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} -fgnu89-inline @gol
170 -aux-info @var{filename} @gol
171 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
172 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
173 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
174 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
175 -fsigned-bitfields -fsigned-char @gol
176 -funsigned-bitfields -funsigned-char}
178 @item C++ Language Options
179 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
180 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
181 -fconserve-space -ffriend-injection @gol
182 -fno-elide-constructors @gol
183 -fno-enforce-eh-specs @gol
184 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
185 -fno-implicit-templates @gol
186 -fno-implicit-inline-templates @gol
187 -fno-implement-inlines -fms-extensions @gol
188 -fno-nonansi-builtins -fno-operator-names @gol
189 -fno-optional-diags -fpermissive @gol
190 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
191 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
192 -fno-default-inline -fvisibility-inlines-hidden @gol
193 -Wabi -Wctor-dtor-privacy @gol
194 -Wnon-virtual-dtor -Wreorder @gol
195 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
196 -Wno-non-template-friend -Wold-style-cast @gol
197 -Woverloaded-virtual -Wno-pmf-conversions @gol
200 @item Objective-C and Objective-C++ Language Options
201 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
202 Objective-C and Objective-C++ Dialects}.
203 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
204 -fgnu-runtime -fnext-runtime @gol
205 -fno-nil-receivers @gol
206 -fobjc-call-cxx-cdtors @gol
207 -fobjc-direct-dispatch @gol
208 -fobjc-exceptions @gol
210 -freplace-objc-classes @gol
213 -Wassign-intercept @gol
214 -Wno-protocol -Wselector @gol
215 -Wstrict-selector-match @gol
216 -Wundeclared-selector}
218 @item Language Independent Options
219 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
220 @gccoptlist{-fmessage-length=@var{n} @gol
221 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
222 -fdiagnostics-show-option}
224 @item Warning Options
225 @xref{Warning Options,,Options to Request or Suppress Warnings}.
226 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
227 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
228 -Wno-attributes -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
229 -Wchar-subscripts -Wclobbered -Wcomment @gol
230 -Wconversion -Wcoverage-mismatch -Wno-deprecated-declarations @gol
231 -Wdisabled-optimization -Wno-div-by-zero @gol
232 -Wempty-body -Wno-endif-labels @gol
233 -Werror -Werror=* @gol
234 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
235 -Wno-format-extra-args -Wformat-nonliteral @gol
236 -Wformat-security -Wformat-y2k @gol
237 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
238 -Wimport -Wno-import -Winit-self -Winline @gol
239 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
240 -Winvalid-pch -Wlarger-than-@var{len} -Wunsafe-loop-optimizations @gol
241 -Wlogical-op -Wlong-long @gol
242 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
243 -Wmissing-format-attribute -Wmissing-include-dirs @gol
244 -Wmissing-noreturn @gol
245 -Wno-multichar -Wnonnull -Wno-overflow @gol
246 -Woverlength-strings -Wpacked -Wpadded @gol
247 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
248 -Wredundant-decls @gol
249 -Wreturn-type -Wsequence-point -Wshadow @gol
250 -Wsign-compare -Wstack-protector @gol
251 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
252 -Wstrict-overflow -Wstrict-overflow=@var{n} @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 @gol
258 -Wvariadic-macros -Wvla @gol
259 -Wvolatile-register-var -Wwrite-strings}
261 @item C-only Warning Options
262 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
263 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
264 -Wold-style-declaration -Wold-style-definition @gol
265 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
266 -Wdeclaration-after-statement -Wpointer-sign}
268 @item Debugging Options
269 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
270 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
271 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
272 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
273 -fdump-ipa-all -fdump-ipa-cgraph @gol
275 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
276 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
277 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
278 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
280 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
283 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
284 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-nrv -fdump-tree-vect @gol
289 -fdump-tree-sink @gol
290 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-salias @gol
292 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
294 -ftree-vectorizer-verbose=@var{n} @gol
295 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
296 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
297 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
298 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
299 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
300 -ftest-coverage -ftime-report -fvar-tracking @gol
301 -g -g@var{level} -gcoff -gdwarf-2 @gol
302 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
303 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
304 -print-multi-directory -print-multi-lib @gol
305 -print-prog-name=@var{program} -print-search-dirs -Q @gol
306 -print-sysroot-headers-suffix @gol
309 @item Optimization Options
310 @xref{Optimize Options,,Options that Control Optimization}.
311 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
312 -falign-labels=@var{n} -falign-loops=@var{n} @gol
313 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
314 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
315 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
316 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
317 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
318 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
319 -fexpensive-optimizations -ffast-math -ffloat-store @gol
320 -fforce-addr -fforward-propagate -ffunction-sections @gol
321 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
322 -fcrossjumping -fif-conversion -fif-conversion2 @gol
323 -finline-functions -finline-functions-called-once @gol
324 -finline-limit=@var{n} -fkeep-inline-functions @gol
325 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
326 -fmodulo-sched -fno-branch-count-reg @gol
327 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
328 -fno-function-cse -fno-guess-branch-probability @gol
329 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
330 -funsafe-math-optimizations -funsafe-loop-optimizations @gol
331 -ffinite-math-only -fno-signed-zeros @gol
332 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
333 -fomit-frame-pointer -foptimize-register-move @gol
334 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
335 -fprofile-generate -fprofile-use @gol
336 -fregmove -frename-registers @gol
337 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
338 -frerun-cse-after-loop @gol
339 -frounding-math -frtl-abstract-sequences @gol
340 -fschedule-insns -fschedule-insns2 @gol
341 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
342 -fsched-spec-load-dangerous @gol
343 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
344 -fsched2-use-superblocks @gol
345 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
346 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
347 -fno-split-wide-types -fstack-protector -fstack-protector-all @gol
348 -fstrict-aliasing -fstrict-overflow -ftracer -fthread-jumps @gol
349 -funroll-all-loops -funroll-loops -fpeel-loops @gol
350 -fsplit-ivs-in-unroller -funswitch-loops @gol
351 -fvariable-expansion-in-unroller @gol
352 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
353 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
354 -fcheck-data-deps @gol
355 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
356 -ftree-ch -ftree-sra -ftree-ter -ftree-fre -ftree-vectorize @gol
357 -ftree-vect-loop-version -ftree-salias -fipa-pta -fweb @gol
358 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
359 --param @var{name}=@var{value}
360 -O -O0 -O1 -O2 -O3 -Os}
362 @item Preprocessor Options
363 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
364 @gccoptlist{-A@var{question}=@var{answer} @gol
365 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
366 -C -dD -dI -dM -dN @gol
367 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
368 -idirafter @var{dir} @gol
369 -include @var{file} -imacros @var{file} @gol
370 -iprefix @var{file} -iwithprefix @var{dir} @gol
371 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
372 -imultilib @var{dir} -isysroot @var{dir} @gol
373 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
374 -P -fworking-directory -remap @gol
375 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
376 -Xpreprocessor @var{option}}
378 @item Assembler Option
379 @xref{Assembler Options,,Passing Options to the Assembler}.
380 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
383 @xref{Link Options,,Options for Linking}.
384 @gccoptlist{@var{object-file-name} -l@var{library} @gol
385 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
386 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
387 -Wl,@var{option} -Xlinker @var{option} @gol
390 @item Directory Options
391 @xref{Directory Options,,Options for Directory Search}.
392 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
393 -specs=@var{file} -I- --sysroot=@var{dir}}
396 @c I wrote this xref this way to avoid overfull hbox. -- rms
397 @xref{Target Options}.
398 @gccoptlist{-V @var{version} -b @var{machine}}
400 @item Machine Dependent Options
401 @xref{Submodel Options,,Hardware Models and Configurations}.
402 @c This list is ordered alphanumerically by subsection name.
403 @c Try and put the significant identifier (CPU or system) first,
404 @c so users have a clue at guessing where the ones they want will be.
407 @gccoptlist{-EB -EL @gol
408 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
409 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
412 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
413 -mabi=@var{name} @gol
414 -mapcs-stack-check -mno-apcs-stack-check @gol
415 -mapcs-float -mno-apcs-float @gol
416 -mapcs-reentrant -mno-apcs-reentrant @gol
417 -msched-prolog -mno-sched-prolog @gol
418 -mlittle-endian -mbig-endian -mwords-little-endian @gol
419 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
420 -mthumb-interwork -mno-thumb-interwork @gol
421 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
422 -mstructure-size-boundary=@var{n} @gol
423 -mabort-on-noreturn @gol
424 -mlong-calls -mno-long-calls @gol
425 -msingle-pic-base -mno-single-pic-base @gol
426 -mpic-register=@var{reg} @gol
427 -mnop-fun-dllimport @gol
428 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
429 -mpoke-function-name @gol
431 -mtpcs-frame -mtpcs-leaf-frame @gol
432 -mcaller-super-interworking -mcallee-super-interworking @gol
436 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
437 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
439 @emph{Blackfin Options}
440 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
441 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
442 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
443 -mno-id-shared-library -mshared-library-id=@var{n} @gol
444 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
445 -msep-data -mno-sep-data -mlong-calls -mno-long-calls}
448 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
449 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
450 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
451 -mstack-align -mdata-align -mconst-align @gol
452 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
453 -melf -maout -melinux -mlinux -sim -sim2 @gol
454 -mmul-bug-workaround -mno-mul-bug-workaround}
457 @gccoptlist{-mmac -mpush-args}
459 @emph{Darwin Options}
460 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
461 -arch_only -bind_at_load -bundle -bundle_loader @gol
462 -client_name -compatibility_version -current_version @gol
464 -dependency-file -dylib_file -dylinker_install_name @gol
465 -dynamic -dynamiclib -exported_symbols_list @gol
466 -filelist -flat_namespace -force_cpusubtype_ALL @gol
467 -force_flat_namespace -headerpad_max_install_names @gol
469 -image_base -init -install_name -keep_private_externs @gol
470 -multi_module -multiply_defined -multiply_defined_unused @gol
471 -noall_load -no_dead_strip_inits_and_terms @gol
472 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
473 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
474 -private_bundle -read_only_relocs -sectalign @gol
475 -sectobjectsymbols -whyload -seg1addr @gol
476 -sectcreate -sectobjectsymbols -sectorder @gol
477 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
478 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
479 -segprot -segs_read_only_addr -segs_read_write_addr @gol
480 -single_module -static -sub_library -sub_umbrella @gol
481 -twolevel_namespace -umbrella -undefined @gol
482 -unexported_symbols_list -weak_reference_mismatches @gol
483 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
484 -mkernel -mone-byte-bool}
486 @emph{DEC Alpha Options}
487 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
488 -mieee -mieee-with-inexact -mieee-conformant @gol
489 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
490 -mtrap-precision=@var{mode} -mbuild-constants @gol
491 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
492 -mbwx -mmax -mfix -mcix @gol
493 -mfloat-vax -mfloat-ieee @gol
494 -mexplicit-relocs -msmall-data -mlarge-data @gol
495 -msmall-text -mlarge-text @gol
496 -mmemory-latency=@var{time}}
498 @emph{DEC Alpha/VMS Options}
499 @gccoptlist{-mvms-return-codes}
502 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
503 -mhard-float -msoft-float @gol
504 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
505 -mdouble -mno-double @gol
506 -mmedia -mno-media -mmuladd -mno-muladd @gol
507 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
508 -mlinked-fp -mlong-calls -malign-labels @gol
509 -mlibrary-pic -macc-4 -macc-8 @gol
510 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
511 -moptimize-membar -mno-optimize-membar @gol
512 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
513 -mvliw-branch -mno-vliw-branch @gol
514 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
515 -mno-nested-cond-exec -mtomcat-stats @gol
519 @emph{GNU/Linux Options}
520 @gccoptlist{-muclibc}
522 @emph{H8/300 Options}
523 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
526 @gccoptlist{-march=@var{architecture-type} @gol
527 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
528 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
529 -mfixed-range=@var{register-range} @gol
530 -mjump-in-delay -mlinker-opt -mlong-calls @gol
531 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
532 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
533 -mno-jump-in-delay -mno-long-load-store @gol
534 -mno-portable-runtime -mno-soft-float @gol
535 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
536 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
537 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
538 -munix=@var{unix-std} -nolibdld -static -threads}
540 @emph{i386 and x86-64 Options}
541 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
542 -mfpmath=@var{unit} @gol
543 -masm=@var{dialect} -mno-fancy-math-387 @gol
544 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
545 -mno-wide-multiply -mrtd -malign-double @gol
546 -mpreferred-stack-boundary=@var{num} -mcx16 -msahf @gol
547 -mmmx -msse -msse2 -msse3 -mssse3 -msse4a -m3dnow -mpopcnt -mabm @gol
548 -mthreads -mno-align-stringops -minline-all-stringops @gol
549 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
550 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
552 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
553 -mcmodel=@var{code-model} @gol
554 -m32 -m64 -mlarge-data-threshold=@var{num}}
557 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
558 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
559 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
560 -minline-float-divide-max-throughput @gol
561 -minline-int-divide-min-latency @gol
562 -minline-int-divide-max-throughput @gol
563 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
564 -mno-dwarf2-asm -mearly-stop-bits @gol
565 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
566 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
567 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
568 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
569 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
570 -mno-sched-prefer-non-data-spec-insns @gol
571 -mno-sched-prefer-non-control-spec-insns @gol
572 -mno-sched-count-spec-in-critical-path}
574 @emph{M32R/D Options}
575 @gccoptlist{-m32r2 -m32rx -m32r @gol
577 -malign-loops -mno-align-loops @gol
578 -missue-rate=@var{number} @gol
579 -mbranch-cost=@var{number} @gol
580 -mmodel=@var{code-size-model-type} @gol
581 -msdata=@var{sdata-type} @gol
582 -mno-flush-func -mflush-func=@var{name} @gol
583 -mno-flush-trap -mflush-trap=@var{number} @gol
587 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
589 @emph{M680x0 Options}
590 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
591 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
592 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
593 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
594 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
595 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
596 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
597 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
599 @emph{M68hc1x Options}
600 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
601 -mauto-incdec -minmax -mlong-calls -mshort @gol
602 -msoft-reg-count=@var{count}}
605 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
606 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
607 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
608 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
609 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
612 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
613 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
614 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
615 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
616 -mfp32 -mfp64 -mhard-float -msoft-float @gol
617 -msingle-float -mdouble-float -mdsp -mdspr2 -mpaired-single -mips3d @gol
618 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
619 -G@var{num} -membedded-data -mno-embedded-data @gol
620 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
621 -msplit-addresses -mno-split-addresses @gol
622 -mexplicit-relocs -mno-explicit-relocs @gol
623 -mcheck-zero-division -mno-check-zero-division @gol
624 -mdivide-traps -mdivide-breaks @gol
625 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
626 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
627 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
628 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
629 -mfix-sb1 -mno-fix-sb1 @gol
630 -mflush-func=@var{func} -mno-flush-func @gol
631 -mbranch-likely -mno-branch-likely @gol
632 -mfp-exceptions -mno-fp-exceptions @gol
633 -mvr4130-align -mno-vr4130-align}
636 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
637 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
638 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
639 -mno-base-addresses -msingle-exit -mno-single-exit}
641 @emph{MN10300 Options}
642 @gccoptlist{-mmult-bug -mno-mult-bug @gol
643 -mam33 -mno-am33 @gol
644 -mam33-2 -mno-am33-2 @gol
645 -mreturn-pointer-on-d0 @gol
649 @gccoptlist{-mno-crt0 -mbacc -msim @gol
650 -march=@var{cpu-type} }
652 @emph{PDP-11 Options}
653 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
654 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
655 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
656 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
657 -mbranch-expensive -mbranch-cheap @gol
658 -msplit -mno-split -munix-asm -mdec-asm}
660 @emph{PowerPC Options}
661 See RS/6000 and PowerPC Options.
663 @emph{RS/6000 and PowerPC Options}
664 @gccoptlist{-mcpu=@var{cpu-type} @gol
665 -mtune=@var{cpu-type} @gol
666 -mpower -mno-power -mpower2 -mno-power2 @gol
667 -mpowerpc -mpowerpc64 -mno-powerpc @gol
668 -maltivec -mno-altivec @gol
669 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
670 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
671 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
672 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mdfp -mno-dfp @gol
673 -mnew-mnemonics -mold-mnemonics @gol
674 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
675 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
676 -malign-power -malign-natural @gol
677 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
678 -mstring -mno-string -mupdate -mno-update @gol
679 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
680 -mstrict-align -mno-strict-align -mrelocatable @gol
681 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
682 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
683 -mdynamic-no-pic -maltivec -mswdiv @gol
684 -mprioritize-restricted-insns=@var{priority} @gol
685 -msched-costly-dep=@var{dependence_type} @gol
686 -minsert-sched-nops=@var{scheme} @gol
687 -mcall-sysv -mcall-netbsd @gol
688 -maix-struct-return -msvr4-struct-return @gol
689 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
690 -misel -mno-isel @gol
691 -misel=yes -misel=no @gol
693 -mspe=yes -mspe=no @gol
694 -mvrsave -mno-vrsave @gol
695 -mmulhw -mno-mulhw @gol
696 -mdlmzb -mno-dlmzb @gol
697 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
698 -mprototype -mno-prototype @gol
699 -msim -mmvme -mads -myellowknife -memb -msdata @gol
700 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
702 @emph{S/390 and zSeries Options}
703 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
704 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
705 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
706 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
707 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
708 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
709 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
712 @gccoptlist{-mel -mel @gol
717 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
718 -m4-nofpu -m4-single-only -m4-single -m4 @gol
719 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
720 -m5-64media -m5-64media-nofpu @gol
721 -m5-32media -m5-32media-nofpu @gol
722 -m5-compact -m5-compact-nofpu @gol
723 -mb -ml -mdalign -mrelax @gol
724 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
725 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
726 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
727 -mdivsi3_libfunc=@var{name} @gol
728 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
732 @gccoptlist{-mcpu=@var{cpu-type} @gol
733 -mtune=@var{cpu-type} @gol
734 -mcmodel=@var{code-model} @gol
735 -m32 -m64 -mapp-regs -mno-app-regs @gol
736 -mfaster-structs -mno-faster-structs @gol
737 -mfpu -mno-fpu -mhard-float -msoft-float @gol
738 -mhard-quad-float -msoft-quad-float @gol
739 -mimpure-text -mno-impure-text -mlittle-endian @gol
740 -mstack-bias -mno-stack-bias @gol
741 -munaligned-doubles -mno-unaligned-doubles @gol
742 -mv8plus -mno-v8plus -mvis -mno-vis
743 -threads -pthreads -pthread}
746 @gccoptlist{-mwarn-reloc -merror-reloc @gol
747 -msafe-dma -munsafe-dma @gol
749 -msmall-mem -mlarge-mem -mstdmain @gol
750 -mfixed-range=@var{register-range}}
752 @emph{System V Options}
753 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
755 @emph{TMS320C3x/C4x Options}
756 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
757 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
758 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
759 -mparallel-insns -mparallel-mpy -mpreserve-float}
762 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
763 -mprolog-function -mno-prolog-function -mspace @gol
764 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
765 -mapp-regs -mno-app-regs @gol
766 -mdisable-callt -mno-disable-callt @gol
772 @gccoptlist{-mg -mgnu -munix}
774 @emph{VxWorks Options}
775 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
776 -Xbind-lazy -Xbind-now}
778 @emph{x86-64 Options}
779 See i386 and x86-64 Options.
781 @emph{Xstormy16 Options}
784 @emph{Xtensa Options}
785 @gccoptlist{-mconst16 -mno-const16 @gol
786 -mfused-madd -mno-fused-madd @gol
787 -mtext-section-literals -mno-text-section-literals @gol
788 -mtarget-align -mno-target-align @gol
789 -mlongcalls -mno-longcalls}
791 @emph{zSeries Options}
792 See S/390 and zSeries Options.
794 @item Code Generation Options
795 @xref{Code Gen Options,,Options for Code Generation Conventions}.
796 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
797 -ffixed-@var{reg} -fexceptions @gol
798 -fnon-call-exceptions -funwind-tables @gol
799 -fasynchronous-unwind-tables @gol
800 -finhibit-size-directive -finstrument-functions @gol
801 -fno-common -fno-ident @gol
802 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
803 -fno-jump-tables @gol
804 -frecord-gcc-switches @gol
805 -freg-struct-return -fshort-enums @gol
806 -fshort-double -fshort-wchar @gol
807 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
808 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
809 -fargument-alias -fargument-noalias @gol
810 -fargument-noalias-global -fargument-noalias-anything
811 -fleading-underscore -ftls-model=@var{model} @gol
812 -ftrapv -fwrapv -fbounds-check @gol
817 * Overall Options:: Controlling the kind of output:
818 an executable, object files, assembler files,
819 or preprocessed source.
820 * C Dialect Options:: Controlling the variant of C language compiled.
821 * C++ Dialect Options:: Variations on C++.
822 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
824 * Language Independent Options:: Controlling how diagnostics should be
826 * Warning Options:: How picky should the compiler be?
827 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
828 * Optimize Options:: How much optimization?
829 * Preprocessor Options:: Controlling header files and macro definitions.
830 Also, getting dependency information for Make.
831 * Assembler Options:: Passing options to the assembler.
832 * Link Options:: Specifying libraries and so on.
833 * Directory Options:: Where to find header files and libraries.
834 Where to find the compiler executable files.
835 * Spec Files:: How to pass switches to sub-processes.
836 * Target Options:: Running a cross-compiler, or an old version of GCC.
839 @node Overall Options
840 @section Options Controlling the Kind of Output
842 Compilation can involve up to four stages: preprocessing, compilation
843 proper, assembly and linking, always in that order. GCC is capable of
844 preprocessing and compiling several files either into several
845 assembler input files, or into one assembler input file; then each
846 assembler input file produces an object file, and linking combines all
847 the object files (those newly compiled, and those specified as input)
848 into an executable file.
850 @cindex file name suffix
851 For any given input file, the file name suffix determines what kind of
856 C source code which must be preprocessed.
859 C source code which should not be preprocessed.
862 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.
869 Objective-C source code which should not be preprocessed.
873 Objective-C++ source code. Note that you must link with the @file{libobjc}
874 library to make an Objective-C++ program work. Note that @samp{.M} refers
875 to a literal capital M@.
878 Objective-C++ source code which should not be preprocessed.
881 C, C++, Objective-C or Objective-C++ header file to be turned into a
886 @itemx @var{file}.cxx
887 @itemx @var{file}.cpp
888 @itemx @var{file}.CPP
889 @itemx @var{file}.c++
891 C++ source code which must be preprocessed. Note that in @samp{.cxx},
892 the last two letters must both be literally @samp{x}. Likewise,
893 @samp{.C} refers to a literal capital C@.
897 Objective-C++ source code which must be preprocessed.
900 Objective-C++ source code which should not be preprocessed.
904 C++ header file to be turned into a precompiled header.
907 @itemx @var{file}.for
908 @itemx @var{file}.FOR
909 Fixed form Fortran source code which should not be preprocessed.
912 @itemx @var{file}.fpp
913 @itemx @var{file}.FPP
914 Fixed form Fortran source code which must be preprocessed (with the traditional
918 @itemx @var{file}.f95
919 Free form Fortran source code which should not be preprocessed.
922 @itemx @var{file}.F95
923 Free form Fortran source code which must be preprocessed (with the
924 traditional preprocessor).
926 @c FIXME: Descriptions of Java file types.
933 Ada source code file which contains a library unit declaration (a
934 declaration of a package, subprogram, or generic, or a generic
935 instantiation), or a library unit renaming declaration (a package,
936 generic, or subprogram renaming declaration). Such files are also
939 @itemx @var{file}.adb
940 Ada source code file containing a library unit body (a subprogram or
941 package body). Such files are also called @dfn{bodies}.
943 @c GCC also knows about some suffixes for languages not yet included:
954 Assembler code which must be preprocessed.
957 An object file to be fed straight into linking.
958 Any file name with no recognized suffix is treated this way.
962 You can specify the input language explicitly with the @option{-x} option:
965 @item -x @var{language}
966 Specify explicitly the @var{language} for the following input files
967 (rather than letting the compiler choose a default based on the file
968 name suffix). This option applies to all following input files until
969 the next @option{-x} option. Possible values for @var{language} are:
971 c c-header c-cpp-output
972 c++ c++-header c++-cpp-output
973 objective-c objective-c-header objective-c-cpp-output
974 objective-c++ objective-c++-header objective-c++-cpp-output
975 assembler assembler-with-cpp
983 Turn off any specification of a language, so that subsequent files are
984 handled according to their file name suffixes (as they are if @option{-x}
985 has not been used at all).
987 @item -pass-exit-codes
988 @opindex pass-exit-codes
989 Normally the @command{gcc} program will exit with the code of 1 if any
990 phase of the compiler returns a non-success return code. If you specify
991 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
992 numerically highest error produced by any phase that returned an error
993 indication. The C, C++, and Fortran frontends return 4, if an internal
994 compiler error is encountered.
997 If you only want some of the stages of compilation, you can use
998 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
999 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1000 @command{gcc} is to stop. Note that some combinations (for example,
1001 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1006 Compile or assemble the source files, but do not link. The linking
1007 stage simply is not done. The ultimate output is in the form of an
1008 object file for each source file.
1010 By default, the object file name for a source file is made by replacing
1011 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1013 Unrecognized input files, not requiring compilation or assembly, are
1018 Stop after the stage of compilation proper; do not assemble. The output
1019 is in the form of an assembler code file for each non-assembler input
1022 By default, the assembler file name for a source file is made by
1023 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1025 Input files that don't require compilation are ignored.
1029 Stop after the preprocessing stage; do not run the compiler proper. The
1030 output is in the form of preprocessed source code, which is sent to the
1033 Input files which don't require preprocessing are ignored.
1035 @cindex output file option
1038 Place output in file @var{file}. This applies regardless to whatever
1039 sort of output is being produced, whether it be an executable file,
1040 an object file, an assembler file or preprocessed C code.
1042 If @option{-o} is not specified, the default is to put an executable
1043 file in @file{a.out}, the object file for
1044 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1045 assembler file in @file{@var{source}.s}, a precompiled header file in
1046 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1051 Print (on standard error output) the commands executed to run the stages
1052 of compilation. Also print the version number of the compiler driver
1053 program and of the preprocessor and the compiler proper.
1057 Like @option{-v} except the commands are not executed and all command
1058 arguments are quoted. This is useful for shell scripts to capture the
1059 driver-generated command lines.
1063 Use pipes rather than temporary files for communication between the
1064 various stages of compilation. This fails to work on some systems where
1065 the assembler is unable to read from a pipe; but the GNU assembler has
1070 If you are compiling multiple source files, this option tells the driver
1071 to pass all the source files to the compiler at once (for those
1072 languages for which the compiler can handle this). This will allow
1073 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1074 language for which this is supported is C@. If you pass source files for
1075 multiple languages to the driver, using this option, the driver will invoke
1076 the compiler(s) that support IMA once each, passing each compiler all the
1077 source files appropriate for it. For those languages that do not support
1078 IMA this option will be ignored, and the compiler will be invoked once for
1079 each source file in that language. If you use this option in conjunction
1080 with @option{-save-temps}, the compiler will generate multiple
1082 (one for each source file), but only one (combined) @file{.o} or
1087 Print (on the standard output) a description of the command line options
1088 understood by @command{gcc}. If the @option{-v} option is also specified
1089 then @option{--help} will also be passed on to the various processes
1090 invoked by @command{gcc}, so that they can display the command line options
1091 they accept. If the @option{-Wextra} option has also been specified
1092 (prior to the @option{--help} option), then command line options which
1093 have no documentation associated with them will also be displayed.
1096 @opindex target-help
1097 Print (on the standard output) a description of target-specific command
1098 line options for each tool.
1100 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1101 Print (on the standard output) a description of the command line
1102 options understood by the compiler that fit into a specific class.
1103 The class can be one of @var{optimizers}, @var{warnings}, @var{target}
1107 @item @var{optimizers}
1108 This will display all of the optimization options supported by the
1111 @item @var{warnings}
1112 This will display all of the options controlling warning messages
1113 produced by the compiler.
1116 This will display target-specific options. Unlike the
1117 @option{--target-help} option however, target-specific options of the
1118 linker and assembler will not be displayed. This is because those
1119 tools do not currently support the extended @option{--help=} syntax.
1122 This will display the values recognized by the @option{--param}
1126 It is possible to further refine the output of the @option{--help=}
1127 option by adding a comma separated list of qualifiers after the
1128 class. These can be any from the following list:
1132 Display only those options which are undocumented.
1135 Display options which take an argument that appears after an equal
1136 sign in the same continuous piece of text, such as:
1137 @samp{--help=target}.
1140 Display options which take an argument that appears as a separate word
1141 following the original option, such as: @samp{-o output-file}.
1144 Thus for example to display all the undocumented target-specific
1145 switches supported by the compiler the following can be used:
1148 --help=target,undocumented
1151 The sense of a qualifier can be inverted by prefixing it with the
1152 @var{^} character, so for example to display all binary warning
1153 options (i.e. ones that are either on or off and that do not take an
1154 argument), which have a description the following can be used:
1157 --help=warnings,^joined,^undocumented
1160 A class can also be used as a qualifier, although this usually
1161 restricts the output by so much that there is nothing to display. One
1162 case where it does work however is when one of the classes is
1163 @var{target}. So for example to display all the target-specific
1164 optimization options the following can be used:
1167 --help=target,optimizers
1170 The @option{--help=} option can be repeated on the command line. Each
1171 successive use will display its requested class of options, skipping
1172 those that have already been displayed.
1174 If the @option{-Q} option appears on the command line before the
1175 @option{--help=} option, then the descriptive text displayed by
1176 @option{--help=} is changed. Instead of describing the displayed
1177 options, an indication is given as to whether the option is enabled,
1178 disabled or set to a specific value (assuming that the compiler
1179 knows this at the point where the @option{--help=} option is used).
1181 Here is a truncated example from the ARM port of @command{gcc}:
1184 % gcc -Q -mabi=2 --help=target -c
1185 The following options are target specific:
1187 -mabort-on-noreturn [disabled]
1191 The output is sensitive to the effects of previous command line
1192 options, so for example it is possible to find out which optimizations
1193 are enabled at @option{-O2} by using:
1196 -O2 --help=optimizers
1199 Alternatively you can discover which binary optimizations are enabled
1200 by @option{-O3} by using:
1203 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1204 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1205 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1210 Display the version number and copyrights of the invoked GCC@.
1212 @include @value{srcdir}/../libiberty/at-file.texi
1216 @section Compiling C++ Programs
1218 @cindex suffixes for C++ source
1219 @cindex C++ source file suffixes
1220 C++ source files conventionally use one of the suffixes @samp{.C},
1221 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1222 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1223 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1224 files with these names and compiles them as C++ programs even if you
1225 call the compiler the same way as for compiling C programs (usually
1226 with the name @command{gcc}).
1230 However, the use of @command{gcc} does not add the C++ library.
1231 @command{g++} is a program that calls GCC and treats @samp{.c},
1232 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1233 files unless @option{-x} is used, and automatically specifies linking
1234 against the C++ library. This program is also useful when
1235 precompiling a C header file with a @samp{.h} extension for use in C++
1236 compilations. On many systems, @command{g++} is also installed with
1237 the name @command{c++}.
1239 @cindex invoking @command{g++}
1240 When you compile C++ programs, you may specify many of the same
1241 command-line options that you use for compiling programs in any
1242 language; or command-line options meaningful for C and related
1243 languages; or options that are meaningful only for C++ programs.
1244 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1245 explanations of options for languages related to C@.
1246 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1247 explanations of options that are meaningful only for C++ programs.
1249 @node C Dialect Options
1250 @section Options Controlling C Dialect
1251 @cindex dialect options
1252 @cindex language dialect options
1253 @cindex options, dialect
1255 The following options control the dialect of C (or languages derived
1256 from C, such as C++, Objective-C and Objective-C++) that the compiler
1260 @cindex ANSI support
1264 In C mode, support all ISO C90 programs. In C++ mode,
1265 remove GNU extensions that conflict with ISO C++.
1267 This turns off certain features of GCC that are incompatible with ISO
1268 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1269 such as the @code{asm} and @code{typeof} keywords, and
1270 predefined macros such as @code{unix} and @code{vax} that identify the
1271 type of system you are using. It also enables the undesirable and
1272 rarely used ISO trigraph feature. For the C compiler,
1273 it disables recognition of C++ style @samp{//} comments as well as
1274 the @code{inline} keyword.
1276 The alternate keywords @code{__asm__}, @code{__extension__},
1277 @code{__inline__} and @code{__typeof__} continue to work despite
1278 @option{-ansi}. You would not want to use them in an ISO C program, of
1279 course, but it is useful to put them in header files that might be included
1280 in compilations done with @option{-ansi}. Alternate predefined macros
1281 such as @code{__unix__} and @code{__vax__} are also available, with or
1282 without @option{-ansi}.
1284 The @option{-ansi} option does not cause non-ISO programs to be
1285 rejected gratuitously. For that, @option{-pedantic} is required in
1286 addition to @option{-ansi}. @xref{Warning Options}.
1288 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1289 option is used. Some header files may notice this macro and refrain
1290 from declaring certain functions or defining certain macros that the
1291 ISO standard doesn't call for; this is to avoid interfering with any
1292 programs that might use these names for other things.
1294 Functions which would normally be built in but do not have semantics
1295 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1296 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1297 built-in functions provided by GCC}, for details of the functions
1302 Determine the language standard. This option is currently only
1303 supported when compiling C or C++. A value for this option must be
1304 provided; possible values are
1309 ISO C90 (same as @option{-ansi}).
1311 @item iso9899:199409
1312 ISO C90 as modified in amendment 1.
1318 ISO C99. Note that this standard is not yet fully supported; see
1319 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1320 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1323 Default, ISO C90 plus GNU extensions (including some C99 features).
1327 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1328 this will become the default. The name @samp{gnu9x} is deprecated.
1331 The 1998 ISO C++ standard plus amendments.
1334 The same as @option{-std=c++98} plus GNU extensions. This is the
1335 default for C++ code.
1338 The working draft of the upcoming ISO C++0x standard. This option
1339 enables experimental features that are likely to be included in
1340 C++0x. The working draft is constantly changing, and any feature that is
1341 enabled by this flag may be removed from future versions of GCC if it is
1342 not part of the C++0x standard.
1345 The same as @option{-std=c++0x} plus GNU extensions. As with
1346 @option{-std=c++0x}, this option enables experimental features that may
1347 be removed in future versions of GCC.
1350 Even when this option is not specified, you can still use some of the
1351 features of newer standards in so far as they do not conflict with
1352 previous C standards. For example, you may use @code{__restrict__} even
1353 when @option{-std=c99} is not specified.
1355 The @option{-std} options specifying some version of ISO C have the same
1356 effects as @option{-ansi}, except that features that were not in ISO C90
1357 but are in the specified version (for example, @samp{//} comments and
1358 the @code{inline} keyword in ISO C99) are not disabled.
1360 @xref{Standards,,Language Standards Supported by GCC}, for details of
1361 these standard versions.
1363 @item -fgnu89-inline
1364 @opindex fgnu89-inline
1365 The option @option{-fgnu89-inline} tells GCC to use the traditional
1366 GNU semantics for @code{inline} functions when in C99 mode.
1367 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1368 is accepted and ignored by GCC versions 4.1.3 up to but not including
1369 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1370 C99 mode. Using this option is roughly equivalent to adding the
1371 @code{gnu_inline} function attribute to all inline functions
1372 (@pxref{Function Attributes}).
1374 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1375 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1376 specifies the default behavior). This option was first supported in
1377 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1379 The preprocesor macros @code{__GNUC_GNU_INLINE__} and
1380 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1381 in effect for @code{inline} functions. @xref{Common Predefined
1382 Macros,,,cpp,The C Preprocessor}.
1384 @item -aux-info @var{filename}
1386 Output to the given filename prototyped declarations for all functions
1387 declared and/or defined in a translation unit, including those in header
1388 files. This option is silently ignored in any language other than C@.
1390 Besides declarations, the file indicates, in comments, the origin of
1391 each declaration (source file and line), whether the declaration was
1392 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1393 @samp{O} for old, respectively, in the first character after the line
1394 number and the colon), and whether it came from a declaration or a
1395 definition (@samp{C} or @samp{F}, respectively, in the following
1396 character). In the case of function definitions, a K&R-style list of
1397 arguments followed by their declarations is also provided, inside
1398 comments, after the declaration.
1402 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1403 keyword, so that code can use these words as identifiers. You can use
1404 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1405 instead. @option{-ansi} implies @option{-fno-asm}.
1407 In C++, this switch only affects the @code{typeof} keyword, since
1408 @code{asm} and @code{inline} are standard keywords. You may want to
1409 use the @option{-fno-gnu-keywords} flag instead, which has the same
1410 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1411 switch only affects the @code{asm} and @code{typeof} keywords, since
1412 @code{inline} is a standard keyword in ISO C99.
1415 @itemx -fno-builtin-@var{function}
1416 @opindex fno-builtin
1417 @cindex built-in functions
1418 Don't recognize built-in functions that do not begin with
1419 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1420 functions provided by GCC}, for details of the functions affected,
1421 including those which are not built-in functions when @option{-ansi} or
1422 @option{-std} options for strict ISO C conformance are used because they
1423 do not have an ISO standard meaning.
1425 GCC normally generates special code to handle certain built-in functions
1426 more efficiently; for instance, calls to @code{alloca} may become single
1427 instructions that adjust the stack directly, and calls to @code{memcpy}
1428 may become inline copy loops. The resulting code is often both smaller
1429 and faster, but since the function calls no longer appear as such, you
1430 cannot set a breakpoint on those calls, nor can you change the behavior
1431 of the functions by linking with a different library. In addition,
1432 when a function is recognized as a built-in function, GCC may use
1433 information about that function to warn about problems with calls to
1434 that function, or to generate more efficient code, even if the
1435 resulting code still contains calls to that function. For example,
1436 warnings are given with @option{-Wformat} for bad calls to
1437 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1438 known not to modify global memory.
1440 With the @option{-fno-builtin-@var{function}} option
1441 only the built-in function @var{function} is
1442 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1443 function is named this is not built-in in this version of GCC, this
1444 option is ignored. There is no corresponding
1445 @option{-fbuiltin-@var{function}} option; if you wish to enable
1446 built-in functions selectively when using @option{-fno-builtin} or
1447 @option{-ffreestanding}, you may define macros such as:
1450 #define abs(n) __builtin_abs ((n))
1451 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1456 @cindex hosted environment
1458 Assert that compilation takes place in a hosted environment. This implies
1459 @option{-fbuiltin}. A hosted environment is one in which the
1460 entire standard library is available, and in which @code{main} has a return
1461 type of @code{int}. Examples are nearly everything except a kernel.
1462 This is equivalent to @option{-fno-freestanding}.
1464 @item -ffreestanding
1465 @opindex ffreestanding
1466 @cindex hosted environment
1468 Assert that compilation takes place in a freestanding environment. This
1469 implies @option{-fno-builtin}. A freestanding environment
1470 is one in which the standard library may not exist, and program startup may
1471 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1472 This is equivalent to @option{-fno-hosted}.
1474 @xref{Standards,,Language Standards Supported by GCC}, for details of
1475 freestanding and hosted environments.
1479 @cindex openmp parallel
1480 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1481 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1482 compiler generates parallel code according to the OpenMP Application
1483 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1485 @item -fms-extensions
1486 @opindex fms-extensions
1487 Accept some non-standard constructs used in Microsoft header files.
1489 Some cases of unnamed fields in structures and unions are only
1490 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1491 fields within structs/unions}, for details.
1495 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1496 options for strict ISO C conformance) implies @option{-trigraphs}.
1498 @item -no-integrated-cpp
1499 @opindex no-integrated-cpp
1500 Performs a compilation in two passes: preprocessing and compiling. This
1501 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1502 @option{-B} option. The user supplied compilation step can then add in
1503 an additional preprocessing step after normal preprocessing but before
1504 compiling. The default is to use the integrated cpp (internal cpp)
1506 The semantics of this option will change if "cc1", "cc1plus", and
1507 "cc1obj" are merged.
