
11301 @opindex muninit-const-in-rodata
11302 @opindex mno-uninit-const-in-rodata
11303 Put uninitialized @code{const} variables in the read-only data section.
11304 This option is only meaningful in conjunction with @option{-membedded-data}.
11306 @item -msplit-addresses
11307 @itemx -mno-split-addresses
11308 @opindex msplit-addresses
11309 @opindex mno-split-addresses
11310 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
11311 relocation operators. This option has been superseded by
11312 @option{-mexplicit-relocs} but is retained for backwards compatibility.
11314 @item -mexplicit-relocs
11315 @itemx -mno-explicit-relocs
11316 @opindex mexplicit-relocs
11317 @opindex mno-explicit-relocs
11318 Use (do not use) assembler relocation operators when dealing with symbolic
11319 addresses. The alternative, selected by @option{-mno-explicit-relocs},
11320 is to use assembler macros instead.
11322 @option{-mexplicit-relocs} is the default if GCC was configured
11323 to use an assembler that supports relocation operators.
11325 @item -mcheck-zero-division
11326 @itemx -mno-check-zero-division
11327 @opindex mcheck-zero-division
11328 @opindex mno-check-zero-division
11329 Trap (do not trap) on integer division by zero.
11331 The default is @option{-mcheck-zero-division}.
11333 @item -mdivide-traps
11334 @itemx -mdivide-breaks
11335 @opindex mdivide-traps
11336 @opindex mdivide-breaks
11337 MIPS systems check for division by zero by generating either a
11338 conditional trap or a break instruction. Using traps results in
11339 smaller code, but is only supported on MIPS II and later. Also, some
11340 versions of the Linux kernel have a bug that prevents trap from
11341 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
11342 allow conditional traps on architectures that support them and
11343 @option{-mdivide-breaks} to force the use of breaks.
11345 The default is usually @option{-mdivide-traps}, but this can be
11346 overridden at configure time using @option{--with-divide=breaks}.
11347 Divide-by-zero checks can be completely disabled using
11348 @option{-mno-check-zero-division}.
11353 @opindex mno-memcpy
11354 Force (do not force) the use of @code{memcpy()} for non-trivial block
11355 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
11356 most constant-sized copies.
11359 @itemx -mno-long-calls
11360 @opindex mlong-calls
11361 @opindex mno-long-calls
11362 Disable (do not disable) use of the @code{jal} instruction. Calling
11363 functions using @code{jal} is more efficient but requires the caller
11364 and callee to be in the same 256 megabyte segment.
11366 This option has no effect on abicalls code. The default is
11367 @option{-mno-long-calls}.
11373 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
11374 instructions, as provided by the R4650 ISA@.
11377 @itemx -mno-fused-madd
11378 @opindex mfused-madd
11379 @opindex mno-fused-madd
11380 Enable (disable) use of the floating point multiply-accumulate
11381 instructions, when they are available. The default is
11382 @option{-mfused-madd}.
11384 When multiply-accumulate instructions are used, the intermediate
11385 product is calculated to infinite precision and is not subject to
11386 the FCSR Flush to Zero bit. This may be undesirable in some
11391 Tell the MIPS assembler to not run its preprocessor over user
11392 assembler files (with a @samp{.s} suffix) when assembling them.
11395 @itemx -mno-fix-r4000
11396 @opindex mfix-r4000
11397 @opindex mno-fix-r4000
11398 Work around certain R4000 CPU errata:
11401 A double-word or a variable shift may give an incorrect result if executed
11402 immediately after starting an integer division.
11404 A double-word or a variable shift may give an incorrect result if executed
11405 while an integer multiplication is in progress.
11407 An integer division may give an incorrect result if started in a delay slot
11408 of a taken branch or a jump.
11412 @itemx -mno-fix-r4400
11413 @opindex mfix-r4400
11414 @opindex mno-fix-r4400
11415 Work around certain R4400 CPU errata:
11418 A double-word or a variable shift may give an incorrect result if executed
11419 immediately after starting an integer division.
11423 @itemx -mno-fix-vr4120
11424 @opindex mfix-vr4120
11425 Work around certain VR4120 errata:
11428 @code{dmultu} does not always produce the correct result.
11430 @code{div} and @code{ddiv} do not always produce the correct result if one
11431 of the operands is negative.
11433 The workarounds for the division errata rely on special functions in
11434 @file{libgcc.a}. At present, these functions are only provided by
11435 the @code{mips64vr*-elf} configurations.
11437 Other VR4120 errata require a nop to be inserted between certain pairs of
11438 instructions. These errata are handled by the assembler, not by GCC itself.
11441 @opindex mfix-vr4130
11442 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
11443 workarounds are implemented by the assembler rather than by GCC,
11444 although GCC will avoid using @code{mflo} and @code{mfhi} if the
11445 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
11446 instructions are available instead.
11449 @itemx -mno-fix-sb1
11451 Work around certain SB-1 CPU core errata.
11452 (This flag currently works around the SB-1 revision 2
11453 ``F1'' and ``F2'' floating point errata.)
11455 @item -mflush-func=@var{func}
11456 @itemx -mno-flush-func
11457 @opindex mflush-func
11458 Specifies the function to call to flush the I and D caches, or to not
11459 call any such function. If called, the function must take the same
11460 arguments as the common @code{_flush_func()}, that is, the address of the
11461 memory range for which the cache is being flushed, the size of the
11462 memory range, and the number 3 (to flush both caches). The default
11463 depends on the target GCC was configured for, but commonly is either
11464 @samp{_flush_func} or @samp{__cpu_flush}.
11466 @item -mbranch-likely
11467 @itemx -mno-branch-likely
11468 @opindex mbranch-likely
11469 @opindex mno-branch-likely
11470 Enable or disable use of Branch Likely instructions, regardless of the
11471 default for the selected architecture. By default, Branch Likely
11472 instructions may be generated if they are supported by the selected
11473 architecture. An exception is for the MIPS32 and MIPS64 architectures
11474 and processors which implement those architectures; for those, Branch
11475 Likely instructions will not be generated by default because the MIPS32
11476 and MIPS64 architectures specifically deprecate their use.
11478 @item -mfp-exceptions
11479 @itemx -mno-fp-exceptions
11480 @opindex mfp-exceptions
11481 Specifies whether FP exceptions are enabled. This affects how we schedule
11482 FP instructions for some processors. The default is that FP exceptions are
11485 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
11486 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
11489 @item -mvr4130-align
11490 @itemx -mno-vr4130-align
11491 @opindex mvr4130-align
11492 The VR4130 pipeline is two-way superscalar, but can only issue two
11493 instructions together if the first one is 8-byte aligned. When this
11494 option is enabled, GCC will align pairs of instructions that it
11495 thinks should execute in parallel.
11497 This option only has an effect when optimizing for the VR4130.
11498 It normally makes code faster, but at the expense of making it bigger.
11499 It is enabled by default at optimization level @option{-O3}.
11503 @subsection MMIX Options
11504 @cindex MMIX Options
11506 These options are defined for the MMIX:
11510 @itemx -mno-libfuncs
11512 @opindex mno-libfuncs
11513 Specify that intrinsic library functions are being compiled, passing all
11514 values in registers, no matter the size.
11517 @itemx -mno-epsilon
11519 @opindex mno-epsilon
11520 Generate floating-point comparison instructions that compare with respect
11521 to the @code{rE} epsilon register.
11523 @item -mabi=mmixware
11525 @opindex mabi-mmixware
11527 Generate code that passes function parameters and return values that (in
11528 the called function) are seen as registers @code{$0} and up, as opposed to
11529 the GNU ABI which uses global registers @code{$231} and up.
11531 @item -mzero-extend
11532 @itemx -mno-zero-extend
11533 @opindex mzero-extend
11534 @opindex mno-zero-extend
11535 When reading data from memory in sizes shorter than 64 bits, use (do not
11536 use) zero-extending load instructions by default, rather than
11537 sign-extending ones.
11540 @itemx -mno-knuthdiv
11542 @opindex mno-knuthdiv
11543 Make the result of a division yielding a remainder have the same sign as
11544 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
11545 remainder follows the sign of the dividend. Both methods are
11546 arithmetically valid, the latter being almost exclusively used.
11548 @item -mtoplevel-symbols
11549 @itemx -mno-toplevel-symbols
11550 @opindex mtoplevel-symbols
11551 @opindex mno-toplevel-symbols
11552 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
11553 code can be used with the @code{PREFIX} assembly directive.
11557 Generate an executable in the ELF format, rather than the default
11558 @samp{mmo} format used by the @command{mmix} simulator.
11560 @item -mbranch-predict
11561 @itemx -mno-branch-predict
11562 @opindex mbranch-predict
11563 @opindex mno-branch-predict
11564 Use (do not use) the probable-branch instructions, when static branch
11565 prediction indicates a probable branch.
11567 @item -mbase-addresses
11568 @itemx -mno-base-addresses
11569 @opindex mbase-addresses
11570 @opindex mno-base-addresses
11571 Generate (do not generate) code that uses @emph{base addresses}. Using a
11572 base address automatically generates a request (handled by the assembler
11573 and the linker) for a constant to be set up in a global register. The
11574 register is used for one or more base address requests within the range 0
11575 to 255 from the value held in the register. The generally leads to short
11576 and fast code, but the number of different data items that can be
11577 addressed is limited. This means that a program that uses lots of static
11578 data may require @option{-mno-base-addresses}.
11580 @item -msingle-exit
11581 @itemx -mno-single-exit
11582 @opindex msingle-exit
11583 @opindex mno-single-exit
11584 Force (do not force) generated code to have a single exit point in each
11588 @node MN10300 Options
11589 @subsection MN10300 Options
11590 @cindex MN10300 options
11592 These @option{-m} options are defined for Matsushita MN10300 architectures:
11597 Generate code to avoid bugs in the multiply instructions for the MN10300
11598 processors. This is the default.
11600 @item -mno-mult-bug
11601 @opindex mno-mult-bug
11602 Do not generate code to avoid bugs in the multiply instructions for the
11603 MN10300 processors.
11607 Generate code which uses features specific to the AM33 processor.
11611 Do not generate code which uses features specific to the AM33 processor. This
11614 @item -mreturn-pointer-on-d0
11615 @opindex mreturn-pointer-on-d0
11616 When generating a function which returns a pointer, return the pointer
11617 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
11618 only in a0, and attempts to call such functions without a prototype
11619 would result in errors. Note that this option is on by default; use
11620 @option{-mno-return-pointer-on-d0} to disable it.
11624 Do not link in the C run-time initialization object file.
11628 Indicate to the linker that it should perform a relaxation optimization pass
11629 to shorten branches, calls and absolute memory addresses. This option only
11630 has an effect when used on the command line for the final link step.
11632 This option makes symbolic debugging impossible.
11636 @subsection MT Options
11639 These @option{-m} options are defined for Morpho MT architectures:
11643 @item -march=@var{cpu-type}
11645 Generate code that will run on @var{cpu-type}, which is the name of a system
11646 representing a certain processor type. Possible values for
11647 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11648 @samp{ms1-16-003} and @samp{ms2}.
11650 When this option is not used, the default is @option{-march=ms1-16-002}.
11654 Use byte loads and stores when generating code.
11658 Do not use byte loads and stores when generating code.
11662 Use simulator runtime
11666 Do not link in the C run-time initialization object file
11667 @file{crti.o}. Other run-time initialization and termination files
11668 such as @file{startup.o} and @file{exit.o} are still included on the
11669 linker command line.
11673 @node PDP-11 Options
11674 @subsection PDP-11 Options
11675 @cindex PDP-11 Options
11677 These options are defined for the PDP-11:
11682 Use hardware FPP floating point. This is the default. (FIS floating
11683 point on the PDP-11/40 is not supported.)
11686 @opindex msoft-float
11687 Do not use hardware floating point.
11691 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11695 Return floating-point results in memory. This is the default.
11699 Generate code for a PDP-11/40.
11703 Generate code for a PDP-11/45. This is the default.
11707 Generate code for a PDP-11/10.
11709 @item -mbcopy-builtin
11710 @opindex bcopy-builtin
11711 Use inline @code{movmemhi} patterns for copying memory. This is the
11716 Do not use inline @code{movmemhi} patterns for copying memory.
11722 Use 16-bit @code{int}. This is the default.
11728 Use 32-bit @code{int}.
11731 @itemx -mno-float32
11733 @opindex mno-float32
11734 Use 64-bit @code{float}. This is the default.
11737 @itemx -mno-float64
11739 @opindex mno-float64
11740 Use 32-bit @code{float}.
11744 Use @code{abshi2} pattern. This is the default.
11748 Do not use @code{abshi2} pattern.
11750 @item -mbranch-expensive
11751 @opindex mbranch-expensive
11752 Pretend that branches are expensive. This is for experimenting with
11753 code generation only.
11755 @item -mbranch-cheap
11756 @opindex mbranch-cheap
11757 Do not pretend that branches are expensive. This is the default.
11761 Generate code for a system with split I&D@.
11765 Generate code for a system without split I&D@. This is the default.
11769 Use Unix assembler syntax. This is the default when configured for
11770 @samp{pdp11-*-bsd}.
11774 Use DEC assembler syntax. This is the default when configured for any
11775 PDP-11 target other than @samp{pdp11-*-bsd}.
11778 @node PowerPC Options
11779 @subsection PowerPC Options
11780 @cindex PowerPC options
11782 These are listed under @xref{RS/6000 and PowerPC Options}.
11784 @node RS/6000 and PowerPC Options
11785 @subsection IBM RS/6000 and PowerPC Options
11786 @cindex RS/6000 and PowerPC Options
11787 @cindex IBM RS/6000 and PowerPC Options
11789 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11796 @itemx -mno-powerpc
11797 @itemx -mpowerpc-gpopt
11798 @itemx -mno-powerpc-gpopt
11799 @itemx -mpowerpc-gfxopt
11800 @itemx -mno-powerpc-gfxopt
11802 @itemx -mno-powerpc64
11806 @itemx -mno-popcntb
11814 @opindex mno-power2
11816 @opindex mno-powerpc
11817 @opindex mpowerpc-gpopt
11818 @opindex mno-powerpc-gpopt
11819 @opindex mpowerpc-gfxopt
11820 @opindex mno-powerpc-gfxopt
11821 @opindex mpowerpc64
11822 @opindex mno-powerpc64
11826 @opindex mno-popcntb
11830 @opindex mno-mfpgpr
11831 GCC supports two related instruction set architectures for the
11832 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11833 instructions supported by the @samp{rios} chip set used in the original
11834 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11835 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11836 the IBM 4xx, 6xx, and follow-on microprocessors.
11838 Neither architecture is a subset of the other. However there is a
11839 large common subset of instructions supported by both. An MQ
11840 register is included in processors supporting the POWER architecture.
11842 You use these options to specify which instructions are available on the
11843 processor you are using. The default value of these options is
11844 determined when configuring GCC@. Specifying the
11845 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11846 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11847 rather than the options listed above.
11849 The @option{-mpower} option allows GCC to generate instructions that
11850 are found only in the POWER architecture and to use the MQ register.
11851 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11852 to generate instructions that are present in the POWER2 architecture but
11853 not the original POWER architecture.
11855 The @option{-mpowerpc} option allows GCC to generate instructions that
11856 are found only in the 32-bit subset of the PowerPC architecture.
11857 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11858 GCC to use the optional PowerPC architecture instructions in the
11859 General Purpose group, including floating-point square root. Specifying
11860 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11861 use the optional PowerPC architecture instructions in the Graphics
11862 group, including floating-point select.
11864 The @option{-mmfcrf} option allows GCC to generate the move from
11865 condition register field instruction implemented on the POWER4
11866 processor and other processors that support the PowerPC V2.01
11868 The @option{-mpopcntb} option allows GCC to generate the popcount and
11869 double precision FP reciprocal estimate instruction implemented on the
11870 POWER5 processor and other processors that support the PowerPC V2.02
11872 The @option{-mfprnd} option allows GCC to generate the FP round to
11873 integer instructions implemented on the POWER5+ processor and other
11874 processors that support the PowerPC V2.03 architecture.
11875 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
11876 general purpose register instructions implemented on the POWER6X
11877 processor and other processors that support the extended PowerPC V2.05
11880 The @option{-mpowerpc64} option allows GCC to generate the additional
11881 64-bit instructions that are found in the full PowerPC64 architecture
11882 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11883 @option{-mno-powerpc64}.
11885 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11886 will use only the instructions in the common subset of both
11887 architectures plus some special AIX common-mode calls, and will not use
11888 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11889 permits GCC to use any instruction from either architecture and to
11890 allow use of the MQ register; specify this for the Motorola MPC601.
11892 @item -mnew-mnemonics
11893 @itemx -mold-mnemonics
11894 @opindex mnew-mnemonics
11895 @opindex mold-mnemonics
11896 Select which mnemonics to use in the generated assembler code. With
11897 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11898 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11899 assembler mnemonics defined for the POWER architecture. Instructions
11900 defined in only one architecture have only one mnemonic; GCC uses that
11901 mnemonic irrespective of which of these options is specified.
11903 GCC defaults to the mnemonics appropriate for the architecture in
11904 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11905 value of these option. Unless you are building a cross-compiler, you
11906 should normally not specify either @option{-mnew-mnemonics} or
11907 @option{-mold-mnemonics}, but should instead accept the default.
11909 @item -mcpu=@var{cpu_type}
11911 Set architecture type, register usage, choice of mnemonics, and
11912 instruction scheduling parameters for machine type @var{cpu_type}.
11913 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11914 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11915 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11916 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11917 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11918 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11919 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11920 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
11921 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
11922 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11924 @option{-mcpu=common} selects a completely generic processor. Code
11925 generated under this option will run on any POWER or PowerPC processor.
11926 GCC will use only the instructions in the common subset of both
11927 architectures, and will not use the MQ register. GCC assumes a generic
11928 processor model for scheduling purposes.
11930 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11931 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11932 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11933 types, with an appropriate, generic processor model assumed for
11934 scheduling purposes.
11936 The other options specify a specific processor. Code generated under
11937 those options will run best on that processor, and may not run at all on
11940 The @option{-mcpu} options automatically enable or disable the
11943 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
11944 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
11945 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
11947 The particular options set for any particular CPU will vary between
11948 compiler versions, depending on what setting seems to produce optimal
11949 code for that CPU; it doesn't necessarily reflect the actual hardware's
11950 capabilities. If you wish to set an individual option to a particular
11951 value, you may specify it after the @option{-mcpu} option, like
11952 @samp{-mcpu=970 -mno-altivec}.
11954 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11955 not enabled or disabled by the @option{-mcpu} option at present because
11956 AIX does not have full support for these options. You may still
11957 enable or disable them individually if you're sure it'll work in your
11960 @item -mtune=@var{cpu_type}
11962 Set the instruction scheduling parameters for machine type
11963 @var{cpu_type}, but do not set the architecture type, register usage, or
11964 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11965 values for @var{cpu_type} are used for @option{-mtune} as for
11966 @option{-mcpu}. If both are specified, the code generated will use the
11967 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11968 scheduling parameters set by @option{-mtune}.
11974 Generate code to compute division as reciprocal estimate and iterative
11975 refinement, creating opportunities for increased throughput. This
11976 feature requires: optional PowerPC Graphics instruction set for single
11977 precision and FRE instruction for double precision, assuming divides
11978 cannot generate user-visible traps, and the domain values not include
11979 Infinities, denormals or zero denominator.
11982 @itemx -mno-altivec
11984 @opindex mno-altivec
11985 Generate code that uses (does not use) AltiVec instructions, and also
11986 enable the use of built-in functions that allow more direct access to
11987 the AltiVec instruction set. You may also need to set
11988 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11994 @opindex mno-vrsave
11995 Generate VRSAVE instructions when generating AltiVec code.
11998 @opindex msecure-plt
11999 Generate code that allows ld and ld.so to build executables and shared
12000 libraries with non-exec .plt and .got sections. This is a PowerPC
12001 32-bit SYSV ABI option.
12005 Generate code that uses a BSS .plt section that ld.so fills in, and
12006 requires .plt and .got sections that are both writable and executable.
12007 This is a PowerPC 32-bit SYSV ABI option.
12013 This switch enables or disables the generation of ISEL instructions.
12015 @item -misel=@var{yes/no}
12016 This switch has been deprecated. Use @option{-misel} and
12017 @option{-mno-isel} instead.
12023 This switch enables or disables the generation of SPE simd
12026 @item -mspe=@var{yes/no}
12027 This option has been deprecated. Use @option{-mspe} and
12028 @option{-mno-spe} instead.
12030 @item -mfloat-gprs=@var{yes/single/double/no}
12031 @itemx -mfloat-gprs
12032 @opindex mfloat-gprs
12033 This switch enables or disables the generation of floating point
12034 operations on the general purpose registers for architectures that
12037 The argument @var{yes} or @var{single} enables the use of
12038 single-precision floating point operations.
12040 The argument @var{double} enables the use of single and
12041 double-precision floating point operations.
12043 The argument @var{no} disables floating point operations on the
12044 general purpose registers.
12046 This option is currently only available on the MPC854x.
12052 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
12053 targets (including GNU/Linux). The 32-bit environment sets int, long
12054 and pointer to 32 bits and generates code that runs on any PowerPC
12055 variant. The 64-bit environment sets int to 32 bits and long and
12056 pointer to 64 bits, and generates code for PowerPC64, as for
12057 @option{-mpowerpc64}.
12060 @itemx -mno-fp-in-toc
12061 @itemx -mno-sum-in-toc
12062 @itemx -mminimal-toc
12064 @opindex mno-fp-in-toc
12065 @opindex mno-sum-in-toc
12066 @opindex mminimal-toc
12067 Modify generation of the TOC (Table Of Contents), which is created for
12068 every executable file. The @option{-mfull-toc} option is selected by
12069 default. In that case, GCC will allocate at least one TOC entry for
12070 each unique non-automatic variable reference in your program. GCC
12071 will also place floating-point constants in the TOC@. However, only
12072 16,384 entries are available in the TOC@.
12074 If you receive a linker error message that saying you have overflowed
12075 the available TOC space, you can reduce the amount of TOC space used
12076 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
12077 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
12078 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
12079 generate code to calculate the sum of an address and a constant at
12080 run-time instead of putting that sum into the TOC@. You may specify one
12081 or both of these options. Each causes GCC to produce very slightly
12082 slower and larger code at the expense of conserving TOC space.
12084 If you still run out of space in the TOC even when you specify both of
12085 these options, specify @option{-mminimal-toc} instead. This option causes
12086 GCC to make only one TOC entry for every file. When you specify this
12087 option, GCC will produce code that is slower and larger but which
12088 uses extremely little TOC space. You may wish to use this option
12089 only on files that contain less frequently executed code.
12095 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
12096 @code{long} type, and the infrastructure needed to support them.
12097 Specifying @option{-maix64} implies @option{-mpowerpc64} and
12098 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
12099 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
12102 @itemx -mno-xl-compat
12103 @opindex mxl-compat
12104 @opindex mno-xl-compat
12105 Produce code that conforms more closely to IBM XL compiler semantics
12106 when using AIX-compatible ABI. Pass floating-point arguments to
12107 prototyped functions beyond the register save area (RSA) on the stack
12108 in addition to argument FPRs. Do not assume that most significant
12109 double in 128-bit long double value is properly rounded when comparing
12110 values and converting to double. Use XL symbol names for long double
12113 The AIX calling convention was extended but not initially documented to
12114 handle an obscure K&R C case of calling a function that takes the
12115 address of its arguments with fewer arguments than declared. IBM XL
12116 compilers access floating point arguments which do not fit in the
12117 RSA from the stack when a subroutine is compiled without
12118 optimization. Because always storing floating-point arguments on the
12119 stack is inefficient and rarely needed, this option is not enabled by
12120 default and only is necessary when calling subroutines compiled by IBM
12121 XL compilers without optimization.
12125 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
12126 application written to use message passing with special startup code to
12127 enable the application to run. The system must have PE installed in the
12128 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
12129 must be overridden with the @option{-specs=} option to specify the
12130 appropriate directory location. The Parallel Environment does not
12131 support threads, so the @option{-mpe} option and the @option{-pthread}
12132 option are incompatible.
12134 @item -malign-natural
12135 @itemx -malign-power
12136 @opindex malign-natural
12137 @opindex malign-power
12138 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
12139 @option{-malign-natural} overrides the ABI-defined alignment of larger
12140 types, such as floating-point doubles, on their natural size-based boundary.
12141 The option @option{-malign-power} instructs GCC to follow the ABI-specified
12142 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
12144 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
12148 @itemx -mhard-float
12149 @opindex msoft-float
12150 @opindex mhard-float
12151 Generate code that does not use (uses) the floating-point register set.
12152 Software floating point emulation is provided if you use the
12153 @option{-msoft-float} option, and pass the option to GCC when linking.
12156 @itemx -mno-multiple
12158 @opindex mno-multiple
12159 Generate code that uses (does not use) the load multiple word
12160 instructions and the store multiple word instructions. These
12161 instructions are generated by default on POWER systems, and not
12162 generated on PowerPC systems. Do not use @option{-mmultiple} on little
12163 endian PowerPC systems, since those instructions do not work when the
12164 processor is in little endian mode. The exceptions are PPC740 and
12165 PPC750 which permit the instructions usage in little endian mode.
12170 @opindex mno-string
12171 Generate code that uses (does not use) the load string instructions
12172 and the store string word instructions to save multiple registers and
12173 do small block moves. These instructions are generated by default on
12174 POWER systems, and not generated on PowerPC systems. Do not use
12175 @option{-mstring} on little endian PowerPC systems, since those
12176 instructions do not work when the processor is in little endian mode.
12177 The exceptions are PPC740 and PPC750 which permit the instructions
12178 usage in little endian mode.
12183 @opindex mno-update
12184 Generate code that uses (does not use) the load or store instructions
12185 that update the base register to the address of the calculated memory
12186 location. These instructions are generated by default. If you use
12187 @option{-mno-update}, there is a small window between the time that the
12188 stack pointer is updated and the address of the previous frame is
12189 stored, which means code that walks the stack frame across interrupts or
12190 signals may get corrupted data.
12193 @itemx -mno-fused-madd
12194 @opindex mfused-madd
12195 @opindex mno-fused-madd
12196 Generate code that uses (does not use) the floating point multiply and
12197 accumulate instructions. These instructions are generated by default if
12198 hardware floating is used.
12204 Generate code that uses (does not use) the half-word multiply and
12205 multiply-accumulate instructions on the IBM 405 and 440 processors.
12206 These instructions are generated by default when targetting those
12213 Generate code that uses (does not use) the string-search @samp{dlmzb}
12214 instruction on the IBM 405 and 440 processors. This instruction is
12215 generated by default when targetting those processors.
12217 @item -mno-bit-align
12219 @opindex mno-bit-align
12220 @opindex mbit-align
12221 On System V.4 and embedded PowerPC systems do not (do) force structures
12222 and unions that contain bit-fields to be aligned to the base type of the
12225 For example, by default a structure containing nothing but 8
12226 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
12227 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
12228 the structure would be aligned to a 1 byte boundary and be one byte in
12231 @item -mno-strict-align
12232 @itemx -mstrict-align
12233 @opindex mno-strict-align
12234 @opindex mstrict-align
12235 On System V.4 and embedded PowerPC systems do not (do) assume that
12236 unaligned memory references will be handled by the system.