1509 @cindex traditional C language
1510 @cindex C language, traditional
1512 @itemx -traditional-cpp
1513 @opindex traditional-cpp
1514 @opindex traditional
1515 Formerly, these options caused GCC to attempt to emulate a pre-standard
1516 C compiler. They are now only supported with the @option{-E} switch.
1517 The preprocessor continues to support a pre-standard mode. See the GNU
1518 CPP manual for details.
1520 @item -fcond-mismatch
1521 @opindex fcond-mismatch
1522 Allow conditional expressions with mismatched types in the second and
1523 third arguments. The value of such an expression is void. This option
1524 is not supported for C++.
1526 @item -flax-vector-conversions
1527 @opindex flax-vector-conversions
1528 Allow implicit conversions between vectors with differing numbers of
1529 elements and/or incompatible element types. This option should not be
1532 @item -funsigned-char
1533 @opindex funsigned-char
1534 Let the type @code{char} be unsigned, like @code{unsigned char}.
1536 Each kind of machine has a default for what @code{char} should
1537 be. It is either like @code{unsigned char} by default or like
1538 @code{signed char} by default.
1540 Ideally, a portable program should always use @code{signed char} or
1541 @code{unsigned char} when it depends on the signedness of an object.
1542 But many programs have been written to use plain @code{char} and
1543 expect it to be signed, or expect it to be unsigned, depending on the
1544 machines they were written for. This option, and its inverse, let you
1545 make such a program work with the opposite default.
1547 The type @code{char} is always a distinct type from each of
1548 @code{signed char} or @code{unsigned char}, even though its behavior
1549 is always just like one of those two.
1552 @opindex fsigned-char
1553 Let the type @code{char} be signed, like @code{signed char}.
1555 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1556 the negative form of @option{-funsigned-char}. Likewise, the option
1557 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1559 @item -fsigned-bitfields
1560 @itemx -funsigned-bitfields
1561 @itemx -fno-signed-bitfields
1562 @itemx -fno-unsigned-bitfields
1563 @opindex fsigned-bitfields
1564 @opindex funsigned-bitfields
1565 @opindex fno-signed-bitfields
1566 @opindex fno-unsigned-bitfields
1567 These options control whether a bit-field is signed or unsigned, when the
1568 declaration does not use either @code{signed} or @code{unsigned}. By
1569 default, such a bit-field is signed, because this is consistent: the
1570 basic integer types such as @code{int} are signed types.
1573 @node C++ Dialect Options
1574 @section Options Controlling C++ Dialect
1576 @cindex compiler options, C++
1577 @cindex C++ options, command line
1578 @cindex options, C++
1579 This section describes the command-line options that are only meaningful
1580 for C++ programs; but you can also use most of the GNU compiler options
1581 regardless of what language your program is in. For example, you
1582 might compile a file @code{firstClass.C} like this:
1585 g++ -g -frepo -O -c firstClass.C
1589 In this example, only @option{-frepo} is an option meant
1590 only for C++ programs; you can use the other options with any
1591 language supported by GCC@.
1593 Here is a list of options that are @emph{only} for compiling C++ programs:
1597 @item -fabi-version=@var{n}
1598 @opindex fabi-version
1599 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1600 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1601 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1602 the version that conforms most closely to the C++ ABI specification.
1603 Therefore, the ABI obtained using version 0 will change as ABI bugs
1606 The default is version 2.
1608 @item -fno-access-control
1609 @opindex fno-access-control
1610 Turn off all access checking. This switch is mainly useful for working
1611 around bugs in the access control code.
1615 Check that the pointer returned by @code{operator new} is non-null
1616 before attempting to modify the storage allocated. This check is
1617 normally unnecessary because the C++ standard specifies that
1618 @code{operator new} will only return @code{0} if it is declared
1619 @samp{throw()}, in which case the compiler will always check the
1620 return value even without this option. In all other cases, when
1621 @code{operator new} has a non-empty exception specification, memory
1622 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1623 @samp{new (nothrow)}.
1625 @item -fconserve-space
1626 @opindex fconserve-space
1627 Put uninitialized or runtime-initialized global variables into the
1628 common segment, as C does. This saves space in the executable at the
1629 cost of not diagnosing duplicate definitions. If you compile with this
1630 flag and your program mysteriously crashes after @code{main()} has
1631 completed, you may have an object that is being destroyed twice because
1632 two definitions were merged.
1634 This option is no longer useful on most targets, now that support has
1635 been added for putting variables into BSS without making them common.
1637 @item -ffriend-injection
1638 @opindex ffriend-injection
1639 Inject friend functions into the enclosing namespace, so that they are
1640 visible outside the scope of the class in which they are declared.
1641 Friend functions were documented to work this way in the old Annotated
1642 C++ Reference Manual, and versions of G++ before 4.1 always worked
1643 that way. However, in ISO C++ a friend function which is not declared
1644 in an enclosing scope can only be found using argument dependent
1645 lookup. This option causes friends to be injected as they were in
1648 This option is for compatibility, and may be removed in a future
1651 @item -fno-elide-constructors
1652 @opindex fno-elide-constructors
1653 The C++ standard allows an implementation to omit creating a temporary
1654 which is only used to initialize another object of the same type.
1655 Specifying this option disables that optimization, and forces G++ to
1656 call the copy constructor in all cases.
1658 @item -fno-enforce-eh-specs
1659 @opindex fno-enforce-eh-specs
1660 Don't generate code to check for violation of exception specifications
1661 at runtime. This option violates the C++ standard, but may be useful
1662 for reducing code size in production builds, much like defining
1663 @samp{NDEBUG}. This does not give user code permission to throw
1664 exceptions in violation of the exception specifications; the compiler
1665 will still optimize based on the specifications, so throwing an
1666 unexpected exception will result in undefined behavior.
1669 @itemx -fno-for-scope
1671 @opindex fno-for-scope
1672 If @option{-ffor-scope} is specified, the scope of variables declared in
1673 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1674 as specified by the C++ standard.
1675 If @option{-fno-for-scope} is specified, the scope of variables declared in
1676 a @i{for-init-statement} extends to the end of the enclosing scope,
1677 as was the case in old versions of G++, and other (traditional)
1678 implementations of C++.
1680 The default if neither flag is given to follow the standard,
1681 but to allow and give a warning for old-style code that would
1682 otherwise be invalid, or have different behavior.
1684 @item -fno-gnu-keywords
1685 @opindex fno-gnu-keywords
1686 Do not recognize @code{typeof} as a keyword, so that code can use this
1687 word as an identifier. You can use the keyword @code{__typeof__} instead.
1688 @option{-ansi} implies @option{-fno-gnu-keywords}.
1690 @item -fno-implicit-templates
1691 @opindex fno-implicit-templates
1692 Never emit code for non-inline templates which are instantiated
1693 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1694 @xref{Template Instantiation}, for more information.
1696 @item -fno-implicit-inline-templates
1697 @opindex fno-implicit-inline-templates
1698 Don't emit code for implicit instantiations of inline templates, either.
1699 The default is to handle inlines differently so that compiles with and
1700 without optimization will need the same set of explicit instantiations.
1702 @item -fno-implement-inlines
1703 @opindex fno-implement-inlines
1704 To save space, do not emit out-of-line copies of inline functions
1705 controlled by @samp{#pragma implementation}. This will cause linker
1706 errors if these functions are not inlined everywhere they are called.
1708 @item -fms-extensions
1709 @opindex fms-extensions
1710 Disable pedantic warnings about constructs used in MFC, such as implicit
1711 int and getting a pointer to member function via non-standard syntax.
1713 @item -fno-nonansi-builtins
1714 @opindex fno-nonansi-builtins
1715 Disable built-in declarations of functions that are not mandated by
1716 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1717 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1719 @item -fno-operator-names
1720 @opindex fno-operator-names
1721 Do not treat the operator name keywords @code{and}, @code{bitand},
1722 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1723 synonyms as keywords.
1725 @item -fno-optional-diags
1726 @opindex fno-optional-diags
1727 Disable diagnostics that the standard says a compiler does not need to
1728 issue. Currently, the only such diagnostic issued by G++ is the one for
1729 a name having multiple meanings within a class.
1732 @opindex fpermissive
1733 Downgrade some diagnostics about nonconformant code from errors to
1734 warnings. Thus, using @option{-fpermissive} will allow some
1735 nonconforming code to compile.
1739 Enable automatic template instantiation at link time. This option also
1740 implies @option{-fno-implicit-templates}. @xref{Template
1741 Instantiation}, for more information.
1745 Disable generation of information about every class with virtual
1746 functions for use by the C++ runtime type identification features
1747 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1748 of the language, you can save some space by using this flag. Note that
1749 exception handling uses the same information, but it will generate it as
1750 needed. The @samp{dynamic_cast} operator can still be used for casts that
1751 do not require runtime type information, i.e. casts to @code{void *} or to
1752 unambiguous base classes.
1756 Emit statistics about front-end processing at the end of the compilation.
1757 This information is generally only useful to the G++ development team.
1759 @item -ftemplate-depth-@var{n}
1760 @opindex ftemplate-depth
1761 Set the maximum instantiation depth for template classes to @var{n}.
1762 A limit on the template instantiation depth is needed to detect
1763 endless recursions during template class instantiation. ANSI/ISO C++
1764 conforming programs must not rely on a maximum depth greater than 17.
1766 @item -fno-threadsafe-statics
1767 @opindex fno-threadsafe-statics
1768 Do not emit the extra code to use the routines specified in the C++
1769 ABI for thread-safe initialization of local statics. You can use this
1770 option to reduce code size slightly in code that doesn't need to be
1773 @item -fuse-cxa-atexit
1774 @opindex fuse-cxa-atexit
1775 Register destructors for objects with static storage duration with the
1776 @code{__cxa_atexit} function rather than the @code{atexit} function.
1777 This option is required for fully standards-compliant handling of static
1778 destructors, but will only work if your C library supports
1779 @code{__cxa_atexit}.
1781 @item -fno-use-cxa-get-exception-ptr
1782 @opindex fno-use-cxa-get-exception-ptr
1783 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1784 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1785 if the runtime routine is not available.
1787 @item -fvisibility-inlines-hidden
1788 @opindex fvisibility-inlines-hidden
1789 This switch declares that the user does not attempt to compare
1790 pointers to inline methods where the addresses of the two functions
1791 were taken in different shared objects.
1793 The effect of this is that GCC may, effectively, mark inline methods with
1794 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1795 appear in the export table of a DSO and do not require a PLT indirection
1796 when used within the DSO@. Enabling this option can have a dramatic effect
1797 on load and link times of a DSO as it massively reduces the size of the
1798 dynamic export table when the library makes heavy use of templates.
1800 The behavior of this switch is not quite the same as marking the
1801 methods as hidden directly, because it does not affect static variables
1802 local to the function or cause the compiler to deduce that
1803 the function is defined in only one shared object.
1805 You may mark a method as having a visibility explicitly to negate the
1806 effect of the switch for that method. For example, if you do want to
1807 compare pointers to a particular inline method, you might mark it as
1808 having default visibility. Marking the enclosing class with explicit
1809 visibility will have no effect.
1811 Explicitly instantiated inline methods are unaffected by this option
1812 as their linkage might otherwise cross a shared library boundary.
1813 @xref{Template Instantiation}.
1817 Do not use weak symbol support, even if it is provided by the linker.
1818 By default, G++ will use weak symbols if they are available. This
1819 option exists only for testing, and should not be used by end-users;
1820 it will result in inferior code and has no benefits. This option may
1821 be removed in a future release of G++.
1825 Do not search for header files in the standard directories specific to
1826 C++, but do still search the other standard directories. (This option
1827 is used when building the C++ library.)
1830 In addition, these optimization, warning, and code generation options
1831 have meanings only for C++ programs:
1834 @item -fno-default-inline
1835 @opindex fno-default-inline
1836 Do not assume @samp{inline} for functions defined inside a class scope.
1837 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1838 functions will have linkage like inline functions; they just won't be
1841 @item -Wabi @r{(C++ only)}
1843 Warn when G++ generates code that is probably not compatible with the
1844 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1845 all such cases, there are probably some cases that are not warned about,
1846 even though G++ is generating incompatible code. There may also be
1847 cases where warnings are emitted even though the code that is generated
1850 You should rewrite your code to avoid these warnings if you are
1851 concerned about the fact that code generated by G++ may not be binary
1852 compatible with code generated by other compilers.
1854 The known incompatibilities at this point include:
1859 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1860 pack data into the same byte as a base class. For example:
1863 struct A @{ virtual void f(); int f1 : 1; @};
1864 struct B : public A @{ int f2 : 1; @};
1868 In this case, G++ will place @code{B::f2} into the same byte
1869 as@code{A::f1}; other compilers will not. You can avoid this problem
1870 by explicitly padding @code{A} so that its size is a multiple of the
1871 byte size on your platform; that will cause G++ and other compilers to
1872 layout @code{B} identically.
1875 Incorrect handling of tail-padding for virtual bases. G++ does not use
1876 tail padding when laying out virtual bases. For example:
1879 struct A @{ virtual void f(); char c1; @};
1880 struct B @{ B(); char c2; @};
1881 struct C : public A, public virtual B @{@};
1885 In this case, G++ will not place @code{B} into the tail-padding for
1886 @code{A}; other compilers will. You can avoid this problem by
1887 explicitly padding @code{A} so that its size is a multiple of its
1888 alignment (ignoring virtual base classes); that will cause G++ and other
1889 compilers to layout @code{C} identically.
1892 Incorrect handling of bit-fields with declared widths greater than that
1893 of their underlying types, when the bit-fields appear in a union. For
1897 union U @{ int i : 4096; @};
1901 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1902 union too small by the number of bits in an @code{int}.
1905 Empty classes can be placed at incorrect offsets. For example:
1915 struct C : public B, public A @{@};
1919 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1920 it should be placed at offset zero. G++ mistakenly believes that the
1921 @code{A} data member of @code{B} is already at offset zero.
1924 Names of template functions whose types involve @code{typename} or
1925 template template parameters can be mangled incorrectly.
1928 template <typename Q>
1929 void f(typename Q::X) @{@}
1931 template <template <typename> class Q>
1932 void f(typename Q<int>::X) @{@}
1936 Instantiations of these templates may be mangled incorrectly.
1940 @item -Wctor-dtor-privacy @r{(C++ only)}
1941 @opindex Wctor-dtor-privacy
1942 Warn when a class seems unusable because all the constructors or
1943 destructors in that class are private, and it has neither friends nor
1944 public static member functions.
1946 @item -Wnon-virtual-dtor @r{(C++ only)}
1947 @opindex Wnon-virtual-dtor
1948 Warn when a class appears to be polymorphic, thereby requiring a virtual
1949 destructor, yet it declares a non-virtual one. This warning is also
1950 enabled if -Weffc++ is specified.
1952 @item -Wreorder @r{(C++ only)}
1954 @cindex reordering, warning
1955 @cindex warning for reordering of member initializers
1956 Warn when the order of member initializers given in the code does not
1957 match the order in which they must be executed. For instance:
1963 A(): j (0), i (1) @{ @}
1967 The compiler will rearrange the member initializers for @samp{i}
1968 and @samp{j} to match the declaration order of the members, emitting
1969 a warning to that effect. This warning is enabled by @option{-Wall}.
1972 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1975 @item -Weffc++ @r{(C++ only)}
1977 Warn about violations of the following style guidelines from Scott Meyers'
1978 @cite{Effective C++} book:
1982 Item 11: Define a copy constructor and an assignment operator for classes
1983 with dynamically allocated memory.
1986 Item 12: Prefer initialization to assignment in constructors.
1989 Item 14: Make destructors virtual in base classes.
1992 Item 15: Have @code{operator=} return a reference to @code{*this}.
1995 Item 23: Don't try to return a reference when you must return an object.
1999 Also warn about violations of the following style guidelines from
2000 Scott Meyers' @cite{More Effective C++} book:
2004 Item 6: Distinguish between prefix and postfix forms of increment and
2005 decrement operators.
2008 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2012 When selecting this option, be aware that the standard library
2013 headers do not obey all of these guidelines; use @samp{grep -v}
2014 to filter out those warnings.
2016 @item -Wno-deprecated @r{(C++ only)}
2017 @opindex Wno-deprecated
2018 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
2020 @item -Wstrict-null-sentinel @r{(C++ only)}
2021 @opindex Wstrict-null-sentinel
2022 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2023 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2024 to @code{__null}. Although it is a null pointer constant not a null pointer,
2025 it is guaranteed to of the same size as a pointer. But this use is
2026 not portable across different compilers.
2028 @item -Wno-non-template-friend @r{(C++ only)}
2029 @opindex Wno-non-template-friend
2030 Disable warnings when non-templatized friend functions are declared
2031 within a template. Since the advent of explicit template specification
2032 support in G++, if the name of the friend is an unqualified-id (i.e.,
2033 @samp{friend foo(int)}), the C++ language specification demands that the
2034 friend declare or define an ordinary, nontemplate function. (Section
2035 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2036 could be interpreted as a particular specialization of a templatized
2037 function. Because this non-conforming behavior is no longer the default
2038 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2039 check existing code for potential trouble spots and is on by default.
2040 This new compiler behavior can be turned off with
2041 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2042 but disables the helpful warning.
2044 @item -Wold-style-cast @r{(C++ only)}
2045 @opindex Wold-style-cast
2046 Warn if an old-style (C-style) cast to a non-void type is used within
2047 a C++ program. The new-style casts (@samp{dynamic_cast},
2048 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2049 less vulnerable to unintended effects and much easier to search for.
2051 @item -Woverloaded-virtual @r{(C++ only)}
2052 @opindex Woverloaded-virtual
2053 @cindex overloaded virtual fn, warning
2054 @cindex warning for overloaded virtual fn
2055 Warn when a function declaration hides virtual functions from a
2056 base class. For example, in:
2063 struct B: public A @{
2068 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2076 will fail to compile.
2078 @item -Wno-pmf-conversions @r{(C++ only)}
2079 @opindex Wno-pmf-conversions
2080 Disable the diagnostic for converting a bound pointer to member function
2083 @item -Wsign-promo @r{(C++ only)}
2084 @opindex Wsign-promo
2085 Warn when overload resolution chooses a promotion from unsigned or
2086 enumerated type to a signed type, over a conversion to an unsigned type of
2087 the same size. Previous versions of G++ would try to preserve
2088 unsignedness, but the standard mandates the current behavior.
2093 A& operator = (int);
2103 In this example, G++ will synthesize a default @samp{A& operator =
2104 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2107 @node Objective-C and Objective-C++ Dialect Options
2108 @section Options Controlling Objective-C and Objective-C++ Dialects
2110 @cindex compiler options, Objective-C and Objective-C++
2111 @cindex Objective-C and Objective-C++ options, command line
2112 @cindex options, Objective-C and Objective-C++
2113 (NOTE: This manual does not describe the Objective-C and Objective-C++
2114 languages themselves. See @xref{Standards,,Language Standards
2115 Supported by GCC}, for references.)
2117 This section describes the command-line options that are only meaningful
2118 for Objective-C and Objective-C++ programs, but you can also use most of
2119 the language-independent GNU compiler options.
2120 For example, you might compile a file @code{some_class.m} like this:
2123 gcc -g -fgnu-runtime -O -c some_class.m
2127 In this example, @option{-fgnu-runtime} is an option meant only for
2128 Objective-C and Objective-C++ programs; you can use the other options with
2129 any language supported by GCC@.
2131 Note that since Objective-C is an extension of the C language, Objective-C
2132 compilations may also use options specific to the C front-end (e.g.,
2133 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2134 C++-specific options (e.g., @option{-Wabi}).
2136 Here is a list of options that are @emph{only} for compiling Objective-C
2137 and Objective-C++ programs:
2140 @item -fconstant-string-class=@var{class-name}
2141 @opindex fconstant-string-class
2142 Use @var{class-name} as the name of the class to instantiate for each
2143 literal string specified with the syntax @code{@@"@dots{}"}. The default
2144 class name is @code{NXConstantString} if the GNU runtime is being used, and
2145 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2146 @option{-fconstant-cfstrings} option, if also present, will override the
2147 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2148 to be laid out as constant CoreFoundation strings.
2151 @opindex fgnu-runtime
2152 Generate object code compatible with the standard GNU Objective-C
2153 runtime. This is the default for most types of systems.
2155 @item -fnext-runtime
2156 @opindex fnext-runtime
2157 Generate output compatible with the NeXT runtime. This is the default
2158 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2159 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2162 @item -fno-nil-receivers
2163 @opindex fno-nil-receivers
2164 Assume that all Objective-C message dispatches (e.g.,
2165 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2166 is not @code{nil}. This allows for more efficient entry points in the runtime
2167 to be used. Currently, this option is only available in conjunction with
2168 the NeXT runtime on Mac OS X 10.3 and later.
2170 @item -fobjc-call-cxx-cdtors
2171 @opindex fobjc-call-cxx-cdtors
2172 For each Objective-C class, check if any of its instance variables is a
2173 C++ object with a non-trivial default constructor. If so, synthesize a
2174 special @code{- (id) .cxx_construct} instance method that will run
2175 non-trivial default constructors on any such instance variables, in order,
2176 and then return @code{self}. Similarly, check if any instance variable
2177 is a C++ object with a non-trivial destructor, and if so, synthesize a
2178 special @code{- (void) .cxx_destruct} method that will run
2179 all such default destructors, in reverse order.
2181 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2182 thusly generated will only operate on instance variables declared in the
2183 current Objective-C class, and not those inherited from superclasses. It
2184 is the responsibility of the Objective-C runtime to invoke all such methods
2185 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2186 will be invoked by the runtime immediately after a new object
2187 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2188 be invoked immediately before the runtime deallocates an object instance.
2190 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2191 support for invoking the @code{- (id) .cxx_construct} and
2192 @code{- (void) .cxx_destruct} methods.
2194 @item -fobjc-direct-dispatch
2195 @opindex fobjc-direct-dispatch
2196 Allow fast jumps to the message dispatcher. On Darwin this is
2197 accomplished via the comm page.
2199 @item -fobjc-exceptions
2200 @opindex fobjc-exceptions
2201 Enable syntactic support for structured exception handling in Objective-C,
2202 similar to what is offered by C++ and Java. This option is
2203 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2212 @@catch (AnObjCClass *exc) @{
2219 @@catch (AnotherClass *exc) @{
2222 @@catch (id allOthers) @{
2232 The @code{@@throw} statement may appear anywhere in an Objective-C or
2233 Objective-C++ program; when used inside of a @code{@@catch} block, the
2234 @code{@@throw} may appear without an argument (as shown above), in which case
2235 the object caught by the @code{@@catch} will be rethrown.
2237 Note that only (pointers to) Objective-C objects may be thrown and
2238 caught using this scheme. When an object is thrown, it will be caught
2239 by the nearest @code{@@catch} clause capable of handling objects of that type,
2240 analogously to how @code{catch} blocks work in C++ and Java. A
2241 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2242 any and all Objective-C exceptions not caught by previous @code{@@catch}
2245 The @code{@@finally} clause, if present, will be executed upon exit from the
2246 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2247 regardless of whether any exceptions are thrown, caught or rethrown
2248 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2249 of the @code{finally} clause in Java.
2251 There are several caveats to using the new exception mechanism:
2255 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2256 idioms provided by the @code{NSException} class, the new
2257 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2258 systems, due to additional functionality needed in the (NeXT) Objective-C
2262 As mentioned above, the new exceptions do not support handling
2263 types other than Objective-C objects. Furthermore, when used from
2264 Objective-C++, the Objective-C exception model does not interoperate with C++
2265 exceptions at this time. This means you cannot @code{@@throw} an exception
2266 from Objective-C and @code{catch} it in C++, or vice versa
2267 (i.e., @code{throw @dots{} @@catch}).
2270 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2271 blocks for thread-safe execution:
2274 @@synchronized (ObjCClass *guard) @{
2279 Upon entering the @code{@@synchronized} block, a thread of execution shall
2280 first check whether a lock has been placed on the corresponding @code{guard}
2281 object by another thread. If it has, the current thread shall wait until
2282 the other thread relinquishes its lock. Once @code{guard} becomes available,
2283 the current thread will place its own lock on it, execute the code contained in
2284 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2285 making @code{guard} available to other threads).
2287 Unlike Java, Objective-C does not allow for entire methods to be marked
2288 @code{@@synchronized}. Note that throwing exceptions out of
2289 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2290 to be unlocked properly.
2294 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2296 @item -freplace-objc-classes
2297 @opindex freplace-objc-classes
2298 Emit a special marker instructing @command{ld(1)} not to statically link in
2299 the resulting object file, and allow @command{dyld(1)} to load it in at
2300 run time instead. This is used in conjunction with the Fix-and-Continue
2301 debugging mode, where the object file in question may be recompiled and
2302 dynamically reloaded in the course of program execution, without the need
2303 to restart the program itself. Currently, Fix-and-Continue functionality
2304 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2309 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2310 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2311 compile time) with static class references that get initialized at load time,
2312 which improves run-time performance. Specifying the @option{-fzero-link} flag
2313 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2314 to be retained. This is useful in Zero-Link debugging mode, since it allows
2315 for individual class implementations to be modified during program execution.
2319 Dump interface declarations for all classes seen in the source file to a
2320 file named @file{@var{sourcename}.decl}.
2322 @item -Wassign-intercept
2323 @opindex Wassign-intercept
2324 Warn whenever an Objective-C assignment is being intercepted by the
2328 @opindex Wno-protocol
2329 If a class is declared to implement a protocol, a warning is issued for
2330 every method in the protocol that is not implemented by the class. The
2331 default behavior is to issue a warning for every method not explicitly
2332 implemented in the class, even if a method implementation is inherited
2333 from the superclass. If you use the @option{-Wno-protocol} option, then
2334 methods inherited from the superclass are considered to be implemented,
2335 and no warning is issued for them.
2339 Warn if multiple methods of different types for the same selector are
2340 found during compilation. The check is performed on the list of methods
2341 in the final stage of compilation. Additionally, a check is performed
2342 for each selector appearing in a @code{@@selector(@dots{})}
2343 expression, and a corresponding method for that selector has been found
2344 during compilation. Because these checks scan the method table only at
2345 the end of compilation, these warnings are not produced if the final
2346 stage of compilation is not reached, for example because an error is
2347 found during compilation, or because the @option{-fsyntax-only} option is
2350 @item -Wstrict-selector-match
2351 @opindex Wstrict-selector-match
2352 Warn if multiple methods with differing argument and/or return types are
2353 found for a given selector when attempting to send a message using this
2354 selector to a receiver of type @code{id} or @code{Class}. When this flag
2355 is off (which is the default behavior), the compiler will omit such warnings
2356 if any differences found are confined to types which share the same size
2359 @item -Wundeclared-selector
2360 @opindex Wundeclared-selector
2361 Warn if a @code{@@selector(@dots{})} expression referring to an
2362 undeclared selector is found. A selector is considered undeclared if no
2363 method with that name has been declared before the
2364 @code{@@selector(@dots{})} expression, either explicitly in an
2365 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2366 an @code{@@implementation} section. This option always performs its
2367 checks as soon as a @code{@@selector(@dots{})} expression is found,
2368 while @option{-Wselector} only performs its checks in the final stage of
2369 compilation. This also enforces the coding style convention
2370 that methods and selectors must be declared before being used.
2372 @item -print-objc-runtime-info
2373 @opindex print-objc-runtime-info
2374 Generate C header describing the largest structure that is passed by
2379 @node Language Independent Options
2380 @section Options to Control Diagnostic Messages Formatting
2381 @cindex options to control diagnostics formatting
2382 @cindex diagnostic messages
2383 @cindex message formatting
2385 Traditionally, diagnostic messages have been formatted irrespective of
2386 the output device's aspect (e.g.@: its width, @dots{}). The options described
2387 below can be used to control the diagnostic messages formatting
2388 algorithm, e.g.@: how many characters per line, how often source location
2389 information should be reported. Right now, only the C++ front end can
2390 honor these options. However it is expected, in the near future, that
2391 the remaining front ends would be able to digest them correctly.
2394 @item -fmessage-length=@var{n}
2395 @opindex fmessage-length
2396 Try to format error messages so that they fit on lines of about @var{n}
2397 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2398 the front ends supported by GCC@. If @var{n} is zero, then no
2399 line-wrapping will be done; each error message will appear on a single
2402 @opindex fdiagnostics-show-location
2403 @item -fdiagnostics-show-location=once
2404 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2405 reporter to emit @emph{once} source location information; that is, in
2406 case the message is too long to fit on a single physical line and has to
2407 be wrapped, the source location won't be emitted (as prefix) again,
2408 over and over, in subsequent continuation lines. This is the default
2411 @item -fdiagnostics-show-location=every-line
2412 Only meaningful in line-wrapping mode. Instructs the diagnostic
2413 messages reporter to emit the same source location information (as
2414 prefix) for physical lines that result from the process of breaking
2415 a message which is too long to fit on a single line.
2417 @item -fdiagnostics-show-option
2418 @opindex fdiagnostics-show-option
2419 This option instructs the diagnostic machinery to add text to each
2420 diagnostic emitted, which indicates which command line option directly
2421 controls that diagnostic, when such an option is known to the
2422 diagnostic machinery.
2424 @item -Wcoverage-mismatch
2425 @opindex Wcoverage-mismatch
2426 Warn if feedback profiles do not match when using the
2427 @option{-fprofile-use} option.
2428 If a source file was changed between @option{-fprofile-gen} and
2429 @option{-fprofile-use}, the files with the profile feedback can fail
2430 to match the source file and GCC can not use the profile feedback
2431 information. By default, GCC emits an error message in this case.
2432 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2433 error. GCC does not use appropriate feedback profiles, so using this
2434 option can result in poorly optimized code. This option is useful
2435 only in the case of very minor changes such as bug fixes to an
2440 @node Warning Options
2441 @section Options to Request or Suppress Warnings
2442 @cindex options to control warnings
2443 @cindex warning messages
2444 @cindex messages, warning
2445 @cindex suppressing warnings
2447 Warnings are diagnostic messages that report constructions which
2448 are not inherently erroneous but which are risky or suggest there
2449 may have been an error.
2451 You can request many specific warnings with options beginning @samp{-W},
2452 for example @option{-Wimplicit} to request warnings on implicit
2453 declarations. Each of these specific warning options also has a
2454 negative form beginning @samp{-Wno-} to turn off warnings;
2455 for example, @option{-Wno-implicit}. This manual lists only one of the
2456 two forms, whichever is not the default.
2458 The following options control the amount and kinds of warnings produced
2459 by GCC; for further, language-specific options also refer to
2460 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2464 @cindex syntax checking
2466 @opindex fsyntax-only
2467 Check the code for syntax errors, but don't do anything beyond that.
2471 Issue all the warnings demanded by strict ISO C and ISO C++;
2472 reject all programs that use forbidden extensions, and some other
2473 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2474 version of the ISO C standard specified by any @option{-std} option used.
2476 Valid ISO C and ISO C++ programs should compile properly with or without
2477 this option (though a rare few will require @option{-ansi} or a
2478 @option{-std} option specifying the required version of ISO C)@. However,
2479 without this option, certain GNU extensions and traditional C and C++
2480 features are supported as well. With this option, they are rejected.
2482 @option{-pedantic} does not cause warning messages for use of the
2483 alternate keywords whose names begin and end with @samp{__}. Pedantic
2484 warnings are also disabled in the expression that follows
2485 @code{__extension__}. However, only system header files should use
2486 these escape routes; application programs should avoid them.
2487 @xref{Alternate Keywords}.
2489 Some users try to use @option{-pedantic} to check programs for strict ISO
2490 C conformance. They soon find that it does not do quite what they want:
2491 it finds some non-ISO practices, but not all---only those for which
2492 ISO C @emph{requires} a diagnostic, and some others for which
2493 diagnostics have been added.
2495 A feature to report any failure to conform to ISO C might be useful in
2496 some instances, but would require considerable additional work and would
2497 be quite different from @option{-pedantic}. We don't have plans to
2498 support such a feature in the near future.
2500 Where the standard specified with @option{-std} represents a GNU
2501 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2502 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2503 extended dialect is based. Warnings from @option{-pedantic} are given
2504 where they are required by the base standard. (It would not make sense
2505 for such warnings to be given only for features not in the specified GNU
2506 C dialect, since by definition the GNU dialects of C include all
2507 features the compiler supports with the given option, and there would be
2508 nothing to warn about.)
2510 @item -pedantic-errors
2511 @opindex pedantic-errors
2512 Like @option{-pedantic}, except that errors are produced rather than
2517 Inhibit all warning messages.
2521 Inhibit warning messages about the use of @samp{#import}.
2523 @item -Wchar-subscripts
2524 @opindex Wchar-subscripts
2525 Warn if an array subscript has type @code{char}. This is a common cause
2526 of error, as programmers often forget that this type is signed on some
2528 This warning is enabled by @option{-Wall}.
2532 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2533 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2534 This warning is enabled by @option{-Wall}.
2536 @item -Wfatal-errors
2537 @opindex Wfatal-errors
2538 This option causes the compiler to abort compilation on the first error
2539 occurred rather than trying to keep going and printing further error
2544 @opindex ffreestanding
2545 @opindex fno-builtin
2546 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2547 the arguments supplied have types appropriate to the format string
2548 specified, and that the conversions specified in the format string make
2549 sense. This includes standard functions, and others specified by format
2550 attributes (@pxref{Function Attributes}), in the @code{printf},
2551 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2552 not in the C standard) families (or other target-specific families).
2553 Which functions are checked without format attributes having been
2554 specified depends on the standard version selected, and such checks of
2555 functions without the attribute specified are disabled by
2556 @option{-ffreestanding} or @option{-fno-builtin}.
2558 The formats are checked against the format features supported by GNU
2559 libc version 2.2. These include all ISO C90 and C99 features, as well
2560 as features from the Single Unix Specification and some BSD and GNU
2561 extensions. Other library implementations may not support all these
2562 features; GCC does not support warning about features that go beyond a
2563 particular library's limitations. However, if @option{-pedantic} is used
2564 with @option{-Wformat}, warnings will be given about format features not
2565 in the selected standard version (but not for @code{strfmon} formats,
2566 since those are not in any version of the C standard). @xref{C Dialect
2567 Options,,Options Controlling C Dialect}.
2569 Since @option{-Wformat} also checks for null format arguments for
2570 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2572 @option{-Wformat} is included in @option{-Wall}. For more control over some
2573 aspects of format checking, the options @option{-Wformat-y2k},
2574 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2575 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2576 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2579 @opindex Wformat-y2k
2580 If @option{-Wformat} is specified, also warn about @code{strftime}
2581 formats which may yield only a two-digit year.
2583 @item -Wno-format-extra-args
2584 @opindex Wno-format-extra-args
2585 If @option{-Wformat} is specified, do not warn about excess arguments to a
2586 @code{printf} or @code{scanf} format function. The C standard specifies
2587 that such arguments are ignored.
2589 Where the unused arguments lie between used arguments that are
2590 specified with @samp{$} operand number specifications, normally
2591 warnings are still given, since the implementation could not know what
2592 type to pass to @code{va_arg} to skip the unused arguments. However,
2593 in the case of @code{scanf} formats, this option will suppress the
2594 warning if the unused arguments are all pointers, since the Single
2595 Unix Specification says that such unused arguments are allowed.
2597 @item -Wno-format-zero-length
2598 @opindex Wno-format-zero-length
2599 If @option{-Wformat} is specified, do not warn about zero-length formats.
2600 The C standard specifies that zero-length formats are allowed.
2602 @item -Wformat-nonliteral
2603 @opindex Wformat-nonliteral
2604 If @option{-Wformat} is specified, also warn if the format string is not a
2605 string literal and so cannot be checked, unless the format function
2606 takes its format arguments as a @code{va_list}.
2608 @item -Wformat-security
2609 @opindex Wformat-security
2610 If @option{-Wformat} is specified, also warn about uses of format
2611 functions that represent possible security problems. At present, this
2612 warns about calls to @code{printf} and @code{scanf} functions where the
2613 format string is not a string literal and there are no format arguments,
2614 as in @code{printf (foo);}. This may be a security hole if the format
2615 string came from untrusted input and contains @samp{%n}. (This is
2616 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2617 in future warnings may be added to @option{-Wformat-security} that are not
2618 included in @option{-Wformat-nonliteral}.)
2622 Enable @option{-Wformat} plus format checks not included in
2623 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2624 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2628 Warn about passing a null pointer for arguments marked as
2629 requiring a non-null value by the @code{nonnull} function attribute.
2631 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2632 can be disabled with the @option{-Wno-nonnull} option.
2634 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2636 Warn about uninitialized variables which are initialized with themselves.
2637 Note this option can only be used with the @option{-Wuninitialized} option,
2638 which in turn only works with @option{-O1} and above.
2640 For example, GCC will warn about @code{i} being uninitialized in the
2641 following snippet only when @option{-Winit-self} has been specified:
2652 @item -Wimplicit-int
2653 @opindex Wimplicit-int
2654 Warn when a declaration does not specify a type.
2655 This warning is enabled by @option{-Wall}.
2657 @item -Wimplicit-function-declaration
2658 @opindex Wimplicit-function-declaration
2659 @opindex Wno-implicit-function-declaration
2660 Give a warning whenever a function is used before being declared. In
2661 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2662 enabled by default and it is made into an error by
2663 @option{-pedantic-errors}. This warning is also enabled by
2668 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2669 This warning is enabled by @option{-Wall}.
2673 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2674 function with external linkage, returning int, taking either zero
2675 arguments, two, or three arguments of appropriate types.
2676 This warning is enabled by @option{-Wall}.
2678 @item -Wmissing-braces
2679 @opindex Wmissing-braces
2680 Warn if an aggregate or union initializer is not fully bracketed. In
2681 the following example, the initializer for @samp{a} is not fully
2682 bracketed, but that for @samp{b} is fully bracketed.
2685 int a[2][2] = @{ 0, 1, 2, 3 @};
2686 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2689 This warning is enabled by @option{-Wall}.
2691 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2692 @opindex Wmissing-include-dirs
2693 Warn if a user-supplied include directory does not exist.
2696 @opindex Wparentheses
2697 Warn if parentheses are omitted in certain contexts, such
2698 as when there is an assignment in a context where a truth value
2699 is expected, or when operators are nested whose precedence people
2700 often get confused about.
2702 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2703 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2704 interpretation from that of ordinary mathematical notation.
2706 Also warn about constructions where there may be confusion to which
2707 @code{if} statement an @code{else} branch belongs. Here is an example of
2722 In C/C++, every @code{else} branch belongs to the innermost possible
2723 @code{if} statement, which in this example is @code{if (b)}. This is
2724 often not what the programmer expected, as illustrated in the above
2725 example by indentation the programmer chose. When there is the
2726 potential for this confusion, GCC will issue a warning when this flag
2727 is specified. To eliminate the warning, add explicit braces around
2728 the innermost @code{if} statement so there is no way the @code{else}
2729 could belong to the enclosing @code{if}. The resulting code would
2746 This warning is enabled by @option{-Wall}.
2748 @item -Wsequence-point
2749 @opindex Wsequence-point
2750 Warn about code that may have undefined semantics because of violations
2751 of sequence point rules in the C and C++ standards.
2753 The C and C++ standards defines the order in which expressions in a C/C++
2754 program are evaluated in terms of @dfn{sequence points}, which represent
2755 a partial ordering between the execution of parts of the program: those
2756 executed before the sequence point, and those executed after it. These
2757 occur after the evaluation of a full expression (one which is not part
2758 of a larger expression), after the evaluation of the first operand of a
2759 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2760 function is called (but after the evaluation of its arguments and the
2761 expression denoting the called function), and in certain other places.
2762 Other than as expressed by the sequence point rules, the order of
2763 evaluation of subexpressions of an expression is not specified. All
2764 these rules describe only a partial order rather than a total order,
2765 since, for example, if two functions are called within one expression
2766 with no sequence point between them, the order in which the functions
2767 are called is not specified. However, the standards committee have
2768 ruled that function calls do not overlap.
2770 It is not specified when between sequence points modifications to the
2771 values of objects take effect. Programs whose behavior depends on this
2772 have undefined behavior; the C and C++ standards specify that ``Between
2773 the previous and next sequence point an object shall have its stored
2774 value modified at most once by the evaluation of an expression.
2775 Furthermore, the prior value shall be read only to determine the value
2776 to be stored.''. If a program breaks these rules, the results on any
2777 particular implementation are entirely unpredictable.
2779 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2780 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2781 diagnosed by this option, and it may give an occasional false positive
2782 result, but in general it has been found fairly effective at detecting
2783 this sort of problem in programs.