12238 @item -mrelocatable
12239 @itemx -mno-relocatable
12240 @opindex mrelocatable
12241 @opindex mno-relocatable
12242 On embedded PowerPC systems generate code that allows (does not allow)
12243 the program to be relocated to a different address at runtime. If you
12244 use @option{-mrelocatable} on any module, all objects linked together must
12245 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
12247 @item -mrelocatable-lib
12248 @itemx -mno-relocatable-lib
12249 @opindex mrelocatable-lib
12250 @opindex mno-relocatable-lib
12251 On embedded PowerPC systems generate code that allows (does not allow)
12252 the program to be relocated to a different address at runtime. Modules
12253 compiled with @option{-mrelocatable-lib} can be linked with either modules
12254 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
12255 with modules compiled with the @option{-mrelocatable} options.
12261 On System V.4 and embedded PowerPC systems do not (do) assume that
12262 register 2 contains a pointer to a global area pointing to the addresses
12263 used in the program.
12266 @itemx -mlittle-endian
12268 @opindex mlittle-endian
12269 On System V.4 and embedded PowerPC systems compile code for the
12270 processor in little endian mode. The @option{-mlittle-endian} option is
12271 the same as @option{-mlittle}.
12274 @itemx -mbig-endian
12276 @opindex mbig-endian
12277 On System V.4 and embedded PowerPC systems compile code for the
12278 processor in big endian mode. The @option{-mbig-endian} option is
12279 the same as @option{-mbig}.
12281 @item -mdynamic-no-pic
12282 @opindex mdynamic-no-pic
12283 On Darwin and Mac OS X systems, compile code so that it is not
12284 relocatable, but that its external references are relocatable. The
12285 resulting code is suitable for applications, but not shared
12288 @item -mprioritize-restricted-insns=@var{priority}
12289 @opindex mprioritize-restricted-insns
12290 This option controls the priority that is assigned to
12291 dispatch-slot restricted instructions during the second scheduling
12292 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
12293 @var{no/highest/second-highest} priority to dispatch slot restricted
12296 @item -msched-costly-dep=@var{dependence_type}
12297 @opindex msched-costly-dep
12298 This option controls which dependences are considered costly
12299 by the target during instruction scheduling. The argument
12300 @var{dependence_type} takes one of the following values:
12301 @var{no}: no dependence is costly,
12302 @var{all}: all dependences are costly,
12303 @var{true_store_to_load}: a true dependence from store to load is costly,
12304 @var{store_to_load}: any dependence from store to load is costly,
12305 @var{number}: any dependence which latency >= @var{number} is costly.
12307 @item -minsert-sched-nops=@var{scheme}
12308 @opindex minsert-sched-nops
12309 This option controls which nop insertion scheme will be used during
12310 the second scheduling pass. The argument @var{scheme} takes one of the
12312 @var{no}: Don't insert nops.
12313 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
12314 according to the scheduler's grouping.
12315 @var{regroup_exact}: Insert nops to force costly dependent insns into
12316 separate groups. Insert exactly as many nops as needed to force an insn
12317 to a new group, according to the estimated processor grouping.
12318 @var{number}: Insert nops to force costly dependent insns into
12319 separate groups. Insert @var{number} nops to force an insn to a new group.
12322 @opindex mcall-sysv
12323 On System V.4 and embedded PowerPC systems compile code using calling
12324 conventions that adheres to the March 1995 draft of the System V
12325 Application Binary Interface, PowerPC processor supplement. This is the
12326 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
12328 @item -mcall-sysv-eabi
12329 @opindex mcall-sysv-eabi
12330 Specify both @option{-mcall-sysv} and @option{-meabi} options.
12332 @item -mcall-sysv-noeabi
12333 @opindex mcall-sysv-noeabi
12334 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
12336 @item -mcall-solaris
12337 @opindex mcall-solaris
12338 On System V.4 and embedded PowerPC systems compile code for the Solaris
12342 @opindex mcall-linux
12343 On System V.4 and embedded PowerPC systems compile code for the
12344 Linux-based GNU system.
12348 On System V.4 and embedded PowerPC systems compile code for the
12349 Hurd-based GNU system.
12351 @item -mcall-netbsd
12352 @opindex mcall-netbsd
12353 On System V.4 and embedded PowerPC systems compile code for the
12354 NetBSD operating system.
12356 @item -maix-struct-return
12357 @opindex maix-struct-return
12358 Return all structures in memory (as specified by the AIX ABI)@.
12360 @item -msvr4-struct-return
12361 @opindex msvr4-struct-return
12362 Return structures smaller than 8 bytes in registers (as specified by the
12365 @item -mabi=@var{abi-type}
12367 Extend the current ABI with a particular extension, or remove such extension.
12368 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
12369 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
12373 Extend the current ABI with SPE ABI extensions. This does not change
12374 the default ABI, instead it adds the SPE ABI extensions to the current
12378 @opindex mabi=no-spe
12379 Disable Booke SPE ABI extensions for the current ABI@.
12381 @item -mabi=ibmlongdouble
12382 @opindex mabi=ibmlongdouble
12383 Change the current ABI to use IBM extended precision long double.
12384 This is a PowerPC 32-bit SYSV ABI option.
12386 @item -mabi=ieeelongdouble
12387 @opindex mabi=ieeelongdouble
12388 Change the current ABI to use IEEE extended precision long double.
12389 This is a PowerPC 32-bit Linux ABI option.
12392 @itemx -mno-prototype
12393 @opindex mprototype
12394 @opindex mno-prototype
12395 On System V.4 and embedded PowerPC systems assume that all calls to
12396 variable argument functions are properly prototyped. Otherwise, the
12397 compiler must insert an instruction before every non prototyped call to
12398 set or clear bit 6 of the condition code register (@var{CR}) to
12399 indicate whether floating point values were passed in the floating point
12400 registers in case the function takes a variable arguments. With
12401 @option{-mprototype}, only calls to prototyped variable argument functions
12402 will set or clear the bit.
12406 On embedded PowerPC systems, assume that the startup module is called
12407 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
12408 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
12413 On embedded PowerPC systems, assume that the startup module is called
12414 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
12419 On embedded PowerPC systems, assume that the startup module is called
12420 @file{crt0.o} and the standard C libraries are @file{libads.a} and
12423 @item -myellowknife
12424 @opindex myellowknife
12425 On embedded PowerPC systems, assume that the startup module is called
12426 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
12431 On System V.4 and embedded PowerPC systems, specify that you are
12432 compiling for a VxWorks system.
12436 Specify that you are compiling for the WindISS simulation environment.
12440 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
12441 header to indicate that @samp{eabi} extended relocations are used.
12447 On System V.4 and embedded PowerPC systems do (do not) adhere to the
12448 Embedded Applications Binary Interface (eabi) which is a set of
12449 modifications to the System V.4 specifications. Selecting @option{-meabi}
12450 means that the stack is aligned to an 8 byte boundary, a function
12451 @code{__eabi} is called to from @code{main} to set up the eabi
12452 environment, and the @option{-msdata} option can use both @code{r2} and
12453 @code{r13} to point to two separate small data areas. Selecting
12454 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
12455 do not call an initialization function from @code{main}, and the
12456 @option{-msdata} option will only use @code{r13} to point to a single
12457 small data area. The @option{-meabi} option is on by default if you
12458 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
12461 @opindex msdata=eabi
12462 On System V.4 and embedded PowerPC systems, put small initialized
12463 @code{const} global and static data in the @samp{.sdata2} section, which
12464 is pointed to by register @code{r2}. Put small initialized
12465 non-@code{const} global and static data in the @samp{.sdata} section,
12466 which is pointed to by register @code{r13}. Put small uninitialized
12467 global and static data in the @samp{.sbss} section, which is adjacent to
12468 the @samp{.sdata} section. The @option{-msdata=eabi} option is
12469 incompatible with the @option{-mrelocatable} option. The
12470 @option{-msdata=eabi} option also sets the @option{-memb} option.
12473 @opindex msdata=sysv
12474 On System V.4 and embedded PowerPC systems, put small global and static
12475 data in the @samp{.sdata} section, which is pointed to by register
12476 @code{r13}. Put small uninitialized global and static data in the
12477 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
12478 The @option{-msdata=sysv} option is incompatible with the
12479 @option{-mrelocatable} option.
12481 @item -msdata=default
12483 @opindex msdata=default
12485 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
12486 compile code the same as @option{-msdata=eabi}, otherwise compile code the
12487 same as @option{-msdata=sysv}.
12490 @opindex msdata-data
12491 On System V.4 and embedded PowerPC systems, put small global
12492 data in the @samp{.sdata} section. Put small uninitialized global
12493 data in the @samp{.sbss} section. Do not use register @code{r13}
12494 to address small data however. This is the default behavior unless
12495 other @option{-msdata} options are used.
12499 @opindex msdata=none
12501 On embedded PowerPC systems, put all initialized global and static data
12502 in the @samp{.data} section, and all uninitialized data in the
12503 @samp{.bss} section.
12507 @cindex smaller data references (PowerPC)
12508 @cindex .sdata/.sdata2 references (PowerPC)
12509 On embedded PowerPC systems, put global and static items less than or
12510 equal to @var{num} bytes into the small data or bss sections instead of
12511 the normal data or bss section. By default, @var{num} is 8. The
12512 @option{-G @var{num}} switch is also passed to the linker.
12513 All modules should be compiled with the same @option{-G @var{num}} value.
12516 @itemx -mno-regnames
12518 @opindex mno-regnames
12519 On System V.4 and embedded PowerPC systems do (do not) emit register
12520 names in the assembly language output using symbolic forms.
12523 @itemx -mno-longcall
12525 @opindex mno-longcall
12526 By default assume that all calls are far away so that a longer more
12527 expensive calling sequence is required. This is required for calls
12528 further than 32 megabytes (33,554,432 bytes) from the current location.
12529 A short call will be generated if the compiler knows
12530 the call cannot be that far away. This setting can be overridden by
12531 the @code{shortcall} function attribute, or by @code{#pragma
12534 Some linkers are capable of detecting out-of-range calls and generating
12535 glue code on the fly. On these systems, long calls are unnecessary and
12536 generate slower code. As of this writing, the AIX linker can do this,
12537 as can the GNU linker for PowerPC/64. It is planned to add this feature
12538 to the GNU linker for 32-bit PowerPC systems as well.
12540 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
12541 callee, L42'', plus a ``branch island'' (glue code). The two target
12542 addresses represent the callee and the ``branch island''. The
12543 Darwin/PPC linker will prefer the first address and generate a ``bl
12544 callee'' if the PPC ``bl'' instruction will reach the callee directly;
12545 otherwise, the linker will generate ``bl L42'' to call the ``branch
12546 island''. The ``branch island'' is appended to the body of the
12547 calling function; it computes the full 32-bit address of the callee
12550 On Mach-O (Darwin) systems, this option directs the compiler emit to
12551 the glue for every direct call, and the Darwin linker decides whether
12552 to use or discard it.
12554 In the future, we may cause GCC to ignore all longcall specifications
12555 when the linker is known to generate glue.
12559 Adds support for multithreading with the @dfn{pthreads} library.
12560 This option sets flags for both the preprocessor and linker.
12564 @node S/390 and zSeries Options
12565 @subsection S/390 and zSeries Options
12566 @cindex S/390 and zSeries Options
12568 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
12572 @itemx -msoft-float
12573 @opindex mhard-float
12574 @opindex msoft-float
12575 Use (do not use) the hardware floating-point instructions and registers
12576 for floating-point operations. When @option{-msoft-float} is specified,
12577 functions in @file{libgcc.a} will be used to perform floating-point
12578 operations. When @option{-mhard-float} is specified, the compiler
12579 generates IEEE floating-point instructions. This is the default.
12581 @item -mlong-double-64
12582 @itemx -mlong-double-128
12583 @opindex mlong-double-64
12584 @opindex mlong-double-128
12585 These switches control the size of @code{long double} type. A size
12586 of 64bit makes the @code{long double} type equivalent to the @code{double}
12587 type. This is the default.
12590 @itemx -mno-backchain
12591 @opindex mbackchain
12592 @opindex mno-backchain
12593 Store (do not store) the address of the caller's frame as backchain pointer
12594 into the callee's stack frame.
12595 A backchain may be needed to allow debugging using tools that do not understand
12596 DWARF-2 call frame information.
12597 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
12598 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
12599 the backchain is placed into the topmost word of the 96/160 byte register
12602 In general, code compiled with @option{-mbackchain} is call-compatible with
12603 code compiled with @option{-mmo-backchain}; however, use of the backchain
12604 for debugging purposes usually requires that the whole binary is built with
12605 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
12606 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12607 to build a linux kernel use @option{-msoft-float}.
12609 The default is to not maintain the backchain.
12611 @item -mpacked-stack
12612 @item -mno-packed-stack
12613 @opindex mpacked-stack
12614 @opindex mno-packed-stack
12615 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
12616 specified, the compiler uses the all fields of the 96/160 byte register save
12617 area only for their default purpose; unused fields still take up stack space.
12618 When @option{-mpacked-stack} is specified, register save slots are densely
12619 packed at the top of the register save area; unused space is reused for other
12620 purposes, allowing for more efficient use of the available stack space.
12621 However, when @option{-mbackchain} is also in effect, the topmost word of
12622 the save area is always used to store the backchain, and the return address
12623 register is always saved two words below the backchain.
12625 As long as the stack frame backchain is not used, code generated with
12626 @option{-mpacked-stack} is call-compatible with code generated with
12627 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
12628 S/390 or zSeries generated code that uses the stack frame backchain at run
12629 time, not just for debugging purposes. Such code is not call-compatible
12630 with code compiled with @option{-mpacked-stack}. Also, note that the
12631 combination of @option{-mbackchain},
12632 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12633 to build a linux kernel use @option{-msoft-float}.
12635 The default is to not use the packed stack layout.
12638 @itemx -mno-small-exec
12639 @opindex msmall-exec
12640 @opindex mno-small-exec
12641 Generate (or do not generate) code using the @code{bras} instruction
12642 to do subroutine calls.
12643 This only works reliably if the total executable size does not
12644 exceed 64k. The default is to use the @code{basr} instruction instead,
12645 which does not have this limitation.
12651 When @option{-m31} is specified, generate code compliant to the
12652 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12653 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12654 particular to generate 64-bit instructions. For the @samp{s390}
12655 targets, the default is @option{-m31}, while the @samp{s390x}
12656 targets default to @option{-m64}.
12662 When @option{-mzarch} is specified, generate code using the
12663 instructions available on z/Architecture.
12664 When @option{-mesa} is specified, generate code using the
12665 instructions available on ESA/390. Note that @option{-mesa} is
12666 not possible with @option{-m64}.
12667 When generating code compliant to the GNU/Linux for S/390 ABI,
12668 the default is @option{-mesa}. When generating code compliant
12669 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12675 Generate (or do not generate) code using the @code{mvcle} instruction
12676 to perform block moves. When @option{-mno-mvcle} is specified,
12677 use a @code{mvc} loop instead. This is the default unless optimizing for
12684 Print (or do not print) additional debug information when compiling.
12685 The default is to not print debug information.
12687 @item -march=@var{cpu-type}
12689 Generate code that will run on @var{cpu-type}, which is the name of a system
12690 representing a certain processor type. Possible values for
12691 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12692 When generating code using the instructions available on z/Architecture,
12693 the default is @option{-march=z900}. Otherwise, the default is
12694 @option{-march=g5}.
12696 @item -mtune=@var{cpu-type}
12698 Tune to @var{cpu-type} everything applicable about the generated code,
12699 except for the ABI and the set of available instructions.
12700 The list of @var{cpu-type} values is the same as for @option{-march}.
12701 The default is the value used for @option{-march}.
12704 @itemx -mno-tpf-trace
12705 @opindex mtpf-trace
12706 @opindex mno-tpf-trace
12707 Generate code that adds (does not add) in TPF OS specific branches to trace
12708 routines in the operating system. This option is off by default, even
12709 when compiling for the TPF OS@.
12712 @itemx -mno-fused-madd
12713 @opindex mfused-madd
12714 @opindex mno-fused-madd
12715 Generate code that uses (does not use) the floating point multiply and
12716 accumulate instructions. These instructions are generated by default if
12717 hardware floating point is used.
12719 @item -mwarn-framesize=@var{framesize}
12720 @opindex mwarn-framesize
12721 Emit a warning if the current function exceeds the given frame size. Because
12722 this is a compile time check it doesn't need to be a real problem when the program
12723 runs. It is intended to identify functions which most probably cause
12724 a stack overflow. It is useful to be used in an environment with limited stack
12725 size e.g.@: the linux kernel.
12727 @item -mwarn-dynamicstack
12728 @opindex mwarn-dynamicstack
12729 Emit a warning if the function calls alloca or uses dynamically
12730 sized arrays. This is generally a bad idea with a limited stack size.
12732 @item -mstack-guard=@var{stack-guard}
12733 @item -mstack-size=@var{stack-size}
12734 @opindex mstack-guard
12735 @opindex mstack-size
12736 These arguments always have to be used in conjunction. If they are present the s390
12737 back end emits additional instructions in the function prologue which trigger a trap
12738 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12739 (remember that the stack on s390 grows downward). These options are intended to
12740 be used to help debugging stack overflow problems. The additionally emitted code
12741 causes only little overhead and hence can also be used in production like systems
12742 without greater performance degradation. The given values have to be exact
12743 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12745 In order to be efficient the extra code makes the assumption that the stack starts
12746 at an address aligned to the value given by @var{stack-size}.
12749 @node Score Options
12750 @subsection Score Options
12751 @cindex Score Options
12753 These options are defined for Score implementations:
12758 Compile code for little endian mode.
12762 Compile code for big endian mode. This is the default.
12766 Enable the use of multiply-accumulate instructions. Disabled by default.
12770 Specify the SCORE5U of the target architecture.
12774 Specify the SCORE7 of the target architecture. This is the default.
12778 @subsection SH Options
12780 These @samp{-m} options are defined for the SH implementations:
12785 Generate code for the SH1.
12789 Generate code for the SH2.
12792 Generate code for the SH2e.
12796 Generate code for the SH3.
12800 Generate code for the SH3e.
12804 Generate code for the SH4 without a floating-point unit.
12806 @item -m4-single-only
12807 @opindex m4-single-only
12808 Generate code for the SH4 with a floating-point unit that only
12809 supports single-precision arithmetic.
12813 Generate code for the SH4 assuming the floating-point unit is in
12814 single-precision mode by default.
12818 Generate code for the SH4.
12822 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12823 floating-point unit is not used.
12825 @item -m4a-single-only
12826 @opindex m4a-single-only
12827 Generate code for the SH4a, in such a way that no double-precision
12828 floating point operations are used.
12831 @opindex m4a-single
12832 Generate code for the SH4a assuming the floating-point unit is in
12833 single-precision mode by default.
12837 Generate code for the SH4a.
12841 Same as @option{-m4a-nofpu}, except that it implicitly passes
12842 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12843 instructions at the moment.
12847 Compile code for the processor in big endian mode.
12851 Compile code for the processor in little endian mode.
12855 Align doubles at 64-bit boundaries. Note that this changes the calling
12856 conventions, and thus some functions from the standard C library will
12857 not work unless you recompile it first with @option{-mdalign}.
12861 Shorten some address references at link time, when possible; uses the
12862 linker option @option{-relax}.
12866 Use 32-bit offsets in @code{switch} tables. The default is to use
12871 Enable the use of the instruction @code{fmovd}.
12875 Comply with the calling conventions defined by Renesas.
12879 Comply with the calling conventions defined by Renesas.
12883 Comply with the calling conventions defined for GCC before the Renesas
12884 conventions were available. This option is the default for all
12885 targets of the SH toolchain except for @samp{sh-symbianelf}.
12888 @opindex mnomacsave
12889 Mark the @code{MAC} register as call-clobbered, even if
12890 @option{-mhitachi} is given.
12894 Increase IEEE-compliance of floating-point code.
12895 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12896 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12897 comparisons of NANs / infinities incurs extra overhead in every
12898 floating point comparison, therefore the default is set to
12899 @option{-ffinite-math-only}.
12901 @item -minline-ic_invalidate
12902 @opindex minline-ic_invalidate
12903 Inline code to invalidate instruction cache entries after setting up
12904 nested function trampolines.
12905 This option has no effect if -musermode is in effect and the selected
12906 code generation option (e.g. -m4) does not allow the use of the icbi
12908 If the selected code generation option does not allow the use of the icbi
12909 instruction, and -musermode is not in effect, the inlined code will
12910 manipulate the instruction cache address array directly with an associative
12911 write. This not only requires privileged mode, but it will also
12912 fail if the cache line had been mapped via the TLB and has become unmapped.
12916 Dump instruction size and location in the assembly code.
12919 @opindex mpadstruct
12920 This option is deprecated. It pads structures to multiple of 4 bytes,
12921 which is incompatible with the SH ABI@.
12925 Optimize for space instead of speed. Implied by @option{-Os}.
12928 @opindex mprefergot
12929 When generating position-independent code, emit function calls using
12930 the Global Offset Table instead of the Procedure Linkage Table.
12934 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
12935 if the inlined code would not work in user mode.
12936 This is the default when the target is @code{sh-*-linux*}.
12938 @item -multcost=@var{number}
12939 @opindex multcost=@var{number}
12940 Set the cost to assume for a multiply insn.
12942 @item -mdiv=@var{strategy}
12943 @opindex mdiv=@var{strategy}
12944 Set the division strategy to use for SHmedia code. @var{strategy} must be
12945 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12946 inv:call2, inv:fp .
12947 "fp" performs the operation in floating point. This has a very high latency,
12948 but needs only a few instructions, so it might be a good choice if
12949 your code has enough easily exploitable ILP to allow the compiler to
12950 schedule the floating point instructions together with other instructions.
12951 Division by zero causes a floating point exception.
12952 "inv" uses integer operations to calculate the inverse of the divisor,
12953 and then multiplies the dividend with the inverse. This strategy allows
12954 cse and hoisting of the inverse calculation. Division by zero calculates
12955 an unspecified result, but does not trap.
12956 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12957 have been found, or if the entire operation has been hoisted to the same
12958 place, the last stages of the inverse calculation are intertwined with the
12959 final multiply to reduce the overall latency, at the expense of using a few
12960 more instructions, and thus offering fewer scheduling opportunities with
12962 "call" calls a library function that usually implements the inv:minlat
12964 This gives high code density for m5-*media-nofpu compilations.
12965 "call2" uses a different entry point of the same library function, where it
12966 assumes that a pointer to a lookup table has already been set up, which
12967 exposes the pointer load to cse / code hoisting optimizations.
12968 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12969 code generation, but if the code stays unoptimized, revert to the "call",
12970 "call2", or "fp" strategies, respectively. Note that the
12971 potentially-trapping side effect of division by zero is carried by a
12972 separate instruction, so it is possible that all the integer instructions
12973 are hoisted out, but the marker for the side effect stays where it is.
12974 A recombination to fp operations or a call is not possible in that case.
12975 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12976 that the inverse calculation was nor separated from the multiply, they speed
12977 up division where the dividend fits into 20 bits (plus sign where applicable),
12978 by inserting a test to skip a number of operations in this case; this test
12979 slows down the case of larger dividends. inv20u assumes the case of a such
12980 a small dividend to be unlikely, and inv20l assumes it to be likely.
12982 @item -mdivsi3_libfunc=@var{name}
12983 @opindex mdivsi3_libfunc=@var{name}
12984 Set the name of the library function used for 32 bit signed division to
12985 @var{name}. This only affect the name used in the call and inv:call
12986 division strategies, and the compiler will still expect the same
12987 sets of input/output/clobbered registers as if this option was not present.
12989 @item -madjust-unroll
12990 @opindex madjust-unroll
12991 Throttle unrolling to avoid thrashing target registers.
12992 This option only has an effect if the gcc code base supports the
12993 TARGET_ADJUST_UNROLL_MAX target hook.
12995 @item -mindexed-addressing
12996 @opindex mindexed-addressing
12997 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12998 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12999 semantics for the indexed addressing mode. The architecture allows the
13000 implementation of processors with 64 bit MMU, which the OS could use to
13001 get 32 bit addressing, but since no current hardware implementation supports
13002 this or any other way to make the indexed addressing mode safe to use in
13003 the 32 bit ABI, the default is -mno-indexed-addressing.
13005 @item -mgettrcost=@var{number}
13006 @opindex mgettrcost=@var{number}
13007 Set the cost assumed for the gettr instruction to @var{number}.
13008 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
13012 Assume pt* instructions won't trap. This will generally generate better
13013 scheduled code, but is unsafe on current hardware. The current architecture
13014 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
13015 This has the unintentional effect of making it unsafe to schedule ptabs /
13016 ptrel before a branch, or hoist it out of a loop. For example,
13017 __do_global_ctors, a part of libgcc that runs constructors at program
13018 startup, calls functions in a list which is delimited by @minus{}1. With the
13019 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
13020 That means that all the constructors will be run a bit quicker, but when
13021 the loop comes to the end of the list, the program crashes because ptabs
13022 loads @minus{}1 into a target register. Since this option is unsafe for any
13023 hardware implementing the current architecture specification, the default
13024 is -mno-pt-fixed. Unless the user specifies a specific cost with
13025 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
13026 this deters register allocation using target registers for storing
13029 @item -minvalid-symbols
13030 @opindex minvalid-symbols
13031 Assume symbols might be invalid. Ordinary function symbols generated by
13032 the compiler will always be valid to load with movi/shori/ptabs or
13033 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
13034 to generate symbols that will cause ptabs / ptrel to trap.
13035 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
13036 It will then prevent cross-basic-block cse, hoisting and most scheduling
13037 of symbol loads. The default is @option{-mno-invalid-symbols}.
13040 @node SPARC Options
13041 @subsection SPARC Options
13042 @cindex SPARC options
13044 These @samp{-m} options are supported on the SPARC:
13047 @item -mno-app-regs
13049 @opindex mno-app-regs
13051 Specify @option{-mapp-regs} to generate output using the global registers
13052 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
13055 To be fully SVR4 ABI compliant at the cost of some performance loss,
13056 specify @option{-mno-app-regs}. You should compile libraries and system
13057 software with this option.
13060 @itemx -mhard-float
13062 @opindex mhard-float
13063 Generate output containing floating point instructions. This is the
13067 @itemx -msoft-float
13069 @opindex msoft-float
13070 Generate output containing library calls for floating point.
13071 @strong{Warning:} the requisite libraries are not available for all SPARC
13072 targets. Normally the facilities of the machine's usual C compiler are
13073 used, but this cannot be done directly in cross-compilation. You must make
13074 your own arrangements to provide suitable library functions for
13075 cross-compilation. The embedded targets @samp{sparc-*-aout} and
13076 @samp{sparclite-*-*} do provide software floating point support.