2785 The standard is worded confusingly, therefore there is some debate
2786 over the precise meaning of the sequence point rules in subtle cases.
2787 Links to discussions of the problem, including proposed formal
2788 definitions, may be found on the GCC readings page, at
2789 @w{@uref{http://gcc.gnu.org/readings.html}}.
2791 This warning is enabled by @option{-Wall} for C and C++.
2794 @opindex Wreturn-type
2795 Warn whenever a function is defined with a return-type that defaults to
2796 @code{int}. Also warn about any @code{return} statement with no
2797 return-value in a function whose return-type is not @code{void}.
2799 Also warn if the return type of a function has a type qualifier
2800 such as @code{const}. For ISO C such a type qualifier has no effect,
2801 since the value returned by a function is not an lvalue.
2802 For C++, the warning is only emitted for scalar types or @code{void}.
2803 ISO C prohibits qualified @code{void} return types on function
2804 definitions, so such return types always receive a warning
2805 even without this option.
2807 For C++, a function without return type always produces a diagnostic
2808 message, even when @option{-Wno-return-type} is specified. The only
2809 exceptions are @samp{main} and functions defined in system headers.
2811 This warning is enabled by @option{-Wall}.
2815 Warn whenever a @code{switch} statement has an index of enumerated type
2816 and lacks a @code{case} for one or more of the named codes of that
2817 enumeration. (The presence of a @code{default} label prevents this
2818 warning.) @code{case} labels outside the enumeration range also
2819 provoke warnings when this option is used.
2820 This warning is enabled by @option{-Wall}.
2822 @item -Wswitch-default
2823 @opindex Wswitch-switch
2824 Warn whenever a @code{switch} statement does not have a @code{default}
2828 @opindex Wswitch-enum
2829 Warn whenever a @code{switch} statement has an index of enumerated type
2830 and lacks a @code{case} for one or more of the named codes of that
2831 enumeration. @code{case} labels outside the enumeration range also
2832 provoke warnings when this option is used.
2836 Warn if any trigraphs are encountered that might change the meaning of
2837 the program (trigraphs within comments are not warned about).
2838 This warning is enabled by @option{-Wall}.
2840 @item -Wunused-function
2841 @opindex Wunused-function
2842 Warn whenever a static function is declared but not defined or a
2843 non-inline static function is unused.
2844 This warning is enabled by @option{-Wall}.
2846 @item -Wunused-label
2847 @opindex Wunused-label
2848 Warn whenever a label is declared but not used.
2849 This warning is enabled by @option{-Wall}.
2851 To suppress this warning use the @samp{unused} attribute
2852 (@pxref{Variable Attributes}).
2854 @item -Wunused-parameter
2855 @opindex Wunused-parameter
2856 Warn whenever a function parameter is unused aside from its declaration.
2858 To suppress this warning use the @samp{unused} attribute
2859 (@pxref{Variable Attributes}).
2861 @item -Wunused-variable
2862 @opindex Wunused-variable
2863 Warn whenever a local variable or non-constant static variable is unused
2864 aside from its declaration.
2865 This warning is enabled by @option{-Wall}.
2867 To suppress this warning use the @samp{unused} attribute
2868 (@pxref{Variable Attributes}).
2870 @item -Wunused-value
2871 @opindex Wunused-value
2872 Warn whenever a statement computes a result that is explicitly not
2873 used. To suppress this warning cast the unused expression to
2874 @samp{void}. This includes an expression-statement or the left-hand
2875 side of a comma expression that contains no side effects. For example,
2876 an expression such as @samp{x[i,j]} will cause a warning, while
2877 @samp{x[(void)i,j]} will not.
2879 This warning is enabled by @option{-Wall}.
2883 All the above @option{-Wunused} options combined.
2885 In order to get a warning about an unused function parameter, you must
2886 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2887 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2889 @item -Wuninitialized
2890 @opindex Wuninitialized
2891 Warn if an automatic variable is used without first being initialized or
2892 if a variable may be clobbered by a @code{setjmp} call.
2894 These warnings are possible only in optimizing compilation,
2895 because they require data flow information that is computed only
2896 when optimizing. If you do not specify @option{-O}, you will not get
2897 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2898 requiring @option{-O}.
2900 If you want to warn about code which uses the uninitialized value of the
2901 variable in its own initializer, use the @option{-Winit-self} option.
2903 These warnings occur for individual uninitialized or clobbered
2904 elements of structure, union or array variables as well as for
2905 variables which are uninitialized or clobbered as a whole. They do
2906 not occur for variables or elements declared @code{volatile}. Because
2907 these warnings depend on optimization, the exact variables or elements
2908 for which there are warnings will depend on the precise optimization
2909 options and version of GCC used.
2911 Note that there may be no warning about a variable that is used only
2912 to compute a value that itself is never used, because such
2913 computations may be deleted by data flow analysis before the warnings
2916 These warnings are made optional because GCC is not smart
2917 enough to see all the reasons why the code might be correct
2918 despite appearing to have an error. Here is one example of how
2939 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2940 always initialized, but GCC doesn't know this. Here is
2941 another common case:
2946 if (change_y) save_y = y, y = new_y;
2948 if (change_y) y = save_y;
2953 This has no bug because @code{save_y} is used only if it is set.
2955 @cindex @code{longjmp} warnings
2956 This option also warns when a non-volatile automatic variable might be
2957 changed by a call to @code{longjmp}. These warnings as well are possible
2958 only in optimizing compilation.
2960 The compiler sees only the calls to @code{setjmp}. It cannot know
2961 where @code{longjmp} will be called; in fact, a signal handler could
2962 call it at any point in the code. As a result, you may get a warning
2963 even when there is in fact no problem because @code{longjmp} cannot
2964 in fact be called at the place which would cause a problem.
2966 Some spurious warnings can be avoided if you declare all the functions
2967 you use that never return as @code{noreturn}. @xref{Function
2970 This warning is enabled by @option{-Wall}.
2972 @item -Wunknown-pragmas
2973 @opindex Wunknown-pragmas
2974 @cindex warning for unknown pragmas
2975 @cindex unknown pragmas, warning
2976 @cindex pragmas, warning of unknown
2977 Warn when a #pragma directive is encountered which is not understood by
2978 GCC@. If this command line option is used, warnings will even be issued
2979 for unknown pragmas in system header files. This is not the case if
2980 the warnings were only enabled by the @option{-Wall} command line option.
2983 @opindex Wno-pragmas
2985 Do not warn about misuses of pragmas, such as incorrect parameters,
2986 invalid syntax, or conflicts between pragmas. See also
2987 @samp{-Wunknown-pragmas}.
2989 @item -Wstrict-aliasing
2990 @opindex Wstrict-aliasing
2991 This option is only active when @option{-fstrict-aliasing} is active.
2992 It warns about code which might break the strict aliasing rules that the
2993 compiler is using for optimization. The warning does not catch all
2994 cases, but does attempt to catch the more common pitfalls. It is
2995 included in @option{-Wall}.
2997 @item -Wstrict-aliasing=2
2998 @opindex Wstrict-aliasing=2
2999 This option is only active when @option{-fstrict-aliasing} is active.
3000 It warns about code which might break the strict aliasing rules that the
3001 compiler is using for optimization. This warning catches more cases than
3002 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
3003 cases that are safe.
3005 @item -Wstrict-overflow
3006 @item -Wstrict-overflow=@var{n}
3007 @opindex -Wstrict-overflow
3008 This option is only active when @option{-fstrict-overflow} is active.
3009 It warns about cases where the compiler optimizes based on the
3010 assumption that signed overflow does not occur. Note that it does not
3011 warn about all cases where the code might overflow: it only warns
3012 about cases where the compiler implements some optimization. Thus
3013 this warning depends on the optimization level.
3015 An optimization which assumes that signed overflow does not occur is
3016 perfectly safe if the values of the variables involved are such that
3017 overflow never does, in fact, occur. Therefore this warning can
3018 easily give a false positive: a warning about code which is not
3019 actually a problem. To help focus on important issues, several
3020 warning levels are defined.
3023 @item -Wstrict-overflow=1
3024 Warn about cases which are both questionable and easy to avoid. For
3025 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3026 compiler will simplify this to @code{1}. This level of
3027 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3028 are not, and must be explicitly requested.
3030 @item -Wstrict-overflow=2
3031 Also warn about other cases where a comparison is simplified to a
3032 constant. For example: @code{abs (x) >= 0}. This can only be
3033 simplified when @option{-fstrict-overflow} is in effect, because
3034 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3035 zero. @option{-Wstrict-overflow} (with no level) is the same as
3036 @option{-Wstrict-overflow=2}.
3038 @item -Wstrict-overflow=3
3039 Also warn about other cases where a comparison is simplified. For
3040 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3042 @item -Wstrict-overflow=4
3043 Also warn about other simplifications not covered by the above cases.
3044 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3046 @item -Wstrict-overflow=5
3047 Also warn about cases where the compiler reduces the magnitude of a
3048 constant involved in a comparison. For example: @code{x + 2 > y} will
3049 be simplified to @code{x + 1 >= y}. This is reported only at the
3050 highest warning level because this simplification applies to many
3051 comparisons, so this warning level will give a very large number of
3055 @item -Warray-bounds
3056 @opindex Wno-array-bounds
3057 @opindex Warray-bounds
3058 This option is only active when @option{-ftree-vrp} is active
3059 (default for -O2 and above). It warns about subscripts to arrays
3060 that are always out of bounds. This warning is enabled by @option{-Wall}.
3064 All of the above @samp{-W} options combined. This enables all the
3065 warnings about constructions that some users consider questionable, and
3066 that are easy to avoid (or modify to prevent the warning), even in
3067 conjunction with macros. This also enables some language-specific
3068 warnings described in @ref{C++ Dialect Options} and
3069 @ref{Objective-C and Objective-C++ Dialect Options}.
3072 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
3073 Some of them warn about constructions that users generally do not
3074 consider questionable, but which occasionally you might wish to check
3075 for; others warn about constructions that are necessary or hard to avoid
3076 in some cases, and there is no simple way to modify the code to suppress
3083 (This option used to be called @option{-W}. The older name is still
3084 supported, but the newer name is more descriptive.) Print extra warning
3085 messages for these events:
3089 A function can return either with or without a value. (Falling
3090 off the end of the function body is considered returning without
3091 a value.) For example, this function would evoke such a
3105 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
3108 Storage-class specifiers like @code{static} are not the first things
3109 in a declaration. According to the C Standard, this usage is
3110 obsolescent. This warning can be independently controlled by
3111 @option{-Wold-style-declaration}.
3114 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
3118 A comparison between signed and unsigned values could produce an
3119 incorrect result when the signed value is converted to unsigned.
3120 (But don't warn if @option{-Wno-sign-compare} is also specified.)
3123 An aggregate has an initializer which does not initialize all members.
3124 This warning can be independently controlled by
3125 @option{-Wmissing-field-initializers}.
3128 An initialized field without side effects is overridden when using
3129 designated initializers (@pxref{Designated Inits, , Designated
3130 Initializers}). This warning can be independently controlled by
3131 @option{-Woverride-init}.
3134 A function parameter is declared without a type specifier in K&R-style
3135 functions. This warning can be independently controlled by
3136 @option{-Wmissing-parameter-type}.
3139 An empty body occurs in an @samp{if} or @samp{else} statement. This
3140 warning can be independently controlled by @option{-Wempty-body}.
3143 A pointer is compared against integer zero with @samp{<}, @samp{<=},
3144 @samp{>}, or @samp{>=}.
3147 A variable might be changed by @samp{longjmp} or @samp{vfork}.
3148 This warning can be independently controlled by @option{-Wclobbered}.
3150 @item @r{(C++ only)}
3151 An enumerator and a non-enumerator both appear in a conditional expression.
3153 @item @r{(C++ only)}
3154 A non-static reference or non-static @samp{const} member appears in a
3155 class without constructors.
3157 @item @r{(C++ only)}
3158 Ambiguous virtual bases.
3160 @item @r{(C++ only)}
3161 Subscripting an array which has been declared @samp{register}.
3163 @item @r{(C++ only)}
3164 Taking the address of a variable which has been declared @samp{register}.
3166 @item @r{(C++ only)}
3167 A base class is not initialized in a derived class' copy constructor.
3170 @item -Wno-div-by-zero
3171 @opindex Wno-div-by-zero
3172 @opindex Wdiv-by-zero
3173 Do not warn about compile-time integer division by zero. Floating point
3174 division by zero is not warned about, as it can be a legitimate way of
3175 obtaining infinities and NaNs.
3177 @item -Wsystem-headers
3178 @opindex Wsystem-headers
3179 @cindex warnings from system headers
3180 @cindex system headers, warnings from
3181 Print warning messages for constructs found in system header files.
3182 Warnings from system headers are normally suppressed, on the assumption
3183 that they usually do not indicate real problems and would only make the
3184 compiler output harder to read. Using this command line option tells
3185 GCC to emit warnings from system headers as if they occurred in user
3186 code. However, note that using @option{-Wall} in conjunction with this
3187 option will @emph{not} warn about unknown pragmas in system
3188 headers---for that, @option{-Wunknown-pragmas} must also be used.
3191 @opindex Wfloat-equal
3192 Warn if floating point values are used in equality comparisons.
3194 The idea behind this is that sometimes it is convenient (for the
3195 programmer) to consider floating-point values as approximations to
3196 infinitely precise real numbers. If you are doing this, then you need
3197 to compute (by analyzing the code, or in some other way) the maximum or
3198 likely maximum error that the computation introduces, and allow for it
3199 when performing comparisons (and when producing output, but that's a
3200 different problem). In particular, instead of testing for equality, you
3201 would check to see whether the two values have ranges that overlap; and
3202 this is done with the relational operators, so equality comparisons are
3205 @item -Wtraditional @r{(C only)}
3206 @opindex Wtraditional
3207 Warn about certain constructs that behave differently in traditional and
3208 ISO C@. Also warn about ISO C constructs that have no traditional C
3209 equivalent, and/or problematic constructs which should be avoided.
3213 Macro parameters that appear within string literals in the macro body.
3214 In traditional C macro replacement takes place within string literals,
3215 but does not in ISO C@.
3218 In traditional C, some preprocessor directives did not exist.
3219 Traditional preprocessors would only consider a line to be a directive
3220 if the @samp{#} appeared in column 1 on the line. Therefore
3221 @option{-Wtraditional} warns about directives that traditional C
3222 understands but would ignore because the @samp{#} does not appear as the
3223 first character on the line. It also suggests you hide directives like
3224 @samp{#pragma} not understood by traditional C by indenting them. Some
3225 traditional implementations would not recognize @samp{#elif}, so it
3226 suggests avoiding it altogether.
3229 A function-like macro that appears without arguments.
3232 The unary plus operator.
3235 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3236 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3237 constants.) Note, these suffixes appear in macros defined in the system
3238 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3239 Use of these macros in user code might normally lead to spurious
3240 warnings, however GCC's integrated preprocessor has enough context to
3241 avoid warning in these cases.
3244 A function declared external in one block and then used after the end of
3248 A @code{switch} statement has an operand of type @code{long}.
3251 A non-@code{static} function declaration follows a @code{static} one.
3252 This construct is not accepted by some traditional C compilers.
3255 The ISO type of an integer constant has a different width or
3256 signedness from its traditional type. This warning is only issued if
3257 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3258 typically represent bit patterns, are not warned about.
3261 Usage of ISO string concatenation is detected.
3264 Initialization of automatic aggregates.
3267 Identifier conflicts with labels. Traditional C lacks a separate
3268 namespace for labels.
3271 Initialization of unions. If the initializer is zero, the warning is
3272 omitted. This is done under the assumption that the zero initializer in
3273 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3274 initializer warnings and relies on default initialization to zero in the
3278 Conversions by prototypes between fixed/floating point values and vice
3279 versa. The absence of these prototypes when compiling with traditional
3280 C would cause serious problems. This is a subset of the possible
3281 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3284 Use of ISO C style function definitions. This warning intentionally is
3285 @emph{not} issued for prototype declarations or variadic functions
3286 because these ISO C features will appear in your code when using
3287 libiberty's traditional C compatibility macros, @code{PARAMS} and
3288 @code{VPARAMS}. This warning is also bypassed for nested functions
3289 because that feature is already a GCC extension and thus not relevant to
3290 traditional C compatibility.
3293 @item -Wtraditional-conversion @r{(C only)}
3294 @opindex Wtraditional-conversion
3295 Warn if a prototype causes a type conversion that is different from what
3296 would happen to the same argument in the absence of a prototype. This
3297 includes conversions of fixed point to floating and vice versa, and
3298 conversions changing the width or signedness of a fixed point argument
3299 except when the same as the default promotion.
3301 @item -Wdeclaration-after-statement @r{(C only)}
3302 @opindex Wdeclaration-after-statement
3303 Warn when a declaration is found after a statement in a block. This
3304 construct, known from C++, was introduced with ISO C99 and is by default
3305 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3306 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3310 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3312 @item -Wno-endif-labels
3313 @opindex Wno-endif-labels
3314 @opindex Wendif-labels
3315 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3319 Warn whenever a local variable shadows another local variable, parameter or
3320 global variable or whenever a built-in function is shadowed.
3322 @item -Wlarger-than-@var{len}
3323 @opindex Wlarger-than
3324 Warn whenever an object of larger than @var{len} bytes is defined.
3326 @item -Wunsafe-loop-optimizations
3327 @opindex Wunsafe-loop-optimizations
3328 Warn if the loop cannot be optimized because the compiler could not
3329 assume anything on the bounds of the loop indices. With
3330 @option{-funsafe-loop-optimizations} warn if the compiler made
3333 @item -Wpointer-arith
3334 @opindex Wpointer-arith
3335 Warn about anything that depends on the ``size of'' a function type or
3336 of @code{void}. GNU C assigns these types a size of 1, for
3337 convenience in calculations with @code{void *} pointers and pointers
3338 to functions. In C++, warn also when an arithmetic operation involves
3339 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3341 @item -Wbad-function-cast @r{(C only)}
3342 @opindex Wbad-function-cast
3343 Warn whenever a function call is cast to a non-matching type.
3344 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3347 Warn about ISO C constructs that are outside of the common subset of
3348 ISO C and ISO C++, e.g.@: request for implicit conversion from
3349 @code{void *} to a pointer to non-@code{void} type.
3351 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3352 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3353 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3354 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3358 Warn whenever a pointer is cast so as to remove a type qualifier from
3359 the target type. For example, warn if a @code{const char *} is cast
3360 to an ordinary @code{char *}.
3363 @opindex Wcast-align
3364 Warn whenever a pointer is cast such that the required alignment of the
3365 target is increased. For example, warn if a @code{char *} is cast to
3366 an @code{int *} on machines where integers can only be accessed at
3367 two- or four-byte boundaries.
3369 @item -Wwrite-strings
3370 @opindex Wwrite-strings
3371 When compiling C, give string constants the type @code{const
3372 char[@var{length}]} so that
3373 copying the address of one into a non-@code{const} @code{char *}
3374 pointer will get a warning; when compiling C++, warn about the
3375 deprecated conversion from string literals to @code{char *}. This
3376 warning, by default, is enabled for C++ programs.
3377 These warnings will help you find at
3378 compile time code that can try to write into a string constant, but
3379 only if you have been very careful about using @code{const} in
3380 declarations and prototypes. Otherwise, it will just be a nuisance;
3381 this is why we did not make @option{-Wall} request these warnings.
3385 Warn for variables that might be changed by @samp{longjmp} or
3386 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3389 @opindex Wconversion
3390 Warn for implicit conversions that may alter a value. This includes
3391 conversions between real and integer, like @code{abs (x)} when
3392 @code{x} is @code{double}; conversions between signed and unsigned,
3393 like @code{unsigned ui = -1}; and conversions to smaller types, like
3394 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3395 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3396 changed by the conversion like in @code{abs (2.0)}.
3399 @opindex Wempty-body
3400 An empty body occurs in an @samp{if} or @samp{else} statement.
3401 This warning is also enabled by @option{-Wextra}.
3403 @item -Wsign-compare
3404 @opindex Wsign-compare
3405 @cindex warning for comparison of signed and unsigned values
3406 @cindex comparison of signed and unsigned values, warning
3407 @cindex signed and unsigned values, comparison warning
3408 Warn when a comparison between signed and unsigned values could produce
3409 an incorrect result when the signed value is converted to unsigned.
3410 This warning is also enabled by @option{-Wextra}; to get the other warnings
3411 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3415 @opindex Wno-address
3416 Warn about suspicious uses of memory addresses. These include using
3417 the address of a function in a conditional expression, such as
3418 @code{void func(void); if (func)}, and comparisons against the memory
3419 address of a string literal, such as @code{if (x == "abc")}. Such
3420 uses typically indicate a programmer error: the address of a function
3421 always evaluates to true, so their use in a conditional usually
3422 indicate that the programmer forgot the parentheses in a function
3423 call; and comparisons against string literals result in unspecified
3424 behavior and are not portable in C, so they usually indicate that the
3425 programmer intended to use @code{strcmp}. This warning is enabled by
3429 @opindex Wlogical-op
3430 @opindex Wno-logical-op
3431 Warn about suspicious uses of logical operators in expressions.
3432 This includes using logical operators in contexts where a
3433 bit-wise operator is likely to be expected.
3435 @item -Waggregate-return
3436 @opindex Waggregate-return
3437 Warn if any functions that return structures or unions are defined or
3438 called. (In languages where you can return an array, this also elicits
3441 @item -Wno-attributes
3442 @opindex Wno-attributes
3443 @opindex Wattributes
3444 Do not warn if an unexpected @code{__attribute__} is used, such as
3445 unrecognized attributes, function attributes applied to variables,
3446 etc. This will not stop errors for incorrect use of supported
3449 @item -Wstrict-prototypes @r{(C only)}
3450 @opindex Wstrict-prototypes
3451 Warn if a function is declared or defined without specifying the
3452 argument types. (An old-style function definition is permitted without
3453 a warning if preceded by a declaration which specifies the argument
3456 @item -Wold-style-declaration @r{(C only)}
3457 @opindex Wold-style-declaration
3458 Warn for obsolescent usages, according to the C Standard, in a
3459 declaration. For example, warn if storage-class specifiers like
3460 @code{static} are not the first things in a declaration. This warning
3461 is also enabled by @option{-Wextra}.
3463 @item -Wold-style-definition @r{(C only)}
3464 @opindex Wold-style-definition
3465 Warn if an old-style function definition is used. A warning is given
3466 even if there is a previous prototype.
3468 @item -Wmissing-parameter-type @r{(C only)}
3469 @opindex Wmissing-parameter-type
3470 A function parameter is declared without a type specifier in K&R-style
3477 This warning is also enabled by @option{-Wextra}.
3479 @item -Wmissing-prototypes @r{(C only)}
3480 @opindex Wmissing-prototypes
3481 Warn if a global function is defined without a previous prototype
3482 declaration. This warning is issued even if the definition itself
3483 provides a prototype. The aim is to detect global functions that fail
3484 to be declared in header files.
3486 @item -Wmissing-declarations @r{(C and C++ only)}
3487 @opindex Wmissing-declarations
3488 Warn if a global function is defined without a previous declaration.
3489 Do so even if the definition itself provides a prototype.
3490 Use this option to detect global functions that are not declared in
3491 header files. In C++, no warnings are issued for function templates,
3492 or for inline functions, or for functions in anonymous namespaces.
3494 @item -Wmissing-field-initializers
3495 @opindex Wmissing-field-initializers
3498 Warn if a structure's initializer has some fields missing. For
3499 example, the following code would cause such a warning, because
3500 @code{x.h} is implicitly zero:
3503 struct s @{ int f, g, h; @};
3504 struct s x = @{ 3, 4 @};
3507 This option does not warn about designated initializers, so the following
3508 modification would not trigger a warning:
3511 struct s @{ int f, g, h; @};
3512 struct s x = @{ .f = 3, .g = 4 @};
3515 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3516 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3518 @item -Wmissing-noreturn
3519 @opindex Wmissing-noreturn
3520 Warn about functions which might be candidates for attribute @code{noreturn}.
3521 Note these are only possible candidates, not absolute ones. Care should
3522 be taken to manually verify functions actually do not ever return before
3523 adding the @code{noreturn} attribute, otherwise subtle code generation
3524 bugs could be introduced. You will not get a warning for @code{main} in
3525 hosted C environments.
3527 @item -Wmissing-format-attribute
3528 @opindex Wmissing-format-attribute
3530 Warn about function pointers which might be candidates for @code{format}
3531 attributes. Note these are only possible candidates, not absolute ones.
3532 GCC will guess that function pointers with @code{format} attributes that
3533 are used in assignment, initialization, parameter passing or return
3534 statements should have a corresponding @code{format} attribute in the
3535 resulting type. I.e.@: the left-hand side of the assignment or
3536 initialization, the type of the parameter variable, or the return type
3537 of the containing function respectively should also have a @code{format}
3538 attribute to avoid the warning.
3540 GCC will also warn about function definitions which might be
3541 candidates for @code{format} attributes. Again, these are only
3542 possible candidates. GCC will guess that @code{format} attributes
3543 might be appropriate for any function that calls a function like
3544 @code{vprintf} or @code{vscanf}, but this might not always be the
3545 case, and some functions for which @code{format} attributes are
3546 appropriate may not be detected.
3548 @item -Wno-multichar
3549 @opindex Wno-multichar
3551 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3552 Usually they indicate a typo in the user's code, as they have
3553 implementation-defined values, and should not be used in portable code.
3555 @item -Wnormalized=<none|id|nfc|nfkc>
3556 @opindex Wnormalized
3559 @cindex character set, input normalization
3560 In ISO C and ISO C++, two identifiers are different if they are
3561 different sequences of characters. However, sometimes when characters
3562 outside the basic ASCII character set are used, you can have two
3563 different character sequences that look the same. To avoid confusion,
3564 the ISO 10646 standard sets out some @dfn{normalization rules} which
3565 when applied ensure that two sequences that look the same are turned into
3566 the same sequence. GCC can warn you if you are using identifiers which
3567 have not been normalized; this option controls that warning.
3569 There are four levels of warning that GCC supports. The default is
3570 @option{-Wnormalized=nfc}, which warns about any identifier which is
3571 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3572 recommended form for most uses.
3574 Unfortunately, there are some characters which ISO C and ISO C++ allow
3575 in identifiers that when turned into NFC aren't allowable as
3576 identifiers. That is, there's no way to use these symbols in portable
3577 ISO C or C++ and have all your identifiers in NFC.
3578 @option{-Wnormalized=id} suppresses the warning for these characters.
3579 It is hoped that future versions of the standards involved will correct
3580 this, which is why this option is not the default.
3582 You can switch the warning off for all characters by writing
3583 @option{-Wnormalized=none}. You would only want to do this if you
3584 were using some other normalization scheme (like ``D''), because
3585 otherwise you can easily create bugs that are literally impossible to see.
3587 Some characters in ISO 10646 have distinct meanings but look identical
3588 in some fonts or display methodologies, especially once formatting has
3589 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3590 LETTER N'', will display just like a regular @code{n} which has been
3591 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3592 normalization scheme to convert all these into a standard form as
3593 well, and GCC will warn if your code is not in NFKC if you use
3594 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3595 about every identifier that contains the letter O because it might be
3596 confused with the digit 0, and so is not the default, but may be
3597 useful as a local coding convention if the programming environment is
3598 unable to be fixed to display these characters distinctly.
3600 @item -Wno-deprecated-declarations
3601 @opindex Wno-deprecated-declarations
3602 Do not warn about uses of functions (@pxref{Function Attributes}),
3603 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3604 Attributes}) marked as deprecated by using the @code{deprecated}
3608 @opindex Wno-overflow
3609 Do not warn about compile-time overflow in constant expressions.
3611 @item -Woverride-init
3612 @opindex Woverride-init
3615 Warn if an initialized field without side effects is overridden when
3616 using designated initializers (@pxref{Designated Inits, , Designated
3619 This warning is included in @option{-Wextra}. To get other
3620 @option{-Wextra} warnings without this one, use @samp{-Wextra
3621 -Wno-override-init}.
3625 Warn if a structure is given the packed attribute, but the packed
3626 attribute has no effect on the layout or size of the structure.
3627 Such structures may be mis-aligned for little benefit. For
3628 instance, in this code, the variable @code{f.x} in @code{struct bar}
3629 will be misaligned even though @code{struct bar} does not itself
3630 have the packed attribute:
3637 @} __attribute__((packed));
3647 Warn if padding is included in a structure, either to align an element
3648 of the structure or to align the whole structure. Sometimes when this
3649 happens it is possible to rearrange the fields of the structure to
3650 reduce the padding and so make the structure smaller.
3652 @item -Wredundant-decls
3653 @opindex Wredundant-decls
3654 Warn if anything is declared more than once in the same scope, even in
3655 cases where multiple declaration is valid and changes nothing.
3657 @item -Wnested-externs @r{(C only)}
3658 @opindex Wnested-externs
3659 Warn if an @code{extern} declaration is encountered within a function.
3661 @item -Wunreachable-code
3662 @opindex Wunreachable-code
3663 Warn if the compiler detects that code will never be executed.
3665 This option is intended to warn when the compiler detects that at
3666 least a whole line of source code will never be executed, because
3667 some condition is never satisfied or because it is after a
3668 procedure that never returns.
3670 It is possible for this option to produce a warning even though there
3671 are circumstances under which part of the affected line can be executed,
3672 so care should be taken when removing apparently-unreachable code.
3674 For instance, when a function is inlined, a warning may mean that the
3675 line is unreachable in only one inlined copy of the function.
3677 This option is not made part of @option{-Wall} because in a debugging
3678 version of a program there is often substantial code which checks
3679 correct functioning of the program and is, hopefully, unreachable
3680 because the program does work. Another common use of unreachable
3681 code is to provide behavior which is selectable at compile-time.
3685 Warn if a function can not be inlined and it was declared as inline.
3686 Even with this option, the compiler will not warn about failures to
3687 inline functions declared in system headers.
3689 The compiler uses a variety of heuristics to determine whether or not
3690 to inline a function. For example, the compiler takes into account
3691 the size of the function being inlined and the amount of inlining
3692 that has already been done in the current function. Therefore,
3693 seemingly insignificant changes in the source program can cause the
3694 warnings produced by @option{-Winline} to appear or disappear.
3696 @item -Wno-invalid-offsetof @r{(C++ only)}
3697 @opindex Wno-invalid-offsetof
3698 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3699 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3700 to a non-POD type is undefined. In existing C++ implementations,
3701 however, @samp{offsetof} typically gives meaningful results even when
3702 applied to certain kinds of non-POD types. (Such as a simple
3703 @samp{struct} that fails to be a POD type only by virtue of having a
3704 constructor.) This flag is for users who are aware that they are
3705 writing nonportable code and who have deliberately chosen to ignore the
3708 The restrictions on @samp{offsetof} may be relaxed in a future version
3709 of the C++ standard.
3711 @item -Wno-int-to-pointer-cast @r{(C only)}
3712 @opindex Wno-int-to-pointer-cast
3713 Suppress warnings from casts to pointer type of an integer of a
3716 @item -Wno-pointer-to-int-cast @r{(C only)}
3717 @opindex Wno-pointer-to-int-cast
3718 Suppress warnings from casts from a pointer to an integer type of a
3722 @opindex Winvalid-pch
3723 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3724 the search path but can't be used.
3728 @opindex Wno-long-long
3729 Warn if @samp{long long} type is used. This is default. To inhibit
3730 the warning messages, use @option{-Wno-long-long}. Flags
3731 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3732 only when @option{-pedantic} flag is used.
3734 @item -Wvariadic-macros
3735 @opindex Wvariadic-macros
3736 @opindex Wno-variadic-macros
3737 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3738 alternate syntax when in pedantic ISO C99 mode. This is default.
3739 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3744 Warn if variable length array is used in the code.
3745 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
3746 the variable length array.
3748 @item -Wvolatile-register-var
3749 @opindex Wvolatile-register-var
3750 @opindex Wno-volatile-register-var
3751 Warn if a register variable is declared volatile. The volatile
3752 modifier does not inhibit all optimizations that may eliminate reads
3753 and/or writes to register variables.
3755 @item -Wdisabled-optimization
3756 @opindex Wdisabled-optimization
3757 Warn if a requested optimization pass is disabled. This warning does
3758 not generally indicate that there is anything wrong with your code; it
3759 merely indicates that GCC's optimizers were unable to handle the code
3760 effectively. Often, the problem is that your code is too big or too
3761 complex; GCC will refuse to optimize programs when the optimization
3762 itself is likely to take inordinate amounts of time.
3764 @item -Wpointer-sign
3765 @opindex Wpointer-sign
3766 @opindex Wno-pointer-sign
3767 Warn for pointer argument passing or assignment with different signedness.
3768 This option is only supported for C and Objective-C@. It is implied by
3769 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3770 @option{-Wno-pointer-sign}.
3774 Make all warnings into errors.
3778 Make the specified warning into an errors. The specifier for a
3779 warning is appended, for example @option{-Werror=switch} turns the
3780 warnings controlled by @option{-Wswitch} into errors. This switch
3781 takes a negative form, to be used to negate @option{-Werror} for
3782 specific warnings, for example @option{-Wno-error=switch} makes
3783 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3784 is in effect. You can use the @option{-fdiagnostics-show-option}
3785 option to have each controllable warning amended with the option which
3786 controls it, to determine what to use with this option.
3788 Note that specifying @option{-Werror=}@var{foo} automatically implies
3789 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3792 @item -Wstack-protector
3793 @opindex Wstack-protector
3794 This option is only active when @option{-fstack-protector} is active. It
3795 warns about functions that will not be protected against stack smashing.
3797 @item -Woverlength-strings
3798 @opindex Woverlength-strings
3799 Warn about string constants which are longer than the ``minimum
3800 maximum'' length specified in the C standard. Modern compilers
3801 generally allow string constants which are much longer than the
3802 standard's minimum limit, but very portable programs should avoid
3803 using longer strings.
3805 The limit applies @emph{after} string constant concatenation, and does
3806 not count the trailing NUL@. In C89, the limit was 509 characters; in
3807 C99, it was raised to 4095. C++98 does not specify a normative
3808 minimum maximum, so we do not diagnose overlength strings in C++@.
3810 This option is implied by @option{-pedantic}, and can be disabled with
3811 @option{-Wno-overlength-strings}.
3814 @node Debugging Options
3815 @section Options for Debugging Your Program or GCC
3816 @cindex options, debugging
3817 @cindex debugging information options
3819 GCC has various special options that are used for debugging
3820 either your program or GCC:
3825 Produce debugging information in the operating system's native format
3826 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3829 On most systems that use stabs format, @option{-g} enables use of extra
3830 debugging information that only GDB can use; this extra information
3831 makes debugging work better in GDB but will probably make other debuggers
3833 refuse to read the program. If you want to control for certain whether
3834 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3835 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3837 GCC allows you to use @option{-g} with
3838 @option{-O}. The shortcuts taken by optimized code may occasionally
3839 produce surprising results: some variables you declared may not exist
3840 at all; flow of control may briefly move where you did not expect it;
3841 some statements may not be executed because they compute constant
3842 results or their values were already at hand; some statements may
3843 execute in different places because they were moved out of loops.
3845 Nevertheless it proves possible to debug optimized output. This makes
3846 it reasonable to use the optimizer for programs that might have bugs.
3848 The following options are useful when GCC is generated with the
3849 capability for more than one debugging format.
3853 Produce debugging information for use by GDB@. This means to use the
3854 most expressive format available (DWARF 2, stabs, or the native format
3855 if neither of those are supported), including GDB extensions if at all
3860 Produce debugging information in stabs format (if that is supported),
3861 without GDB extensions. This is the format used by DBX on most BSD
3862 systems. On MIPS, Alpha and System V Release 4 systems this option
3863 produces stabs debugging output which is not understood by DBX or SDB@.
3864 On System V Release 4 systems this option requires the GNU assembler.
3866 @item -feliminate-unused-debug-symbols
3867 @opindex feliminate-unused-debug-symbols
3868 Produce debugging information in stabs format (if that is supported),
3869 for only symbols that are actually used.
3871 @item -femit-class-debug-always
3872 Instead of emitting debugging information for a C++ class in only one
3873 object file, emit it in all object files using the class. This option
3874 should be used only with debuggers that are unable to handle the way GCC
3875 normally emits debugging information for classes because using this
3876 option will increase the size of debugging information by as much as a
3881 Produce debugging information in stabs format (if that is supported),
3882 using GNU extensions understood only by the GNU debugger (GDB)@. The
3883 use of these extensions is likely to make other debuggers crash or
3884 refuse to read the program.
3888 Produce debugging information in COFF format (if that is supported).
3889 This is the format used by SDB on most System V systems prior to
3894 Produce debugging information in XCOFF format (if that is supported).
3895 This is the format used by the DBX debugger on IBM RS/6000 systems.
3899 Produce debugging information in XCOFF format (if that is supported),
3900 using GNU extensions understood only by the GNU debugger (GDB)@. The
3901 use of these extensions is likely to make other debuggers crash or
3902 refuse to read the program, and may cause assemblers other than the GNU
3903 assembler (GAS) to fail with an error.
3907 Produce debugging information in DWARF version 2 format (if that is
3908 supported). This is the format used by DBX on IRIX 6. With this
3909 option, GCC uses features of DWARF version 3 when they are useful;
3910 version 3 is upward compatible with version 2, but may still cause
3911 problems for older debuggers.
3915 Produce debugging information in VMS debug format (if that is
3916 supported). This is the format used by DEBUG on VMS systems.
3919 @itemx -ggdb@var{level}
3920 @itemx -gstabs@var{level}
3921 @itemx -gcoff@var{level}
3922 @itemx -gxcoff@var{level}
3923 @itemx -gvms@var{level}
3924 Request debugging information and also use @var{level} to specify how
3925 much information. The default level is 2.
3927 Level 0 produces no debug information at all. Thus, @option{-g0} negates
3930 Level 1 produces minimal information, enough for making backtraces in
3931 parts of the program that you don't plan to debug. This includes
3932 descriptions of functions and external variables, but no information
3933 about local variables and no line numbers.
3935 Level 3 includes extra information, such as all the macro definitions
3936 present in the program. Some debuggers support macro expansion when
3937 you use @option{-g3}.
3939 @option{-gdwarf-2} does not accept a concatenated debug level, because
3940 GCC used to support an option @option{-gdwarf} that meant to generate
3941 debug information in version 1 of the DWARF format (which is very
3942 different from version 2), and it would have been too confusing. That
3943 debug format is long obsolete, but the option cannot be changed now.
3944 Instead use an additional @option{-g@var{level}} option to change the
3945 debug level for DWARF2.
3947 @item -feliminate-dwarf2-dups
3948 @opindex feliminate-dwarf2-dups
3949 Compress DWARF2 debugging information by eliminating duplicated
3950 information about each symbol. This option only makes sense when
3951 generating DWARF2 debugging information with @option{-gdwarf-2}.
3953 @cindex @command{prof}
3956 Generate extra code to write profile information suitable for the
3957 analysis program @command{prof}. You must use this option when compiling
3958 the source files you want data about, and you must also use it when
3961 @cindex @command{gprof}
3964 Generate extra code to write profile information suitable for the
3965 analysis program @command{gprof}. You must use this option when compiling
3966 the source files you want data about, and you must also use it when
3971 Makes the compiler print out each function name as it is compiled, and
3972 print some statistics about each pass when it finishes.
3975 @opindex ftime-report
3976 Makes the compiler print some statistics about the time consumed by each
3977 pass when it finishes.
3980 @opindex fmem-report
3981 Makes the compiler print some statistics about permanent memory
3982 allocation when it finishes.
3984 @item -fpre-ipa-mem-report
3985 @opindex fpre-ipa-mem-report
3986 @item -fpost-ipa-mem-report
3987 @opindex fpost-ipa-mem-report
3988 Makes the compiler print some statistics about permanent memory
3989 allocation before or after interprocedural optimization.
3991 @item -fprofile-arcs
3992 @opindex fprofile-arcs
3993 Add code so that program flow @dfn{arcs} are instrumented. During
3994 execution the program records how many times each branch and call is
3995 executed and how many times it is taken or returns. When the compiled
3996 program exits it saves this data to a file called
3997 @file{@var{auxname}.gcda} for each source file. The data may be used for
3998 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3999 test coverage analysis (@option{-ftest-coverage}). Each object file's
4000 @var{auxname} is generated from the name of the output file, if
4001 explicitly specified and it is not the final executable, otherwise it is
4002 the basename of the source file. In both cases any suffix is removed
4003 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4004 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4005 @xref{Cross-profiling}.