13078 @option{-msoft-float} changes the calling convention in the output file;
13079 therefore, it is only useful if you compile @emph{all} of a program with
13080 this option. In particular, you need to compile @file{libgcc.a}, the
13081 library that comes with GCC, with @option{-msoft-float} in order for
13084 @item -mhard-quad-float
13085 @opindex mhard-quad-float
13086 Generate output containing quad-word (long double) floating point
13089 @item -msoft-quad-float
13090 @opindex msoft-quad-float
13091 Generate output containing library calls for quad-word (long double)
13092 floating point instructions. The functions called are those specified
13093 in the SPARC ABI@. This is the default.
13095 As of this writing, there are no SPARC implementations that have hardware
13096 support for the quad-word floating point instructions. They all invoke
13097 a trap handler for one of these instructions, and then the trap handler
13098 emulates the effect of the instruction. Because of the trap handler overhead,
13099 this is much slower than calling the ABI library routines. Thus the
13100 @option{-msoft-quad-float} option is the default.
13102 @item -mno-unaligned-doubles
13103 @itemx -munaligned-doubles
13104 @opindex mno-unaligned-doubles
13105 @opindex munaligned-doubles
13106 Assume that doubles have 8 byte alignment. This is the default.
13108 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
13109 alignment only if they are contained in another type, or if they have an
13110 absolute address. Otherwise, it assumes they have 4 byte alignment.
13111 Specifying this option avoids some rare compatibility problems with code
13112 generated by other compilers. It is not the default because it results
13113 in a performance loss, especially for floating point code.
13115 @item -mno-faster-structs
13116 @itemx -mfaster-structs
13117 @opindex mno-faster-structs
13118 @opindex mfaster-structs
13119 With @option{-mfaster-structs}, the compiler assumes that structures
13120 should have 8 byte alignment. This enables the use of pairs of
13121 @code{ldd} and @code{std} instructions for copies in structure
13122 assignment, in place of twice as many @code{ld} and @code{st} pairs.
13123 However, the use of this changed alignment directly violates the SPARC
13124 ABI@. Thus, it's intended only for use on targets where the developer
13125 acknowledges that their resulting code will not be directly in line with
13126 the rules of the ABI@.
13128 @item -mimpure-text
13129 @opindex mimpure-text
13130 @option{-mimpure-text}, used in addition to @option{-shared}, tells
13131 the compiler to not pass @option{-z text} to the linker when linking a
13132 shared object. Using this option, you can link position-dependent
13133 code into a shared object.
13135 @option{-mimpure-text} suppresses the ``relocations remain against
13136 allocatable but non-writable sections'' linker error message.
13137 However, the necessary relocations will trigger copy-on-write, and the
13138 shared object is not actually shared across processes. Instead of
13139 using @option{-mimpure-text}, you should compile all source code with
13140 @option{-fpic} or @option{-fPIC}.
13142 This option is only available on SunOS and Solaris.
13144 @item -mcpu=@var{cpu_type}
13146 Set the instruction set, register set, and instruction scheduling parameters
13147 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
13148 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
13149 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
13150 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
13151 @samp{ultrasparc3}, and @samp{niagara}.
13153 Default instruction scheduling parameters are used for values that select
13154 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
13155 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
13157 Here is a list of each supported architecture and their supported
13162 v8: supersparc, hypersparc
13163 sparclite: f930, f934, sparclite86x
13165 v9: ultrasparc, ultrasparc3, niagara
13168 By default (unless configured otherwise), GCC generates code for the V7
13169 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
13170 additionally optimizes it for the Cypress CY7C602 chip, as used in the
13171 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
13172 SPARCStation 1, 2, IPX etc.
13174 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
13175 architecture. The only difference from V7 code is that the compiler emits
13176 the integer multiply and integer divide instructions which exist in SPARC-V8
13177 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
13178 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
13181 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
13182 the SPARC architecture. This adds the integer multiply, integer divide step
13183 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
13184 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
13185 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
13186 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
13187 MB86934 chip, which is the more recent SPARClite with FPU@.
13189 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
13190 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
13191 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
13192 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
13193 optimizes it for the TEMIC SPARClet chip.
13195 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
13196 architecture. This adds 64-bit integer and floating-point move instructions,
13197 3 additional floating-point condition code registers and conditional move
13198 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
13199 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
13200 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
13201 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
13202 @option{-mcpu=niagara}, the compiler additionally optimizes it for
13203 Sun UltraSPARC T1 chips.
13205 @item -mtune=@var{cpu_type}
13207 Set the instruction scheduling parameters for machine type
13208 @var{cpu_type}, but do not set the instruction set or register set that the
13209 option @option{-mcpu=@var{cpu_type}} would.
13211 The same values for @option{-mcpu=@var{cpu_type}} can be used for
13212 @option{-mtune=@var{cpu_type}}, but the only useful values are those
13213 that select a particular cpu implementation. Those are @samp{cypress},
13214 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
13215 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
13216 @samp{ultrasparc3}, and @samp{niagara}.
13221 @opindex mno-v8plus
13222 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
13223 difference from the V8 ABI is that the global and out registers are
13224 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
13225 mode for all SPARC-V9 processors.
13231 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
13232 Visual Instruction Set extensions. The default is @option{-mno-vis}.
13235 These @samp{-m} options are supported in addition to the above
13236 on SPARC-V9 processors in 64-bit environments:
13239 @item -mlittle-endian
13240 @opindex mlittle-endian
13241 Generate code for a processor running in little-endian mode. It is only
13242 available for a few configurations and most notably not on Solaris and Linux.
13248 Generate code for a 32-bit or 64-bit environment.
13249 The 32-bit environment sets int, long and pointer to 32 bits.
13250 The 64-bit environment sets int to 32 bits and long and pointer
13253 @item -mcmodel=medlow
13254 @opindex mcmodel=medlow
13255 Generate code for the Medium/Low code model: 64-bit addresses, programs
13256 must be linked in the low 32 bits of memory. Programs can be statically
13257 or dynamically linked.
13259 @item -mcmodel=medmid
13260 @opindex mcmodel=medmid
13261 Generate code for the Medium/Middle code model: 64-bit addresses, programs
13262 must be linked in the low 44 bits of memory, the text and data segments must
13263 be less than 2GB in size and the data segment must be located within 2GB of
13266 @item -mcmodel=medany
13267 @opindex mcmodel=medany
13268 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
13269 may be linked anywhere in memory, the text and data segments must be less
13270 than 2GB in size and the data segment must be located within 2GB of the
13273 @item -mcmodel=embmedany
13274 @opindex mcmodel=embmedany
13275 Generate code for the Medium/Anywhere code model for embedded systems:
13276 64-bit addresses, the text and data segments must be less than 2GB in
13277 size, both starting anywhere in memory (determined at link time). The
13278 global register %g4 points to the base of the data segment. Programs
13279 are statically linked and PIC is not supported.
13282 @itemx -mno-stack-bias
13283 @opindex mstack-bias
13284 @opindex mno-stack-bias
13285 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
13286 frame pointer if present, are offset by @minus{}2047 which must be added back
13287 when making stack frame references. This is the default in 64-bit mode.
13288 Otherwise, assume no such offset is present.
13291 These switches are supported in addition to the above on Solaris:
13296 Add support for multithreading using the Solaris threads library. This
13297 option sets flags for both the preprocessor and linker. This option does
13298 not affect the thread safety of object code produced by the compiler or
13299 that of libraries supplied with it.
13303 Add support for multithreading using the POSIX threads library. This
13304 option sets flags for both the preprocessor and linker. This option does
13305 not affect the thread safety of object code produced by the compiler or
13306 that of libraries supplied with it.
13310 This is a synonym for @option{-pthreads}.
13314 @subsection SPU Options
13315 @cindex SPU options
13317 These @samp{-m} options are supported on the SPU:
13321 @itemx -merror-reloc
13322 @opindex mwarn-reloc
13323 @opindex merror-reloc
13325 The loader for SPU does not handle dynamic relocations. By default, GCC
13326 will give an error when it generates code that requires a dynamic
13327 relocation. @option{-mno-error-reloc} disables the error,
13328 @option{-mwarn-reloc} will generate a warning instead.
13331 @itemx -munsafe-dma
13333 @opindex munsafe-dma
13335 Instructions which initiate or test completion of DMA must not be
13336 reordered with respect to loads and stores of the memory which is being
13337 accessed. Users typically address this problem using the volatile
13338 keyword, but that can lead to inefficient code in places where the
13339 memory is known to not change. Rather than mark the memory as volatile
13340 we treat the DMA instructions as potentially effecting all memory. With
13341 @option{-munsafe-dma} users must use the volatile keyword to protect
13344 @item -mbranch-hints
13345 @opindex mbranch-hints
13347 By default, GCC will generate a branch hint instruction to avoid
13348 pipeline stalls for always taken or probably taken branches. A hint
13349 will not be generated closer than 8 instructions away from its branch.
13350 There is little reason to disable them, except for debugging purposes,
13351 or to make an object a little bit smaller.
13355 @opindex msmall-mem
13356 @opindex mlarge-mem
13358 By default, GCC generates code assuming that addresses are never larger
13359 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
13360 a full 32 bit address.
13365 By default, GCC links against startup code that assumes the SPU-style
13366 main function interface (which has an unconventional parameter list).
13367 With @option{-mstdmain}, GCC will link your program against startup
13368 code that assumes a C99-style interface to @code{main}, including a
13369 local copy of @code{argv} strings.
13371 @item -mfixed-range=@var{register-range}
13372 @opindex mfixed-range
13373 Generate code treating the given register range as fixed registers.
13374 A fixed register is one that the register allocator can not use. This is
13375 useful when compiling kernel code. A register range is specified as
13376 two registers separated by a dash. Multiple register ranges can be
13377 specified separated by a comma.
13381 @node System V Options
13382 @subsection Options for System V
13384 These additional options are available on System V Release 4 for
13385 compatibility with other compilers on those systems:
13390 Create a shared object.
13391 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
13395 Identify the versions of each tool used by the compiler, in a
13396 @code{.ident} assembler directive in the output.
13400 Refrain from adding @code{.ident} directives to the output file (this is
13403 @item -YP,@var{dirs}
13405 Search the directories @var{dirs}, and no others, for libraries
13406 specified with @option{-l}.
13408 @item -Ym,@var{dir}
13410 Look in the directory @var{dir} to find the M4 preprocessor.
13411 The assembler uses this option.
13412 @c This is supposed to go with a -Yd for predefined M4 macro files, but
13413 @c the generic assembler that comes with Solaris takes just -Ym.
13416 @node TMS320C3x/C4x Options
13417 @subsection TMS320C3x/C4x Options
13418 @cindex TMS320C3x/C4x Options
13420 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
13424 @item -mcpu=@var{cpu_type}
13426 Set the instruction set, register set, and instruction scheduling
13427 parameters for machine type @var{cpu_type}. Supported values for
13428 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
13429 @samp{c44}. The default is @samp{c40} to generate code for the
13434 @itemx -msmall-memory
13436 @opindex mbig-memory
13438 @opindex msmall-memory
13440 Generates code for the big or small memory model. The small memory
13441 model assumed that all data fits into one 64K word page. At run-time
13442 the data page (DP) register must be set to point to the 64K page
13443 containing the .bss and .data program sections. The big memory model is
13444 the default and requires reloading of the DP register for every direct
13451 Allow (disallow) allocation of general integer operands into the block
13452 count register BK@.
13458 Enable (disable) generation of code using decrement and branch,
13459 DBcond(D), instructions. This is enabled by default for the C4x. To be
13460 on the safe side, this is disabled for the C3x, since the maximum
13461 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
13462 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
13463 that it can utilize the decrement and branch instruction, but will give
13464 up if there is more than one memory reference in the loop. Thus a loop
13465 where the loop counter is decremented can generate slightly more
13466 efficient code, in cases where the RPTB instruction cannot be utilized.
13468 @item -mdp-isr-reload
13470 @opindex mdp-isr-reload
13472 Force the DP register to be saved on entry to an interrupt service
13473 routine (ISR), reloaded to point to the data section, and restored on
13474 exit from the ISR@. This should not be required unless someone has
13475 violated the small memory model by modifying the DP register, say within
13482 For the C3x use the 24-bit MPYI instruction for integer multiplies
13483 instead of a library call to guarantee 32-bit results. Note that if one
13484 of the operands is a constant, then the multiplication will be performed
13485 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
13486 then squaring operations are performed inline instead of a library call.
13489 @itemx -mno-fast-fix
13491 @opindex mno-fast-fix
13492 The C3x/C4x FIX instruction to convert a floating point value to an
13493 integer value chooses the nearest integer less than or equal to the
13494 floating point value rather than to the nearest integer. Thus if the
13495 floating point number is negative, the result will be incorrectly
13496 truncated an additional code is necessary to detect and correct this
13497 case. This option can be used to disable generation of the additional
13498 code required to correct the result.
13504 Enable (disable) generation of repeat block sequences using the RPTB
13505 instruction for zero overhead looping. The RPTB construct is only used
13506 for innermost loops that do not call functions or jump across the loop
13507 boundaries. There is no advantage having nested RPTB loops due to the
13508 overhead required to save and restore the RC, RS, and RE registers.
13509 This is enabled by default with @option{-O2}.
13511 @item -mrpts=@var{count}
13515 Enable (disable) the use of the single instruction repeat instruction
13516 RPTS@. If a repeat block contains a single instruction, and the loop
13517 count can be guaranteed to be less than the value @var{count}, GCC will
13518 emit a RPTS instruction instead of a RPTB@. If no value is specified,
13519 then a RPTS will be emitted even if the loop count cannot be determined
13520 at compile time. Note that the repeated instruction following RPTS does
13521 not have to be reloaded from memory each iteration, thus freeing up the
13522 CPU buses for operands. However, since interrupts are blocked by this
13523 instruction, it is disabled by default.