4007 @cindex @command{gcov}
4011 This option is used to compile and link code instrumented for coverage
4012 analysis. The option is a synonym for @option{-fprofile-arcs}
4013 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4014 linking). See the documentation for those options for more details.
4019 Compile the source files with @option{-fprofile-arcs} plus optimization
4020 and code generation options. For test coverage analysis, use the
4021 additional @option{-ftest-coverage} option. You do not need to profile
4022 every source file in a program.
4025 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4026 (the latter implies the former).
4029 Run the program on a representative workload to generate the arc profile
4030 information. This may be repeated any number of times. You can run
4031 concurrent instances of your program, and provided that the file system
4032 supports locking, the data files will be correctly updated. Also
4033 @code{fork} calls are detected and correctly handled (double counting
4037 For profile-directed optimizations, compile the source files again with
4038 the same optimization and code generation options plus
4039 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4040 Control Optimization}).
4043 For test coverage analysis, use @command{gcov} to produce human readable
4044 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4045 @command{gcov} documentation for further information.
4049 With @option{-fprofile-arcs}, for each function of your program GCC
4050 creates a program flow graph, then finds a spanning tree for the graph.
4051 Only arcs that are not on the spanning tree have to be instrumented: the
4052 compiler adds code to count the number of times that these arcs are
4053 executed. When an arc is the only exit or only entrance to a block, the
4054 instrumentation code can be added to the block; otherwise, a new basic
4055 block must be created to hold the instrumentation code.
4058 @item -ftest-coverage
4059 @opindex ftest-coverage
4060 Produce a notes file that the @command{gcov} code-coverage utility
4061 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4062 show program coverage. Each source file's note file is called
4063 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4064 above for a description of @var{auxname} and instructions on how to
4065 generate test coverage data. Coverage data will match the source files
4066 more closely, if you do not optimize.
4068 @item -d@var{letters}
4069 @item -fdump-rtl-@var{pass}
4071 Says to make debugging dumps during compilation at times specified by
4072 @var{letters}. This is used for debugging the RTL-based passes of the
4073 compiler. The file names for most of the dumps are made by appending a
4074 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4075 from the name of the output file, if explicitly specified and it is not
4076 an executable, otherwise it is the basename of the source file.
4078 Most debug dumps can be enabled either passing a letter to the @option{-d}
4079 option, or with a long @option{-fdump-rtl} switch; here are the possible
4080 letters for use in @var{letters} and @var{pass}, and their meanings:
4085 Annotate the assembler output with miscellaneous debugging information.
4088 @itemx -fdump-rtl-bbro
4090 @opindex fdump-rtl-bbro
4091 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4094 @itemx -fdump-rtl-combine
4096 @opindex fdump-rtl-combine
4097 Dump after the RTL instruction combination pass, to the file
4098 @file{@var{file}.129r.combine}.
4101 @itemx -fdump-rtl-ce1
4102 @itemx -fdump-rtl-ce2
4104 @opindex fdump-rtl-ce1
4105 @opindex fdump-rtl-ce2
4106 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4107 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4108 and @option{-fdump-rtl-ce2} enable dumping after the second if
4109 conversion, to the file @file{@var{file}.130r.ce2}.
4112 @itemx -fdump-rtl-btl
4113 @itemx -fdump-rtl-dbr
4115 @opindex fdump-rtl-btl
4116 @opindex fdump-rtl-dbr
4117 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4118 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4119 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4120 scheduling, to @file{@var{file}.36.dbr}.
4124 Dump all macro definitions, at the end of preprocessing, in addition to
4128 @itemx -fdump-rtl-ce3
4130 @opindex fdump-rtl-ce3
4131 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4134 @itemx -fdump-rtl-cfg
4135 @itemx -fdump-rtl-life
4137 @opindex fdump-rtl-cfg
4138 @opindex fdump-rtl-life
4139 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4140 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4141 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4142 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4145 @itemx -fdump-rtl-greg
4147 @opindex fdump-rtl-greg
4148 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4151 @itemx -fdump-rtl-gcse
4152 @itemx -fdump-rtl-bypass
4154 @opindex fdump-rtl-gcse
4155 @opindex fdump-rtl-bypass
4156 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4157 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4158 enable dumping after jump bypassing and control flow optimizations, to
4159 @file{@var{file}.115r.bypass}.
4162 @itemx -fdump-rtl-eh
4164 @opindex fdump-rtl-eh
4165 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4168 @itemx -fdump-rtl-sibling
4170 @opindex fdump-rtl-sibling
4171 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4174 @itemx -fdump-rtl-jump
4176 @opindex fdump-rtl-jump
4177 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4180 @itemx -fdump-rtl-stack
4182 @opindex fdump-rtl-stack
4183 Dump after conversion from GCC's "flat register file" registers to the
4184 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4187 @itemx -fdump-rtl-lreg
4189 @opindex fdump-rtl-lreg
4190 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4193 @itemx -fdump-rtl-loop2
4195 @opindex fdump-rtl-loop2
4196 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4197 loop optimization pass, to @file{@var{file}.119r.loop2},
4198 @file{@var{file}.120r.loop2_init},
4199 @file{@var{file}.121r.loop2_invariant}, and
4200 @file{@var{file}.125r.loop2_done}.
4203 @itemx -fdump-rtl-sms
4205 @opindex fdump-rtl-sms
4206 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4209 @itemx -fdump-rtl-mach
4211 @opindex fdump-rtl-mach
4212 Dump after performing the machine dependent reorganization pass, to
4213 @file{@var{file}.155r.mach}.
4216 @itemx -fdump-rtl-rnreg
4218 @opindex fdump-rtl-rnreg
4219 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4222 @itemx -fdump-rtl-regmove
4224 @opindex fdump-rtl-regmove
4225 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4228 @itemx -fdump-rtl-postreload
4230 @opindex fdump-rtl-postreload
4231 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4234 @itemx -fdump-rtl-expand
4236 @opindex fdump-rtl-expand
4237 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4240 @itemx -fdump-rtl-sched2
4242 @opindex fdump-rtl-sched2
4243 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4246 @itemx -fdump-rtl-cse
4248 @opindex fdump-rtl-cse
4249 Dump after CSE (including the jump optimization that sometimes follows
4250 CSE), to @file{@var{file}.113r.cse}.
4253 @itemx -fdump-rtl-sched1
4255 @opindex fdump-rtl-sched1
4256 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4259 @itemx -fdump-rtl-cse2
4261 @opindex fdump-rtl-cse2
4262 Dump after the second CSE pass (including the jump optimization that
4263 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4266 @itemx -fdump-rtl-tracer
4268 @opindex fdump-rtl-tracer
4269 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4272 @itemx -fdump-rtl-vpt
4273 @itemx -fdump-rtl-vartrack
4275 @opindex fdump-rtl-vpt
4276 @opindex fdump-rtl-vartrack
4277 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4278 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4279 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4280 to @file{@var{file}.154r.vartrack}.
4283 @itemx -fdump-rtl-flow2
4285 @opindex fdump-rtl-flow2
4286 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4289 @itemx -fdump-rtl-peephole2
4291 @opindex fdump-rtl-peephole2
4292 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4295 @itemx -fdump-rtl-web
4297 @opindex fdump-rtl-web
4298 Dump after live range splitting, to @file{@var{file}.126r.web}.
4301 @itemx -fdump-rtl-all
4303 @opindex fdump-rtl-all
4304 Produce all the dumps listed above.
4308 Produce a core dump whenever an error occurs.
4312 Print statistics on memory usage, at the end of the run, to
4317 Annotate the assembler output with a comment indicating which
4318 pattern and alternative was used. The length of each instruction is
4323 Dump the RTL in the assembler output as a comment before each instruction.
4324 Also turns on @option{-dp} annotation.
4328 For each of the other indicated dump files (either with @option{-d} or
4329 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4330 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4334 Just generate RTL for a function instead of compiling it. Usually used
4335 with @samp{r} (@option{-fdump-rtl-expand}).
4339 Dump debugging information during parsing, to standard error.
4343 @opindex fdump-noaddr
4344 When doing debugging dumps (see @option{-d} option above), suppress
4345 address output. This makes it more feasible to use diff on debugging
4346 dumps for compiler invocations with different compiler binaries and/or
4347 different text / bss / data / heap / stack / dso start locations.
4349 @item -fdump-unnumbered
4350 @opindex fdump-unnumbered
4351 When doing debugging dumps (see @option{-d} option above), suppress instruction
4352 numbers, line number note and address output. This makes it more feasible to
4353 use diff on debugging dumps for compiler invocations with different
4354 options, in particular with and without @option{-g}.
4356 @item -fdump-translation-unit @r{(C++ only)}
4357 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4358 @opindex fdump-translation-unit
4359 Dump a representation of the tree structure for the entire translation
4360 unit to a file. The file name is made by appending @file{.tu} to the
4361 source file name. If the @samp{-@var{options}} form is used, @var{options}
4362 controls the details of the dump as described for the
4363 @option{-fdump-tree} options.
4365 @item -fdump-class-hierarchy @r{(C++ only)}
4366 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4367 @opindex fdump-class-hierarchy
4368 Dump a representation of each class's hierarchy and virtual function
4369 table layout to a file. The file name is made by appending @file{.class}
4370 to the source file name. If the @samp{-@var{options}} form is used,
4371 @var{options} controls the details of the dump as described for the
4372 @option{-fdump-tree} options.
4374 @item -fdump-ipa-@var{switch}
4376 Control the dumping at various stages of inter-procedural analysis
4377 language tree to a file. The file name is generated by appending a switch
4378 specific suffix to the source file name. The following dumps are possible:
4382 Enables all inter-procedural analysis dumps; currently the only produced
4383 dump is the @samp{cgraph} dump.
4386 Dumps information about call-graph optimization, unused function removal,
4387 and inlining decisions.
4390 @item -fdump-tree-@var{switch}
4391 @itemx -fdump-tree-@var{switch}-@var{options}
4393 Control the dumping at various stages of processing the intermediate
4394 language tree to a file. The file name is generated by appending a switch
4395 specific suffix to the source file name. If the @samp{-@var{options}}
4396 form is used, @var{options} is a list of @samp{-} separated options that
4397 control the details of the dump. Not all options are applicable to all
4398 dumps, those which are not meaningful will be ignored. The following
4399 options are available
4403 Print the address of each node. Usually this is not meaningful as it
4404 changes according to the environment and source file. Its primary use
4405 is for tying up a dump file with a debug environment.
4407 Inhibit dumping of members of a scope or body of a function merely
4408 because that scope has been reached. Only dump such items when they
4409 are directly reachable by some other path. When dumping pretty-printed
4410 trees, this option inhibits dumping the bodies of control structures.
4412 Print a raw representation of the tree. By default, trees are
4413 pretty-printed into a C-like representation.
4415 Enable more detailed dumps (not honored by every dump option).
4417 Enable dumping various statistics about the pass (not honored by every dump
4420 Enable showing basic block boundaries (disabled in raw dumps).
4422 Enable showing virtual operands for every statement.
4424 Enable showing line numbers for statements.
4426 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4428 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4431 The following tree dumps are possible:
4435 Dump before any tree based optimization, to @file{@var{file}.original}.
4438 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4441 Dump after function inlining, to @file{@var{file}.inlined}.
4444 @opindex fdump-tree-gimple
4445 Dump each function before and after the gimplification pass to a file. The
4446 file name is made by appending @file{.gimple} to the source file name.
4449 @opindex fdump-tree-cfg
4450 Dump the control flow graph of each function to a file. The file name is
4451 made by appending @file{.cfg} to the source file name.
4454 @opindex fdump-tree-vcg
4455 Dump the control flow graph of each function to a file in VCG format. The
4456 file name is made by appending @file{.vcg} to the source file name. Note
4457 that if the file contains more than one function, the generated file cannot
4458 be used directly by VCG@. You will need to cut and paste each function's
4459 graph into its own separate file first.
4462 @opindex fdump-tree-ch
4463 Dump each function after copying loop headers. The file name is made by
4464 appending @file{.ch} to the source file name.
4467 @opindex fdump-tree-ssa
4468 Dump SSA related information to a file. The file name is made by appending
4469 @file{.ssa} to the source file name.
4472 @opindex fdump-tree-salias
4473 Dump structure aliasing variable information to a file. This file name
4474 is made by appending @file{.salias} to the source file name.
4477 @opindex fdump-tree-alias
4478 Dump aliasing information for each function. The file name is made by
4479 appending @file{.alias} to the source file name.
4482 @opindex fdump-tree-ccp
4483 Dump each function after CCP@. The file name is made by appending
4484 @file{.ccp} to the source file name.
4487 @opindex fdump-tree-storeccp
4488 Dump each function after STORE-CCP. The file name is made by appending
4489 @file{.storeccp} to the source file name.
4492 @opindex fdump-tree-pre
4493 Dump trees after partial redundancy elimination. The file name is made
4494 by appending @file{.pre} to the source file name.
4497 @opindex fdump-tree-fre
4498 Dump trees after full redundancy elimination. The file name is made
4499 by appending @file{.fre} to the source file name.
4502 @opindex fdump-tree-copyprop
4503 Dump trees after copy propagation. The file name is made
4504 by appending @file{.copyprop} to the source file name.
4506 @item store_copyprop
4507 @opindex fdump-tree-store_copyprop
4508 Dump trees after store copy-propagation. The file name is made
4509 by appending @file{.store_copyprop} to the source file name.
4512 @opindex fdump-tree-dce
4513 Dump each function after dead code elimination. The file name is made by
4514 appending @file{.dce} to the source file name.
4517 @opindex fdump-tree-mudflap
4518 Dump each function after adding mudflap instrumentation. The file name is
4519 made by appending @file{.mudflap} to the source file name.
4522 @opindex fdump-tree-sra
4523 Dump each function after performing scalar replacement of aggregates. The
4524 file name is made by appending @file{.sra} to the source file name.
4527 @opindex fdump-tree-sink
4528 Dump each function after performing code sinking. The file name is made
4529 by appending @file{.sink} to the source file name.
4532 @opindex fdump-tree-dom
4533 Dump each function after applying dominator tree optimizations. The file
4534 name is made by appending @file{.dom} to the source file name.
4537 @opindex fdump-tree-dse
4538 Dump each function after applying dead store elimination. The file
4539 name is made by appending @file{.dse} to the source file name.
4542 @opindex fdump-tree-phiopt
4543 Dump each function after optimizing PHI nodes into straightline code. The file
4544 name is made by appending @file{.phiopt} to the source file name.
4547 @opindex fdump-tree-forwprop
4548 Dump each function after forward propagating single use variables. The file
4549 name is made by appending @file{.forwprop} to the source file name.
4552 @opindex fdump-tree-copyrename
4553 Dump each function after applying the copy rename optimization. The file
4554 name is made by appending @file{.copyrename} to the source file name.
4557 @opindex fdump-tree-nrv
4558 Dump each function after applying the named return value optimization on
4559 generic trees. The file name is made by appending @file{.nrv} to the source
4563 @opindex fdump-tree-vect
4564 Dump each function after applying vectorization of loops. The file name is
4565 made by appending @file{.vect} to the source file name.
4568 @opindex fdump-tree-vrp
4569 Dump each function after Value Range Propagation (VRP). The file name
4570 is made by appending @file{.vrp} to the source file name.
4573 @opindex fdump-tree-all
4574 Enable all the available tree dumps with the flags provided in this option.
4577 @item -ftree-vectorizer-verbose=@var{n}
4578 @opindex ftree-vectorizer-verbose
4579 This option controls the amount of debugging output the vectorizer prints.
4580 This information is written to standard error, unless
4581 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4582 in which case it is output to the usual dump listing file, @file{.vect}.
4583 For @var{n}=0 no diagnostic information is reported.
4584 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4585 and the total number of loops that got vectorized.
4586 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4587 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4588 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4589 level that @option{-fdump-tree-vect-stats} uses.
4590 Higher verbosity levels mean either more information dumped for each
4591 reported loop, or same amount of information reported for more loops:
4592 If @var{n}=3, alignment related information is added to the reports.
4593 If @var{n}=4, data-references related information (e.g. memory dependences,
4594 memory access-patterns) is added to the reports.
4595 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4596 that did not pass the first analysis phase (i.e. may not be countable, or
4597 may have complicated control-flow).
4598 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4599 For @var{n}=7, all the information the vectorizer generates during its
4600 analysis and transformation is reported. This is the same verbosity level
4601 that @option{-fdump-tree-vect-details} uses.
4603 @item -frandom-seed=@var{string}
4604 @opindex frandom-string
4605 This option provides a seed that GCC uses when it would otherwise use
4606 random numbers. It is used to generate certain symbol names
4607 that have to be different in every compiled file. It is also used to
4608 place unique stamps in coverage data files and the object files that
4609 produce them. You can use the @option{-frandom-seed} option to produce
4610 reproducibly identical object files.
4612 The @var{string} should be different for every file you compile.
4614 @item -fsched-verbose=@var{n}
4615 @opindex fsched-verbose
4616 On targets that use instruction scheduling, this option controls the
4617 amount of debugging output the scheduler prints. This information is
4618 written to standard error, unless @option{-dS} or @option{-dR} is
4619 specified, in which case it is output to the usual dump
4620 listing file, @file{.sched} or @file{.sched2} respectively. However
4621 for @var{n} greater than nine, the output is always printed to standard
4624 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4625 same information as @option{-dRS}. For @var{n} greater than one, it
4626 also output basic block probabilities, detailed ready list information
4627 and unit/insn info. For @var{n} greater than two, it includes RTL
4628 at abort point, control-flow and regions info. And for @var{n} over
4629 four, @option{-fsched-verbose} also includes dependence info.
4633 Store the usual ``temporary'' intermediate files permanently; place them
4634 in the current directory and name them based on the source file. Thus,
4635 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4636 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4637 preprocessed @file{foo.i} output file even though the compiler now
4638 normally uses an integrated preprocessor.
4640 When used in combination with the @option{-x} command line option,
4641 @option{-save-temps} is sensible enough to avoid over writing an
4642 input source file with the same extension as an intermediate file.
4643 The corresponding intermediate file may be obtained by renaming the
4644 source file before using @option{-save-temps}.
4648 Report the CPU time taken by each subprocess in the compilation
4649 sequence. For C source files, this is the compiler proper and assembler
4650 (plus the linker if linking is done). The output looks like this:
4657 The first number on each line is the ``user time'', that is time spent
4658 executing the program itself. The second number is ``system time'',
4659 time spent executing operating system routines on behalf of the program.
4660 Both numbers are in seconds.
4662 @item -fvar-tracking
4663 @opindex fvar-tracking
4664 Run variable tracking pass. It computes where variables are stored at each
4665 position in code. Better debugging information is then generated
4666 (if the debugging information format supports this information).
4668 It is enabled by default when compiling with optimization (@option{-Os},
4669 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4670 the debug info format supports it.
4672 @item -print-file-name=@var{library}
4673 @opindex print-file-name
4674 Print the full absolute name of the library file @var{library} that
4675 would be used when linking---and don't do anything else. With this
4676 option, GCC does not compile or link anything; it just prints the
4679 @item -print-multi-directory
4680 @opindex print-multi-directory
4681 Print the directory name corresponding to the multilib selected by any
4682 other switches present in the command line. This directory is supposed
4683 to exist in @env{GCC_EXEC_PREFIX}.
4685 @item -print-multi-lib
4686 @opindex print-multi-lib
4687 Print the mapping from multilib directory names to compiler switches
4688 that enable them. The directory name is separated from the switches by
4689 @samp{;}, and each switch starts with an @samp{@@} instead of the
4690 @samp{-}, without spaces between multiple switches. This is supposed to
4691 ease shell-processing.
4693 @item -print-prog-name=@var{program}
4694 @opindex print-prog-name
4695 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4697 @item -print-libgcc-file-name
4698 @opindex print-libgcc-file-name
4699 Same as @option{-print-file-name=libgcc.a}.
4701 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4702 but you do want to link with @file{libgcc.a}. You can do
4705 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4708 @item -print-search-dirs
4709 @opindex print-search-dirs
4710 Print the name of the configured installation directory and a list of
4711 program and library directories @command{gcc} will search---and don't do anything else.
4713 This is useful when @command{gcc} prints the error message
4714 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4715 To resolve this you either need to put @file{cpp0} and the other compiler
4716 components where @command{gcc} expects to find them, or you can set the environment
4717 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4718 Don't forget the trailing @samp{/}.
4719 @xref{Environment Variables}.
4721 @item -print-sysroot-headers-suffix
4722 @opindex print-sysroot-headers-suffix
4723 Print the suffix added to the target sysroot when searching for
4724 headers, or give an error if the compiler is not configured with such
4725 a suffix---and don't do anything else.
4728 @opindex dumpmachine
4729 Print the compiler's target machine (for example,
4730 @samp{i686-pc-linux-gnu})---and don't do anything else.
4733 @opindex dumpversion
4734 Print the compiler version (for example, @samp{3.0})---and don't do
4739 Print the compiler's built-in specs---and don't do anything else. (This
4740 is used when GCC itself is being built.) @xref{Spec Files}.
4742 @item -feliminate-unused-debug-types
4743 @opindex feliminate-unused-debug-types
4744 Normally, when producing DWARF2 output, GCC will emit debugging
4745 information for all types declared in a compilation
4746 unit, regardless of whether or not they are actually used
4747 in that compilation unit. Sometimes this is useful, such as
4748 if, in the debugger, you want to cast a value to a type that is
4749 not actually used in your program (but is declared). More often,
4750 however, this results in a significant amount of wasted space.
4751 With this option, GCC will avoid producing debug symbol output
4752 for types that are nowhere used in the source file being compiled.
4755 @node Optimize Options
4756 @section Options That Control Optimization
4757 @cindex optimize options
4758 @cindex options, optimization
4760 These options control various sorts of optimizations.
4762 Without any optimization option, the compiler's goal is to reduce the
4763 cost of compilation and to make debugging produce the expected
4764 results. Statements are independent: if you stop the program with a
4765 breakpoint between statements, you can then assign a new value to any
4766 variable or change the program counter to any other statement in the
4767 function and get exactly the results you would expect from the source
4770 Turning on optimization flags makes the compiler attempt to improve
4771 the performance and/or code size at the expense of compilation time
4772 and possibly the ability to debug the program.
4774 The compiler performs optimization based on the knowledge it has of
4775 the program. Optimization levels @option{-O} and above, in
4776 particular, enable @emph{unit-at-a-time} mode, which allows the
4777 compiler to consider information gained from later functions in
4778 the file when compiling a function. Compiling multiple files at
4779 once to a single output file in @emph{unit-at-a-time} mode allows
4780 the compiler to use information gained from all of the files when
4781 compiling each of them.
4783 Not all optimizations are controlled directly by a flag. Only
4784 optimizations that have a flag are listed.
4791 Optimize. Optimizing compilation takes somewhat more time, and a lot
4792 more memory for a large function.
4794 With @option{-O}, the compiler tries to reduce code size and execution
4795 time, without performing any optimizations that take a great deal of
4798 @option{-O} turns on the following optimization flags:
4799 @gccoptlist{-fdefer-pop @gol
4800 -fdelayed-branch @gol
4801 -fguess-branch-probability @gol
4802 -fcprop-registers @gol
4803 -fif-conversion @gol
4804 -fif-conversion2 @gol
4805 -fsplit-wide-types @gol
4808 -ftree-dominator-opts @gol
4812 -ftree-copyrename @gol
4815 -funit-at-a-time @gol
4818 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4819 where doing so does not interfere with debugging.
4823 Optimize even more. GCC performs nearly all supported optimizations
4824 that do not involve a space-speed tradeoff. The compiler does not
4825 perform loop unrolling or function inlining when you specify @option{-O2}.
4826 As compared to @option{-O}, this option increases both compilation time
4827 and the performance of the generated code.
4829 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4830 also turns on the following optimization flags:
4831 @gccoptlist{-fthread-jumps @gol
4833 -foptimize-sibling-calls @gol
4834 -fcse-follow-jumps -fcse-skip-blocks @gol
4835 -fgcse -fgcse-lm @gol
4836 -fexpensive-optimizations @gol
4837 -frerun-cse-after-loop @gol
4840 -fschedule-insns -fschedule-insns2 @gol
4841 -fsched-interblock -fsched-spec @gol
4843 -fstrict-aliasing -fstrict-overflow @gol
4844 -fdelete-null-pointer-checks @gol
4845 -freorder-blocks -freorder-functions @gol
4846 -falign-functions -falign-jumps @gol
4847 -falign-loops -falign-labels @gol
4851 Please note the warning under @option{-fgcse} about
4852 invoking @option{-O2} on programs that use computed gotos.
4856 Optimize yet more. @option{-O3} turns on all optimizations specified by
4857 @option{-O2} and also turns on the @option{-finline-functions},
4858 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4862 Reduce compilation time and make debugging produce the expected
4863 results. This is the default.
4867 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4868 do not typically increase code size. It also performs further
4869 optimizations designed to reduce code size.
4871 @option{-Os} disables the following optimization flags:
4872 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4873 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4874 -fprefetch-loop-arrays -ftree-vect-loop-version}
4876 If you use multiple @option{-O} options, with or without level numbers,
4877 the last such option is the one that is effective.
4880 Options of the form @option{-f@var{flag}} specify machine-independent
4881 flags. Most flags have both positive and negative forms; the negative
4882 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4883 below, only one of the forms is listed---the one you typically will
4884 use. You can figure out the other form by either removing @samp{no-}
4887 The following options control specific optimizations. They are either
4888 activated by @option{-O} options or are related to ones that are. You
4889 can use the following flags in the rare cases when ``fine-tuning'' of
4890 optimizations to be performed is desired.
4893 @item -fno-default-inline
4894 @opindex fno-default-inline
4895 Do not make member functions inline by default merely because they are
4896 defined inside the class scope (C++ only). Otherwise, when you specify
4897 @w{@option{-O}}, member functions defined inside class scope are compiled
4898 inline by default; i.e., you don't need to add @samp{inline} in front of
4899 the member function name.
4901 @item -fno-defer-pop
4902 @opindex fno-defer-pop
4903 Always pop the arguments to each function call as soon as that function
4904 returns. For machines which must pop arguments after a function call,
4905 the compiler normally lets arguments accumulate on the stack for several
4906 function calls and pops them all at once.
4908 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4911 @opindex fforce-addr
4912 Force memory address constants to be copied into registers before
4913 doing arithmetic on them.
4915 @item -fforward-propagate
4916 @opindex fforward-propagate
4917 Perform a forward propagation pass on RTL. The pass tries to combine two
4918 instructions and checks if the result can be simplified. If loop unrolling
4919 is active, two passes are performed and the second is scheduled after
4922 This option is enabled by default at optimization levels @option{-O2},
4923 @option{-O3}, @option{-Os}.
4925 @item -fomit-frame-pointer
4926 @opindex fomit-frame-pointer
4927 Don't keep the frame pointer in a register for functions that
4928 don't need one. This avoids the instructions to save, set up and
4929 restore frame pointers; it also makes an extra register available
4930 in many functions. @strong{It also makes debugging impossible on
4933 On some machines, such as the VAX, this flag has no effect, because
4934 the standard calling sequence automatically handles the frame pointer
4935 and nothing is saved by pretending it doesn't exist. The
4936 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4937 whether a target machine supports this flag. @xref{Registers,,Register
4938 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4940 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4942 @item -foptimize-sibling-calls
4943 @opindex foptimize-sibling-calls
4944 Optimize sibling and tail recursive calls.
4946 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4950 Don't pay attention to the @code{inline} keyword. Normally this option
4951 is used to keep the compiler from expanding any functions inline.
4952 Note that if you are not optimizing, no functions can be expanded inline.
4954 @item -finline-functions
4955 @opindex finline-functions
4956 Integrate all simple functions into their callers. The compiler
4957 heuristically decides which functions are simple enough to be worth
4958 integrating in this way.
4960 If all calls to a given function are integrated, and the function is
4961 declared @code{static}, then the function is normally not output as
4962 assembler code in its own right.
4964 Enabled at level @option{-O3}.
4966 @item -finline-functions-called-once
4967 @opindex finline-functions-called-once
4968 Consider all @code{static} functions called once for inlining into their
4969 caller even if they are not marked @code{inline}. If a call to a given
4970 function is integrated, then the function is not output as assembler code
4973 Enabled if @option{-funit-at-a-time} is enabled.
4975 @item -fearly-inlining
4976 @opindex fearly-inlining
4977 Inline functions marked by @code{always_inline} and functions whose body seems
4978 smaller than the function call overhead early before doing
4979 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4980 makes profiling significantly cheaper and usually inlining faster on programs
4981 having large chains of nested wrapper functions.
4985 @item -finline-limit=@var{n}
4986 @opindex finline-limit
4987 By default, GCC limits the size of functions that can be inlined. This flag
4988 allows the control of this limit for functions that are explicitly marked as
4989 inline (i.e., marked with the inline keyword or defined within the class
4990 definition in c++). @var{n} is the size of functions that can be inlined in
4991 number of pseudo instructions (not counting parameter handling). The default
4992 value of @var{n} is 600.
4993 Increasing this value can result in more inlined code at
4994 the cost of compilation time and memory consumption. Decreasing usually makes
4995 the compilation faster and less code will be inlined (which presumably
4996 means slower programs). This option is particularly useful for programs that
4997 use inlining heavily such as those based on recursive templates with C++.
4999 Inlining is actually controlled by a number of parameters, which may be
5000 specified individually by using @option{--param @var{name}=@var{value}}.
5001 The @option{-finline-limit=@var{n}} option sets some of these parameters
5005 @item max-inline-insns-single
5006 is set to @var{n}/2.
5007 @item max-inline-insns-auto
5008 is set to @var{n}/2.
5009 @item min-inline-insns
5010 is set to 130 or @var{n}/4, whichever is smaller.
5011 @item max-inline-insns-rtl
5015 See below for a documentation of the individual
5016 parameters controlling inlining.
5018 @emph{Note:} pseudo instruction represents, in this particular context, an
5019 abstract measurement of function's size. In no way does it represent a count
5020 of assembly instructions and as such its exact meaning might change from one
5021 release to an another.
5023 @item -fkeep-inline-functions
5024 @opindex fkeep-inline-functions
5025 In C, emit @code{static} functions that are declared @code{inline}
5026 into the object file, even if the function has been inlined into all
5027 of its callers. This switch does not affect functions using the
5028 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5029 inline functions into the object file.
5031 @item -fkeep-static-consts
5032 @opindex fkeep-static-consts
5033 Emit variables declared @code{static const} when optimization isn't turned
5034 on, even if the variables aren't referenced.
5036 GCC enables this option by default. If you want to force the compiler to
5037 check if the variable was referenced, regardless of whether or not
5038 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5040 @item -fmerge-constants
5041 Attempt to merge identical constants (string constants and floating point
5042 constants) across compilation units.
5044 This option is the default for optimized compilation if the assembler and
5045 linker support it. Use @option{-fno-merge-constants} to inhibit this
5048 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5050 @item -fmerge-all-constants
5051 Attempt to merge identical constants and identical variables.
5053 This option implies @option{-fmerge-constants}. In addition to
5054 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5055 arrays or initialized constant variables with integral or floating point
5056 types. Languages like C or C++ require each non-automatic variable to
5057 have distinct location, so using this option will result in non-conforming
5060 @item -fmodulo-sched
5061 @opindex fmodulo-sched
5062 Perform swing modulo scheduling immediately before the first scheduling
5063 pass. This pass looks at innermost loops and reorders their
5064 instructions by overlapping different iterations.
5066 @item -fno-branch-count-reg
5067 @opindex fno-branch-count-reg
5068 Do not use ``decrement and branch'' instructions on a count register,
5069 but instead generate a sequence of instructions that decrement a
5070 register, compare it against zero, then branch based upon the result.
5071 This option is only meaningful on architectures that support such
5072 instructions, which include x86, PowerPC, IA-64 and S/390.
5074 The default is @option{-fbranch-count-reg}.
5076 @item -fno-function-cse
5077 @opindex fno-function-cse
5078 Do not put function addresses in registers; make each instruction that
5079 calls a constant function contain the function's address explicitly.
5081 This option results in less efficient code, but some strange hacks
5082 that alter the assembler output may be confused by the optimizations
5083 performed when this option is not used.
5085 The default is @option{-ffunction-cse}
5087 @item -fno-zero-initialized-in-bss
5088 @opindex fno-zero-initialized-in-bss
5089 If the target supports a BSS section, GCC by default puts variables that
5090 are initialized to zero into BSS@. This can save space in the resulting
5093 This option turns off this behavior because some programs explicitly
5094 rely on variables going to the data section. E.g., so that the
5095 resulting executable can find the beginning of that section and/or make
5096 assumptions based on that.
5098 The default is @option{-fzero-initialized-in-bss}.
5100 @item -fbounds-check
5101 @opindex fbounds-check
5102 For front-ends that support it, generate additional code to check that
5103 indices used to access arrays are within the declared range. This is
5104 currently only supported by the Java and Fortran front-ends, where
5105 this option defaults to true and false respectively.
5107 @item -fmudflap -fmudflapth -fmudflapir
5111 @cindex bounds checking
5113 For front-ends that support it (C and C++), instrument all risky
5114 pointer/array dereferencing operations, some standard library
5115 string/heap functions, and some other associated constructs with
5116 range/validity tests. Modules so instrumented should be immune to
5117 buffer overflows, invalid heap use, and some other classes of C/C++
5118 programming errors. The instrumentation relies on a separate runtime
5119 library (@file{libmudflap}), which will be linked into a program if
5120 @option{-fmudflap} is given at link time. Run-time behavior of the
5121 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5122 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5125 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5126 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5127 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5128 instrumentation should ignore pointer reads. This produces less
5129 instrumentation (and therefore faster execution) and still provides
5130 some protection against outright memory corrupting writes, but allows
5131 erroneously read data to propagate within a program.
5133 @item -fthread-jumps
5134 @opindex fthread-jumps
5135 Perform optimizations where we check to see if a jump branches to a
5136 location where another comparison subsumed by the first is found. If
5137 so, the first branch is redirected to either the destination of the
5138 second branch or a point immediately following it, depending on whether
5139 the condition is known to be true or false.
5141 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5143 @item -fsplit-wide-types
5144 @opindex -fsplit-wide-types
5145 When using a type that occupies multiple registers, such as @code{long
5146 long} on a 32-bit system, split the registers apart and allocate them
5147 independently. This normally generates better code for those types,
5148 but may make debugging more difficult.
5150 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5153 @item -fcse-follow-jumps
5154 @opindex fcse-follow-jumps
5155 In common subexpression elimination, scan through jump instructions
5156 when the target of the jump is not reached by any other path. For
5157 example, when CSE encounters an @code{if} statement with an
5158 @code{else} clause, CSE will follow the jump when the condition
5161 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5163 @item -fcse-skip-blocks
5164 @opindex fcse-skip-blocks
5165 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5166 follow jumps which conditionally skip over blocks. When CSE
5167 encounters a simple @code{if} statement with no else clause,
5168 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5169 body of the @code{if}.
5171 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5173 @item -frerun-cse-after-loop
5174 @opindex frerun-cse-after-loop
5175 Re-run common subexpression elimination after loop optimizations has been
5178 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5182 Perform a global common subexpression elimination pass.
5183 This pass also performs global constant and copy propagation.
5185 @emph{Note:} When compiling a program using computed gotos, a GCC
5186 extension, you may get better runtime performance if you disable
5187 the global common subexpression elimination pass by adding
5188 @option{-fno-gcse} to the command line.
5190 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5194 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5195 attempt to move loads which are only killed by stores into themselves. This
5196 allows a loop containing a load/store sequence to be changed to a load outside
5197 the loop, and a copy/store within the loop.
5199 Enabled by default when gcse is enabled.
5203 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5204 global common subexpression elimination. This pass will attempt to move
5205 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5206 loops containing a load/store sequence can be changed to a load before
5207 the loop and a store after the loop.
5209 Not enabled at any optimization level.
5213 When @option{-fgcse-las} is enabled, the global common subexpression
5214 elimination pass eliminates redundant loads that come after stores to the
5215 same memory location (both partial and full redundancies).
5217 Not enabled at any optimization level.
5219 @item -fgcse-after-reload
5220 @opindex fgcse-after-reload
5221 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5222 pass is performed after reload. The purpose of this pass is to cleanup
5225 @item -funsafe-loop-optimizations
5226 @opindex funsafe-loop-optimizations
5227 If given, the loop optimizer will assume that loop indices do not
5228 overflow, and that the loops with nontrivial exit condition are not
5229 infinite. This enables a wider range of loop optimizations even if
5230 the loop optimizer itself cannot prove that these assumptions are valid.
5231 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5232 if it finds this kind of loop.
5234 @item -fcrossjumping
5235 @opindex crossjumping
5236 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5237 resulting code may or may not perform better than without cross-jumping.
5239 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5241 @item -fif-conversion
5242 @opindex if-conversion
5243 Attempt to transform conditional jumps into branch-less equivalents. This
5244 include use of conditional moves, min, max, set flags and abs instructions, and
5245 some tricks doable by standard arithmetics. The use of conditional execution
5246 on chips where it is available is controlled by @code{if-conversion2}.
5248 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5250 @item -fif-conversion2
5251 @opindex if-conversion2
5252 Use conditional execution (where available) to transform conditional jumps into
5253 branch-less equivalents.
5255 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5257 @item -fdelete-null-pointer-checks
5258 @opindex fdelete-null-pointer-checks
5259 Use global dataflow analysis to identify and eliminate useless checks
5260 for null pointers. The compiler assumes that dereferencing a null
5261 pointer would have halted the program. If a pointer is checked after
5262 it has already been dereferenced, it cannot be null.
5264 In some environments, this assumption is not true, and programs can
5265 safely dereference null pointers. Use
5266 @option{-fno-delete-null-pointer-checks} to disable this optimization
5267 for programs which depend on that behavior.
5269 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5271 @item -fexpensive-optimizations
5272 @opindex fexpensive-optimizations
5273 Perform a number of minor optimizations that are relatively expensive.
5275 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5277 @item -foptimize-register-move
5279 @opindex foptimize-register-move
5281 Attempt to reassign register numbers in move instructions and as
5282 operands of other simple instructions in order to maximize the amount of
5283 register tying. This is especially helpful on machines with two-operand
5286 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5289 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5291 @item -fdelayed-branch
5292 @opindex fdelayed-branch
5293 If supported for the target machine, attempt to reorder instructions
5294 to exploit instruction slots available after delayed branch
5297 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5299 @item -fschedule-insns
5300 @opindex fschedule-insns
5301 If supported for the target machine, attempt to reorder instructions to
5302 eliminate execution stalls due to required data being unavailable. This
5303 helps machines that have slow floating point or memory load instructions
5304 by allowing other instructions to be issued until the result of the load
5305 or floating point instruction is required.
5307 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5309 @item -fschedule-insns2
5310 @opindex fschedule-insns2
5311 Similar to @option{-fschedule-insns}, but requests an additional pass of
5312 instruction scheduling after register allocation has been done. This is
5313 especially useful on machines with a relatively small number of
5314 registers and where memory load instructions take more than one cycle.
5316 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5318 @item -fno-sched-interblock
5319 @opindex fno-sched-interblock
5320 Don't schedule instructions across basic blocks. This is normally
5321 enabled by default when scheduling before register allocation, i.e.@:
5322 with @option{-fschedule-insns} or at @option{-O2} or higher.
5324 @item -fno-sched-spec
5325 @opindex fno-sched-spec
5326 Don't allow speculative motion of non-load instructions. This is normally
5327 enabled by default when scheduling before register allocation, i.e.@:
5328 with @option{-fschedule-insns} or at @option{-O2} or higher.
5330 @item -fsched-spec-load
5331 @opindex fsched-spec-load
5332 Allow speculative motion of some load instructions. This only makes
5333 sense when scheduling before register allocation, i.e.@: with
5334 @option{-fschedule-insns} or at @option{-O2} or higher.
5336 @item -fsched-spec-load-dangerous
5337 @opindex fsched-spec-load-dangerous
5338 Allow speculative motion of more load instructions. This only makes
5339 sense when scheduling before register allocation, i.e.@: with
5340 @option{-fschedule-insns} or at @option{-O2} or higher.
5342 @item -fsched-stalled-insns=@var{n}
5343 @opindex fsched-stalled-insns
5344 Define how many insns (if any) can be moved prematurely from the queue
5345 of stalled insns into the ready list, during the second scheduling pass.
5347 @item -fsched-stalled-insns-dep=@var{n}
5348 @opindex fsched-stalled-insns-dep
5349 Define how many insn groups (cycles) will be examined for a dependency
5350 on a stalled insn that is candidate for premature removal from the queue
5351 of stalled insns. Has an effect only during the second scheduling pass,
5352 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5354 @item -fsched2-use-superblocks
5355 @opindex fsched2-use-superblocks
5356 When scheduling after register allocation, do use superblock scheduling
5357 algorithm. Superblock scheduling allows motion across basic block boundaries
5358 resulting on faster schedules. This option is experimental, as not all machine
5359 descriptions used by GCC model the CPU closely enough to avoid unreliable
5360 results from the algorithm.
5362 This only makes sense when scheduling after register allocation, i.e.@: with
5363 @option{-fschedule-insns2} or at @option{-O2} or higher.
5365 @item -fsched2-use-traces
5366 @opindex fsched2-use-traces
5367 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5368 allocation and additionally perform code duplication in order to increase the
5369 size of superblocks using tracer pass. See @option{-ftracer} for details on
5372 This mode should produce faster but significantly longer programs. Also
5373 without @option{-fbranch-probabilities} the traces constructed may not
5374 match the reality and hurt the performance. This only makes
5375 sense when scheduling after register allocation, i.e.@: with
5376 @option{-fschedule-insns2} or at @option{-O2} or higher.
5380 Eliminates redundant extension instructions and move the non redundant
5381 ones to optimal placement using LCM.
5383 @item -freschedule-modulo-scheduled-loops
5384 @opindex fscheduling-in-modulo-scheduled-loops
5385 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5386 we may want to prevent the later scheduling passes from changing its schedule, we use this
5387 option to control that.
5389 @item -fcaller-saves
5390 @opindex fcaller-saves
5391 Enable values to be allocated in registers that will be clobbered by
5392 function calls, by emitting extra instructions to save and restore the
5393 registers around such calls. Such allocation is done only when it
5394 seems to result in better code than would otherwise be produced.
5396 This option is always enabled by default on certain machines, usually
5397 those which have no call-preserved registers to use instead.
5399 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5402 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5403 enabled by default at @option{-O2} and @option{-O3}.
5406 Perform Full Redundancy Elimination (FRE) on trees. The difference
5407 between FRE and PRE is that FRE only considers expressions
5408 that are computed on all paths leading to the redundant computation.
5409 This analysis faster than PRE, though it exposes fewer redundancies.
5410 This flag is enabled by default at @option{-O} and higher.
5412 @item -ftree-copy-prop
5413 Perform copy propagation on trees. This pass eliminates unnecessary
5414 copy operations. This flag is enabled by default at @option{-O} and
5417 @item -ftree-store-copy-prop
5418 Perform copy propagation of memory loads and stores. This pass
5419 eliminates unnecessary copy operations in memory references
5420 (structures, global variables, arrays, etc). This flag is enabled by
5421 default at @option{-O2} and higher.
5424 Perform structural alias analysis on trees. This flag
5425 is enabled by default at @option{-O} and higher.
5428 Perform interprocedural pointer analysis.
5431 Perform forward store motion on trees. This flag is
5432 enabled by default at @option{-O} and higher.
5435 Perform sparse conditional constant propagation (CCP) on trees. This
5436 pass only operates on local scalar variables and is enabled by default
5437 at @option{-O} and higher.
5439 @item -ftree-store-ccp
5440 Perform sparse conditional constant propagation (CCP) on trees. This
5441 pass operates on both local scalar variables and memory stores and
5442 loads (global variables, structures, arrays, etc). This flag is
5443 enabled by default at @option{-O2} and higher.
5446 Perform dead code elimination (DCE) on trees. This flag is enabled by
5447 default at @option{-O} and higher.
5449 @item -ftree-dominator-opts
5450 Perform a variety of simple scalar cleanups (constant/copy
5451 propagation, redundancy elimination, range propagation and expression
5452 simplification) based on a dominator tree traversal. This also
5453 performs jump threading (to reduce jumps to jumps). This flag is
5454 enabled by default at @option{-O} and higher.
5457 Perform loop header copying on trees. This is beneficial since it increases
5458 effectiveness of code motion optimizations. It also saves one jump. This flag
5459 is enabled by default at @option{-O} and higher. It is not enabled
5460 for @option{-Os}, since it usually increases code size.
5462 @item -ftree-loop-optimize
5463 Perform loop optimizations on trees. This flag is enabled by default
5464 at @option{-O} and higher.
5466 @item -ftree-loop-linear
5467 Perform linear loop transformations on tree. This flag can improve cache
5468 performance and allow further loop optimizations to take place.
5470 @item -fcheck-data-deps
5471 Compare the results of several data dependence analyzers. This option
5472 is used for debugging the data dependence analyzers.
5474 @item -ftree-loop-im
5475 Perform loop invariant motion on trees. This pass moves only invariants that
5476 would be hard to handle at RTL level (function calls, operations that expand to
5477 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5478 operands of conditions that are invariant out of the loop, so that we can use
5479 just trivial invariantness analysis in loop unswitching. The pass also includes
5482 @item -ftree-loop-ivcanon
5483 Create a canonical counter for number of iterations in the loop for that
5484 determining number of iterations requires complicated analysis. Later
5485 optimizations then may determine the number easily. Useful especially
5486 in connection with unrolling.
5489 Perform induction variable optimizations (strength reduction, induction
5490 variable merging and induction variable elimination) on trees.
5493 Perform scalar replacement of aggregates. This pass replaces structure
5494 references with scalars to prevent committing structures to memory too
5495 early. This flag is enabled by default at @option{-O} and higher.
5497 @item -ftree-copyrename
5498 Perform copy renaming on trees. This pass attempts to rename compiler
5499 temporaries to other variables at copy locations, usually resulting in
5500 variable names which more closely resemble the original variables. This flag
5501 is enabled by default at @option{-O} and higher.
5504 Perform temporary expression replacement during the SSA->normal phase. Single
5505 use/single def temporaries are replaced at their use location with their
5506 defining expression. This results in non-GIMPLE code, but gives the expanders
5507 much more complex trees to work on resulting in better RTL generation. This is
5508 enabled by default at @option{-O} and higher.
5510 @item -ftree-vectorize
5511 Perform loop vectorization on trees.
5513 @item -ftree-vect-loop-version
5514 @opindex ftree-vect-loop-version
5515 Perform loop versioning when doing loop vectorization on trees. When a loop
5516 appears to be vectorizable except that data alignment or data dependence cannot
5517 be determined at compile time then vectorized and non-vectorized versions of
5518 the loop are generated along with runtime checks for alignment or dependence
5519 to control which version is executed. This option is enabled by default
5520 except at level @option{-Os} where it is disabled.
5523 Perform Value Range Propagation on trees. This is similar to the
5524 constant propagation pass, but instead of values, ranges of values are
5525 propagated. This allows the optimizers to remove unnecessary range
5526 checks like array bound checks and null pointer checks. This is
5527 enabled by default at @option{-O2} and higher. Null pointer check
5528 elimination is only done if @option{-fdelete-null-pointer-checks} is
5533 Perform tail duplication to enlarge superblock size. This transformation
5534 simplifies the control flow of the function allowing other optimizations to do
5537 @item -funroll-loops
5538 @opindex funroll-loops
5539 Unroll loops whose number of iterations can be determined at compile
5540 time or upon entry to the loop. @option{-funroll-loops} implies
5541 @option{-frerun-cse-after-loop}. This option makes code larger,
5542 and may or may not make it run faster.
5544 @item -funroll-all-loops
5545 @opindex funroll-all-loops
5546 Unroll all loops, even if their number of iterations is uncertain when
5547 the loop is entered. This usually makes programs run more slowly.
5548 @option{-funroll-all-loops} implies the same options as
5549 @option{-funroll-loops},
5551 @item -fsplit-ivs-in-unroller
5552 @opindex -fsplit-ivs-in-unroller
5553 Enables expressing of values of induction variables in later iterations
5554 of the unrolled loop using the value in the first iteration. This breaks
5555 long dependency chains, thus improving efficiency of the scheduling passes.
5557 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5558 same effect. However in cases the loop body is more complicated than
5559 a single basic block, this is not reliable. It also does not work at all
5560 on some of the architectures due to restrictions in the CSE pass.
5562 This optimization is enabled by default.
5564 @item -fvariable-expansion-in-unroller
5565 @opindex -fvariable-expansion-in-unroller
5566 With this option, the compiler will create multiple copies of some
5567 local variables when unrolling a loop which can result in superior code.
5569 @item -fprefetch-loop-arrays
5570 @opindex fprefetch-loop-arrays
5571 If supported by the target machine, generate instructions to prefetch
5572 memory to improve the performance of loops that access large arrays.
5574 This option may generate better or worse code; results are highly
5575 dependent on the structure of loops within the source code.
5577 Disabled at level @option{-Os}.
5580 @itemx -fno-peephole2
5581 @opindex fno-peephole
5582 @opindex fno-peephole2
5583 Disable any machine-specific peephole optimizations. The difference
5584 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5585 are implemented in the compiler; some targets use one, some use the
5586 other, a few use both.
5588 @option{-fpeephole} is enabled by default.
5589 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5591 @item -fno-guess-branch-probability
5592 @opindex fno-guess-branch-probability
5593 Do not guess branch probabilities using heuristics.
5595 GCC will use heuristics to guess branch probabilities if they are
5596 not provided by profiling feedback (@option{-fprofile-arcs}). These
5597 heuristics are based on the control flow graph. If some branch probabilities
5598 are specified by @samp{__builtin_expect}, then the heuristics will be
5599 used to guess branch probabilities for the rest of the control flow graph,
5600 taking the @samp{__builtin_expect} info into account. The interactions
5601 between the heuristics and @samp{__builtin_expect} can be complex, and in
5602 some cases, it may be useful to disable the heuristics so that the effects
5603 of @samp{__builtin_expect} are easier to understand.
5605 The default is @option{-fguess-branch-probability} at levels
5606 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5608 @item -freorder-blocks
5609 @opindex freorder-blocks
5610 Reorder basic blocks in the compiled function in order to reduce number of
5611 taken branches and improve code locality.
5613 Enabled at levels @option{-O2}, @option{-O3}.
5615 @item -freorder-blocks-and-partition
5616 @opindex freorder-blocks-and-partition
5617 In addition to reordering basic blocks in the compiled function, in order
5618 to reduce number of taken branches, partitions hot and cold basic blocks
5619 into separate sections of the assembly and .o files, to improve
5620 paging and cache locality performance.
5622 This optimization is automatically turned off in the presence of
5623 exception handling, for linkonce sections, for functions with a user-defined
5624 section attribute and on any architecture that does not support named
5627 @item -freorder-functions
5628 @opindex freorder-functions
5629 Reorder functions in the object file in order to
5630 improve code locality. This is implemented by using special
5631 subsections @code{.text.hot} for most frequently executed functions and
5632 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5633 the linker so object file format must support named sections and linker must
5634 place them in a reasonable way.
5636 Also profile feedback must be available in to make this option effective. See
5637 @option{-fprofile-arcs} for details.
5639 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5641 @item -fstrict-aliasing
5642 @opindex fstrict-aliasing
5643 Allows the compiler to assume the strictest aliasing rules applicable to
5644 the language being compiled. For C (and C++), this activates
5645 optimizations based on the type of expressions. In particular, an
5646 object of one type is assumed never to reside at the same address as an
5647 object of a different type, unless the types are almost the same. For
5648 example, an @code{unsigned int} can alias an @code{int}, but not a
5649 @code{void*} or a @code{double}. A character type may alias any other
5652 Pay special attention to code like this:
5665 The practice of reading from a different union member than the one most
5666 recently written to (called ``type-punning'') is common. Even with
5667 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5668 is accessed through the union type. So, the code above will work as
5669 expected. However, this code might not:
5680 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5682 @item -fstrict-overflow
5683 @opindex fstrict-overflow
5684 Allow the compiler to assume strict signed overflow rules, depending
5685 on the language being compiled. For C (and C++) this means that
5686 overflow when doing arithmetic with signed numbers is undefined, which
5687 means that the compiler may assume that it will not happen. This
5688 permits various optimizations. For example, the compiler will assume
5689 that an expression like @code{i + 10 > i} will always be true for
5690 signed @code{i}. This assumption is only valid if signed overflow is
5691 undefined, as the expression is false if @code{i + 10} overflows when
5692 using twos complement arithmetic. When this option is in effect any
5693 attempt to determine whether an operation on signed numbers will
5694 overflow must be written carefully to not actually involve overflow.
5696 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
5697 that signed overflow is fully defined: it wraps. When
5698 @option{-fwrapv} is used, there is no difference between
5699 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
5700 @option{-fwrapv} certain types of overflow are permitted. For
5701 example, if the compiler gets an overflow when doing arithmetic on
5702 constants, the overflowed value can still be used with
5703 @option{-fwrapv}, but not otherwise.
5705 The @option{-fstrict-overflow} option is enabled at levels
5706 @option{-O2}, @option{-O3}, @option{-Os}.
5708 @item -falign-functions
5709 @itemx -falign-functions=@var{n}
5710 @opindex falign-functions
5711 Align the start of functions to the next power-of-two greater than
5712 @var{n}, skipping up to @var{n} bytes. For instance,
5713 @option{-falign-functions=32} aligns functions to the next 32-byte
5714 boundary, but @option{-falign-functions=24} would align to the next
5715 32-byte boundary only if this can be done by skipping 23 bytes or less.
5717 @option{-fno-align-functions} and @option{-falign-functions=1} are
5718 equivalent and mean that functions will not be aligned.
5720 Some assemblers only support this flag when @var{n} is a power of two;
5721 in that case, it is rounded up.
5723 If @var{n} is not specified or is zero, use a machine-dependent default.
5725 Enabled at levels @option{-O2}, @option{-O3}.
5727 @item -falign-labels
5728 @itemx -falign-labels=@var{n}
5729 @opindex falign-labels
5730 Align all branch targets to a power-of-two boundary, skipping up to
5731 @var{n} bytes like @option{-falign-functions}. This option can easily
5732 make code slower, because it must insert dummy operations for when the
5733 branch target is reached in the usual flow of the code.
5735 @option{-fno-align-labels} and @option{-falign-labels=1} are
5736 equivalent and mean that labels will not be aligned.
5738 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5739 are greater than this value, then their values are used instead.
5741 If @var{n} is not specified or is zero, use a machine-dependent default
5742 which is very likely to be @samp{1}, meaning no alignment.
5744 Enabled at levels @option{-O2}, @option{-O3}.
5747 @itemx -falign-loops=@var{n}
5748 @opindex falign-loops
5749 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5750 like @option{-falign-functions}. The hope is that the loop will be
5751 executed many times, which will make up for any execution of the dummy
5754 @option{-fno-align-loops} and @option{-falign-loops=1} are
5755 equivalent and mean that loops will not be aligned.
5757 If @var{n} is not specified or is zero, use a machine-dependent default.
5759 Enabled at levels @option{-O2}, @option{-O3}.
5762 @itemx -falign-jumps=@var{n}
5763 @opindex falign-jumps
5764 Align branch targets to a power-of-two boundary, for branch targets
5765 where the targets can only be reached by jumping, skipping up to @var{n}
5766 bytes like @option{-falign-functions}. In this case, no dummy operations
5769 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5770 equivalent and mean that loops will not be aligned.
5772 If @var{n} is not specified or is zero, use a machine-dependent default.
5774 Enabled at levels @option{-O2}, @option{-O3}.
5776 @item -funit-at-a-time
5777 @opindex funit-at-a-time
5778 Parse the whole compilation unit before starting to produce code.
5779 This allows some extra optimizations to take place but consumes
5780 more memory (in general). There are some compatibility issues
5781 with @emph{unit-at-a-time} mode:
5784 enabling @emph{unit-at-a-time} mode may change the order
5785 in which functions, variables, and top-level @code{asm} statements
5786 are emitted, and will likely break code relying on some particular
5787 ordering. The majority of such top-level @code{asm} statements,
5788 though, can be replaced by @code{section} attributes. The
5789 @option{fno-toplevel-reorder} option may be used to keep the ordering
5790 used in the input file, at the cost of some optimizations.
5793 @emph{unit-at-a-time} mode removes unreferenced static variables
5794 and functions. This may result in undefined references
5795 when an @code{asm} statement refers directly to variables or functions
5796 that are otherwise unused. In that case either the variable/function
5797 shall be listed as an operand of the @code{asm} statement operand or,
5798 in the case of top-level @code{asm} statements the attribute @code{used}
5799 shall be used on the declaration.
5802 Static functions now can use non-standard passing conventions that
5803 may break @code{asm} statements calling functions directly. Again,
5804 attribute @code{used} will prevent this behavior.
5807 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5808 but this scheme may not be supported by future releases of GCC@.
5810 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5812 @item -fno-toplevel-reorder
5813 Do not reorder top-level functions, variables, and @code{asm}
5814 statements. Output them in the same order that they appear in the
5815 input file. When this option is used, unreferenced static variables
5816 will not be removed. This option is intended to support existing code
5817 which relies on a particular ordering. For new code, it is better to
5822 Constructs webs as commonly used for register allocation purposes and assign
5823 each web individual pseudo register. This allows the register allocation pass
5824 to operate on pseudos directly, but also strengthens several other optimization
5825 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5826 however, make debugging impossible, since variables will no longer stay in a
5829 Enabled by default with @option{-funroll-loops}.
5831 @item -fwhole-program
5832 @opindex fwhole-program
5833 Assume that the current compilation unit represents whole program being
5834 compiled. All public functions and variables with the exception of @code{main}
5835 and those merged by attribute @code{externally_visible} become static functions
5836 and in a affect gets more aggressively optimized by interprocedural optimizers.
5837 While this option is equivalent to proper use of @code{static} keyword for
5838 programs consisting of single file, in combination with option
5839 @option{--combine} this flag can be used to compile most of smaller scale C
5840 programs since the functions and variables become local for the whole combined
5841 compilation unit, not for the single source file itself.
5844 @item -fno-cprop-registers
5845 @opindex fno-cprop-registers
5846 After register allocation and post-register allocation instruction splitting,
5847 we perform a copy-propagation pass to try to reduce scheduling dependencies
5848 and occasionally eliminate the copy.
5850 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5852 @item -fprofile-generate
5853 @opindex fprofile-generate
5855 Enable options usually used for instrumenting application to produce
5856 profile useful for later recompilation with profile feedback based
5857 optimization. You must use @option{-fprofile-generate} both when
5858 compiling and when linking your program.
5860 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5863 @opindex fprofile-use
5864 Enable profile feedback directed optimizations, and optimizations
5865 generally profitable only with profile feedback available.
5867 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5868 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5870 By default, GCC emits an error message if the feedback profiles do not
5871 match the source code. This error can be turned into a warning by using
5872 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
5876 The following options control compiler behavior regarding floating
5877 point arithmetic. These options trade off between speed and
5878 correctness. All must be specifically enabled.
5882 @opindex ffloat-store
5883 Do not store floating point variables in registers, and inhibit other
5884 options that might change whether a floating point value is taken from a
5887 @cindex floating point precision
5888 This option prevents undesirable excess precision on machines such as
5889 the 68000 where the floating registers (of the 68881) keep more
5890 precision than a @code{double} is supposed to have. Similarly for the
5891 x86 architecture. For most programs, the excess precision does only
5892 good, but a few programs rely on the precise definition of IEEE floating
5893 point. Use @option{-ffloat-store} for such programs, after modifying
5894 them to store all pertinent intermediate computations into variables.
5898 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5899 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5900 @option{-fno-rounding-math}, @option{-fno-signaling-nans},
5901 @option{-fno-signed-zeros} and @option{fcx-limited-range}.
5903 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5905 This option should never be turned on by any @option{-O} option since
5906 it can result in incorrect output for programs which depend on
5907 an exact implementation of IEEE or ISO rules/specifications for
5910 @item -fno-math-errno
5911 @opindex fno-math-errno
5912 Do not set ERRNO after calling math functions that are executed
5913 with a single instruction, e.g., sqrt. A program that relies on
5914 IEEE exceptions for math error handling may want to use this flag
5915 for speed while maintaining IEEE arithmetic compatibility.
5917 This option should never be turned on by any @option{-O} option since
5918 it can result in incorrect output for programs which depend on
5919 an exact implementation of IEEE or ISO rules/specifications for
5922 The default is @option{-fmath-errno}.
5924 On Darwin systems, the math library never sets @code{errno}. There is
5925 therefore no reason for the compiler to consider the possibility that
5926 it might, and @option{-fno-math-errno} is the default.
5928 @item -funsafe-math-optimizations
5929 @opindex funsafe-math-optimizations
5930 Allow optimizations for floating-point arithmetic that (a) assume
5931 that arguments and results are valid and (b) may violate IEEE or
5932 ANSI standards. When used at link-time, it may include libraries
5933 or startup files that change the default FPU control word or other
5934 similar optimizations.
5936 This option should never be turned on by any @option{-O} option since
5937 it can result in incorrect output for programs which depend on
5938 an exact implementation of IEEE or ISO rules/specifications for
5941 The default is @option{-fno-unsafe-math-optimizations}.
5943 @item -ffinite-math-only
5944 @opindex ffinite-math-only
5945 Allow optimizations for floating-point arithmetic that assume
5946 that arguments and results are not NaNs or +-Infs.
5948 This option should never be turned on by any @option{-O} option since
5949 it can result in incorrect output for programs which depend on
5950 an exact implementation of IEEE or ISO rules/specifications.
5952 The default is @option{-fno-finite-math-only}.
5954 @item -fno-signed-zeros
5955 @opindex fno-signed-zeros
5956 Allow optimizations for floating point arithmetic that ignore the
5957 signedness of zero. IEEE arithmetic specifies the behavior of
5958 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
5959 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
5960 This option implies that the sign of a zero result isn't significant.
5962 The default is @option{-fsigned-zeros}.
5964 @item -fno-trapping-math
5965 @opindex fno-trapping-math
5966 Compile code assuming that floating-point operations cannot generate
5967 user-visible traps. These traps include division by zero, overflow,
5968 underflow, inexact result and invalid operation. This option implies
5969 @option{-fno-signaling-nans}. Setting this option may allow faster
5970 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5972 This option should never be turned on by any @option{-O} option since
5973 it can result in incorrect output for programs which depend on
5974 an exact implementation of IEEE or ISO rules/specifications for
5977 The default is @option{-ftrapping-math}.
5979 @item -frounding-math
5980 @opindex frounding-math
5981 Disable transformations and optimizations that assume default floating
5982 point rounding behavior. This is round-to-zero for all floating point
5983 to integer conversions, and round-to-nearest for all other arithmetic
5984 truncations. This option should be specified for programs that change
5985 the FP rounding mode dynamically, or that may be executed with a
5986 non-default rounding mode. This option disables constant folding of
5987 floating point expressions at compile-time (which may be affected by
5988 rounding mode) and arithmetic transformations that are unsafe in the
5989 presence of sign-dependent rounding modes.
5991 The default is @option{-fno-rounding-math}.
5993 This option is experimental and does not currently guarantee to
5994 disable all GCC optimizations that are affected by rounding mode.
5995 Future versions of GCC may provide finer control of this setting
5996 using C99's @code{FENV_ACCESS} pragma. This command line option
5997 will be used to specify the default state for @code{FENV_ACCESS}.
5999 @item -frtl-abstract-sequences
6000 @opindex frtl-abstract-sequences
6001 It is a size optimization method. This option is to find identical
6002 sequences of code, which can be turned into pseudo-procedures and
6003 then replace all occurrences with calls to the newly created
6004 subroutine. It is kind of an opposite of @option{-finline-functions}.
6005 This optimization runs at RTL level.
6007 @item -fsignaling-nans
6008 @opindex fsignaling-nans
6009 Compile code assuming that IEEE signaling NaNs may generate user-visible
6010 traps during floating-point operations. Setting this option disables
6011 optimizations that may change the number of exceptions visible with
6012 signaling NaNs. This option implies @option{-ftrapping-math}.
6014 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6017 The default is @option{-fno-signaling-nans}.
6019 This option is experimental and does not currently guarantee to
6020 disable all GCC optimizations that affect signaling NaN behavior.
6022 @item -fsingle-precision-constant
6023 @opindex fsingle-precision-constant
6024 Treat floating point constant as single precision constant instead of
6025 implicitly converting it to double precision constant.
6027 @item -fcx-limited-range
6028 @itemx -fno-cx-limited-range
6029 @opindex fcx-limited-range
6030 @opindex fno-cx-limited-range
6031 When enabled, this option states that a range reduction step is not
6032 needed when performing complex division. The default is
6033 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
6035 This option controls the default setting of the ISO C99
6036 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6041 The following options control optimizations that may improve
6042 performance, but are not enabled by any @option{-O} options. This
6043 section includes experimental options that may produce broken code.
6046 @item -fbranch-probabilities
6047 @opindex fbranch-probabilities
6048 After running a program compiled with @option{-fprofile-arcs}
6049 (@pxref{Debugging Options,, Options for Debugging Your Program or
6050 @command{gcc}}), you can compile it a second time using
6051 @option{-fbranch-probabilities}, to improve optimizations based on
6052 the number of times each branch was taken. When the program
6053 compiled with @option{-fprofile-arcs} exits it saves arc execution
6054 counts to a file called @file{@var{sourcename}.gcda} for each source
6055 file. The information in this data file is very dependent on the
6056 structure of the generated code, so you must use the same source code
6057 and the same optimization options for both compilations.
6059 With @option{-fbranch-probabilities}, GCC puts a
6060 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6061 These can be used to improve optimization. Currently, they are only
6062 used in one place: in @file{reorg.c}, instead of guessing which path a
6063 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6064 exactly determine which path is taken more often.
6066 @item -fprofile-values
6067 @opindex fprofile-values
6068 If combined with @option{-fprofile-arcs}, it adds code so that some
6069 data about values of expressions in the program is gathered.
6071 With @option{-fbranch-probabilities}, it reads back the data gathered
6072 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6073 notes to instructions for their later usage in optimizations.
6075 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6079 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6080 a code to gather information about values of expressions.
6082 With @option{-fbranch-probabilities}, it reads back the data gathered
6083 and actually performs the optimizations based on them.
6084 Currently the optimizations include specialization of division operation
6085 using the knowledge about the value of the denominator.
6087 @item -frename-registers
6088 @opindex frename-registers
6089 Attempt to avoid false dependencies in scheduled code by making use
6090 of registers left over after register allocation. This optimization
6091 will most benefit processors with lots of registers. Depending on the
6092 debug information format adopted by the target, however, it can
6093 make debugging impossible, since variables will no longer stay in
6094 a ``home register''.
6096 Enabled by default with @option{-funroll-loops}.
6100 Perform tail duplication to enlarge superblock size. This transformation
6101 simplifies the control flow of the function allowing other optimizations to do
6104 Enabled with @option{-fprofile-use}.
6106 @item -funroll-loops
6107 @opindex funroll-loops
6108 Unroll loops whose number of iterations can be determined at compile time or
6109 upon entry to the loop. @option{-funroll-loops} implies
6110 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6111 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6112 small constant number of iterations). This option makes code larger, and may
6113 or may not make it run faster.
6115 Enabled with @option{-fprofile-use}.
6117 @item -funroll-all-loops
6118 @opindex funroll-all-loops
6119 Unroll all loops, even if their number of iterations is uncertain when
6120 the loop is entered. This usually makes programs run more slowly.
6121 @option{-funroll-all-loops} implies the same options as
6122 @option{-funroll-loops}.
6125 @opindex fpeel-loops
6126 Peels the loops for that there is enough information that they do not
6127 roll much (from profile feedback). It also turns on complete loop peeling
6128 (i.e.@: complete removal of loops with small constant number of iterations).
6130 Enabled with @option{-fprofile-use}.
6132 @item -fmove-loop-invariants
6133 @opindex fmove-loop-invariants
6134 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6135 at level @option{-O1}
6137 @item -funswitch-loops
6138 @opindex funswitch-loops
6139 Move branches with loop invariant conditions out of the loop, with duplicates
6140 of the loop on both branches (modified according to result of the condition).
6142 @item -ffunction-sections
6143 @itemx -fdata-sections
6144 @opindex ffunction-sections
6145 @opindex fdata-sections
6146 Place each function or data item into its own section in the output
6147 file if the target supports arbitrary sections. The name of the
6148 function or the name of the data item determines the section's name
6151 Use these options on systems where the linker can perform optimizations
6152 to improve locality of reference in the instruction space. Most systems
6153 using the ELF object format and SPARC processors running Solaris 2 have
6154 linkers with such optimizations. AIX may have these optimizations in
6157 Only use these options when there are significant benefits from doing
6158 so. When you specify these options, the assembler and linker will
6159 create larger object and executable files and will also be slower.
6160 You will not be able to use @code{gprof} on all systems if you
6161 specify this option and you may have problems with debugging if
6162 you specify both this option and @option{-g}.
6164 @item -fbranch-target-load-optimize
6165 @opindex fbranch-target-load-optimize
6166 Perform branch target register load optimization before prologue / epilogue
6168 The use of target registers can typically be exposed only during reload,
6169 thus hoisting loads out of loops and doing inter-block scheduling needs
6170 a separate optimization pass.
6172 @item -fbranch-target-load-optimize2
6173 @opindex fbranch-target-load-optimize2
6174 Perform branch target register load optimization after prologue / epilogue
6177 @item -fbtr-bb-exclusive
6178 @opindex fbtr-bb-exclusive
6179 When performing branch target register load optimization, don't reuse
6180 branch target registers in within any basic block.
6182 @item -fstack-protector
6183 Emit extra code to check for buffer overflows, such as stack smashing
6184 attacks. This is done by adding a guard variable to functions with
6185 vulnerable objects. This includes functions that call alloca, and
6186 functions with buffers larger than 8 bytes. The guards are initialized
6187 when a function is entered and then checked when the function exits.
6188 If a guard check fails, an error message is printed and the program exits.
6190 @item -fstack-protector-all
6191 Like @option{-fstack-protector} except that all functions are protected.
6193 @item -fsection-anchors
6194 @opindex fsection-anchors
6195 Try to reduce the number of symbolic address calculations by using
6196 shared ``anchor'' symbols to address nearby objects. This transformation
6197 can help to reduce the number of GOT entries and GOT accesses on some
6200 For example, the implementation of the following function @code{foo}:
6204 int foo (void) @{ return a + b + c; @}
6207 would usually calculate the addresses of all three variables, but if you
6208 compile it with @option{-fsection-anchors}, it will access the variables
6209 from a common anchor point instead. The effect is similar to the
6210 following pseudocode (which isn't valid C):
6215 register int *xr = &x;
6216 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6220 Not all targets support this option.
6222 @item --param @var{name}=@var{value}
6224 In some places, GCC uses various constants to control the amount of
6225 optimization that is done. For example, GCC will not inline functions
6226 that contain more that a certain number of instructions. You can
6227 control some of these constants on the command-line using the
6228 @option{--param} option.
6230 The names of specific parameters, and the meaning of the values, are
6231 tied to the internals of the compiler, and are subject to change
6232 without notice in future releases.
6234 In each case, the @var{value} is an integer. The allowable choices for
6235 @var{name} are given in the following table:
6238 @item salias-max-implicit-fields
6239 The maximum number of fields in a variable without direct
6240 structure accesses for which structure aliasing will consider trying
6241 to track each field. The default is 5
6243 @item salias-max-array-elements
6244 The maximum number of elements an array can have and its elements
6245 still be tracked individually by structure aliasing. The default is 4
6247 @item sra-max-structure-size
6248 The maximum structure size, in bytes, at which the scalar replacement
6249 of aggregates (SRA) optimization will perform block copies. The
6250 default value, 0, implies that GCC will select the most appropriate
6253 @item sra-field-structure-ratio
6254 The threshold ratio (as a percentage) between instantiated fields and
6255 the complete structure size. We say that if the ratio of the number
6256 of bytes in instantiated fields to the number of bytes in the complete
6257 structure exceeds this parameter, then block copies are not used. The
6260 @item max-crossjump-edges
6261 The maximum number of incoming edges to consider for crossjumping.
6262 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6263 the number of edges incoming to each block. Increasing values mean
6264 more aggressive optimization, making the compile time increase with
6265 probably small improvement in executable size.
6267 @item min-crossjump-insns
6268 The minimum number of instructions which must be matched at the end
6269 of two blocks before crossjumping will be performed on them. This
6270 value is ignored in the case where all instructions in the block being
6271 crossjumped from are matched. The default value is 5.
6273 @item max-grow-copy-bb-insns
6274 The maximum code size expansion factor when copying basic blocks
6275 instead of jumping. The expansion is relative to a jump instruction.
6276 The default value is 8.
6278 @item max-goto-duplication-insns
6279 The maximum number of instructions to duplicate to a block that jumps
6280 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6281 passes, GCC factors computed gotos early in the compilation process,
6282 and unfactors them as late as possible. Only computed jumps at the
6283 end of a basic blocks with no more than max-goto-duplication-insns are
6284 unfactored. The default value is 8.
6286 @item max-delay-slot-insn-search
6287 The maximum number of instructions to consider when looking for an
6288 instruction to fill a delay slot. If more than this arbitrary number of
6289 instructions is searched, the time savings from filling the delay slot
6290 will be minimal so stop searching. Increasing values mean more
6291 aggressive optimization, making the compile time increase with probably
6292 small improvement in executable run time.
6294 @item max-delay-slot-live-search
6295 When trying to fill delay slots, the maximum number of instructions to
6296 consider when searching for a block with valid live register
6297 information. Increasing this arbitrarily chosen value means more
6298 aggressive optimization, increasing the compile time. This parameter
6299 should be removed when the delay slot code is rewritten to maintain the
6302 @item max-gcse-memory
6303 The approximate maximum amount of memory that will be allocated in
6304 order to perform the global common subexpression elimination
6305 optimization. If more memory than specified is required, the
6306 optimization will not be done.
6308 @item max-gcse-passes
6309 The maximum number of passes of GCSE to run. The default is 1.
6311 @item max-pending-list-length
6312 The maximum number of pending dependencies scheduling will allow
6313 before flushing the current state and starting over. Large functions
6314 with few branches or calls can create excessively large lists which
6315 needlessly consume memory and resources.
6317 @item max-inline-insns-single
6318 Several parameters control the tree inliner used in gcc.
6319 This number sets the maximum number of instructions (counted in GCC's
6320 internal representation) in a single function that the tree inliner
6321 will consider for inlining. This only affects functions declared
6322 inline and methods implemented in a class declaration (C++).
6323 The default value is 450.
6325 @item max-inline-insns-auto
6326 When you use @option{-finline-functions} (included in @option{-O3}),
6327 a lot of functions that would otherwise not be considered for inlining
6328 by the compiler will be investigated. To those functions, a different
6329 (more restrictive) limit compared to functions declared inline can
6331 The default value is 90.
6333 @item large-function-insns
6334 The limit specifying really large functions. For functions larger than this
6335 limit after inlining inlining is constrained by
6336 @option{--param large-function-growth}. This parameter is useful primarily
6337 to avoid extreme compilation time caused by non-linear algorithms used by the
6339 This parameter is ignored when @option{-funit-at-a-time} is not used.
6340 The default value is 2700.
6342 @item large-function-growth
6343 Specifies maximal growth of large function caused by inlining in percents.
6344 This parameter is ignored when @option{-funit-at-a-time} is not used.
6345 The default value is 100 which limits large function growth to 2.0 times
6348 @item large-unit-insns
6349 The limit specifying large translation unit. Growth caused by inlining of
6350 units larger than this limit is limited by @option{--param inline-unit-growth}.
6351 For small units this might be too tight (consider unit consisting of function A
6352 that is inline and B that just calls A three time. If B is small relative to
6353 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6354 large units consisting of small inlininable functions however the overall unit
6355 growth limit is needed to avoid exponential explosion of code size. Thus for
6356 smaller units, the size is increased to @option{--param large-unit-insns}
6357 before applying @option{--param inline-unit-growth}. The default is 10000
6359 @item inline-unit-growth
6360 Specifies maximal overall growth of the compilation unit caused by inlining.
6361 This parameter is ignored when @option{-funit-at-a-time} is not used.
6362 The default value is 30 which limits unit growth to 1.3 times the original
6365 @item large-stack-frame
6366 The limit specifying large stack frames. While inlining the algorithm is trying
6367 to not grow past this limit too much. Default value is 256 bytes.
6369 @item large-stack-frame-growth
6370 Specifies maximal growth of large stack frames caused by inlining in percents.
6371 The default value is 1000 which limits large stack frame growth to 11 times
6374 @item max-inline-insns-recursive
6375 @itemx max-inline-insns-recursive-auto
6376 Specifies maximum number of instructions out-of-line copy of self recursive inline
6377 function can grow into by performing recursive inlining.
6379 For functions declared inline @option{--param max-inline-insns-recursive} is
6380 taken into account. For function not declared inline, recursive inlining
6381 happens only when @option{-finline-functions} (included in @option{-O3}) is
6382 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6383 default value is 450.
6385 @item max-inline-recursive-depth
6386 @itemx max-inline-recursive-depth-auto
6387 Specifies maximum recursion depth used by the recursive inlining.
6389 For functions declared inline @option{--param max-inline-recursive-depth} is
6390 taken into account. For function not declared inline, recursive inlining
6391 happens only when @option{-finline-functions} (included in @option{-O3}) is
6392 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6393 default value is 450.
6395 @item min-inline-recursive-probability
6396 Recursive inlining is profitable only for function having deep recursion
6397 in average and can hurt for function having little recursion depth by
6398 increasing the prologue size or complexity of function body to other
6401 When profile feedback is available (see @option{-fprofile-generate}) the actual
6402 recursion depth can be guessed from probability that function will recurse via
6403 given call expression. This parameter limits inlining only to call expression
6404 whose probability exceeds given threshold (in percents). The default value is
6407 @item inline-call-cost
6408 Specify cost of call instruction relative to simple arithmetics operations
6409 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6410 functions and at the same time increases size of leaf function that is believed to
6411 reduce function size by being inlined. In effect it increases amount of
6412 inlining for code having large abstraction penalty (many functions that just
6413 pass the arguments to other functions) and decrease inlining for code with low
6414 abstraction penalty. The default value is 16.
6416 @item min-vect-loop-bound
6417 The minimum number of iterations under which a loop will not get vectorized
6418 when @option{-ftree-vectorize} is used. The number of iterations after
6419 vectorization needs to be greater than the value specified by this option
6420 to allow vectorization. The default value is 0.
6422 @item max-unrolled-insns
6423 The maximum number of instructions that a loop should have if that loop
6424 is unrolled, and if the loop is unrolled, it determines how many times
6425 the loop code is unrolled.
6427 @item max-average-unrolled-insns
6428 The maximum number of instructions biased by probabilities of their execution
6429 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6430 it determines how many times the loop code is unrolled.
6432 @item max-unroll-times
6433 The maximum number of unrollings of a single loop.
6435 @item max-peeled-insns
6436 The maximum number of instructions that a loop should have if that loop
6437 is peeled, and if the loop is peeled, it determines how many times
6438 the loop code is peeled.
6440 @item max-peel-times
6441 The maximum number of peelings of a single loop.
6443 @item max-completely-peeled-insns
6444 The maximum number of insns of a completely peeled loop.
6446 @item max-completely-peel-times
6447 The maximum number of iterations of a loop to be suitable for complete peeling.
6449 @item max-unswitch-insns
6450 The maximum number of insns of an unswitched loop.
6452 @item max-unswitch-level
6453 The maximum number of branches unswitched in a single loop.
6456 The minimum cost of an expensive expression in the loop invariant motion.
6458 @item iv-consider-all-candidates-bound
6459 Bound on number of candidates for induction variables below that
6460 all candidates are considered for each use in induction variable
6461 optimizations. Only the most relevant candidates are considered
6462 if there are more candidates, to avoid quadratic time complexity.
6464 @item iv-max-considered-uses
6465 The induction variable optimizations give up on loops that contain more
6466 induction variable uses.
6468 @item iv-always-prune-cand-set-bound
6469 If number of candidates in the set is smaller than this value,
6470 we always try to remove unnecessary ivs from the set during its
6471 optimization when a new iv is added to the set.