13525 @item -mloop-unsigned
13526 @itemx -mno-loop-unsigned
13527 @opindex mloop-unsigned
13528 @opindex mno-loop-unsigned
13529 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
13530 is @math{2^{31} + 1} since these instructions test if the iteration count is
13531 negative to terminate the loop. If the iteration count is unsigned
13532 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
13533 exceeded. This switch allows an unsigned iteration count.
13537 Try to emit an assembler syntax that the TI assembler (asm30) is happy
13538 with. This also enforces compatibility with the API employed by the TI
13539 C3x C compiler. For example, long doubles are passed as structures
13540 rather than in floating point registers.
13546 Generate code that uses registers (stack) for passing arguments to functions.
13547 By default, arguments are passed in registers where possible rather
13548 than by pushing arguments on to the stack.
13550 @item -mparallel-insns
13551 @itemx -mno-parallel-insns
13552 @opindex mparallel-insns
13553 @opindex mno-parallel-insns
13554 Allow the generation of parallel instructions. This is enabled by
13555 default with @option{-O2}.
13557 @item -mparallel-mpy
13558 @itemx -mno-parallel-mpy
13559 @opindex mparallel-mpy
13560 @opindex mno-parallel-mpy
13561 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
13562 provided @option{-mparallel-insns} is also specified. These instructions have
13563 tight register constraints which can pessimize the code generation
13564 of large functions.
13569 @subsection V850 Options
13570 @cindex V850 Options
13572 These @samp{-m} options are defined for V850 implementations:
13576 @itemx -mno-long-calls
13577 @opindex mlong-calls
13578 @opindex mno-long-calls
13579 Treat all calls as being far away (near). If calls are assumed to be
13580 far away, the compiler will always load the functions address up into a
13581 register, and call indirect through the pointer.
13587 Do not optimize (do optimize) basic blocks that use the same index
13588 pointer 4 or more times to copy pointer into the @code{ep} register, and
13589 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
13590 option is on by default if you optimize.
13592 @item -mno-prolog-function
13593 @itemx -mprolog-function
13594 @opindex mno-prolog-function
13595 @opindex mprolog-function
13596 Do not use (do use) external functions to save and restore registers
13597 at the prologue and epilogue of a function. The external functions
13598 are slower, but use less code space if more than one function saves
13599 the same number of registers. The @option{-mprolog-function} option
13600 is on by default if you optimize.
13604 Try to make the code as small as possible. At present, this just turns
13605 on the @option{-mep} and @option{-mprolog-function} options.
13607 @item -mtda=@var{n}
13609 Put static or global variables whose size is @var{n} bytes or less into
13610 the tiny data area that register @code{ep} points to. The tiny data
13611 area can hold up to 256 bytes in total (128 bytes for byte references).
13613 @item -msda=@var{n}
13615 Put static or global variables whose size is @var{n} bytes or less into
13616 the small data area that register @code{gp} points to. The small data
13617 area can hold up to 64 kilobytes.
13619 @item -mzda=@var{n}
13621 Put static or global variables whose size is @var{n} bytes or less into
13622 the first 32 kilobytes of memory.
13626 Specify that the target processor is the V850.
13629 @opindex mbig-switch
13630 Generate code suitable for big switch tables. Use this option only if
13631 the assembler/linker complain about out of range branches within a switch
13636 This option will cause r2 and r5 to be used in the code generated by
13637 the compiler. This setting is the default.
13639 @item -mno-app-regs
13640 @opindex mno-app-regs
13641 This option will cause r2 and r5 to be treated as fixed registers.
13645 Specify that the target processor is the V850E1. The preprocessor
13646 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
13647 this option is used.
13651 Specify that the target processor is the V850E@. The preprocessor
13652 constant @samp{__v850e__} will be defined if this option is used.
13654 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
13655 are defined then a default target processor will be chosen and the
13656 relevant @samp{__v850*__} preprocessor constant will be defined.
13658 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
13659 defined, regardless of which processor variant is the target.
13661 @item -mdisable-callt
13662 @opindex mdisable-callt
13663 This option will suppress generation of the CALLT instruction for the
13664 v850e and v850e1 flavors of the v850 architecture. The default is
13665 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
13670 @subsection VAX Options
13671 @cindex VAX options
13673 These @samp{-m} options are defined for the VAX:
13678 Do not output certain jump instructions (@code{aobleq} and so on)
13679 that the Unix assembler for the VAX cannot handle across long
13684 Do output those jump instructions, on the assumption that you
13685 will assemble with the GNU assembler.
13689 Output code for g-format floating point numbers instead of d-format.
13692 @node x86-64 Options
13693 @subsection x86-64 Options
13694 @cindex x86-64 options
13696 These are listed under @xref{i386 and x86-64 Options}.
13698 @node Xstormy16 Options
13699 @subsection Xstormy16 Options
13700 @cindex Xstormy16 Options
13702 These options are defined for Xstormy16:
13707 Choose startup files and linker script suitable for the simulator.
13710 @node Xtensa Options
13711 @subsection Xtensa Options
13712 @cindex Xtensa Options
13714 These options are supported for Xtensa targets:
13718 @itemx -mno-const16
13720 @opindex mno-const16
13721 Enable or disable use of @code{CONST16} instructions for loading
13722 constant values. The @code{CONST16} instruction is currently not a
13723 standard option from Tensilica. When enabled, @code{CONST16}
13724 instructions are always used in place of the standard @code{L32R}
13725 instructions. The use of @code{CONST16} is enabled by default only if
13726 the @code{L32R} instruction is not available.
13729 @itemx -mno-fused-madd
13730 @opindex mfused-madd
13731 @opindex mno-fused-madd
13732 Enable or disable use of fused multiply/add and multiply/subtract
13733 instructions in the floating-point option. This has no effect if the
13734 floating-point option is not also enabled. Disabling fused multiply/add
13735 and multiply/subtract instructions forces the compiler to use separate
13736 instructions for the multiply and add/subtract operations. This may be
13737 desirable in some cases where strict IEEE 754-compliant results are
13738 required: the fused multiply add/subtract instructions do not round the
13739 intermediate result, thereby producing results with @emph{more} bits of
13740 precision than specified by the IEEE standard. Disabling fused multiply
13741 add/subtract instructions also ensures that the program output is not
13742 sensitive to the compiler's ability to combine multiply and add/subtract
13745 @item -mtext-section-literals
13746 @itemx -mno-text-section-literals
13747 @opindex mtext-section-literals
13748 @opindex mno-text-section-literals
13749 Control the treatment of literal pools. The default is
13750 @option{-mno-text-section-literals}, which places literals in a separate
13751 section in the output file. This allows the literal pool to be placed
13752 in a data RAM/ROM, and it also allows the linker to combine literal
13753 pools from separate object files to remove redundant literals and
13754 improve code size. With @option{-mtext-section-literals}, the literals
13755 are interspersed in the text section in order to keep them as close as
13756 possible to their references. This may be necessary for large assembly
13759 @item -mtarget-align
13760 @itemx -mno-target-align
13761 @opindex mtarget-align
13762 @opindex mno-target-align
13763 When this option is enabled, GCC instructs the assembler to
13764 automatically align instructions to reduce branch penalties at the
13765 expense of some code density. The assembler attempts to widen density
13766 instructions to align branch targets and the instructions following call
13767 instructions. If there are not enough preceding safe density
13768 instructions to align a target, no widening will be performed. The
13769 default is @option{-mtarget-align}. These options do not affect the
13770 treatment of auto-aligned instructions like @code{LOOP}, which the
13771 assembler will always align, either by widening density instructions or
13772 by inserting no-op instructions.
13775 @itemx -mno-longcalls
13776 @opindex mlongcalls
13777 @opindex mno-longcalls
13778 When this option is enabled, GCC instructs the assembler to translate
13779 direct calls to indirect calls unless it can determine that the target
13780 of a direct call is in the range allowed by the call instruction. This
13781 translation typically occurs for calls to functions in other source
13782 files. Specifically, the assembler translates a direct @code{CALL}
13783 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13784 The default is @option{-mno-longcalls}. This option should be used in
13785 programs where the call target can potentially be out of range. This
13786 option is implemented in the assembler, not the compiler, so the
13787 assembly code generated by GCC will still show direct call
13788 instructions---look at the disassembled object code to see the actual
13789 instructions. Note that the assembler will use an indirect call for
13790 every cross-file call, not just those that really will be out of range.
13793 @node zSeries Options
13794 @subsection zSeries Options
13795 @cindex zSeries options
13797 These are listed under @xref{S/390 and zSeries Options}.
13799 @node Code Gen Options
13800 @section Options for Code Generation Conventions
13801 @cindex code generation conventions
13802 @cindex options, code generation
13803 @cindex run-time options
13805 These machine-independent options control the interface conventions
13806 used in code generation.
13808 Most of them have both positive and negative forms; the negative form
13809 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13810 one of the forms is listed---the one which is not the default. You
13811 can figure out the other form by either removing @samp{no-} or adding
13815 @item -fbounds-check
13816 @opindex fbounds-check
13817 For front-ends that support it, generate additional code to check that
13818 indices used to access arrays are within the declared range. This is
13819 currently only supported by the Java and Fortran front-ends, where
13820 this option defaults to true and false respectively.
13824 This option generates traps for signed overflow on addition, subtraction,
13825 multiplication operations.
13829 This option instructs the compiler to assume that signed arithmetic
13830 overflow of addition, subtraction and multiplication wraps around
13831 using twos-complement representation. This flag enables some optimizations
13832 and disables others. This option is enabled by default for the Java
13833 front-end, as required by the Java language specification.
13836 @opindex fexceptions
13837 Enable exception handling. Generates extra code needed to propagate
13838 exceptions. For some targets, this implies GCC will generate frame
13839 unwind information for all functions, which can produce significant data
13840 size overhead, although it does not affect execution. If you do not
13841 specify this option, GCC will enable it by default for languages like
13842 C++ which normally require exception handling, and disable it for
13843 languages like C that do not normally require it. However, you may need
13844 to enable this option when compiling C code that needs to interoperate
13845 properly with exception handlers written in C++. You may also wish to
13846 disable this option if you are compiling older C++ programs that don't
13847 use exception handling.
13849 @item -fnon-call-exceptions
13850 @opindex fnon-call-exceptions
13851 Generate code that allows trapping instructions to throw exceptions.
13852 Note that this requires platform-specific runtime support that does
13853 not exist everywhere. Moreover, it only allows @emph{trapping}
13854 instructions to throw exceptions, i.e.@: memory references or floating
13855 point instructions. It does not allow exceptions to be thrown from
13856 arbitrary signal handlers such as @code{SIGALRM}.
13858 @item -funwind-tables
13859 @opindex funwind-tables
13860 Similar to @option{-fexceptions}, except that it will just generate any needed
13861 static data, but will not affect the generated code in any other way.
13862 You will normally not enable this option; instead, a language processor
13863 that needs this handling would enable it on your behalf.
13865 @item -fasynchronous-unwind-tables
13866 @opindex fasynchronous-unwind-tables
13867 Generate unwind table in dwarf2 format, if supported by target machine. The
13868 table is exact at each instruction boundary, so it can be used for stack
13869 unwinding from asynchronous events (such as debugger or garbage collector).
13871 @item -fpcc-struct-return
13872 @opindex fpcc-struct-return
13873 Return ``short'' @code{struct} and @code{union} values in memory like
13874 longer ones, rather than in registers. This convention is less
13875 efficient, but it has the advantage of allowing intercallability between
13876 GCC-compiled files and files compiled with other compilers, particularly
13877 the Portable C Compiler (pcc).
13879 The precise convention for returning structures in memory depends
13880 on the target configuration macros.
13882 Short structures and unions are those whose size and alignment match
13883 that of some integer type.
13885 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13886 switch is not binary compatible with code compiled with the
13887 @option{-freg-struct-return} switch.
13888 Use it to conform to a non-default application binary interface.
13890 @item -freg-struct-return
13891 @opindex freg-struct-return
13892 Return @code{struct} and @code{union} values in registers when possible.
13893 This is more efficient for small structures than
13894 @option{-fpcc-struct-return}.
13896 If you specify neither @option{-fpcc-struct-return} nor
13897 @option{-freg-struct-return}, GCC defaults to whichever convention is
13898 standard for the target. If there is no standard convention, GCC
13899 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13900 the principal compiler. In those cases, we can choose the standard, and
13901 we chose the more efficient register return alternative.
13903 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13904 switch is not binary compatible with code compiled with the
13905 @option{-fpcc-struct-return} switch.
13906 Use it to conform to a non-default application binary interface.
13908 @item -fshort-enums
13909 @opindex fshort-enums
13910 Allocate to an @code{enum} type only as many bytes as it needs for the
13911 declared range of possible values. Specifically, the @code{enum} type
13912 will be equivalent to the smallest integer type which has enough room.
13914 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13915 code that is not binary compatible with code generated without that switch.
13916 Use it to conform to a non-default application binary interface.
13918 @item -fshort-double
13919 @opindex fshort-double
13920 Use the same size for @code{double} as for @code{float}.
13922 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13923 code that is not binary compatible with code generated without that switch.
13924 Use it to conform to a non-default application binary interface.
13926 @item -fshort-wchar
13927 @opindex fshort-wchar
13928 Override the underlying type for @samp{wchar_t} to be @samp{short
13929 unsigned int} instead of the default for the target. This option is
13930 useful for building programs to run under WINE@.
13932 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13933 code that is not binary compatible with code generated without that switch.