6473 @item scev-max-expr-size
6474 Bound on size of expressions used in the scalar evolutions analyzer.
6475 Large expressions slow the analyzer.
6477 @item omega-max-vars
6478 The maximum number of variables in an Omega constraint system.
6479 The default value is 128.
6481 @item omega-max-geqs
6482 The maximum number of inequalities in an Omega constraint system.
6483 The default value is 256.
6486 The maximum number of equalities in an Omega constraint system.
6487 The default value is 128.
6489 @item omega-max-wild-cards
6490 The maximum number of wildcard variables that the Omega solver will
6491 be able to insert. The default value is 18.
6493 @item omega-hash-table-size
6494 The size of the hash table in the Omega solver. The default value is
6497 @item omega-max-keys
6498 The maximal number of keys used by the Omega solver. The default
6501 @item omega-eliminate-redundant-constraints
6502 When set to 1, use expensive methods to eliminate all redundant
6503 constraints. The default value is 0.
6505 @item vect-max-version-checks
6506 The maximum number of runtime checks that can be performed when doing
6507 loop versioning in the vectorizer. See option ftree-vect-loop-version
6508 for more information.
6510 @item max-iterations-to-track
6512 The maximum number of iterations of a loop the brute force algorithm
6513 for analysis of # of iterations of the loop tries to evaluate.
6515 @item hot-bb-count-fraction
6516 Select fraction of the maximal count of repetitions of basic block in program
6517 given basic block needs to have to be considered hot.
6519 @item hot-bb-frequency-fraction
6520 Select fraction of the maximal frequency of executions of basic block in
6521 function given basic block needs to have to be considered hot
6523 @item max-predicted-iterations
6524 The maximum number of loop iterations we predict statically. This is useful
6525 in cases where function contain single loop with known bound and other loop
6526 with unknown. We predict the known number of iterations correctly, while
6527 the unknown number of iterations average to roughly 10. This means that the
6528 loop without bounds would appear artificially cold relative to the other one.
6530 @item tracer-dynamic-coverage
6531 @itemx tracer-dynamic-coverage-feedback
6533 This value is used to limit superblock formation once the given percentage of
6534 executed instructions is covered. This limits unnecessary code size
6537 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6538 feedback is available. The real profiles (as opposed to statically estimated
6539 ones) are much less balanced allowing the threshold to be larger value.
6541 @item tracer-max-code-growth
6542 Stop tail duplication once code growth has reached given percentage. This is
6543 rather hokey argument, as most of the duplicates will be eliminated later in
6544 cross jumping, so it may be set to much higher values than is the desired code
6547 @item tracer-min-branch-ratio
6549 Stop reverse growth when the reverse probability of best edge is less than this
6550 threshold (in percent).
6552 @item tracer-min-branch-ratio
6553 @itemx tracer-min-branch-ratio-feedback
6555 Stop forward growth if the best edge do have probability lower than this
6558 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6559 compilation for profile feedback and one for compilation without. The value
6560 for compilation with profile feedback needs to be more conservative (higher) in
6561 order to make tracer effective.
6563 @item max-cse-path-length
6565 Maximum number of basic blocks on path that cse considers. The default is 10.
6568 The maximum instructions CSE process before flushing. The default is 1000.
6570 @item max-aliased-vops
6572 Maximum number of virtual operands per statement allowed to represent
6573 aliases before triggering the alias grouping heuristic. Alias
6574 grouping reduces compile times and memory consumption needed for
6575 aliasing at the expense of precision loss in alias information.
6577 @item ggc-min-expand
6579 GCC uses a garbage collector to manage its own memory allocation. This
6580 parameter specifies the minimum percentage by which the garbage
6581 collector's heap should be allowed to expand between collections.
6582 Tuning this may improve compilation speed; it has no effect on code
6585 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6586 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6587 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6588 GCC is not able to calculate RAM on a particular platform, the lower
6589 bound of 30% is used. Setting this parameter and
6590 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6591 every opportunity. This is extremely slow, but can be useful for
6594 @item ggc-min-heapsize
6596 Minimum size of the garbage collector's heap before it begins bothering
6597 to collect garbage. The first collection occurs after the heap expands
6598 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6599 tuning this may improve compilation speed, and has no effect on code
6602 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6603 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6604 with a lower bound of 4096 (four megabytes) and an upper bound of
6605 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6606 particular platform, the lower bound is used. Setting this parameter
6607 very large effectively disables garbage collection. Setting this
6608 parameter and @option{ggc-min-expand} to zero causes a full collection
6609 to occur at every opportunity.
6611 @item max-reload-search-insns
6612 The maximum number of instruction reload should look backward for equivalent
6613 register. Increasing values mean more aggressive optimization, making the
6614 compile time increase with probably slightly better performance. The default
6617 @item max-cselib-memory-locations
6618 The maximum number of memory locations cselib should take into account.
6619 Increasing values mean more aggressive optimization, making the compile time
6620 increase with probably slightly better performance. The default value is 500.
6622 @item max-flow-memory-locations
6623 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6624 The default value is 100.
6626 @item reorder-blocks-duplicate
6627 @itemx reorder-blocks-duplicate-feedback
6629 Used by basic block reordering pass to decide whether to use unconditional
6630 branch or duplicate the code on its destination. Code is duplicated when its
6631 estimated size is smaller than this value multiplied by the estimated size of
6632 unconditional jump in the hot spots of the program.
6634 The @option{reorder-block-duplicate-feedback} is used only when profile
6635 feedback is available and may be set to higher values than
6636 @option{reorder-block-duplicate} since information about the hot spots is more
6639 @item max-sched-ready-insns
6640 The maximum number of instructions ready to be issued the scheduler should
6641 consider at any given time during the first scheduling pass. Increasing
6642 values mean more thorough searches, making the compilation time increase
6643 with probably little benefit. The default value is 100.
6645 @item max-sched-region-blocks
6646 The maximum number of blocks in a region to be considered for
6647 interblock scheduling. The default value is 10.
6649 @item max-sched-region-insns
6650 The maximum number of insns in a region to be considered for
6651 interblock scheduling. The default value is 100.
6654 The minimum probability (in percents) of reaching a source block
6655 for interblock speculative scheduling. The default value is 40.
6657 @item max-sched-extend-regions-iters
6658 The maximum number of iterations through CFG to extend regions.
6659 0 - disable region extension,
6660 N - do at most N iterations.
6661 The default value is 0.
6663 @item max-sched-insn-conflict-delay
6664 The maximum conflict delay for an insn to be considered for speculative motion.
6665 The default value is 3.
6667 @item sched-spec-prob-cutoff
6668 The minimal probability of speculation success (in percents), so that
6669 speculative insn will be scheduled.
6670 The default value is 40.
6672 @item max-last-value-rtl
6674 The maximum size measured as number of RTLs that can be recorded in an expression
6675 in combiner for a pseudo register as last known value of that register. The default
6678 @item integer-share-limit
6679 Small integer constants can use a shared data structure, reducing the
6680 compiler's memory usage and increasing its speed. This sets the maximum
6681 value of a shared integer constant's. The default value is 256.
6683 @item min-virtual-mappings
6684 Specifies the minimum number of virtual mappings in the incremental
6685 SSA updater that should be registered to trigger the virtual mappings
6686 heuristic defined by virtual-mappings-ratio. The default value is
6689 @item virtual-mappings-ratio
6690 If the number of virtual mappings is virtual-mappings-ratio bigger
6691 than the number of virtual symbols to be updated, then the incremental
6692 SSA updater switches to a full update for those symbols. The default
6695 @item ssp-buffer-size
6696 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6697 protection when @option{-fstack-protection} is used.
6699 @item max-jump-thread-duplication-stmts
6700 Maximum number of statements allowed in a block that needs to be
6701 duplicated when threading jumps.
6703 @item max-fields-for-field-sensitive
6704 Maximum number of fields in a structure we will treat in
6705 a field sensitive manner during pointer analysis.
6707 @item prefetch-latency
6708 Estimate on average number of instructions that are executed before
6709 prefetch finishes. The distance we prefetch ahead is proportional
6710 to this constant. Increasing this number may also lead to less
6711 streams being prefetched (see @option{simultaneous-prefetches}).
6713 @item simultaneous-prefetches
6714 Maximum number of prefetches that can run at the same time.
6716 @item l1-cache-line-size
6717 The size of cache line in L1 cache, in bytes.
6720 The number of cache lines in L1 cache.
6722 @item verify-canonical-types
6723 Whether the compiler should verify the ``canonical'' types used for
6724 type equality comparisons within the C++ and Objective-C++ front
6725 ends. Set to 1 (the default when GCC is configured with
6726 --enable-checking) to enable verification, 0 to disable verification
6727 (the default when GCC is configured with --disable-checking).
6732 @node Preprocessor Options
6733 @section Options Controlling the Preprocessor
6734 @cindex preprocessor options
6735 @cindex options, preprocessor
6737 These options control the C preprocessor, which is run on each C source
6738 file before actual compilation.
6740 If you use the @option{-E} option, nothing is done except preprocessing.
6741 Some of these options make sense only together with @option{-E} because
6742 they cause the preprocessor output to be unsuitable for actual
6747 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6748 and pass @var{option} directly through to the preprocessor. If
6749 @var{option} contains commas, it is split into multiple options at the
6750 commas. However, many options are modified, translated or interpreted
6751 by the compiler driver before being passed to the preprocessor, and
6752 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6753 interface is undocumented and subject to change, so whenever possible
6754 you should avoid using @option{-Wp} and let the driver handle the
6757 @item -Xpreprocessor @var{option}
6758 @opindex preprocessor
6759 Pass @var{option} as an option to the preprocessor. You can use this to
6760 supply system-specific preprocessor options which GCC does not know how to
6763 If you want to pass an option that takes an argument, you must use
6764 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6767 @include cppopts.texi
6769 @node Assembler Options
6770 @section Passing Options to the Assembler
6772 @c prevent bad page break with this line
6773 You can pass options to the assembler.
6776 @item -Wa,@var{option}
6778 Pass @var{option} as an option to the assembler. If @var{option}
6779 contains commas, it is split into multiple options at the commas.
6781 @item -Xassembler @var{option}
6783 Pass @var{option} as an option to the assembler. You can use this to
6784 supply system-specific assembler options which GCC does not know how to
6787 If you want to pass an option that takes an argument, you must use
6788 @option{-Xassembler} twice, once for the option and once for the argument.
6793 @section Options for Linking
6794 @cindex link options
6795 @cindex options, linking
6797 These options come into play when the compiler links object files into
6798 an executable output file. They are meaningless if the compiler is
6799 not doing a link step.
6803 @item @var{object-file-name}
6804 A file name that does not end in a special recognized suffix is
6805 considered to name an object file or library. (Object files are
6806 distinguished from libraries by the linker according to the file
6807 contents.) If linking is done, these object files are used as input
6816 If any of these options is used, then the linker is not run, and
6817 object file names should not be used as arguments. @xref{Overall
6821 @item -l@var{library}
6822 @itemx -l @var{library}
6824 Search the library named @var{library} when linking. (The second
6825 alternative with the library as a separate argument is only for
6826 POSIX compliance and is not recommended.)
6828 It makes a difference where in the command you write this option; the
6829 linker searches and processes libraries and object files in the order they
6830 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6831 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6832 to functions in @samp{z}, those functions may not be loaded.
6834 The linker searches a standard list of directories for the library,
6835 which is actually a file named @file{lib@var{library}.a}. The linker
6836 then uses this file as if it had been specified precisely by name.
6838 The directories searched include several standard system directories
6839 plus any that you specify with @option{-L}.
6841 Normally the files found this way are library files---archive files
6842 whose members are object files. The linker handles an archive file by
6843 scanning through it for members which define symbols that have so far
6844 been referenced but not defined. But if the file that is found is an
6845 ordinary object file, it is linked in the usual fashion. The only
6846 difference between using an @option{-l} option and specifying a file name
6847 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6848 and searches several directories.
6852 You need this special case of the @option{-l} option in order to
6853 link an Objective-C or Objective-C++ program.
6856 @opindex nostartfiles
6857 Do not use the standard system startup files when linking.
6858 The standard system libraries are used normally, unless @option{-nostdlib}
6859 or @option{-nodefaultlibs} is used.
6861 @item -nodefaultlibs
6862 @opindex nodefaultlibs
6863 Do not use the standard system libraries when linking.
6864 Only the libraries you specify will be passed to the linker.
6865 The standard startup files are used normally, unless @option{-nostartfiles}
6866 is used. The compiler may generate calls to @code{memcmp},
6867 @code{memset}, @code{memcpy} and @code{memmove}.
6868 These entries are usually resolved by entries in
6869 libc. These entry points should be supplied through some other
6870 mechanism when this option is specified.
6874 Do not use the standard system startup files or libraries when linking.
6875 No startup files and only the libraries you specify will be passed to
6876 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6877 @code{memcpy} and @code{memmove}.
6878 These entries are usually resolved by entries in
6879 libc. These entry points should be supplied through some other
6880 mechanism when this option is specified.
6882 @cindex @option{-lgcc}, use with @option{-nostdlib}
6883 @cindex @option{-nostdlib} and unresolved references
6884 @cindex unresolved references and @option{-nostdlib}
6885 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6886 @cindex @option{-nodefaultlibs} and unresolved references
6887 @cindex unresolved references and @option{-nodefaultlibs}
6888 One of the standard libraries bypassed by @option{-nostdlib} and
6889 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6890 that GCC uses to overcome shortcomings of particular machines, or special
6891 needs for some languages.
6892 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6893 Collection (GCC) Internals},
6894 for more discussion of @file{libgcc.a}.)
6895 In most cases, you need @file{libgcc.a} even when you want to avoid
6896 other standard libraries. In other words, when you specify @option{-nostdlib}
6897 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6898 This ensures that you have no unresolved references to internal GCC
6899 library subroutines. (For example, @samp{__main}, used to ensure C++
6900 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6901 GNU Compiler Collection (GCC) Internals}.)
6905 Produce a position independent executable on targets which support it.
6906 For predictable results, you must also specify the same set of options
6907 that were used to generate code (@option{-fpie}, @option{-fPIE},
6908 or model suboptions) when you specify this option.
6912 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6913 that support it. This instructs the linker to add all symbols, not
6914 only used ones, to the dynamic symbol table. This option is needed
6915 for some uses of @code{dlopen} or to allow obtaining backtraces
6916 from within a program.
6920 Remove all symbol table and relocation information from the executable.
6924 On systems that support dynamic linking, this prevents linking with the shared
6925 libraries. On other systems, this option has no effect.
6929 Produce a shared object which can then be linked with other objects to
6930 form an executable. Not all systems support this option. For predictable
6931 results, you must also specify the same set of options that were used to
6932 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6933 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6934 needs to build supplementary stub code for constructors to work. On
6935 multi-libbed systems, @samp{gcc -shared} must select the correct support
6936 libraries to link against. Failing to supply the correct flags may lead
6937 to subtle defects. Supplying them in cases where they are not necessary
6940 @item -shared-libgcc
6941 @itemx -static-libgcc
6942 @opindex shared-libgcc
6943 @opindex static-libgcc
6944 On systems that provide @file{libgcc} as a shared library, these options
6945 force the use of either the shared or static version respectively.
6946 If no shared version of @file{libgcc} was built when the compiler was
6947 configured, these options have no effect.
6949 There are several situations in which an application should use the
6950 shared @file{libgcc} instead of the static version. The most common
6951 of these is when the application wishes to throw and catch exceptions
6952 across different shared libraries. In that case, each of the libraries
6953 as well as the application itself should use the shared @file{libgcc}.
6955 Therefore, the G++ and GCJ drivers automatically add
6956 @option{-shared-libgcc} whenever you build a shared library or a main
6957 executable, because C++ and Java programs typically use exceptions, so
6958 this is the right thing to do.
6960 If, instead, you use the GCC driver to create shared libraries, you may
6961 find that they will not always be linked with the shared @file{libgcc}.
6962 If GCC finds, at its configuration time, that you have a non-GNU linker
6963 or a GNU linker that does not support option @option{--eh-frame-hdr},
6964 it will link the shared version of @file{libgcc} into shared libraries
6965 by default. Otherwise, it will take advantage of the linker and optimize
6966 away the linking with the shared version of @file{libgcc}, linking with
6967 the static version of libgcc by default. This allows exceptions to
6968 propagate through such shared libraries, without incurring relocation
6969 costs at library load time.
6971 However, if a library or main executable is supposed to throw or catch
6972 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6973 for the languages used in the program, or using the option
6974 @option{-shared-libgcc}, such that it is linked with the shared
6979 Bind references to global symbols when building a shared object. Warn
6980 about any unresolved references (unless overridden by the link editor
6981 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6984 @item -Xlinker @var{option}
6986 Pass @var{option} as an option to the linker. You can use this to
6987 supply system-specific linker options which GCC does not know how to
6990 If you want to pass an option that takes an argument, you must use
6991 @option{-Xlinker} twice, once for the option and once for the argument.
6992 For example, to pass @option{-assert definitions}, you must write
6993 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6994 @option{-Xlinker "-assert definitions"}, because this passes the entire
6995 string as a single argument, which is not what the linker expects.
6997 @item -Wl,@var{option}
6999 Pass @var{option} as an option to the linker. If @var{option} contains
7000 commas, it is split into multiple options at the commas.
7002 @item -u @var{symbol}
7004 Pretend the symbol @var{symbol} is undefined, to force linking of
7005 library modules to define it. You can use @option{-u} multiple times with
7006 different symbols to force loading of additional library modules.
7009 @node Directory Options
7010 @section Options for Directory Search
7011 @cindex directory options
7012 @cindex options, directory search
7015 These options specify directories to search for header files, for
7016 libraries and for parts of the compiler:
7021 Add the directory @var{dir} to the head of the list of directories to be
7022 searched for header files. This can be used to override a system header
7023 file, substituting your own version, since these directories are
7024 searched before the system header file directories. However, you should
7025 not use this option to add directories that contain vendor-supplied
7026 system header files (use @option{-isystem} for that). If you use more than
7027 one @option{-I} option, the directories are scanned in left-to-right
7028 order; the standard system directories come after.
7030 If a standard system include directory, or a directory specified with
7031 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7032 option will be ignored. The directory will still be searched but as a
7033 system directory at its normal position in the system include chain.
7034 This is to ensure that GCC's procedure to fix buggy system headers and
7035 the ordering for the include_next directive are not inadvertently changed.
7036 If you really need to change the search order for system directories,
7037 use the @option{-nostdinc} and/or @option{-isystem} options.
7039 @item -iquote@var{dir}
7041 Add the directory @var{dir} to the head of the list of directories to
7042 be searched for header files only for the case of @samp{#include
7043 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7044 otherwise just like @option{-I}.
7048 Add directory @var{dir} to the list of directories to be searched
7051 @item -B@var{prefix}
7053 This option specifies where to find the executables, libraries,
7054 include files, and data files of the compiler itself.
7056 The compiler driver program runs one or more of the subprograms
7057 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7058 @var{prefix} as a prefix for each program it tries to run, both with and
7059 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7061 For each subprogram to be run, the compiler driver first tries the
7062 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7063 was not specified, the driver tries two standard prefixes, which are
7064 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7065 those results in a file name that is found, the unmodified program
7066 name is searched for using the directories specified in your
7067 @env{PATH} environment variable.
7069 The compiler will check to see if the path provided by the @option{-B}
7070 refers to a directory, and if necessary it will add a directory
7071 separator character at the end of the path.
7073 @option{-B} prefixes that effectively specify directory names also apply
7074 to libraries in the linker, because the compiler translates these
7075 options into @option{-L} options for the linker. They also apply to
7076 includes files in the preprocessor, because the compiler translates these
7077 options into @option{-isystem} options for the preprocessor. In this case,
7078 the compiler appends @samp{include} to the prefix.
7080 The run-time support file @file{libgcc.a} can also be searched for using
7081 the @option{-B} prefix, if needed. If it is not found there, the two
7082 standard prefixes above are tried, and that is all. The file is left
7083 out of the link if it is not found by those means.
7085 Another way to specify a prefix much like the @option{-B} prefix is to use
7086 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7089 As a special kludge, if the path provided by @option{-B} is
7090 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7091 9, then it will be replaced by @file{[dir/]include}. This is to help
7092 with boot-strapping the compiler.
7094 @item -specs=@var{file}
7096 Process @var{file} after the compiler reads in the standard @file{specs}
7097 file, in order to override the defaults that the @file{gcc} driver
7098 program uses when determining what switches to pass to @file{cc1},
7099 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7100 @option{-specs=@var{file}} can be specified on the command line, and they
7101 are processed in order, from left to right.
7103 @item --sysroot=@var{dir}
7105 Use @var{dir} as the logical root directory for headers and libraries.
7106 For example, if the compiler would normally search for headers in
7107 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7108 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7110 If you use both this option and the @option{-isysroot} option, then
7111 the @option{--sysroot} option will apply to libraries, but the
7112 @option{-isysroot} option will apply to header files.
7114 The GNU linker (beginning with version 2.16) has the necessary support
7115 for this option. If your linker does not support this option, the
7116 header file aspect of @option{--sysroot} will still work, but the
7117 library aspect will not.
7121 This option has been deprecated. Please use @option{-iquote} instead for
7122 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7123 Any directories you specify with @option{-I} options before the @option{-I-}
7124 option are searched only for the case of @samp{#include "@var{file}"};
7125 they are not searched for @samp{#include <@var{file}>}.
7127 If additional directories are specified with @option{-I} options after
7128 the @option{-I-}, these directories are searched for all @samp{#include}
7129 directives. (Ordinarily @emph{all} @option{-I} directories are used
7132 In addition, the @option{-I-} option inhibits the use of the current
7133 directory (where the current input file came from) as the first search
7134 directory for @samp{#include "@var{file}"}. There is no way to
7135 override this effect of @option{-I-}. With @option{-I.} you can specify
7136 searching the directory which was current when the compiler was
7137 invoked. That is not exactly the same as what the preprocessor does
7138 by default, but it is often satisfactory.
7140 @option{-I-} does not inhibit the use of the standard system directories
7141 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7148 @section Specifying subprocesses and the switches to pass to them
7151 @command{gcc} is a driver program. It performs its job by invoking a
7152 sequence of other programs to do the work of compiling, assembling and
7153 linking. GCC interprets its command-line parameters and uses these to
7154 deduce which programs it should invoke, and which command-line options
7155 it ought to place on their command lines. This behavior is controlled
7156 by @dfn{spec strings}. In most cases there is one spec string for each
7157 program that GCC can invoke, but a few programs have multiple spec
7158 strings to control their behavior. The spec strings built into GCC can
7159 be overridden by using the @option{-specs=} command-line switch to specify
7162 @dfn{Spec files} are plaintext files that are used to construct spec
7163 strings. They consist of a sequence of directives separated by blank
7164 lines. The type of directive is determined by the first non-whitespace
7165 character on the line and it can be one of the following:
7168 @item %@var{command}
7169 Issues a @var{command} to the spec file processor. The commands that can
7173 @item %include <@var{file}>
7175 Search for @var{file} and insert its text at the current point in the
7178 @item %include_noerr <@var{file}>
7179 @cindex %include_noerr
7180 Just like @samp{%include}, but do not generate an error message if the include
7181 file cannot be found.
7183 @item %rename @var{old_name} @var{new_name}
7185 Rename the spec string @var{old_name} to @var{new_name}.
7189 @item *[@var{spec_name}]:
7190 This tells the compiler to create, override or delete the named spec
7191 string. All lines after this directive up to the next directive or
7192 blank line are considered to be the text for the spec string. If this
7193 results in an empty string then the spec will be deleted. (Or, if the
7194 spec did not exist, then nothing will happened.) Otherwise, if the spec
7195 does not currently exist a new spec will be created. If the spec does
7196 exist then its contents will be overridden by the text of this
7197 directive, unless the first character of that text is the @samp{+}
7198 character, in which case the text will be appended to the spec.
7200 @item [@var{suffix}]:
7201 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7202 and up to the next directive or blank line are considered to make up the
7203 spec string for the indicated suffix. When the compiler encounters an
7204 input file with the named suffix, it will processes the spec string in
7205 order to work out how to compile that file. For example:
7212 This says that any input file whose name ends in @samp{.ZZ} should be
7213 passed to the program @samp{z-compile}, which should be invoked with the
7214 command-line switch @option{-input} and with the result of performing the
7215 @samp{%i} substitution. (See below.)
7217 As an alternative to providing a spec string, the text that follows a
7218 suffix directive can be one of the following:
7221 @item @@@var{language}
7222 This says that the suffix is an alias for a known @var{language}. This is
7223 similar to using the @option{-x} command-line switch to GCC to specify a
7224 language explicitly. For example:
7231 Says that .ZZ files are, in fact, C++ source files.
7234 This causes an error messages saying:
7237 @var{name} compiler not installed on this system.
7241 GCC already has an extensive list of suffixes built into it.
7242 This directive will add an entry to the end of the list of suffixes, but
7243 since the list is searched from the end backwards, it is effectively
7244 possible to override earlier entries using this technique.
7248 GCC has the following spec strings built into it. Spec files can
7249 override these strings or create their own. Note that individual
7250 targets can also add their own spec strings to this list.
7253 asm Options to pass to the assembler
7254 asm_final Options to pass to the assembler post-processor
7255 cpp Options to pass to the C preprocessor
7256 cc1 Options to pass to the C compiler
7257 cc1plus Options to pass to the C++ compiler
7258 endfile Object files to include at the end of the link
7259 link Options to pass to the linker
7260 lib Libraries to include on the command line to the linker
7261 libgcc Decides which GCC support library to pass to the linker
7262 linker Sets the name of the linker
7263 predefines Defines to be passed to the C preprocessor
7264 signed_char Defines to pass to CPP to say whether @code{char} is signed
7266 startfile Object files to include at the start of the link
7269 Here is a small example of a spec file:
7275 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7278 This example renames the spec called @samp{lib} to @samp{old_lib} and
7279 then overrides the previous definition of @samp{lib} with a new one.
7280 The new definition adds in some extra command-line options before
7281 including the text of the old definition.
7283 @dfn{Spec strings} are a list of command-line options to be passed to their
7284 corresponding program. In addition, the spec strings can contain
7285 @samp{%}-prefixed sequences to substitute variable text or to
7286 conditionally insert text into the command line. Using these constructs
7287 it is possible to generate quite complex command lines.
7289 Here is a table of all defined @samp{%}-sequences for spec
7290 strings. Note that spaces are not generated automatically around the
7291 results of expanding these sequences. Therefore you can concatenate them
7292 together or combine them with constant text in a single argument.
7296 Substitute one @samp{%} into the program name or argument.
7299 Substitute the name of the input file being processed.
7302 Substitute the basename of the input file being processed.
7303 This is the substring up to (and not including) the last period
7304 and not including the directory.
7307 This is the same as @samp{%b}, but include the file suffix (text after
7311 Marks the argument containing or following the @samp{%d} as a
7312 temporary file name, so that that file will be deleted if GCC exits
7313 successfully. Unlike @samp{%g}, this contributes no text to the
7316 @item %g@var{suffix}
7317 Substitute a file name that has suffix @var{suffix} and is chosen
7318 once per compilation, and mark the argument in the same way as
7319 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7320 name is now chosen in a way that is hard to predict even when previously
7321 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7322 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7323 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7324 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7325 was simply substituted with a file name chosen once per compilation,
7326 without regard to any appended suffix (which was therefore treated
7327 just like ordinary text), making such attacks more likely to succeed.
7329 @item %u@var{suffix}
7330 Like @samp{%g}, but generates a new temporary file name even if
7331 @samp{%u@var{suffix}} was already seen.
7333 @item %U@var{suffix}
7334 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7335 new one if there is no such last file name. In the absence of any
7336 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7337 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7338 would involve the generation of two distinct file names, one
7339 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7340 simply substituted with a file name chosen for the previous @samp{%u},
7341 without regard to any appended suffix.
7343 @item %j@var{suffix}
7344 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7345 writable, and if save-temps is off; otherwise, substitute the name
7346 of a temporary file, just like @samp{%u}. This temporary file is not
7347 meant for communication between processes, but rather as a junk
7350 @item %|@var{suffix}
7351 @itemx %m@var{suffix}
7352 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7353 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7354 all. These are the two most common ways to instruct a program that it
7355 should read from standard input or write to standard output. If you
7356 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7357 construct: see for example @file{f/lang-specs.h}.
7359 @item %.@var{SUFFIX}
7360 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7361 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7362 terminated by the next space or %.
7365 Marks the argument containing or following the @samp{%w} as the
7366 designated output file of this compilation. This puts the argument
7367 into the sequence of arguments that @samp{%o} will substitute later.
7370 Substitutes the names of all the output files, with spaces
7371 automatically placed around them. You should write spaces
7372 around the @samp{%o} as well or the results are undefined.
7373 @samp{%o} is for use in the specs for running the linker.
7374 Input files whose names have no recognized suffix are not compiled
7375 at all, but they are included among the output files, so they will
7379 Substitutes the suffix for object files. Note that this is
7380 handled specially when it immediately follows @samp{%g, %u, or %U},
7381 because of the need for those to form complete file names. The
7382 handling is such that @samp{%O} is treated exactly as if it had already
7383 been substituted, except that @samp{%g, %u, and %U} do not currently
7384 support additional @var{suffix} characters following @samp{%O} as they would
7385 following, for example, @samp{.o}.
7388 Substitutes the standard macro predefinitions for the
7389 current target machine. Use this when running @code{cpp}.
7392 Like @samp{%p}, but puts @samp{__} before and after the name of each
7393 predefined macro, except for macros that start with @samp{__} or with
7394 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7398 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7399 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7400 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7401 and @option{-imultilib} as necessary.
7404 Current argument is the name of a library or startup file of some sort.
7405 Search for that file in a standard list of directories and substitute
7406 the full name found.
7409 Print @var{str} as an error message. @var{str} is terminated by a newline.
7410 Use this when inconsistent options are detected.
7413 Substitute the contents of spec string @var{name} at this point.
7416 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7418 @item %x@{@var{option}@}
7419 Accumulate an option for @samp{%X}.
7422 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7426 Output the accumulated assembler options specified by @option{-Wa}.
7429 Output the accumulated preprocessor options specified by @option{-Wp}.
7432 Process the @code{asm} spec. This is used to compute the
7433 switches to be passed to the assembler.
7436 Process the @code{asm_final} spec. This is a spec string for
7437 passing switches to an assembler post-processor, if such a program is
7441 Process the @code{link} spec. This is the spec for computing the
7442 command line passed to the linker. Typically it will make use of the
7443 @samp{%L %G %S %D and %E} sequences.
7446 Dump out a @option{-L} option for each directory that GCC believes might
7447 contain startup files. If the target supports multilibs then the
7448 current multilib directory will be prepended to each of these paths.
7451 Process the @code{lib} spec. This is a spec string for deciding which
7452 libraries should be included on the command line to the linker.
7455 Process the @code{libgcc} spec. This is a spec string for deciding
7456 which GCC support library should be included on the command line to the linker.
7459 Process the @code{startfile} spec. This is a spec for deciding which
7460 object files should be the first ones passed to the linker. Typically
7461 this might be a file named @file{crt0.o}.
7464 Process the @code{endfile} spec. This is a spec string that specifies
7465 the last object files that will be passed to the linker.
7468 Process the @code{cpp} spec. This is used to construct the arguments
7469 to be passed to the C preprocessor.
7472 Process the @code{cc1} spec. This is used to construct the options to be
7473 passed to the actual C compiler (@samp{cc1}).
7476 Process the @code{cc1plus} spec. This is used to construct the options to be
7477 passed to the actual C++ compiler (@samp{cc1plus}).
7480 Substitute the variable part of a matched option. See below.
7481 Note that each comma in the substituted string is replaced by
7485 Remove all occurrences of @code{-S} from the command line. Note---this
7486 command is position dependent. @samp{%} commands in the spec string
7487 before this one will see @code{-S}, @samp{%} commands in the spec string
7488 after this one will not.
7490 @item %:@var{function}(@var{args})
7491 Call the named function @var{function}, passing it @var{args}.
7492 @var{args} is first processed as a nested spec string, then split
7493 into an argument vector in the usual fashion. The function returns
7494 a string which is processed as if it had appeared literally as part
7495 of the current spec.
7497 The following built-in spec functions are provided:
7501 The @code{getenv} spec function takes two arguments: an environment
7502 variable name and a string. If the environment variable is not
7503 defined, a fatal error is issued. Otherwise, the return value is the
7504 value of the environment variable concatenated with the string. For
7505 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
7508 %:getenv(TOPDIR /include)
7511 expands to @file{/path/to/top/include}.
7513 @item @code{if-exists}
7514 The @code{if-exists} spec function takes one argument, an absolute
7515 pathname to a file. If the file exists, @code{if-exists} returns the
7516 pathname. Here is a small example of its usage:
7520 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7523 @item @code{if-exists-else}
7524 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7525 spec function, except that it takes two arguments. The first argument is
7526 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7527 returns the pathname. If it does not exist, it returns the second argument.
7528 This way, @code{if-exists-else} can be used to select one file or another,
7529 based on the existence of the first. Here is a small example of its usage:
7533 crt0%O%s %:if-exists(crti%O%s) \
7534 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7537 @item @code{replace-outfile}
7538 The @code{replace-outfile} spec function takes two arguments. It looks for the
7539 first argument in the outfiles array and replaces it with the second argument. Here
7540 is a small example of its usage:
7543 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7549 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7550 If that switch was not specified, this substitutes nothing. Note that
7551 the leading dash is omitted when specifying this option, and it is
7552 automatically inserted if the substitution is performed. Thus the spec
7553 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7554 and would output the command line option @option{-foo}.
7556 @item %W@{@code{S}@}
7557 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7560 @item %@{@code{S}*@}
7561 Substitutes all the switches specified to GCC whose names start
7562 with @code{-S}, but which also take an argument. This is used for
7563 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7564 GCC considers @option{-o foo} as being
7565 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7566 text, including the space. Thus two arguments would be generated.
7568 @item %@{@code{S}*&@code{T}*@}
7569 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7570 (the order of @code{S} and @code{T} in the spec is not significant).
7571 There can be any number of ampersand-separated variables; for each the
7572 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7574 @item %@{@code{S}:@code{X}@}
7575 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7577 @item %@{!@code{S}:@code{X}@}
7578 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7580 @item %@{@code{S}*:@code{X}@}
7581 Substitutes @code{X} if one or more switches whose names start with
7582 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7583 once, no matter how many such switches appeared. However, if @code{%*}
7584 appears somewhere in @code{X}, then @code{X} will be substituted once
7585 for each matching switch, with the @code{%*} replaced by the part of
7586 that switch that matched the @code{*}.
7588 @item %@{.@code{S}:@code{X}@}
7589 Substitutes @code{X}, if processing a file with suffix @code{S}.
7591 @item %@{!.@code{S}:@code{X}@}
7592 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7594 @item %@{,@code{S}:@code{X}@}
7595 Substitutes @code{X}, if processing a file for language @code{S}.
7597 @item %@{!,@code{S}:@code{X}@}
7598 Substitutes @code{X}, if not processing a file for language @code{S}.
7600 @item %@{@code{S}|@code{P}:@code{X}@}
7601 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
7602 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
7603 @code{*} sequences as well, although they have a stronger binding than
7604 the @samp{|}. If @code{%*} appears in @code{X}, all of the
7605 alternatives must be starred, and only the first matching alternative
7608 For example, a spec string like this:
7611 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7614 will output the following command-line options from the following input
7615 command-line options:
7620 -d fred.c -foo -baz -boggle
7621 -d jim.d -bar -baz -boggle
7624 @item %@{S:X; T:Y; :D@}
7626 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7627 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7628 be as many clauses as you need. This may be combined with @code{.},
7629 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
7634 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7635 construct may contain other nested @samp{%} constructs or spaces, or
7636 even newlines. They are processed as usual, as described above.
7637 Trailing white space in @code{X} is ignored. White space may also
7638 appear anywhere on the left side of the colon in these constructs,
7639 except between @code{.} or @code{*} and the corresponding word.
7641 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7642 handled specifically in these constructs. If another value of
7643 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7644 @option{-W} switch is found later in the command line, the earlier
7645 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7646 just one letter, which passes all matching options.
7648 The character @samp{|} at the beginning of the predicate text is used to
7649 indicate that a command should be piped to the following command, but
7650 only if @option{-pipe} is specified.
7652 It is built into GCC which switches take arguments and which do not.
7653 (You might think it would be useful to generalize this to allow each
7654 compiler's spec to say which switches take arguments. But this cannot
7655 be done in a consistent fashion. GCC cannot even decide which input
7656 files have been specified without knowing which switches take arguments,
7657 and it must know which input files to compile in order to tell which
7660 GCC also knows implicitly that arguments starting in @option{-l} are to be
7661 treated as compiler output files, and passed to the linker in their
7662 proper position among the other output files.
7664 @c man begin OPTIONS
7666 @node Target Options
7667 @section Specifying Target Machine and Compiler Version
7668 @cindex target options
7669 @cindex cross compiling
7670 @cindex specifying machine version
7671 @cindex specifying compiler version and target machine
7672 @cindex compiler version, specifying
7673 @cindex target machine, specifying
7675 The usual way to run GCC is to run the executable called @file{gcc}, or
7676 @file{<machine>-gcc} when cross-compiling, or
7677 @file{<machine>-gcc-<version>} to run a version other than the one that
7678 was installed last. Sometimes this is inconvenient, so GCC provides
7679 options that will switch to another cross-compiler or version.
7682 @item -b @var{machine}
7684 The argument @var{machine} specifies the target machine for compilation.
7686 The value to use for @var{machine} is the same as was specified as the
7687 machine type when configuring GCC as a cross-compiler. For
7688 example, if a cross-compiler was configured with @samp{configure
7689 arm-elf}, meaning to compile for an arm processor with elf binaries,
7690 then you would specify @option{-b arm-elf} to run that cross compiler.
7691 Because there are other options beginning with @option{-b}, the
7692 configuration must contain a hyphen.
7694 @item -V @var{version}
7696 The argument @var{version} specifies which version of GCC to run.
7697 This is useful when multiple versions are installed. For example,
7698 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7701 The @option{-V} and @option{-b} options work by running the
7702 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7703 use them if you can just run that directly.
7705 @node Submodel Options
7706 @section Hardware Models and Configurations
7707 @cindex submodel options
7708 @cindex specifying hardware config
7709 @cindex hardware models and configurations, specifying
7710 @cindex machine dependent options
7712 Earlier we discussed the standard option @option{-b} which chooses among
7713 different installed compilers for completely different target
7714 machines, such as VAX vs.@: 68000 vs.@: 80386.
7716 In addition, each of these target machine types can have its own
7717 special options, starting with @samp{-m}, to choose among various
7718 hardware models or configurations---for example, 68010 vs 68020,
7719 floating coprocessor or none. A single installed version of the
7720 compiler can compile for any model or configuration, according to the
7723 Some configurations of the compiler also support additional special
7724 options, usually for compatibility with other compilers on the same
7727 @c This list is ordered alphanumerically by subsection name.
7728 @c It should be the same order and spelling as these options are listed
7729 @c in Machine Dependent Options
7735 * Blackfin Options::
7739 * DEC Alpha Options::
7740 * DEC Alpha/VMS Options::
7742 * GNU/Linux Options::
7745 * i386 and x86-64 Options::
7758 * RS/6000 and PowerPC Options::
7759 * S/390 and zSeries Options::
7764 * System V Options::
7765 * TMS320C3x/C4x Options::
7770 * Xstormy16 Options::
7776 @subsection ARC Options
7779 These options are defined for ARC implementations:
7784 Compile code for little endian mode. This is the default.
7788 Compile code for big endian mode.
7791 @opindex mmangle-cpu
7792 Prepend the name of the cpu to all public symbol names.
7793 In multiple-processor systems, there are many ARC variants with different
7794 instruction and register set characteristics. This flag prevents code
7795 compiled for one cpu to be linked with code compiled for another.
7796 No facility exists for handling variants that are ``almost identical''.
7797 This is an all or nothing option.
7799 @item -mcpu=@var{cpu}
7801 Compile code for ARC variant @var{cpu}.
7802 Which variants are supported depend on the configuration.
7803 All variants support @option{-mcpu=base}, this is the default.
7805 @item -mtext=@var{text-section}
7806 @itemx -mdata=@var{data-section}
7807 @itemx -mrodata=@var{readonly-data-section}
7811 Put functions, data, and readonly data in @var{text-section},
7812 @var{data-section}, and @var{readonly-data-section} respectively
7813 by default. This can be overridden with the @code{section} attribute.
7814 @xref{Variable Attributes}.
7819 @subsection ARM Options
7822 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7826 @item -mabi=@var{name}
7828 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7829 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7832 @opindex mapcs-frame
7833 Generate a stack frame that is compliant with the ARM Procedure Call
7834 Standard for all functions, even if this is not strictly necessary for
7835 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7836 with this option will cause the stack frames not to be generated for
7837 leaf functions. The default is @option{-mno-apcs-frame}.
7841 This is a synonym for @option{-mapcs-frame}.
7844 @c not currently implemented
7845 @item -mapcs-stack-check
7846 @opindex mapcs-stack-check
7847 Generate code to check the amount of stack space available upon entry to
7848 every function (that actually uses some stack space). If there is
7849 insufficient space available then either the function
7850 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7851 called, depending upon the amount of stack space required. The run time
7852 system is required to provide these functions. The default is
7853 @option{-mno-apcs-stack-check}, since this produces smaller code.
7855 @c not currently implemented
7857 @opindex mapcs-float
7858 Pass floating point arguments using the float point registers. This is
7859 one of the variants of the APCS@. This option is recommended if the
7860 target hardware has a floating point unit or if a lot of floating point
7861 arithmetic is going to be performed by the code. The default is
7862 @option{-mno-apcs-float}, since integer only code is slightly increased in
7863 size if @option{-mapcs-float} is used.
7865 @c not currently implemented
7866 @item -mapcs-reentrant
7867 @opindex mapcs-reentrant
7868 Generate reentrant, position independent code. The default is
7869 @option{-mno-apcs-reentrant}.
7872 @item -mthumb-interwork
7873 @opindex mthumb-interwork
7874 Generate code which supports calling between the ARM and Thumb
7875 instruction sets. Without this option the two instruction sets cannot
7876 be reliably used inside one program. The default is
7877 @option{-mno-thumb-interwork}, since slightly larger code is generated
7878 when @option{-mthumb-interwork} is specified.
7880 @item -mno-sched-prolog
7881 @opindex mno-sched-prolog
7882 Prevent the reordering of instructions in the function prolog, or the
7883 merging of those instruction with the instructions in the function's
7884 body. This means that all functions will start with a recognizable set
7885 of instructions (or in fact one of a choice from a small set of
7886 different function prologues), and this information can be used to
7887 locate the start if functions inside an executable piece of code. The
7888 default is @option{-msched-prolog}.
7891 @opindex mhard-float
7892 Generate output containing floating point instructions. This is the
7896 @opindex msoft-float
7897 Generate output containing library calls for floating point.
7898 @strong{Warning:} the requisite libraries are not available for all ARM
7899 targets. Normally the facilities of the machine's usual C compiler are
7900 used, but this cannot be done directly in cross-compilation. You must make
7901 your own arrangements to provide suitable library functions for
7904 @option{-msoft-float} changes the calling convention in the output file;
7905 therefore, it is only useful if you compile @emph{all} of a program with
7906 this option. In particular, you need to compile @file{libgcc.a}, the
7907 library that comes with GCC, with @option{-msoft-float} in order for
7910 @item -mfloat-abi=@var{name}
7912 Specifies which ABI to use for floating point values. Permissible values
7913 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7915 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7916 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7917 of floating point instructions, but still uses the soft-float calling
7920 @item -mlittle-endian
7921 @opindex mlittle-endian
7922 Generate code for a processor running in little-endian mode. This is
7923 the default for all standard configurations.
7926 @opindex mbig-endian
7927 Generate code for a processor running in big-endian mode; the default is
7928 to compile code for a little-endian processor.
7930 @item -mwords-little-endian
7931 @opindex mwords-little-endian
7932 This option only applies when generating code for big-endian processors.
7933 Generate code for a little-endian word order but a big-endian byte
7934 order. That is, a byte order of the form @samp{32107654}. Note: this
7935 option should only be used if you require compatibility with code for
7936 big-endian ARM processors generated by versions of the compiler prior to
7939 @item -mcpu=@var{name}
7941 This specifies the name of the target ARM processor. GCC uses this name
7942 to determine what kind of instructions it can emit when generating
7943 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7944 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7945 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7946 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7947 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7948 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7949 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7950 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7951 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7952 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7953 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7954 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7955 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7956 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
7957 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
7958 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
7960 @itemx -mtune=@var{name}
7962 This option is very similar to the @option{-mcpu=} option, except that
7963 instead of specifying the actual target processor type, and hence
7964 restricting which instructions can be used, it specifies that GCC should
7965 tune the performance of the code as if the target were of the type
7966 specified in this option, but still choosing the instructions that it
7967 will generate based on the cpu specified by a @option{-mcpu=} option.
7968 For some ARM implementations better performance can be obtained by using
7971 @item -march=@var{name}
7973 This specifies the name of the target ARM architecture. GCC uses this
7974 name to determine what kind of instructions it can emit when generating
7975 assembly code. This option can be used in conjunction with or instead
7976 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7977 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7978 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7979 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
7980 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
7982 @item -mfpu=@var{name}
7983 @itemx -mfpe=@var{number}
7984 @itemx -mfp=@var{number}
7988 This specifies what floating point hardware (or hardware emulation) is
7989 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7990 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7991 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7992 with older versions of GCC@.
7994 If @option{-msoft-float} is specified this specifies the format of
7995 floating point values.
7997 @item -mstructure-size-boundary=@var{n}
7998 @opindex mstructure-size-boundary
7999 The size of all structures and unions will be rounded up to a multiple
8000 of the number of bits set by this option. Permissible values are 8, 32
8001 and 64. The default value varies for different toolchains. For the COFF
8002 targeted toolchain the default value is 8. A value of 64 is only allowed
8003 if the underlying ABI supports it.
8005 Specifying the larger number can produce faster, more efficient code, but
8006 can also increase the size of the program. Different values are potentially
8007 incompatible. Code compiled with one value cannot necessarily expect to
8008 work with code or libraries compiled with another value, if they exchange
8009 information using structures or unions.
8011 @item -mabort-on-noreturn
8012 @opindex mabort-on-noreturn
8013 Generate a call to the function @code{abort} at the end of a
8014 @code{noreturn} function. It will be executed if the function tries to
8018 @itemx -mno-long-calls
8019 @opindex mlong-calls
8020 @opindex mno-long-calls
8021 Tells the compiler to perform function calls by first loading the
8022 address of the function into a register and then performing a subroutine
8023 call on this register. This switch is needed if the target function
8024 will lie outside of the 64 megabyte addressing range of the offset based
8025 version of subroutine call instruction.
8027 Even if this switch is enabled, not all function calls will be turned
8028 into long calls. The heuristic is that static functions, functions
8029 which have the @samp{short-call} attribute, functions that are inside
8030 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8031 definitions have already been compiled within the current compilation
8032 unit, will not be turned into long calls. The exception to this rule is
8033 that weak function definitions, functions with the @samp{long-call}
8034 attribute or the @samp{section} attribute, and functions that are within
8035 the scope of a @samp{#pragma long_calls} directive, will always be
8036 turned into long calls.
8038 This feature is not enabled by default. Specifying
8039 @option{-mno-long-calls} will restore the default behavior, as will
8040 placing the function calls within the scope of a @samp{#pragma
8041 long_calls_off} directive. Note these switches have no effect on how
8042 the compiler generates code to handle function calls via function
8045 @item -mnop-fun-dllimport
8046 @opindex mnop-fun-dllimport
8047 Disable support for the @code{dllimport} attribute.
8049 @item -msingle-pic-base
8050 @opindex msingle-pic-base
8051 Treat the register used for PIC addressing as read-only, rather than
8052 loading it in the prologue for each function. The run-time system is
8053 responsible for initializing this register with an appropriate value
8054 before execution begins.
8056 @item -mpic-register=@var{reg}
8057 @opindex mpic-register
8058 Specify the register to be used for PIC addressing. The default is R10
8059 unless stack-checking is enabled, when R9 is used.
8061 @item -mcirrus-fix-invalid-insns
8062 @opindex mcirrus-fix-invalid-insns
8063 @opindex mno-cirrus-fix-invalid-insns
8064 Insert NOPs into the instruction stream to in order to work around
8065 problems with invalid Maverick instruction combinations. This option
8066 is only valid if the @option{-mcpu=ep9312} option has been used to
8067 enable generation of instructions for the Cirrus Maverick floating
8068 point co-processor. This option is not enabled by default, since the
8069 problem is only present in older Maverick implementations. The default
8070 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8073 @item -mpoke-function-name
8074 @opindex mpoke-function-name
8075 Write the name of each function into the text section, directly
8076 preceding the function prologue. The generated code is similar to this:
8080 .ascii "arm_poke_function_name", 0
8083 .word 0xff000000 + (t1 - t0)
8084 arm_poke_function_name
8086 stmfd sp!, @{fp, ip, lr, pc@}
8090 When performing a stack backtrace, code can inspect the value of
8091 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8092 location @code{pc - 12} and the top 8 bits are set, then we know that
8093 there is a function name embedded immediately preceding this location
8094 and has length @code{((pc[-3]) & 0xff000000)}.
8098 Generate code for the Thumb instruction set. The default is to
8099 use the 32-bit ARM instruction set.
8100 This option automatically enables either 16-bit Thumb-1 or
8101 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8102 and @option{-march=@var{name}} options.
8105 @opindex mtpcs-frame
8106 Generate a stack frame that is compliant with the Thumb Procedure Call
8107 Standard for all non-leaf functions. (A leaf function is one that does
8108 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8110 @item -mtpcs-leaf-frame
8111 @opindex mtpcs-leaf-frame
8112 Generate a stack frame that is compliant with the Thumb Procedure Call
8113 Standard for all leaf functions. (A leaf function is one that does
8114 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8116 @item -mcallee-super-interworking
8117 @opindex mcallee-super-interworking
8118 Gives all externally visible functions in the file being compiled an ARM
8119 instruction set header which switches to Thumb mode before executing the
8120 rest of the function. This allows these functions to be called from
8121 non-interworking code.
8123 @item -mcaller-super-interworking
8124 @opindex mcaller-super-interworking
8125 Allows calls via function pointers (including virtual functions) to
8126 execute correctly regardless of whether the target code has been
8127 compiled for interworking or not. There is a small overhead in the cost
8128 of executing a function pointer if this option is enabled.
8130 @item -mtp=@var{name}
8132 Specify the access model for the thread local storage pointer. The valid
8133 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8134 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8135 (supported in the arm6k architecture), and @option{auto}, which uses the
8136 best available method for the selected processor. The default setting is
8142 @subsection AVR Options
8145 These options are defined for AVR implementations:
8148 @item -mmcu=@var{mcu}
8150 Specify ATMEL AVR instruction set or MCU type.
8152 Instruction set avr1 is for the minimal AVR core, not supported by the C
8153 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8154 attiny11, attiny12, attiny15, attiny28).
8156 Instruction set avr2 (default) is for the classic AVR core with up to
8157 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8158 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8159 at90c8534, at90s8535).
8161 Instruction set avr3 is for the classic AVR core with up to 128K program
8162 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8164 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8165 memory space (MCU types: atmega8, atmega83, atmega85).
8167 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8168 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8169 atmega64, atmega128, at43usb355, at94k).
8173 Output instruction sizes to the asm file.
8175 @item -minit-stack=@var{N}
8176 @opindex minit-stack
8177 Specify the initial stack address, which may be a symbol or numeric value,
8178 @samp{__stack} is the default.
8180 @item -mno-interrupts
8181 @opindex mno-interrupts
8182 Generated code is not compatible with hardware interrupts.
8183 Code size will be smaller.
8185 @item -mcall-prologues
8186 @opindex mcall-prologues
8187 Functions prologues/epilogues expanded as call to appropriate
8188 subroutines. Code size will be smaller.
8190 @item -mno-tablejump
8191 @opindex mno-tablejump
8192 Do not generate tablejump insns which sometimes increase code size.
8195 @opindex mtiny-stack
8196 Change only the low 8 bits of the stack pointer.
8200 Assume int to be 8 bit integer. This affects the sizes of all types: A
8201 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8202 and long long will be 4 bytes. Please note that this option does not
8203 comply to the C standards, but it will provide you with smaller code
8207 @node Blackfin Options
8208 @subsection Blackfin Options
8209 @cindex Blackfin Options
8212 @item -momit-leaf-frame-pointer
8213 @opindex momit-leaf-frame-pointer
8214 Don't keep the frame pointer in a register for leaf functions. This
8215 avoids the instructions to save, set up and restore frame pointers and
8216 makes an extra register available in leaf functions. The option
8217 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8218 which might make debugging harder.
8220 @item -mspecld-anomaly
8221 @opindex mspecld-anomaly
8222 When enabled, the compiler will ensure that the generated code does not
8223 contain speculative loads after jump instructions. This option is enabled
8226 @item -mno-specld-anomaly
8227 @opindex mno-specld-anomaly
8228 Don't generate extra code to prevent speculative loads from occurring.
8230 @item -mcsync-anomaly
8231 @opindex mcsync-anomaly
8232 When enabled, the compiler will ensure that the generated code does not
8233 contain CSYNC or SSYNC instructions too soon after conditional branches.
8234 This option is enabled by default.
8236 @item -mno-csync-anomaly
8237 @opindex mno-csync-anomaly
8238 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8239 occurring too soon after a conditional branch.
8243 When enabled, the compiler is free to take advantage of the knowledge that
8244 the entire program fits into the low 64k of memory.
8247 @opindex mno-low-64k
8248 Assume that the program is arbitrarily large. This is the default.
8250 @item -mstack-check-l1
8251 @opindex mstack-check-l1
8252 Do stack checking using information placed into L1 scratchpad memory by the
8255 @item -mid-shared-library
8256 @opindex mid-shared-library
8257 Generate code that supports shared libraries via the library ID method.
8258 This allows for execute in place and shared libraries in an environment
8259 without virtual memory management. This option implies @option{-fPIC}.
8261 @item -mno-id-shared-library
8262 @opindex mno-id-shared-library
8263 Generate code that doesn't assume ID based shared libraries are being used.
8264 This is the default.
8266 @item -mleaf-id-shared-library
8267 @opindex mleaf-id-shared-library
8268 Generate code that supports shared libraries via the library ID method,
8269 but assumes that this library or executable won't link against any other
8270 ID shared libraries. That allows the compiler to use faster code for jumps
8273 @item -mno-leaf-id-shared-library
8274 @opindex mno-leaf-id-shared-library
8275 Do not assume that the code being compiled won't link against any ID shared
8276 libraries. Slower code will be generated for jump and call insns.
8278 @item -mshared-library-id=n
8279 @opindex mshared-library-id
8280 Specified the identification number of the ID based shared library being
8281 compiled. Specifying a value of 0 will generate more compact code, specifying
8282 other values will force the allocation of that number to the current
8283 library but is no more space or time efficient than omitting this option.
8287 Generate code that allows the data segment to be located in a different
8288 area of memory from the text segment. This allows for execute in place in
8289 an environment without virtual memory management by eliminating relocations
8290 against the text section.
8293 @opindex mno-sep-data
8294 Generate code that assumes that the data segment follows the text segment.
8295 This is the default.
8298 @itemx -mno-long-calls
8299 @opindex mlong-calls
8300 @opindex mno-long-calls
8301 Tells the compiler to perform function calls by first loading the
8302 address of the function into a register and then performing a subroutine
8303 call on this register. This switch is needed if the target function
8304 will lie outside of the 24 bit addressing range of the offset based
8305 version of subroutine call instruction.
8307 This feature is not enabled by default. Specifying
8308 @option{-mno-long-calls} will restore the default behavior. Note these
8309 switches have no effect on how the compiler generates code to handle
8310 function calls via function pointers.
8314 @subsection CRIS Options
8315 @cindex CRIS Options
8317 These options are defined specifically for the CRIS ports.
8320 @item -march=@var{architecture-type}
8321 @itemx -mcpu=@var{architecture-type}
8324 Generate code for the specified architecture. The choices for
8325 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8326 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8327 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8330 @item -mtune=@var{architecture-type}
8332 Tune to @var{architecture-type} everything applicable about the generated
8333 code, except for the ABI and the set of available instructions. The
8334 choices for @var{architecture-type} are the same as for
8335 @option{-march=@var{architecture-type}}.
8337 @item -mmax-stack-frame=@var{n}
8338 @opindex mmax-stack-frame
8339 Warn when the stack frame of a function exceeds @var{n} bytes.
8341 @item -melinux-stacksize=@var{n}
8342 @opindex melinux-stacksize
8343 Only available with the @samp{cris-axis-aout} target. Arranges for
8344 indications in the program to the kernel loader that the stack of the
8345 program should be set to @var{n} bytes.
8351 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8352 @option{-march=v3} and @option{-march=v8} respectively.
8354 @item -mmul-bug-workaround
8355 @itemx -mno-mul-bug-workaround
8356 @opindex mmul-bug-workaround
8357 @opindex mno-mul-bug-workaround
8358 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8359 models where it applies. This option is active by default.
8363 Enable CRIS-specific verbose debug-related information in the assembly
8364 code. This option also has the effect to turn off the @samp{#NO_APP}
8365 formatted-code indicator to the assembler at the beginning of the
8370 Do not use condition-code results from previous instruction; always emit
8371 compare and test instructions before use of condition codes.
8373 @item -mno-side-effects
8374 @opindex mno-side-effects
8375 Do not emit instructions with side-effects in addressing modes other than
8379 @itemx -mno-stack-align
8381 @itemx -mno-data-align
8382 @itemx -mconst-align
8383 @itemx -mno-const-align
8384 @opindex mstack-align
8385 @opindex mno-stack-align
8386 @opindex mdata-align
8387 @opindex mno-data-align
8388 @opindex mconst-align
8389 @opindex mno-const-align
8390 These options (no-options) arranges (eliminate arrangements) for the
8391 stack-frame, individual data and constants to be aligned for the maximum
8392 single data access size for the chosen CPU model. The default is to
8393 arrange for 32-bit alignment. ABI details such as structure layout are
8394 not affected by these options.
8402 Similar to the stack- data- and const-align options above, these options
8403 arrange for stack-frame, writable data and constants to all be 32-bit,
8404 16-bit or 8-bit aligned. The default is 32-bit alignment.
8406 @item -mno-prologue-epilogue
8407 @itemx -mprologue-epilogue
8408 @opindex mno-prologue-epilogue
8409 @opindex mprologue-epilogue
8410 With @option{-mno-prologue-epilogue}, the normal function prologue and
8411 epilogue that sets up the stack-frame are omitted and no return
8412 instructions or return sequences are generated in the code. Use this
8413 option only together with visual inspection of the compiled code: no
8414 warnings or errors are generated when call-saved registers must be saved,
8415 or storage for local variable needs to be allocated.
8421 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8422 instruction sequences that load addresses for functions from the PLT part
8423 of the GOT rather than (traditional on other architectures) calls to the
8424 PLT@. The default is @option{-mgotplt}.
8428 Legacy no-op option only recognized with the cris-axis-aout target.
8432 Legacy no-op option only recognized with the cris-axis-elf and
8433 cris-axis-linux-gnu targets.
8437 Only recognized with the cris-axis-aout target, where it selects a
8438 GNU/linux-like multilib, include files and instruction set for
8443 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8447 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8448 to link with input-output functions from a simulator library. Code,
8449 initialized data and zero-initialized data are allocated consecutively.
8453 Like @option{-sim}, but pass linker options to locate initialized data at
8454 0x40000000 and zero-initialized data at 0x80000000.
8458 @subsection CRX Options
8461 These options are defined specifically for the CRX ports.
8467 Enable the use of multiply-accumulate instructions. Disabled by default.
8471 Push instructions will be used to pass outgoing arguments when functions
8472 are called. Enabled by default.
8475 @node Darwin Options
8476 @subsection Darwin Options
8477 @cindex Darwin options
8479 These options are defined for all architectures running the Darwin operating
8482 FSF GCC on Darwin does not create ``fat'' object files; it will create
8483 an object file for the single architecture that it was built to
8484 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8485 @option{-arch} options are used; it does so by running the compiler or
8486 linker multiple times and joining the results together with
8489 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8490 @samp{i686}) is determined by the flags that specify the ISA
8491 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8492 @option{-force_cpusubtype_ALL} option can be used to override this.
8494 The Darwin tools vary in their behavior when presented with an ISA
8495 mismatch. The assembler, @file{as}, will only permit instructions to
8496 be used that are valid for the subtype of the file it is generating,
8497 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8498 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8499 and print an error if asked to create a shared library with a less
8500 restrictive subtype than its input files (for instance, trying to put
8501 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8502 for executables, @file{ld}, will quietly give the executable the most
8503 restrictive subtype of any of its input files.
8508 Add the framework directory @var{dir} to the head of the list of
8509 directories to be searched for header files. These directories are
8510 interleaved with those specified by @option{-I} options and are
8511 scanned in a left-to-right order.
8513 A framework directory is a directory with frameworks in it. A
8514 framework is a directory with a @samp{"Headers"} and/or
8515 @samp{"PrivateHeaders"} directory contained directly in it that ends
8516 in @samp{".framework"}. The name of a framework is the name of this
8517 directory excluding the @samp{".framework"}. Headers associated with
8518 the framework are found in one of those two directories, with
8519 @samp{"Headers"} being searched first. A subframework is a framework
8520 directory that is in a framework's @samp{"Frameworks"} directory.
8521 Includes of subframework headers can only appear in a header of a
8522 framework that contains the subframework, or in a sibling subframework
8523 header. Two subframeworks are siblings if they occur in the same
8524 framework. A subframework should not have the same name as a
8525 framework, a warning will be issued if this is violated. Currently a
8526 subframework cannot have subframeworks, in the future, the mechanism
8527 may be extended to support this. The standard frameworks can be found
8528 in @samp{"/System/Library/Frameworks"} and
8529 @samp{"/Library/Frameworks"}. An example include looks like
8530 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8531 the name of the framework and header.h is found in the
8532 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8534 @item -iframework@var{dir}
8536 Like @option{-F} except the directory is a treated as a system
8537 directory. The main difference between this @option{-iframework} and
8538 @option{-F} is that with @option{-iframework} the compiler does not
8539 warn about constructs contained within header files found via
8540 @var{dir}. This option is valid only for the C family of languages.
8544 Emit debugging information for symbols that are used. For STABS
8545 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8546 This is by default ON@.
8550 Emit debugging information for all symbols and types.
8552 @item -mmacosx-version-min=@var{version}
8553 The earliest version of MacOS X that this executable will run on
8554 is @var{version}. Typical values of @var{version} include @code{10.1},
8555 @code{10.2}, and @code{10.3.9}.
8557 The default for this option is to make choices that seem to be most
8562 Enable kernel development mode. The @option{-mkernel} option sets
8563 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8564 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8565 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8566 applicable. This mode also sets @option{-mno-altivec},
8567 @option{-msoft-float}, @option{-fno-builtin} and
8568 @option{-mlong-branch} for PowerPC targets.
8570 @item -mone-byte-bool
8571 @opindex -mone-byte-bool
8572 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8573 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8574 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8575 option has no effect on x86.
8577 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8578 to generate code that is not binary compatible with code generated
8579 without that switch. Using this switch may require recompiling all
8580 other modules in a program, including system libraries. Use this
8581 switch to conform to a non-default data model.
8583 @item -mfix-and-continue
8584 @itemx -ffix-and-continue
8585 @itemx -findirect-data
8586 @opindex mfix-and-continue
8587 @opindex ffix-and-continue
8588 @opindex findirect-data
8589 Generate code suitable for fast turn around development. Needed to
8590 enable gdb to dynamically load @code{.o} files into already running
8591 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8592 are provided for backwards compatibility.
8596 Loads all members of static archive libraries.
8597 See man ld(1) for more information.
8599 @item -arch_errors_fatal
8600 @opindex arch_errors_fatal
8601 Cause the errors having to do with files that have the wrong architecture
8605 @opindex bind_at_load
8606 Causes the output file to be marked such that the dynamic linker will
8607 bind all undefined references when the file is loaded or launched.
8611 Produce a Mach-o bundle format file.
8612 See man ld(1) for more information.
8614 @item -bundle_loader @var{executable}
8615 @opindex bundle_loader
8616 This option specifies the @var{executable} that will be loading the build
8617 output file being linked. See man ld(1) for more information.
8620 @opindex -dynamiclib
8621 When passed this option, GCC will produce a dynamic library instead of
8622 an executable when linking, using the Darwin @file{libtool} command.
8624 @item -force_cpusubtype_ALL
8625 @opindex -force_cpusubtype_ALL
8626 This causes GCC's output file to have the @var{ALL} subtype, instead of
8627 one controlled by the @option{-mcpu} or @option{-march} option.
8629 @item -allowable_client @var{client_name}
8631 @itemx -compatibility_version
8632 @itemx -current_version
8634 @itemx -dependency-file
8636 @itemx -dylinker_install_name
8638 @itemx -exported_symbols_list
8640 @itemx -flat_namespace
8641 @itemx -force_flat_namespace
8642 @itemx -headerpad_max_install_names
8645 @itemx -install_name
8646 @itemx -keep_private_externs
8647 @itemx -multi_module
8648 @itemx -multiply_defined
8649 @itemx -multiply_defined_unused
8651 @itemx -no_dead_strip_inits_and_terms
8652 @itemx -nofixprebinding
8655 @itemx -noseglinkedit
8656 @itemx -pagezero_size
8658 @itemx -prebind_all_twolevel_modules
8659 @itemx -private_bundle
8660 @itemx -read_only_relocs
8662 @itemx -sectobjectsymbols
8666 @itemx -sectobjectsymbols
8669 @itemx -segs_read_only_addr
8670 @itemx -segs_read_write_addr
8671 @itemx -seg_addr_table
8672 @itemx -seg_addr_table_filename
8675 @itemx -segs_read_only_addr
8676 @itemx -segs_read_write_addr
8677 @itemx -single_module
8680 @itemx -sub_umbrella
8681 @itemx -twolevel_namespace
8684 @itemx -unexported_symbols_list
8685 @itemx -weak_reference_mismatches
8688 @opindex allowable_client
8689 @opindex client_name
8690 @opindex compatibility_version
8691 @opindex current_version
8693 @opindex dependency-file
8695 @opindex dylinker_install_name
8697 @opindex exported_symbols_list
8699 @opindex flat_namespace
8700 @opindex force_flat_namespace
8701 @opindex headerpad_max_install_names
8704 @opindex install_name
8705 @opindex keep_private_externs
8706 @opindex multi_module
8707 @opindex multiply_defined
8708 @opindex multiply_defined_unused
8710 @opindex no_dead_strip_inits_and_terms
8711 @opindex nofixprebinding
8712 @opindex nomultidefs
8714 @opindex noseglinkedit
8715 @opindex pagezero_size
8717 @opindex prebind_all_twolevel_modules
8718 @opindex private_bundle
8719 @opindex read_only_relocs
8721 @opindex sectobjectsymbols
8725 @opindex sectobjectsymbols
8728 @opindex segs_read_only_addr
8729 @opindex segs_read_write_addr
8730 @opindex seg_addr_table
8731 @opindex seg_addr_table_filename
8732 @opindex seglinkedit
8734 @opindex segs_read_only_addr
8735 @opindex segs_read_write_addr
8736 @opindex single_module
8738 @opindex sub_library
8739 @opindex sub_umbrella
8740 @opindex twolevel_namespace
8743 @opindex unexported_symbols_list
8744 @opindex weak_reference_mismatches
8745 @opindex whatsloaded
8747 These options are passed to the Darwin linker. The Darwin linker man page
8748 describes them in detail.
8751 @node DEC Alpha Options
8752 @subsection DEC Alpha Options
8754 These @samp{-m} options are defined for the DEC Alpha implementations:
8757 @item -mno-soft-float
8759 @opindex mno-soft-float
8760 @opindex msoft-float
8761 Use (do not use) the hardware floating-point instructions for
8762 floating-point operations. When @option{-msoft-float} is specified,
8763 functions in @file{libgcc.a} will be used to perform floating-point
8764 operations. Unless they are replaced by routines that emulate the
8765 floating-point operations, or compiled in such a way as to call such
8766 emulations routines, these routines will issue floating-point
8767 operations. If you are compiling for an Alpha without floating-point
8768 operations, you must ensure that the library is built so as not to call
8771 Note that Alpha implementations without floating-point operations are
8772 required to have floating-point registers.
8777 @opindex mno-fp-regs
8778 Generate code that uses (does not use) the floating-point register set.
8779 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8780 register set is not used, floating point operands are passed in integer
8781 registers as if they were integers and floating-point results are passed
8782 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8783 so any function with a floating-point argument or return value called by code
8784 compiled with @option{-mno-fp-regs} must also be compiled with that
8787 A typical use of this option is building a kernel that does not use,
8788 and hence need not save and restore, any floating-point registers.
8792 The Alpha architecture implements floating-point hardware optimized for
8793 maximum performance. It is mostly compliant with the IEEE floating
8794 point standard. However, for full compliance, software assistance is
8795 required. This option generates code fully IEEE compliant code
8796 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8797 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8798 defined during compilation. The resulting code is less efficient but is
8799 able to correctly support denormalized numbers and exceptional IEEE
8800 values such as not-a-number and plus/minus infinity. Other Alpha
8801 compilers call this option @option{-ieee_with_no_inexact}.
8803 @item -mieee-with-inexact
8804 @opindex mieee-with-inexact
8805 This is like @option{-mieee} except the generated code also maintains
8806 the IEEE @var{inexact-flag}. Turning on this option causes the
8807 generated code to implement fully-compliant IEEE math. In addition to
8808 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8809 macro. On some Alpha implementations the resulting code may execute
8810 significantly slower than the code generated by default. Since there is
8811 very little code that depends on the @var{inexact-flag}, you should
8812 normally not specify this option. Other Alpha compilers call this
8813 option @option{-ieee_with_inexact}.
8815 @item -mfp-trap-mode=@var{trap-mode}
8816 @opindex mfp-trap-mode
8817 This option controls what floating-point related traps are enabled.
8818 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8819 The trap mode can be set to one of four values:
8823 This is the default (normal) setting. The only traps that are enabled
8824 are the ones that cannot be disabled in software (e.g., division by zero
8828 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8832 Like @samp{u}, but the instructions are marked to be safe for software
8833 completion (see Alpha architecture manual for details).
8836 Like @samp{su}, but inexact traps are enabled as well.
8839 @item -mfp-rounding-mode=@var{rounding-mode}
8840 @opindex mfp-rounding-mode
8841 Selects the IEEE rounding mode. Other Alpha compilers call this option
8842 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8847 Normal IEEE rounding mode. Floating point numbers are rounded towards
8848 the nearest machine number or towards the even machine number in case
8852 Round towards minus infinity.
8855 Chopped rounding mode. Floating point numbers are rounded towards zero.
8858 Dynamic rounding mode. A field in the floating point control register
8859 (@var{fpcr}, see Alpha architecture reference manual) controls the
8860 rounding mode in effect. The C library initializes this register for
8861 rounding towards plus infinity. Thus, unless your program modifies the
8862 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8865 @item -mtrap-precision=@var{trap-precision}
8866 @opindex mtrap-precision
8867 In the Alpha architecture, floating point traps are imprecise. This
8868 means without software assistance it is impossible to recover from a
8869 floating trap and program execution normally needs to be terminated.
8870 GCC can generate code that can assist operating system trap handlers
8871 in determining the exact location that caused a floating point trap.
8872 Depending on the requirements of an application, different levels of
8873 precisions can be selected:
8877 Program precision. This option is the default and means a trap handler
8878 can only identify which program caused a floating point exception.
8881 Function precision. The trap handler can determine the function that
8882 caused a floating point exception.
8885 Instruction precision. The trap handler can determine the exact
8886 instruction that caused a floating point exception.
8889 Other Alpha compilers provide the equivalent options called
8890 @option{-scope_safe} and @option{-resumption_safe}.
8892 @item -mieee-conformant
8893 @opindex mieee-conformant
8894 This option marks the generated code as IEEE conformant. You must not
8895 use this option unless you also specify @option{-mtrap-precision=i} and either
8896 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8897 is to emit the line @samp{.eflag 48} in the function prologue of the
8898 generated assembly file. Under DEC Unix, this has the effect that
8899 IEEE-conformant math library routines will be linked in.
8901 @item -mbuild-constants
8902 @opindex mbuild-constants
8903 Normally GCC examines a 32- or 64-bit integer constant to
8904 see if it can construct it from smaller constants in two or three
8905 instructions. If it cannot, it will output the constant as a literal and
8906 generate code to load it from the data segment at runtime.
8908 Use this option to require GCC to construct @emph{all} integer constants
8909 using code, even if it takes more instructions (the maximum is six).
8911 You would typically use this option to build a shared library dynamic
8912 loader. Itself a shared library, it must relocate itself in memory
8913 before it can find the variables and constants in its own data segment.
8919 Select whether to generate code to be assembled by the vendor-supplied
8920 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8938 Indicate whether GCC should generate code to use the optional BWX,
8939 CIX, FIX and MAX instruction sets. The default is to use the instruction
8940 sets supported by the CPU type specified via @option{-mcpu=} option or that
8941 of the CPU on which GCC was built if none was specified.
8946 @opindex mfloat-ieee
8947 Generate code that uses (does not use) VAX F and G floating point
8948 arithmetic instead of IEEE single and double precision.
8950 @item -mexplicit-relocs
8951 @itemx -mno-explicit-relocs
8952 @opindex mexplicit-relocs
8953 @opindex mno-explicit-relocs
8954 Older Alpha assemblers provided no way to generate symbol relocations
8955 except via assembler macros. Use of these macros does not allow
8956 optimal instruction scheduling. GNU binutils as of version 2.12
8957 supports a new syntax that allows the compiler to explicitly mark
8958 which relocations should apply to which instructions. This option
8959 is mostly useful for debugging, as GCC detects the capabilities of
8960 the assembler when it is built and sets the default accordingly.
8964 @opindex msmall-data
8965 @opindex mlarge-data
8966 When @option{-mexplicit-relocs} is in effect, static data is
8967 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8968 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8969 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8970 16-bit relocations off of the @code{$gp} register. This limits the
8971 size of the small data area to 64KB, but allows the variables to be
8972 directly accessed via a single instruction.
8974 The default is @option{-mlarge-data}. With this option the data area
8975 is limited to just below 2GB@. Programs that require more than 2GB of
8976 data must use @code{malloc} or @code{mmap} to allocate the data in the
8977 heap instead of in the program's data segment.
8979 When generating code for shared libraries, @option{-fpic} implies
8980 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8984 @opindex msmall-text
8985 @opindex mlarge-text
8986 When @option{-msmall-text} is used, the compiler assumes that the
8987 code of the entire program (or shared library) fits in 4MB, and is
8988 thus reachable with a branch instruction. When @option{-msmall-data}
8989 is used, the compiler can assume that all local symbols share the
8990 same @code{$gp} value, and thus reduce the number of instructions
8991 required for a function call from 4 to 1.
8993 The default is @option{-mlarge-text}.
8995 @item -mcpu=@var{cpu_type}
8997 Set the instruction set and instruction scheduling parameters for
8998 machine type @var{cpu_type}. You can specify either the @samp{EV}
8999 style name or the corresponding chip number. GCC supports scheduling
9000 parameters for the EV4, EV5 and EV6 family of processors and will
9001 choose the default values for the instruction set from the processor
9002 you specify. If you do not specify a processor type, GCC will default
9003 to the processor on which the compiler was built.
9005 Supported values for @var{cpu_type} are
9011 Schedules as an EV4 and has no instruction set extensions.
9015 Schedules as an EV5 and has no instruction set extensions.
9019 Schedules as an EV5 and supports the BWX extension.
9024 Schedules as an EV5 and supports the BWX and MAX extensions.
9028 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9032 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9035 @item -mtune=@var{cpu_type}
9037 Set only the instruction scheduling parameters for machine type
9038 @var{cpu_type}. The instruction set is not changed.
9040 @item -mmemory-latency=@var{time}
9041 @opindex mmemory-latency
9042 Sets the latency the scheduler should assume for typical memory
9043 references as seen by the application. This number is highly
9044 dependent on the memory access patterns used by the application
9045 and the size of the external cache on the machine.
9047 Valid options for @var{time} are
9051 A decimal number representing clock cycles.
9057 The compiler contains estimates of the number of clock cycles for
9058 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9059 (also called Dcache, Scache, and Bcache), as well as to main memory.
9060 Note that L3 is only valid for EV5.
9065 @node DEC Alpha/VMS Options
9066 @subsection DEC Alpha/VMS Options
9068 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9071 @item -mvms-return-codes
9072 @opindex mvms-return-codes
9073 Return VMS condition codes from main. The default is to return POSIX
9074 style condition (e.g.@ error) codes.
9078 @subsection FRV Options
9085 Only use the first 32 general purpose registers.
9090 Use all 64 general purpose registers.
9095 Use only the first 32 floating point registers.
9100 Use all 64 floating point registers
9103 @opindex mhard-float
9105 Use hardware instructions for floating point operations.
9108 @opindex msoft-float
9110 Use library routines for floating point operations.
9115 Dynamically allocate condition code registers.
9120 Do not try to dynamically allocate condition code registers, only
9121 use @code{icc0} and @code{fcc0}.
9126 Change ABI to use double word insns.
9131 Do not use double word instructions.
9136 Use floating point double instructions.
9141 Do not use floating point double instructions.
9146 Use media instructions.
9151 Do not use media instructions.
9156 Use multiply and add/subtract instructions.
9161 Do not use multiply and add/subtract instructions.
9166 Select the FDPIC ABI, that uses function descriptors to represent
9167 pointers to functions. Without any PIC/PIE-related options, it
9168 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9169 assumes GOT entries and small data are within a 12-bit range from the
9170 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9171 are computed with 32 bits.
9174 @opindex minline-plt
9176 Enable inlining of PLT entries in function calls to functions that are
9177 not known to bind locally. It has no effect without @option{-mfdpic}.
9178 It's enabled by default if optimizing for speed and compiling for
9179 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9180 optimization option such as @option{-O3} or above is present in the
9186 Assume a large TLS segment when generating thread-local code.
9191 Do not assume a large TLS segment when generating thread-local code.
9196 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9197 that is known to be in read-only sections. It's enabled by default,
9198 except for @option{-fpic} or @option{-fpie}: even though it may help
9199 make the global offset table smaller, it trades 1 instruction for 4.
9200 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9201 one of which may be shared by multiple symbols, and it avoids the need
9202 for a GOT entry for the referenced symbol, so it's more likely to be a
9203 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9205 @item -multilib-library-pic
9206 @opindex multilib-library-pic
9208 Link with the (library, not FD) pic libraries. It's implied by
9209 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9210 @option{-fpic} without @option{-mfdpic}. You should never have to use
9216 Follow the EABI requirement of always creating a frame pointer whenever
9217 a stack frame is allocated. This option is enabled by default and can
9218 be disabled with @option{-mno-linked-fp}.
9221 @opindex mlong-calls
9223 Use indirect addressing to call functions outside the current
9224 compilation unit. This allows the functions to be placed anywhere
9225 within the 32-bit address space.
9227 @item -malign-labels
9228 @opindex malign-labels
9230 Try to align labels to an 8-byte boundary by inserting nops into the
9231 previous packet. This option only has an effect when VLIW packing
9232 is enabled. It doesn't create new packets; it merely adds nops to
9236 @opindex mlibrary-pic
9238 Generate position-independent EABI code.
9243 Use only the first four media accumulator registers.
9248 Use all eight media accumulator registers.
9253 Pack VLIW instructions.
9258 Do not pack VLIW instructions.
9263 Do not mark ABI switches in e_flags.
9268 Enable the use of conditional-move instructions (default).
9270 This switch is mainly for debugging the compiler and will likely be removed
9271 in a future version.
9273 @item -mno-cond-move
9274 @opindex mno-cond-move
9276 Disable the use of conditional-move instructions.
9278 This switch is mainly for debugging the compiler and will likely be removed
9279 in a future version.
9284 Enable the use of conditional set instructions (default).
9286 This switch is mainly for debugging the compiler and will likely be removed
9287 in a future version.
9292 Disable the use of conditional set instructions.
9294 This switch is mainly for debugging the compiler and will likely be removed
9295 in a future version.