13934 Use it to conform to a non-default application binary interface.
13937 @opindex fno-common
13938 In C, allocate even uninitialized global variables in the data section of the
13939 object file, rather than generating them as common blocks. This has the
13940 effect that if the same variable is declared (without @code{extern}) in
13941 two different compilations, you will get an error when you link them.
13942 The only reason this might be useful is if you wish to verify that the
13943 program will work on other systems which always work this way.
13947 Ignore the @samp{#ident} directive.
13949 @item -finhibit-size-directive
13950 @opindex finhibit-size-directive
13951 Don't output a @code{.size} assembler directive, or anything else that
13952 would cause trouble if the function is split in the middle, and the
13953 two halves are placed at locations far apart in memory. This option is
13954 used when compiling @file{crtstuff.c}; you should not need to use it
13957 @item -fverbose-asm
13958 @opindex fverbose-asm
13959 Put extra commentary information in the generated assembly code to
13960 make it more readable. This option is generally only of use to those
13961 who actually need to read the generated assembly code (perhaps while
13962 debugging the compiler itself).
13964 @option{-fno-verbose-asm}, the default, causes the
13965 extra information to be omitted and is useful when comparing two assembler
13968 @item -frecord-gcc-switches
13969 @opindex frecord-gcc-switches
13970 This switch causes the command line that was used to invoke the
13971 compiler to be recorded into the object file that is being created.
13972 This switch is only implemented on some targets and the exact format
13973 of the recording is target and binary file format dependent, but it
13974 usually takes the form of a section containing ASCII text. This
13975 switch is related to the @option{-fverbose-asm} switch, but that
13976 switch only records information in the assembler output file as
13977 comments, so it never reaches the object file.
13981 @cindex global offset table
13983 Generate position-independent code (PIC) suitable for use in a shared
13984 library, if supported for the target machine. Such code accesses all
13985 constant addresses through a global offset table (GOT)@. The dynamic
13986 loader resolves the GOT entries when the program starts (the dynamic
13987 loader is not part of GCC; it is part of the operating system). If
13988 the GOT size for the linked executable exceeds a machine-specific
13989 maximum size, you get an error message from the linker indicating that
13990 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13991 instead. (These maximums are 8k on the SPARC and 32k
13992 on the m68k and RS/6000. The 386 has no such limit.)
13994 Position-independent code requires special support, and therefore works
13995 only on certain machines. For the 386, GCC supports PIC for System V
13996 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13997 position-independent.
13999 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14004 If supported for the target machine, emit position-independent code,
14005 suitable for dynamic linking and avoiding any limit on the size of the
14006 global offset table. This option makes a difference on the m68k,
14007 PowerPC and SPARC@.
14009 Position-independent code requires special support, and therefore works
14010 only on certain machines.
14012 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14019 These options are similar to @option{-fpic} and @option{-fPIC}, but
14020 generated position independent code can be only linked into executables.
14021 Usually these options are used when @option{-pie} GCC option will be
14022 used during linking.
14024 @item -fno-jump-tables
14025 @opindex fno-jump-tables
14026 Do not use jump tables for switch statements even where it would be
14027 more efficient than other code generation strategies. This option is
14028 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14029 building code which forms part of a dynamic linker and cannot
14030 reference the address of a jump table. On some targets, jump tables
14031 do not require a GOT and this option is not needed.
14033 @item -ffixed-@var{reg}
14035 Treat the register named @var{reg} as a fixed register; generated code
14036 should never refer to it (except perhaps as a stack pointer, frame
14037 pointer or in some other fixed role).
14039 @var{reg} must be the name of a register. The register names accepted
14040 are machine-specific and are defined in the @code{REGISTER_NAMES}
14041 macro in the machine description macro file.
14043 This flag does not have a negative form, because it specifies a
14046 @item -fcall-used-@var{reg}
14047 @opindex fcall-used
14048 Treat the register named @var{reg} as an allocable register that is
14049 clobbered by function calls. It may be allocated for temporaries or
14050 variables that do not live across a call. Functions compiled this way
14051 will not save and restore the register @var{reg}.
14053 It is an error to used this flag with the frame pointer or stack pointer.
14054 Use of this flag for other registers that have fixed pervasive roles in
14055 the machine's execution model will produce disastrous results.
14057 This flag does not have a negative form, because it specifies a
14060 @item -fcall-saved-@var{reg}
14061 @opindex fcall-saved
14062 Treat the register named @var{reg} as an allocable register saved by
14063 functions. It may be allocated even for temporaries or variables that
14064 live across a call. Functions compiled this way will save and restore
14065 the register @var{reg} if they use it.
14067 It is an error to used this flag with the frame pointer or stack pointer.
14068 Use of this flag for other registers that have fixed pervasive roles in
14069 the machine's execution model will produce disastrous results.
14071 A different sort of disaster will result from the use of this flag for
14072 a register in which function values may be returned.
14074 This flag does not have a negative form, because it specifies a
14077 @item -fpack-struct[=@var{n}]
14078 @opindex fpack-struct
14079 Without a value specified, pack all structure members together without
14080 holes. When a value is specified (which must be a small power of two), pack
14081 structure members according to this value, representing the maximum
14082 alignment (that is, objects with default alignment requirements larger than
14083 this will be output potentially unaligned at the next fitting location.
14085 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14086 code that is not binary compatible with code generated without that switch.
14087 Additionally, it makes the code suboptimal.
14088 Use it to conform to a non-default application binary interface.
14090 @item -finstrument-functions
14091 @opindex finstrument-functions
14092 Generate instrumentation calls for entry and exit to functions. Just
14093 after function entry and just before function exit, the following
14094 profiling functions will be called with the address of the current
14095 function and its call site. (On some platforms,
14096 @code{__builtin_return_address} does not work beyond the current
14097 function, so the call site information may not be available to the
14098 profiling functions otherwise.)
14101 void __cyg_profile_func_enter (void *this_fn,
14103 void __cyg_profile_func_exit (void *this_fn,
14107 The first argument is the address of the start of the current function,
14108 which may be looked up exactly in the symbol table.
14110 This instrumentation is also done for functions expanded inline in other
14111 functions. The profiling calls will indicate where, conceptually, the
14112 inline function is entered and exited. This means that addressable
14113 versions of such functions must be available. If all your uses of a
14114 function are expanded inline, this may mean an additional expansion of
14115 code size. If you use @samp{extern inline} in your C code, an
14116 addressable version of such functions must be provided. (This is
14117 normally the case anyways, but if you get lucky and the optimizer always
14118 expands the functions inline, you might have gotten away without
14119 providing static copies.)
14121 A function may be given the attribute @code{no_instrument_function}, in
14122 which case this instrumentation will not be done. This can be used, for
14123 example, for the profiling functions listed above, high-priority
14124 interrupt routines, and any functions from which the profiling functions
14125 cannot safely be called (perhaps signal handlers, if the profiling
14126 routines generate output or allocate memory).
14128 @item -fstack-check
14129 @opindex fstack-check
14130 Generate code to verify that you do not go beyond the boundary of the
14131 stack. You should specify this flag if you are running in an
14132 environment with multiple threads, but only rarely need to specify it in
14133 a single-threaded environment since stack overflow is automatically
14134 detected on nearly all systems if there is only one stack.
14136 Note that this switch does not actually cause checking to be done; the
14137 operating system must do that. The switch causes generation of code
14138 to ensure that the operating system sees the stack being extended.
14140 @item -fstack-limit-register=@var{reg}
14141 @itemx -fstack-limit-symbol=@var{sym}
14142 @itemx -fno-stack-limit
14143 @opindex fstack-limit-register
14144 @opindex fstack-limit-symbol
14145 @opindex fno-stack-limit
14146 Generate code to ensure that the stack does not grow beyond a certain value,
14147 either the value of a register or the address of a symbol. If the stack
14148 would grow beyond the value, a signal is raised. For most targets,
14149 the signal is raised before the stack overruns the boundary, so
14150 it is possible to catch the signal without taking special precautions.
14152 For instance, if the stack starts at absolute address @samp{0x80000000}
14153 and grows downwards, you can use the flags
14154 @option{-fstack-limit-symbol=__stack_limit} and
14155 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
14156 of 128KB@. Note that this may only work with the GNU linker.
14158 @cindex aliasing of parameters
14159 @cindex parameters, aliased
14160 @item -fargument-alias
14161 @itemx -fargument-noalias
14162 @itemx -fargument-noalias-global
14163 @itemx -fargument-noalias-anything
14164 @opindex fargument-alias
14165 @opindex fargument-noalias
14166 @opindex fargument-noalias-global
14167 @opindex fargument-noalias-anything
14168 Specify the possible relationships among parameters and between
14169 parameters and global data.
14171 @option{-fargument-alias} specifies that arguments (parameters) may
14172 alias each other and may alias global storage.@*
14173 @option{-fargument-noalias} specifies that arguments do not alias
14174 each other, but may alias global storage.@*
14175 @option{-fargument-noalias-global} specifies that arguments do not
14176 alias each other and do not alias global storage.
14177 @option{-fargument-noalias-anything} specifies that arguments do not
14178 alias any other storage.
14180 Each language will automatically use whatever option is required by
14181 the language standard. You should not need to use these options yourself.
14183 @item -fleading-underscore
14184 @opindex fleading-underscore
14185 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
14186 change the way C symbols are represented in the object file. One use
14187 is to help link with legacy assembly code.
14189 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
14190 generate code that is not binary compatible with code generated without that
14191 switch. Use it to conform to a non-default application binary interface.
14192 Not all targets provide complete support for this switch.
14194 @item -ftls-model=@var{model}
14195 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
14196 The @var{model} argument should be one of @code{global-dynamic},
14197 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
14199 The default without @option{-fpic} is @code{initial-exec}; with
14200 @option{-fpic} the default is @code{global-dynamic}.
14202 @item -fvisibility=@var{default|internal|hidden|protected}
14203 @opindex fvisibility
14204 Set the default ELF image symbol visibility to the specified option---all
14205 symbols will be marked with this unless overridden within the code.
14206 Using this feature can very substantially improve linking and
14207 load times of shared object libraries, produce more optimized
14208 code, provide near-perfect API export and prevent symbol clashes.
14209 It is @strong{strongly} recommended that you use this in any shared objects
14212 Despite the nomenclature, @code{default} always means public ie;
14213 available to be linked against from outside the shared object.
14214 @code{protected} and @code{internal} are pretty useless in real-world
14215 usage so the only other commonly used option will be @code{hidden}.
14216 The default if @option{-fvisibility} isn't specified is
14217 @code{default}, i.e., make every
14218 symbol public---this causes the same behavior as previous versions of
14221 A good explanation of the benefits offered by ensuring ELF
14222 symbols have the correct visibility is given by ``How To Write
14223 Shared Libraries'' by Ulrich Drepper (which can be found at
14224 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
14225 solution made possible by this option to marking things hidden when
14226 the default is public is to make the default hidden and mark things
14227 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
14228 and @code{__attribute__ ((visibility("default")))} instead of
14229 @code{__declspec(dllexport)} you get almost identical semantics with
14230 identical syntax. This is a great boon to those working with
14231 cross-platform projects.
14233 For those adding visibility support to existing code, you may find
14234 @samp{#pragma GCC visibility} of use. This works by you enclosing
14235 the declarations you wish to set visibility for with (for example)
14236 @samp{#pragma GCC visibility push(hidden)} and
14237 @samp{#pragma GCC visibility pop}.
14238 Bear in mind that symbol visibility should be viewed @strong{as
14239 part of the API interface contract} and thus all new code should
14240 always specify visibility when it is not the default ie; declarations
14241 only for use within the local DSO should @strong{always} be marked explicitly
14242 as hidden as so to avoid PLT indirection overheads---making this
14243 abundantly clear also aids readability and self-documentation of the code.
14244 Note that due to ISO C++ specification requirements, operator new and
14245 operator delete must always be of default visibility.
14247 Be aware that headers from outside your project, in particular system
14248 headers and headers from any other library you use, may not be
14249 expecting to be compiled with visibility other than the default. You
14250 may need to explicitly say @samp{#pragma GCC visibility push(default)}
14251 before including any such headers.
14253 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
14254 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
14255 no modifications. However, this means that calls to @samp{extern}
14256 functions with no explicit visibility will use the PLT, so it is more
14257 effective to use @samp{__attribute ((visibility))} and/or
14258 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
14259 declarations should be treated as hidden.
14261 Note that @samp{-fvisibility} does affect C++ vague linkage
14262 entities. This means that, for instance, an exception class that will
14263 be thrown between DSOs must be explicitly marked with default
14264 visibility so that the @samp{type_info} nodes will be unified between
14267 An overview of these techniques, their benefits and how to use them
14268 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
14274 @node Environment Variables
14275 @section Environment Variables Affecting GCC
14276 @cindex environment variables
14278 @c man begin ENVIRONMENT
14279 This section describes several environment variables that affect how GCC
14280 operates. Some of them work by specifying directories or prefixes to use
14281 when searching for various kinds of files. Some are used to specify other
14282 aspects of the compilation environment.
14284 Note that you can also specify places to search using options such as
14285 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
14286 take precedence over places specified using environment variables, which
14287 in turn take precedence over those specified by the configuration of GCC@.
14288 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
14289 GNU Compiler Collection (GCC) Internals}.
14294 @c @itemx LC_COLLATE
14296 @c @itemx LC_MONETARY
14297 @c @itemx LC_NUMERIC
14302 @c @findex LC_COLLATE
14303 @findex LC_MESSAGES
14304 @c @findex LC_MONETARY
14305 @c @findex LC_NUMERIC
14309 These environment variables control the way that GCC uses
14310 localization information that allow GCC to work with different
14311 national conventions. GCC inspects the locale categories
14312 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
14313 so. These locale categories can be set to any value supported by your
14314 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
14315 Kingdom encoded in UTF-8.
14317 The @env{LC_CTYPE} environment variable specifies character
14318 classification. GCC uses it to determine the character boundaries in
14319 a string; this is needed for some multibyte encodings that contain quote
14320 and escape characters that would otherwise be interpreted as a string
14323 The @env{LC_MESSAGES} environment variable specifies the language to
14324 use in diagnostic messages.
14326 If the @env{LC_ALL} environment variable is set, it overrides the value
14327 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
14328 and @env{LC_MESSAGES} default to the value of the @env{LANG}
14329 environment variable. If none of these variables are set, GCC
14330 defaults to traditional C English behavior.
14334 If @env{TMPDIR} is set, it specifies the directory to use for temporary
14335 files. GCC uses temporary files to hold the output of one stage of
14336 compilation which is to be used as input to the next stage: for example,
14337 the output of the preprocessor, which is the input to the compiler
14340 @item GCC_EXEC_PREFIX
14341 @findex GCC_EXEC_PREFIX
14342 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
14343 names of the subprograms executed by the compiler. No slash is added
14344 when this prefix is combined with the name of a subprogram, but you can
14345 specify a prefix that ends with a slash if you wish.
14347 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
14348 an appropriate prefix to use based on the pathname it was invoked with.
14350 If GCC cannot find the subprogram using the specified prefix, it
14351 tries looking in the usual places for the subprogram.
14353 The default value of @env{GCC_EXEC_PREFIX} is
14354 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
14355 the installed compiler. In many cases @var{prefix} is the value
14356 of @code{prefix} when you ran the @file{configure} script.
14358 Other prefixes specified with @option{-B} take precedence over this prefix.
14360 This prefix is also used for finding files such as @file{crt0.o} that are
14363 In addition, the prefix is used in an unusual way in finding the
14364 directories to search for header files. For each of the standard
14365 directories whose name normally begins with @samp{/usr/local/lib/gcc}
14366 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
14367 replacing that beginning with the specified prefix to produce an
14368 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
14369 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
14370 These alternate directories are searched first; the standard directories
14371 come next. If a standard directory begins with the configured
14372 @var{prefix} then the value of @var{prefix} is replaced by
14373 @env{GCC_EXEC_PREFIX} when looking for header files.
14375 @item COMPILER_PATH
14376 @findex COMPILER_PATH
14377 The value of @env{COMPILER_PATH} is a colon-separated list of
14378 directories, much like @env{PATH}. GCC tries the directories thus
14379 specified when searching for subprograms, if it can't find the
14380 subprograms using @env{GCC_EXEC_PREFIX}.
14383 @findex LIBRARY_PATH
14384 The value of @env{LIBRARY_PATH} is a colon-separated list of
14385 directories, much like @env{PATH}. When configured as a native compiler,
14386 GCC tries the directories thus specified when searching for special
14387 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
14388 using GCC also uses these directories when searching for ordinary
14389 libraries for the @option{-l} option (but directories specified with
14390 @option{-L} come first).
14394 @cindex locale definition
14395 This variable is used to pass locale information to the compiler. One way in
14396 which this information is used is to determine the character set to be used
14397 when character literals, string literals and comments are parsed in C and C++.
14398 When the compiler is configured to allow multibyte characters,
14399 the following values for @env{LANG} are recognized:
14403 Recognize JIS characters.
14405 Recognize SJIS characters.
14407 Recognize EUCJP characters.
14410 If @env{LANG} is not defined, or if it has some other value, then the
14411 compiler will use mblen and mbtowc as defined by the default locale to
14412 recognize and translate multibyte characters.
14416 Some additional environments variables affect the behavior of the
14419 @include cppenv.texi
14423 @node Precompiled Headers
14424 @section Using Precompiled Headers
14425 @cindex precompiled headers
14426 @cindex speed of compilation
14428 Often large projects have many header files that are included in every
14429 source file. The time the compiler takes to process these header files
14430 over and over again can account for nearly all of the time required to
14431 build the project. To make builds faster, GCC allows users to
14432 `precompile' a header file; then, if builds can use the precompiled
14433 header file they will be much faster.
14435 To create a precompiled header file, simply compile it as you would any
14436 other file, if necessary using the @option{-x} option to make the driver
14437 treat it as a C or C++ header file. You will probably want to use a
14438 tool like @command{make} to keep the precompiled header up-to-date when
14439 the headers it contains change.
14441 A precompiled header file will be searched for when @code{#include} is
14442 seen in the compilation. As it searches for the included file
14443 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
14444 compiler looks for a precompiled header in each directory just before it
14445 looks for the include file in that directory. The name searched for is
14446 the name specified in the @code{#include} with @samp{.gch} appended. If
14447 the precompiled header file can't be used, it is ignored.
14449 For instance, if you have @code{#include "all.h"}, and you have
14450 @file{all.h.gch} in the same directory as @file{all.h}, then the
14451 precompiled header file will be used if possible, and the original
14452 header will be used otherwise.
14454 Alternatively, you might decide to put the precompiled header file in a
14455 directory and use @option{-I} to ensure that directory is searched
14456 before (or instead of) the directory containing the original header.
14457 Then, if you want to check that the precompiled header file is always
14458 used, you can put a file of the same name as the original header in this
14459 directory containing an @code{#error} command.
14461 This also works with @option{-include}. So yet another way to use
14462 precompiled headers, good for projects not designed with precompiled
14463 header files in mind, is to simply take most of the header files used by
14464 a project, include them from another header file, precompile that header
14465 file, and @option{-include} the precompiled header. If the header files
14466 have guards against multiple inclusion, they will be skipped because
14467 they've already been included (in the precompiled header).
14469 If you need to precompile the same header file for different
14470 languages, targets, or compiler options, you can instead make a
14471 @emph{directory} named like @file{all.h.gch}, and put each precompiled
14472 header in the directory, perhaps using @option{-o}. It doesn't matter
14473 what you call the files in the directory, every precompiled header in
14474 the directory will be considered. The first precompiled header
14475 encountered in the directory that is valid for this compilation will
14476 be used; they're searched in no particular order.
14478 There are many other possibilities, limited only by your imagination,
14479 good sense, and the constraints of your build system.
14481 A precompiled header file can be used only when these conditions apply:
14485 Only one precompiled header can be used in a particular compilation.
14488 A precompiled header can't be used once the first C token is seen. You
14489 can have preprocessor directives before a precompiled header; you can
14490 even include a precompiled header from inside another header, so long as
14491 there are no C tokens before the @code{#include}.
14494 The precompiled header file must be produced for the same language as
14495 the current compilation. You can't use a C precompiled header for a C++
14499 The precompiled header file must have been produced by the same compiler
14500 binary as the current compilation is using.
14503 Any macros defined before the precompiled header is included must
14504 either be defined in the same way as when the precompiled header was
14505 generated, or must not affect the precompiled header, which usually
14506 means that they don't appear in the precompiled header at all.
14508 The @option{-D} option is one way to define a macro before a
14509 precompiled header is included; using a @code{#define} can also do it.
14510 There are also some options that define macros implicitly, like
14511 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
14514 @item If debugging information is output when using the precompiled
14515 header, using @option{-g} or similar, the same kind of debugging information
14516 must have been output when building the precompiled header. However,
14517 a precompiled header built using @option{-g} can be used in a compilation
14518 when no debugging information is being output.
14520 @item The same @option{-m} options must generally be used when building
14521 and using the precompiled header. @xref{Submodel Options},
14522 for any cases where this rule is relaxed.
14524 @item Each of the following options must be the same when building and using
14525 the precompiled header:
14527 @gccoptlist{-fexceptions -funit-at-a-time}
14530 Some other command-line options starting with @option{-f},
14531 @option{-p}, or @option{-O} must be defined in the same way as when
14532 the precompiled header was generated. At present, it's not clear
14533 which options are safe to change and which are not; the safest choice
14534 is to use exactly the same options when generating and using the
14535 precompiled header. The following are known to be safe:
14537 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
14538 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
14539 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
14544 For all of these except the last, the compiler will automatically
14545 ignore the precompiled header if the conditions aren't met. If you
14546 find an option combination that doesn't work and doesn't cause the
14547 precompiled header to be ignored, please consider filing a bug report,
14550 If you do use differing options when generating and using the
14551 precompiled header, the actual behavior will be a mixture of the
14552 behavior for the options. For instance, if you use @option{-g} to
14553 generate the precompiled header but not when using it, you may or may
14554 not get debugging information for routines in the precompiled header.
14556 @node Running Protoize
14557 @section Running Protoize
14559 The program @code{protoize} is an optional part of GCC@. You can use
14560 it to add prototypes to a program, thus converting the program to ISO
14561 C in one respect. The companion program @code{unprotoize} does the
14562 reverse: it removes argument types from any prototypes that are found.
14564 When you run these programs, you must specify a set of source files as
14565 command line arguments. The conversion programs start out by compiling
14566 these files to see what functions they define. The information gathered
14567 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
14569 After scanning comes actual conversion. The specified files are all
14570 eligible to be converted; any files they include (whether sources or
14571 just headers) are eligible as well.
14573 But not all the eligible files are converted. By default,
14574 @code{protoize} and @code{unprotoize} convert only source and header
14575 files in the current directory. You can specify additional directories
14576 whose files should be converted with the @option{-d @var{directory}}
14577 option. You can also specify particular files to exclude with the
14578 @option{-x @var{file}} option. A file is converted if it is eligible, its
14579 directory name matches one of the specified directory names, and its
14580 name within the directory has not been excluded.
14582 Basic conversion with @code{protoize} consists of rewriting most
14583 function definitions and function declarations to specify the types of
14584 the arguments. The only ones not rewritten are those for varargs
14587 @code{protoize} optionally inserts prototype declarations at the
14588 beginning of the source file, to make them available for any calls that
14589 precede the function's definition. Or it can insert prototype
14590 declarations with block scope in the blocks where undeclared functions
14593 Basic conversion with @code{unprotoize} consists of rewriting most
14594 function declarations to remove any argument types, and rewriting
14595 function definitions to the old-style pre-ISO form.
14597 Both conversion programs print a warning for any function declaration or
14598 definition that they can't convert. You can suppress these warnings
14601 The output from @code{protoize} or @code{unprotoize} replaces the
14602 original source file. The original file is renamed to a name ending
14603 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
14604 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
14605 for DOS) file already exists, then the source file is simply discarded.
14607 @code{protoize} and @code{unprotoize} both depend on GCC itself to
14608 scan the program and collect information about the functions it uses.
14609 So neither of these programs will work until GCC is installed.
14611 Here is a table of the options you can use with @code{protoize} and
14612 @code{unprotoize}. Each option works with both programs unless
14616 @item -B @var{directory}
14617 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
14618 usual directory (normally @file{/usr/local/lib}). This file contains
14619 prototype information about standard system functions. This option
14620 applies only to @code{protoize}.
14622 @item -c @var{compilation-options}
14623 Use @var{compilation-options} as the options when running @command{gcc} to
14624 produce the @samp{.X} files. The special option @option{-aux-info} is
14625 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
14627 Note that the compilation options must be given as a single argument to
14628 @code{protoize} or @code{unprotoize}. If you want to specify several
14629 @command{gcc} options, you must quote the entire set of compilation options
14630 to make them a single word in the shell.
14632 There are certain @command{gcc} arguments that you cannot use, because they
14633 would produce the wrong kind of output. These include @option{-g},
14634 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
14635 the @var{compilation-options}, they are ignored.
14638 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
14639 systems) instead of @samp{.c}. This is convenient if you are converting
14640 a C program to C++. This option applies only to @code{protoize}.
14643 Add explicit global declarations. This means inserting explicit
14644 declarations at the beginning of each source file for each function
14645 that is called in the file and was not declared. These declarations
14646 precede the first function definition that contains a call to an
14647 undeclared function. This option applies only to @code{protoize}.
14649 @item -i @var{string}
14650 Indent old-style parameter declarations with the string @var{string}.
14651 This option applies only to @code{protoize}.
14653 @code{unprotoize} converts prototyped function definitions to old-style
14654 function definitions, where the arguments are declared between the
14655 argument list and the initial @samp{@{}. By default, @code{unprotoize}
14656 uses five spaces as the indentation. If you want to indent with just
14657 one space instead, use @option{-i " "}.
14660 Keep the @samp{.X} files. Normally, they are deleted after conversion
14664 Add explicit local declarations. @code{protoize} with @option{-l} inserts
14665 a prototype declaration for each function in each block which calls the
14666 function without any declaration. This option applies only to
14670 Make no real changes. This mode just prints information about the conversions
14671 that would have been done without @option{-n}.
14674 Make no @samp{.save} files. The original files are simply deleted.
14675 Use this option with caution.
14677 @item -p @var{program}
14678 Use the program @var{program} as the compiler. Normally, the name
14679 @file{gcc} is used.
14682 Work quietly. Most warnings are suppressed.
14685 Print the version number, just like @option{-v} for @command{gcc}.
14688 If you need special compiler options to compile one of your program's
14689 source files, then you should generate that file's @samp{.X} file
14690 specially, by running @command{gcc} on that source file with the
14691 appropriate options and the option @option{-aux-info}. Then run
14692 @code{protoize} on the entire set of files. @code{protoize} will use
14693 the existing @samp{.X} file because it is newer than the source file.
14697 gcc -Dfoo=bar file1.c -aux-info file1.X
14702 You need to include the special files along with the rest in the
14703 @code{protoize} command, even though their @samp{.X} files already
14704 exist, because otherwise they won't get converted.
14706 @xref{Protoize Caveats}, for more information on how to use
14707 @code{protoize} successfully.