9300 Enable the use of conditional execution (default).
9302 This switch is mainly for debugging the compiler and will likely be removed
9303 in a future version.
9305 @item -mno-cond-exec
9306 @opindex mno-cond-exec
9308 Disable the use of conditional execution.
9310 This switch is mainly for debugging the compiler and will likely be removed
9311 in a future version.
9314 @opindex mvliw-branch
9316 Run a pass to pack branches into VLIW instructions (default).
9318 This switch is mainly for debugging the compiler and will likely be removed
9319 in a future version.
9321 @item -mno-vliw-branch
9322 @opindex mno-vliw-branch
9324 Do not run a pass to pack branches into VLIW instructions.
9326 This switch is mainly for debugging the compiler and will likely be removed
9327 in a future version.
9329 @item -mmulti-cond-exec
9330 @opindex mmulti-cond-exec
9332 Enable optimization of @code{&&} and @code{||} in conditional execution
9335 This switch is mainly for debugging the compiler and will likely be removed
9336 in a future version.
9338 @item -mno-multi-cond-exec
9339 @opindex mno-multi-cond-exec
9341 Disable optimization of @code{&&} and @code{||} in conditional execution.
9343 This switch is mainly for debugging the compiler and will likely be removed
9344 in a future version.
9346 @item -mnested-cond-exec
9347 @opindex mnested-cond-exec
9349 Enable nested conditional execution optimizations (default).
9351 This switch is mainly for debugging the compiler and will likely be removed
9352 in a future version.
9354 @item -mno-nested-cond-exec
9355 @opindex mno-nested-cond-exec
9357 Disable nested conditional execution optimizations.
9359 This switch is mainly for debugging the compiler and will likely be removed
9360 in a future version.
9362 @item -moptimize-membar
9363 @opindex moptimize-membar
9365 This switch removes redundant @code{membar} instructions from the
9366 compiler generated code. It is enabled by default.
9368 @item -mno-optimize-membar
9369 @opindex mno-optimize-membar
9371 This switch disables the automatic removal of redundant @code{membar}
9372 instructions from the generated code.
9374 @item -mtomcat-stats
9375 @opindex mtomcat-stats
9377 Cause gas to print out tomcat statistics.
9379 @item -mcpu=@var{cpu}
9382 Select the processor type for which to generate code. Possible values are
9383 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9384 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9388 @node GNU/Linux Options
9389 @subsection GNU/Linux Options
9391 These @samp{-m} options are defined for GNU/Linux targets:
9396 Use the GNU C library instead of uClibc. This is the default except
9397 on @samp{*-*-linux-*uclibc*} targets.
9401 Use uClibc instead of the GNU C library. This is the default on
9402 @samp{*-*-linux-*uclibc*} targets.
9405 @node H8/300 Options
9406 @subsection H8/300 Options
9408 These @samp{-m} options are defined for the H8/300 implementations:
9413 Shorten some address references at link time, when possible; uses the
9414 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9415 ld, Using ld}, for a fuller description.
9419 Generate code for the H8/300H@.
9423 Generate code for the H8S@.
9427 Generate code for the H8S and H8/300H in the normal mode. This switch
9428 must be used either with @option{-mh} or @option{-ms}.
9432 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9436 Make @code{int} data 32 bits by default.
9440 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9441 The default for the H8/300H and H8S is to align longs and floats on 4
9443 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9444 This option has no effect on the H8/300.
9448 @subsection HPPA Options
9449 @cindex HPPA Options
9451 These @samp{-m} options are defined for the HPPA family of computers:
9454 @item -march=@var{architecture-type}
9456 Generate code for the specified architecture. The choices for
9457 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9458 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9459 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9460 architecture option for your machine. Code compiled for lower numbered
9461 architectures will run on higher numbered architectures, but not the
9465 @itemx -mpa-risc-1-1
9466 @itemx -mpa-risc-2-0
9467 @opindex mpa-risc-1-0
9468 @opindex mpa-risc-1-1
9469 @opindex mpa-risc-2-0
9470 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9473 @opindex mbig-switch
9474 Generate code suitable for big switch tables. Use this option only if
9475 the assembler/linker complain about out of range branches within a switch
9478 @item -mjump-in-delay
9479 @opindex mjump-in-delay
9480 Fill delay slots of function calls with unconditional jump instructions
9481 by modifying the return pointer for the function call to be the target
9482 of the conditional jump.
9484 @item -mdisable-fpregs
9485 @opindex mdisable-fpregs
9486 Prevent floating point registers from being used in any manner. This is
9487 necessary for compiling kernels which perform lazy context switching of
9488 floating point registers. If you use this option and attempt to perform
9489 floating point operations, the compiler will abort.
9491 @item -mdisable-indexing
9492 @opindex mdisable-indexing
9493 Prevent the compiler from using indexing address modes. This avoids some
9494 rather obscure problems when compiling MIG generated code under MACH@.
9496 @item -mno-space-regs
9497 @opindex mno-space-regs
9498 Generate code that assumes the target has no space registers. This allows
9499 GCC to generate faster indirect calls and use unscaled index address modes.
9501 Such code is suitable for level 0 PA systems and kernels.
9503 @item -mfast-indirect-calls
9504 @opindex mfast-indirect-calls
9505 Generate code that assumes calls never cross space boundaries. This
9506 allows GCC to emit code which performs faster indirect calls.
9508 This option will not work in the presence of shared libraries or nested
9511 @item -mfixed-range=@var{register-range}
9512 @opindex mfixed-range
9513 Generate code treating the given register range as fixed registers.
9514 A fixed register is one that the register allocator can not use. This is
9515 useful when compiling kernel code. A register range is specified as
9516 two registers separated by a dash. Multiple register ranges can be
9517 specified separated by a comma.
9519 @item -mlong-load-store
9520 @opindex mlong-load-store
9521 Generate 3-instruction load and store sequences as sometimes required by
9522 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9525 @item -mportable-runtime
9526 @opindex mportable-runtime
9527 Use the portable calling conventions proposed by HP for ELF systems.
9531 Enable the use of assembler directives only GAS understands.
9533 @item -mschedule=@var{cpu-type}
9535 Schedule code according to the constraints for the machine type
9536 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9537 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9538 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9539 proper scheduling option for your machine. The default scheduling is
9543 @opindex mlinker-opt
9544 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9545 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9546 linkers in which they give bogus error messages when linking some programs.
9549 @opindex msoft-float
9550 Generate output containing library calls for floating point.
9551 @strong{Warning:} the requisite libraries are not available for all HPPA
9552 targets. Normally the facilities of the machine's usual C compiler are
9553 used, but this cannot be done directly in cross-compilation. You must make
9554 your own arrangements to provide suitable library functions for
9555 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9556 does provide software floating point support.
9558 @option{-msoft-float} changes the calling convention in the output file;
9559 therefore, it is only useful if you compile @emph{all} of a program with
9560 this option. In particular, you need to compile @file{libgcc.a}, the
9561 library that comes with GCC, with @option{-msoft-float} in order for
9566 Generate the predefine, @code{_SIO}, for server IO@. The default is
9567 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9568 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9569 options are available under HP-UX and HI-UX@.
9573 Use GNU ld specific options. This passes @option{-shared} to ld when
9574 building a shared library. It is the default when GCC is configured,
9575 explicitly or implicitly, with the GNU linker. This option does not
9576 have any affect on which ld is called, it only changes what parameters
9577 are passed to that ld. The ld that is called is determined by the
9578 @option{--with-ld} configure option, GCC's program search path, and
9579 finally by the user's @env{PATH}. The linker used by GCC can be printed
9580 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9581 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9585 Use HP ld specific options. This passes @option{-b} to ld when building
9586 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9587 links. It is the default when GCC is configured, explicitly or
9588 implicitly, with the HP linker. This option does not have any affect on
9589 which ld is called, it only changes what parameters are passed to that
9590 ld. The ld that is called is determined by the @option{--with-ld}
9591 configure option, GCC's program search path, and finally by the user's
9592 @env{PATH}. The linker used by GCC can be printed using @samp{which
9593 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9594 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9597 @opindex mno-long-calls
9598 Generate code that uses long call sequences. This ensures that a call
9599 is always able to reach linker generated stubs. The default is to generate
9600 long calls only when the distance from the call site to the beginning
9601 of the function or translation unit, as the case may be, exceeds a
9602 predefined limit set by the branch type being used. The limits for
9603 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9604 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9607 Distances are measured from the beginning of functions when using the
9608 @option{-ffunction-sections} option, or when using the @option{-mgas}
9609 and @option{-mno-portable-runtime} options together under HP-UX with
9612 It is normally not desirable to use this option as it will degrade
9613 performance. However, it may be useful in large applications,
9614 particularly when partial linking is used to build the application.
9616 The types of long calls used depends on the capabilities of the
9617 assembler and linker, and the type of code being generated. The
9618 impact on systems that support long absolute calls, and long pic
9619 symbol-difference or pc-relative calls should be relatively small.
9620 However, an indirect call is used on 32-bit ELF systems in pic code
9621 and it is quite long.
9623 @item -munix=@var{unix-std}
9625 Generate compiler predefines and select a startfile for the specified
9626 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9627 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9628 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9629 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9630 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9633 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9634 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9635 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9636 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9637 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9638 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9640 It is @emph{important} to note that this option changes the interfaces
9641 for various library routines. It also affects the operational behavior
9642 of the C library. Thus, @emph{extreme} care is needed in using this
9645 Library code that is intended to operate with more than one UNIX
9646 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9647 as appropriate. Most GNU software doesn't provide this capability.
9651 Suppress the generation of link options to search libdld.sl when the
9652 @option{-static} option is specified on HP-UX 10 and later.
9656 The HP-UX implementation of setlocale in libc has a dependency on
9657 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9658 when the @option{-static} option is specified, special link options
9659 are needed to resolve this dependency.
9661 On HP-UX 10 and later, the GCC driver adds the necessary options to
9662 link with libdld.sl when the @option{-static} option is specified.
9663 This causes the resulting binary to be dynamic. On the 64-bit port,
9664 the linkers generate dynamic binaries by default in any case. The
9665 @option{-nolibdld} option can be used to prevent the GCC driver from
9666 adding these link options.
9670 Add support for multithreading with the @dfn{dce thread} library
9671 under HP-UX@. This option sets flags for both the preprocessor and
9675 @node i386 and x86-64 Options
9676 @subsection Intel 386 and AMD x86-64 Options
9677 @cindex i386 Options
9678 @cindex x86-64 Options
9679 @cindex Intel 386 Options
9680 @cindex AMD x86-64 Options
9682 These @samp{-m} options are defined for the i386 and x86-64 family of
9686 @item -mtune=@var{cpu-type}
9688 Tune to @var{cpu-type} everything applicable about the generated code, except
9689 for the ABI and the set of available instructions. The choices for
9693 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9694 If you know the CPU on which your code will run, then you should use
9695 the corresponding @option{-mtune} option instead of
9696 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9697 of your application will have, then you should use this option.
9699 As new processors are deployed in the marketplace, the behavior of this
9700 option will change. Therefore, if you upgrade to a newer version of
9701 GCC, the code generated option will change to reflect the processors
9702 that were most common when that version of GCC was released.
9704 There is no @option{-march=generic} option because @option{-march}
9705 indicates the instruction set the compiler can use, and there is no
9706 generic instruction set applicable to all processors. In contrast,
9707 @option{-mtune} indicates the processor (or, in this case, collection of
9708 processors) for which the code is optimized.
9710 This selects the CPU to tune for at compilation time by determining
9711 the processor type of the compiling machine. Using @option{-mtune=native}
9712 will produce code optimized for the local machine under the constraints
9713 of the selected instruction set. Using @option{-march=native} will
9714 enable all instruction subsets supported by the local machine (hence
9715 the result might not run on different machines).
9717 Original Intel's i386 CPU@.
9719 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9721 Intel Pentium CPU with no MMX support.
9723 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9725 Intel PentiumPro CPU@.
9727 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9728 instruction set will be used, so the code will run on all i686 family chips.
9730 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9731 @item pentium3, pentium3m
9732 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9735 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9736 support. Used by Centrino notebooks.
9737 @item pentium4, pentium4m
9738 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9740 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9743 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9744 SSE2 and SSE3 instruction set support.
9746 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
9747 instruction set support.
9749 AMD K6 CPU with MMX instruction set support.
9751 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9752 @item athlon, athlon-tbird
9753 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9755 @item athlon-4, athlon-xp, athlon-mp
9756 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9757 instruction set support.
9758 @item k8, opteron, athlon64, athlon-fx
9759 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9760 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9762 AMD Family 10 core based CPUs with x86-64 instruction set support. (This
9763 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
9764 instruction set extensions.)
9766 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9769 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9770 instruction set support.
9772 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9773 implemented for this chip.)
9775 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9776 implemented for this chip.)
9778 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
9781 While picking a specific @var{cpu-type} will schedule things appropriately
9782 for that particular chip, the compiler will not generate any code that
9783 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9786 @item -march=@var{cpu-type}
9788 Generate instructions for the machine type @var{cpu-type}. The choices
9789 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9790 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9792 @item -mcpu=@var{cpu-type}
9794 A deprecated synonym for @option{-mtune}.
9803 @opindex mpentiumpro
9804 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9805 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9806 These synonyms are deprecated.
9808 @item -mfpmath=@var{unit}
9810 Generate floating point arithmetics for selected unit @var{unit}. The choices
9815 Use the standard 387 floating point coprocessor present majority of chips and
9816 emulated otherwise. Code compiled with this option will run almost everywhere.
9817 The temporary results are computed in 80bit precision instead of precision
9818 specified by the type resulting in slightly different results compared to most
9819 of other chips. See @option{-ffloat-store} for more detailed description.
9821 This is the default choice for i386 compiler.
9824 Use scalar floating point instructions present in the SSE instruction set.
9825 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9826 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9827 instruction set supports only single precision arithmetics, thus the double and
9828 extended precision arithmetics is still done using 387. Later version, present
9829 only in Pentium4 and the future AMD x86-64 chips supports double precision
9832 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9833 or @option{-msse2} switches to enable SSE extensions and make this option
9834 effective. For the x86-64 compiler, these extensions are enabled by default.
9836 The resulting code should be considerably faster in the majority of cases and avoid
9837 the numerical instability problems of 387 code, but may break some existing
9838 code that expects temporaries to be 80bit.
9840 This is the default choice for the x86-64 compiler.
9843 Attempt to utilize both instruction sets at once. This effectively double the
9844 amount of available registers and on chips with separate execution units for
9845 387 and SSE the execution resources too. Use this option with care, as it is
9846 still experimental, because the GCC register allocator does not model separate
9847 functional units well resulting in instable performance.
9850 @item -masm=@var{dialect}
9851 @opindex masm=@var{dialect}
9852 Output asm instructions using selected @var{dialect}. Supported
9853 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9854 not support @samp{intel}.
9859 @opindex mno-ieee-fp
9860 Control whether or not the compiler uses IEEE floating point
9861 comparisons. These handle correctly the case where the result of a
9862 comparison is unordered.
9865 @opindex msoft-float
9866 Generate output containing library calls for floating point.
9867 @strong{Warning:} the requisite libraries are not part of GCC@.
9868 Normally the facilities of the machine's usual C compiler are used, but
9869 this can't be done directly in cross-compilation. You must make your
9870 own arrangements to provide suitable library functions for
9873 On machines where a function returns floating point results in the 80387
9874 register stack, some floating point opcodes may be emitted even if
9875 @option{-msoft-float} is used.
9877 @item -mno-fp-ret-in-387
9878 @opindex mno-fp-ret-in-387
9879 Do not use the FPU registers for return values of functions.
9881 The usual calling convention has functions return values of types
9882 @code{float} and @code{double} in an FPU register, even if there
9883 is no FPU@. The idea is that the operating system should emulate
9886 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9887 in ordinary CPU registers instead.
9889 @item -mno-fancy-math-387
9890 @opindex mno-fancy-math-387
9891 Some 387 emulators do not support the @code{sin}, @code{cos} and
9892 @code{sqrt} instructions for the 387. Specify this option to avoid
9893 generating those instructions. This option is the default on FreeBSD,
9894 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9895 indicates that the target cpu will always have an FPU and so the
9896 instruction will not need emulation. As of revision 2.6.1, these
9897 instructions are not generated unless you also use the
9898 @option{-funsafe-math-optimizations} switch.
9900 @item -malign-double
9901 @itemx -mno-align-double
9902 @opindex malign-double
9903 @opindex mno-align-double
9904 Control whether GCC aligns @code{double}, @code{long double}, and
9905 @code{long long} variables on a two word boundary or a one word
9906 boundary. Aligning @code{double} variables on a two word boundary will
9907 produce code that runs somewhat faster on a @samp{Pentium} at the
9908 expense of more memory.
9910 On x86-64, @option{-malign-double} is enabled by default.
9912 @strong{Warning:} if you use the @option{-malign-double} switch,
9913 structures containing the above types will be aligned differently than
9914 the published application binary interface specifications for the 386
9915 and will not be binary compatible with structures in code compiled
9916 without that switch.
9918 @item -m96bit-long-double
9919 @itemx -m128bit-long-double
9920 @opindex m96bit-long-double
9921 @opindex m128bit-long-double
9922 These switches control the size of @code{long double} type. The i386
9923 application binary interface specifies the size to be 96 bits,
9924 so @option{-m96bit-long-double} is the default in 32 bit mode.
9926 Modern architectures (Pentium and newer) would prefer @code{long double}
9927 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9928 conforming to the ABI, this would not be possible. So specifying a
9929 @option{-m128bit-long-double} will align @code{long double}
9930 to a 16 byte boundary by padding the @code{long double} with an additional
9933 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9934 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9936 Notice that neither of these options enable any extra precision over the x87
9937 standard of 80 bits for a @code{long double}.
9939 @strong{Warning:} if you override the default value for your target ABI, the
9940 structures and arrays containing @code{long double} variables will change
9941 their size as well as function calling convention for function taking
9942 @code{long double} will be modified. Hence they will not be binary
9943 compatible with arrays or structures in code compiled without that switch.
9945 @item -mmlarge-data-threshold=@var{number}
9946 @opindex mlarge-data-threshold=@var{number}
9947 When @option{-mcmodel=medium} is specified, the data greater than
9948 @var{threshold} are placed in large data section. This value must be the
9949 same across all object linked into the binary and defaults to 65535.
9952 @itemx -mno-svr3-shlib
9953 @opindex msvr3-shlib
9954 @opindex mno-svr3-shlib
9955 Control whether GCC places uninitialized local variables into the
9956 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9957 into @code{bss}. These options are meaningful only on System V Release 3.
9961 Use a different function-calling convention, in which functions that
9962 take a fixed number of arguments return with the @code{ret} @var{num}
9963 instruction, which pops their arguments while returning. This saves one
9964 instruction in the caller since there is no need to pop the arguments
9967 You can specify that an individual function is called with this calling
9968 sequence with the function attribute @samp{stdcall}. You can also
9969 override the @option{-mrtd} option by using the function attribute
9970 @samp{cdecl}. @xref{Function Attributes}.
9972 @strong{Warning:} this calling convention is incompatible with the one
9973 normally used on Unix, so you cannot use it if you need to call
9974 libraries compiled with the Unix compiler.
9976 Also, you must provide function prototypes for all functions that
9977 take variable numbers of arguments (including @code{printf});
9978 otherwise incorrect code will be generated for calls to those
9981 In addition, seriously incorrect code will result if you call a
9982 function with too many arguments. (Normally, extra arguments are
9983 harmlessly ignored.)
9985 @item -mregparm=@var{num}
9987 Control how many registers are used to pass integer arguments. By
9988 default, no registers are used to pass arguments, and at most 3
9989 registers can be used. You can control this behavior for a specific
9990 function by using the function attribute @samp{regparm}.
9991 @xref{Function Attributes}.
9993 @strong{Warning:} if you use this switch, and
9994 @var{num} is nonzero, then you must build all modules with the same
9995 value, including any libraries. This includes the system libraries and
9999 @opindex msseregparm
10000 Use SSE register passing conventions for float and double arguments
10001 and return values. You can control this behavior for a specific
10002 function by using the function attribute @samp{sseregparm}.
10003 @xref{Function Attributes}.
10005 @strong{Warning:} if you use this switch then you must build all
10006 modules with the same value, including any libraries. This includes
10007 the system libraries and startup modules.
10009 @item -mstackrealign
10010 @opindex mstackrealign
10011 Realign the stack at entry. On the Intel x86, the
10012 @option{-mstackrealign} option will generate an alternate prologue and
10013 epilogue that realigns the runtime stack. This supports mixing legacy
10014 codes that keep a 4-byte aligned stack with modern codes that keep a
10015 16-byte stack for SSE compatibility. The alternate prologue and
10016 epilogue are slower and bigger than the regular ones, and the
10017 alternate prologue requires an extra scratch register; this lowers the
10018 number of registers available if used in conjunction with the
10019 @code{regparm} attribute. The @option{-mstackrealign} option is
10020 incompatible with the nested function prologue; this is considered a
10021 hard error. See also the attribute @code{force_align_arg_pointer},
10022 applicable to individual functions.
10024 @item -mpreferred-stack-boundary=@var{num}
10025 @opindex mpreferred-stack-boundary
10026 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10027 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10028 the default is 4 (16 bytes or 128 bits).
10030 On Pentium and PentiumPro, @code{double} and @code{long double} values
10031 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10032 suffer significant run time performance penalties. On Pentium III, the
10033 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10034 properly if it is not 16 byte aligned.
10036 To ensure proper alignment of this values on the stack, the stack boundary
10037 must be as aligned as that required by any value stored on the stack.
10038 Further, every function must be generated such that it keeps the stack
10039 aligned. Thus calling a function compiled with a higher preferred
10040 stack boundary from a function compiled with a lower preferred stack
10041 boundary will most likely misalign the stack. It is recommended that
10042 libraries that use callbacks always use the default setting.
10044 This extra alignment does consume extra stack space, and generally
10045 increases code size. Code that is sensitive to stack space usage, such
10046 as embedded systems and operating system kernels, may want to reduce the
10047 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10073 These switches enable or disable the use of instructions in the MMX,
10074 SSE, SSE2, SSE3, SSSE3, SSE4A, ABM or 3DNow! extended instruction sets.
10075 These extensions are also available as built-in functions: see
10076 @ref{X86 Built-in Functions}, for details of the functions enabled and
10077 disabled by these switches.
10079 To have SSE/SSE2 instructions generated automatically from floating-point
10080 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10082 These options will enable GCC to use these extended instructions in
10083 generated code, even without @option{-mfpmath=sse}. Applications which
10084 perform runtime CPU detection must compile separate files for each
10085 supported architecture, using the appropriate flags. In particular,
10086 the file containing the CPU detection code should be compiled without
10091 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10092 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10093 data types. This is useful for high resolution counters that could be updated
10094 by multiple processors (or cores). This instruction is generated as part of
10095 atomic built-in functions: see @ref{Atomic Builtins} for details.
10099 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10100 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10101 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10102 SAHF are load and store instructions, respectively, for certain status flags.
10103 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10104 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10107 @itemx -mno-push-args
10108 @opindex mpush-args
10109 @opindex mno-push-args
10110 Use PUSH operations to store outgoing parameters. This method is shorter
10111 and usually equally fast as method using SUB/MOV operations and is enabled
10112 by default. In some cases disabling it may improve performance because of
10113 improved scheduling and reduced dependencies.
10115 @item -maccumulate-outgoing-args
10116 @opindex maccumulate-outgoing-args
10117 If enabled, the maximum amount of space required for outgoing arguments will be
10118 computed in the function prologue. This is faster on most modern CPUs
10119 because of reduced dependencies, improved scheduling and reduced stack usage
10120 when preferred stack boundary is not equal to 2. The drawback is a notable
10121 increase in code size. This switch implies @option{-mno-push-args}.
10125 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10126 on thread-safe exception handling must compile and link all code with the
10127 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10128 @option{-D_MT}; when linking, it links in a special thread helper library
10129 @option{-lmingwthrd} which cleans up per thread exception handling data.
10131 @item -mno-align-stringops
10132 @opindex mno-align-stringops
10133 Do not align destination of inlined string operations. This switch reduces
10134 code size and improves performance in case the destination is already aligned,
10135 but GCC doesn't know about it.
10137 @item -minline-all-stringops
10138 @opindex minline-all-stringops
10139 By default GCC inlines string operations only when destination is known to be
10140 aligned at least to 4 byte boundary. This enables more inlining, increase code
10141 size, but may improve performance of code that depends on fast memcpy, strlen
10142 and memset for short lengths.
10144 @item -minline-stringops-dynamically
10145 @opindex minline-stringops-dynamically
10146 For string operation of unknown size, inline runtime checks so for small
10147 blocks inline code is used, while for large blocks library call is used.
10149 @item -mstringop-strategy=@var{alg}
10150 @opindex mstringop-strategy=@var{alg}
10151 Overwrite internal decision heuristic about particular algorithm to inline
10152 string operation with. The allowed values are @code{rep_byte},
10153 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10154 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10155 expanding inline loop, @code{libcall} for always expanding library call.
10157 @item -momit-leaf-frame-pointer
10158 @opindex momit-leaf-frame-pointer
10159 Don't keep the frame pointer in a register for leaf functions. This
10160 avoids the instructions to save, set up and restore frame pointers and
10161 makes an extra register available in leaf functions. The option
10162 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10163 which might make debugging harder.
10165 @item -mtls-direct-seg-refs
10166 @itemx -mno-tls-direct-seg-refs
10167 @opindex mtls-direct-seg-refs
10168 Controls whether TLS variables may be accessed with offsets from the
10169 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10170 or whether the thread base pointer must be added. Whether or not this
10171 is legal depends on the operating system, and whether it maps the
10172 segment to cover the entire TLS area.
10174 For systems that use GNU libc, the default is on.
10177 These @samp{-m} switches are supported in addition to the above
10178 on AMD x86-64 processors in 64-bit environments.
10185 Generate code for a 32-bit or 64-bit environment.
10186 The 32-bit environment sets int, long and pointer to 32 bits and
10187 generates code that runs on any i386 system.
10188 The 64-bit environment sets int to 32 bits and long and pointer
10189 to 64 bits and generates code for AMD's x86-64 architecture.
10191 @item -mno-red-zone
10192 @opindex no-red-zone
10193 Do not use a so called red zone for x86-64 code. The red zone is mandated
10194 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10195 stack pointer that will not be modified by signal or interrupt handlers
10196 and therefore can be used for temporary data without adjusting the stack
10197 pointer. The flag @option{-mno-red-zone} disables this red zone.
10199 @item -mcmodel=small
10200 @opindex mcmodel=small
10201 Generate code for the small code model: the program and its symbols must
10202 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10203 Programs can be statically or dynamically linked. This is the default
10206 @item -mcmodel=kernel
10207 @opindex mcmodel=kernel
10208 Generate code for the kernel code model. The kernel runs in the
10209 negative 2 GB of the address space.
10210 This model has to be used for Linux kernel code.
10212 @item -mcmodel=medium
10213 @opindex mcmodel=medium
10214 Generate code for the medium model: The program is linked in the lower 2
10215 GB of the address space but symbols can be located anywhere in the
10216 address space. Programs can be statically or dynamically linked, but
10217 building of shared libraries are not supported with the medium model.
10219 @item -mcmodel=large
10220 @opindex mcmodel=large
10221 Generate code for the large model: This model makes no assumptions
10222 about addresses and sizes of sections.
10225 @node IA-64 Options
10226 @subsection IA-64 Options
10227 @cindex IA-64 Options
10229 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10233 @opindex mbig-endian
10234 Generate code for a big endian target. This is the default for HP-UX@.
10236 @item -mlittle-endian
10237 @opindex mlittle-endian
10238 Generate code for a little endian target. This is the default for AIX5
10244 @opindex mno-gnu-as
10245 Generate (or don't) code for the GNU assembler. This is the default.
10246 @c Also, this is the default if the configure option @option{--with-gnu-as}
10252 @opindex mno-gnu-ld
10253 Generate (or don't) code for the GNU linker. This is the default.
10254 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10259 Generate code that does not use a global pointer register. The result
10260 is not position independent code, and violates the IA-64 ABI@.
10262 @item -mvolatile-asm-stop
10263 @itemx -mno-volatile-asm-stop
10264 @opindex mvolatile-asm-stop
10265 @opindex mno-volatile-asm-stop
10266 Generate (or don't) a stop bit immediately before and after volatile asm
10269 @item -mregister-names
10270 @itemx -mno-register-names
10271 @opindex mregister-names
10272 @opindex mno-register-names
10273 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10274 the stacked registers. This may make assembler output more readable.
10280 Disable (or enable) optimizations that use the small data section. This may
10281 be useful for working around optimizer bugs.
10283 @item -mconstant-gp
10284 @opindex mconstant-gp
10285 Generate code that uses a single constant global pointer value. This is
10286 useful when compiling kernel code.
10290 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10291 This is useful when compiling firmware code.
10293 @item -minline-float-divide-min-latency
10294 @opindex minline-float-divide-min-latency
10295 Generate code for inline divides of floating point values
10296 using the minimum latency algorithm.
10298 @item -minline-float-divide-max-throughput
10299 @opindex minline-float-divide-max-throughput
10300 Generate code for inline divides of floating point values
10301 using the maximum throughput algorithm.
10303 @item -minline-int-divide-min-latency
10304 @opindex minline-int-divide-min-latency
10305 Generate code for inline divides of integer values
10306 using the minimum latency algorithm.
10308 @item -minline-int-divide-max-throughput
10309 @opindex minline-int-divide-max-throughput
10310 Generate code for inline divides of integer values
10311 using the maximum throughput algorithm.
10313 @item -minline-sqrt-min-latency
10314 @opindex minline-sqrt-min-latency
10315 Generate code for inline square roots
10316 using the minimum latency algorithm.
10318 @item -minline-sqrt-max-throughput
10319 @opindex minline-sqrt-max-throughput
10320 Generate code for inline square roots
10321 using the maximum throughput algorithm.
10323 @item -mno-dwarf2-asm
10324 @itemx -mdwarf2-asm
10325 @opindex mno-dwarf2-asm
10326 @opindex mdwarf2-asm
10327 Don't (or do) generate assembler code for the DWARF2 line number debugging
10328 info. This may be useful when not using the GNU assembler.
10330 @item -mearly-stop-bits
10331 @itemx -mno-early-stop-bits
10332 @opindex mearly-stop-bits
10333 @opindex mno-early-stop-bits
10334 Allow stop bits to be placed earlier than immediately preceding the
10335 instruction that triggered the stop bit. This can improve instruction
10336 scheduling, but does not always do so.
10338 @item -mfixed-range=@var{register-range}
10339 @opindex mfixed-range
10340 Generate code treating the given register range as fixed registers.
10341 A fixed register is one that the register allocator can not use. This is
10342 useful when compiling kernel code. A register range is specified as
10343 two registers separated by a dash. Multiple register ranges can be
10344 specified separated by a comma.
10346 @item -mtls-size=@var{tls-size}
10348 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
10351 @item -mtune=@var{cpu-type}
10353 Tune the instruction scheduling for a particular CPU, Valid values are
10354 itanium, itanium1, merced, itanium2, and mckinley.
10360 Add support for multithreading using the POSIX threads library. This
10361 option sets flags for both the preprocessor and linker. It does
10362 not affect the thread safety of object code produced by the compiler or
10363 that of libraries supplied with it. These are HP-UX specific flags.
10369 Generate code for a 32-bit or 64-bit environment.
10370 The 32-bit environment sets int, long and pointer to 32 bits.
10371 The 64-bit environment sets int to 32 bits and long and pointer
10372 to 64 bits. These are HP-UX specific flags.
10374 @item -mno-sched-br-data-spec
10375 @itemx -msched-br-data-spec
10376 @opindex -mno-sched-br-data-spec
10377 @opindex -msched-br-data-spec
10378 (Dis/En)able data speculative scheduling before reload.
10379 This will result in generation of the ld.a instructions and
10380 the corresponding check instructions (ld.c / chk.a).
10381 The default is 'disable'.
10383 @item -msched-ar-data-spec
10384 @itemx -mno-sched-ar-data-spec
10385 @opindex -msched-ar-data-spec
10386 @opindex -mno-sched-ar-data-spec
10387 (En/Dis)able data speculative scheduling after reload.
10388 This will result in generation of the ld.a instructions and
10389 the corresponding check instructions (ld.c / chk.a).
10390 The default is 'enable'.
10392 @item -mno-sched-control-spec
10393 @itemx -msched-control-spec
10394 @opindex -mno-sched-control-spec
10395 @opindex -msched-control-spec
10396 (Dis/En)able control speculative scheduling. This feature is
10397 available only during region scheduling (i.e. before reload).
10398 This will result in generation of the ld.s instructions and
10399 the corresponding check instructions chk.s .
10400 The default is 'disable'.
10402 @item -msched-br-in-data-spec
10403 @itemx -mno-sched-br-in-data-spec
10404 @opindex -msched-br-in-data-spec
10405 @opindex -mno-sched-br-in-data-spec
10406 (En/Dis)able speculative scheduling of the instructions that
10407 are dependent on the data speculative loads before reload.
10408 This is effective only with @option{-msched-br-data-spec} enabled.
10409 The default is 'enable'.
10411 @item -msched-ar-in-data-spec
10412 @itemx -mno-sched-ar-in-data-spec
10413 @opindex -msched-ar-in-data-spec
10414 @opindex -mno-sched-ar-in-data-spec
10415 (En/Dis)able speculative scheduling of the instructions that
10416 are dependent on the data speculative loads after reload.
10417 This is effective only with @option{-msched-ar-data-spec} enabled.
10418 The default is 'enable'.
10420 @item -msched-in-control-spec
10421 @itemx -mno-sched-in-control-spec
10422 @opindex -msched-in-control-spec
10423 @opindex -mno-sched-in-control-spec
10424 (En/Dis)able speculative scheduling of the instructions that
10425 are dependent on the control speculative loads.
10426 This is effective only with @option{-msched-control-spec} enabled.
10427 The default is 'enable'.
10430 @itemx -mno-sched-ldc
10431 @opindex -msched-ldc
10432 @opindex -mno-sched-ldc
10433 (En/Dis)able use of simple data speculation checks ld.c .
10434 If disabled, only chk.a instructions will be emitted to check
10435 data speculative loads.
10436 The default is 'enable'.
10438 @item -mno-sched-control-ldc
10439 @itemx -msched-control-ldc
10440 @opindex -mno-sched-control-ldc
10441 @opindex -msched-control-ldc
10442 (Dis/En)able use of ld.c instructions to check control speculative loads.
10443 If enabled, in case of control speculative load with no speculatively
10444 scheduled dependent instructions this load will be emitted as ld.sa and
10445 ld.c will be used to check it.
10446 The default is 'disable'.
10448 @item -mno-sched-spec-verbose
10449 @itemx -msched-spec-verbose
10450 @opindex -mno-sched-spec-verbose
10451 @opindex -msched-spec-verbose
10452 (Dis/En)able printing of the information about speculative motions.
10454 @item -mno-sched-prefer-non-data-spec-insns
10455 @itemx -msched-prefer-non-data-spec-insns
10456 @opindex -mno-sched-prefer-non-data-spec-insns
10457 @opindex -msched-prefer-non-data-spec-insns
10458 If enabled, data speculative instructions will be chosen for schedule
10459 only if there are no other choices at the moment. This will make
10460 the use of the data speculation much more conservative.
10461 The default is 'disable'.
10463 @item -mno-sched-prefer-non-control-spec-insns
10464 @itemx -msched-prefer-non-control-spec-insns
10465 @opindex -mno-sched-prefer-non-control-spec-insns
10466 @opindex -msched-prefer-non-control-spec-insns
10467 If enabled, control speculative instructions will be chosen for schedule
10468 only if there are no other choices at the moment. This will make
10469 the use of the control speculation much more conservative.
10470 The default is 'disable'.
10472 @item -mno-sched-count-spec-in-critical-path
10473 @itemx -msched-count-spec-in-critical-path
10474 @opindex -mno-sched-count-spec-in-critical-path
10475 @opindex -msched-count-spec-in-critical-path
10476 If enabled, speculative dependencies will be considered during
10477 computation of the instructions priorities. This will make the use of the
10478 speculation a bit more conservative.
10479 The default is 'disable'.
10484 @subsection M32C Options
10485 @cindex M32C options
10488 @item -mcpu=@var{name}
10490 Select the CPU for which code is generated. @var{name} may be one of
10491 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
10492 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
10493 the M32C/80 series.
10497 Specifies that the program will be run on the simulator. This causes
10498 an alternate runtime library to be linked in which supports, for
10499 example, file I/O. You must not use this option when generating
10500 programs that will run on real hardware; you must provide your own
10501 runtime library for whatever I/O functions are needed.
10503 @item -memregs=@var{number}
10505 Specifies the number of memory-based pseudo-registers GCC will use
10506 during code generation. These pseudo-registers will be used like real
10507 registers, so there is a tradeoff between GCC's ability to fit the
10508 code into available registers, and the performance penalty of using
10509 memory instead of registers. Note that all modules in a program must
10510 be compiled with the same value for this option. Because of that, you
10511 must not use this option with the default runtime libraries gcc
10516 @node M32R/D Options
10517 @subsection M32R/D Options
10518 @cindex M32R/D options
10520 These @option{-m} options are defined for Renesas M32R/D architectures:
10525 Generate code for the M32R/2@.
10529 Generate code for the M32R/X@.
10533 Generate code for the M32R@. This is the default.
10535 @item -mmodel=small
10536 @opindex mmodel=small
10537 Assume all objects live in the lower 16MB of memory (so that their addresses
10538 can be loaded with the @code{ld24} instruction), and assume all subroutines
10539 are reachable with the @code{bl} instruction.
10540 This is the default.
10542 The addressability of a particular object can be set with the
10543 @code{model} attribute.
10545 @item -mmodel=medium
10546 @opindex mmodel=medium
10547 Assume objects may be anywhere in the 32-bit address space (the compiler
10548 will generate @code{seth/add3} instructions to load their addresses), and
10549 assume all subroutines are reachable with the @code{bl} instruction.
10551 @item -mmodel=large
10552 @opindex mmodel=large
10553 Assume objects may be anywhere in the 32-bit address space (the compiler
10554 will generate @code{seth/add3} instructions to load their addresses), and
10555 assume subroutines may not be reachable with the @code{bl} instruction
10556 (the compiler will generate the much slower @code{seth/add3/jl}
10557 instruction sequence).
10560 @opindex msdata=none
10561 Disable use of the small data area. Variables will be put into
10562 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10563 @code{section} attribute has been specified).
10564 This is the default.
10566 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10567 Objects may be explicitly put in the small data area with the
10568 @code{section} attribute using one of these sections.
10570 @item -msdata=sdata
10571 @opindex msdata=sdata
10572 Put small global and static data in the small data area, but do not
10573 generate special code to reference them.
10576 @opindex msdata=use
10577 Put small global and static data in the small data area, and generate
10578 special instructions to reference them.
10582 @cindex smaller data references
10583 Put global and static objects less than or equal to @var{num} bytes
10584 into the small data or bss sections instead of the normal data or bss
10585 sections. The default value of @var{num} is 8.
10586 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10587 for this option to have any effect.
10589 All modules should be compiled with the same @option{-G @var{num}} value.
10590 Compiling with different values of @var{num} may or may not work; if it
10591 doesn't the linker will give an error message---incorrect code will not be
10596 Makes the M32R specific code in the compiler display some statistics
10597 that might help in debugging programs.
10599 @item -malign-loops
10600 @opindex malign-loops
10601 Align all loops to a 32-byte boundary.
10603 @item -mno-align-loops
10604 @opindex mno-align-loops
10605 Do not enforce a 32-byte alignment for loops. This is the default.
10607 @item -missue-rate=@var{number}
10608 @opindex missue-rate=@var{number}
10609 Issue @var{number} instructions per cycle. @var{number} can only be 1
10612 @item -mbranch-cost=@var{number}
10613 @opindex mbranch-cost=@var{number}
10614 @var{number} can only be 1 or 2. If it is 1 then branches will be
10615 preferred over conditional code, if it is 2, then the opposite will
10618 @item -mflush-trap=@var{number}
10619 @opindex mflush-trap=@var{number}
10620 Specifies the trap number to use to flush the cache. The default is
10621 12. Valid numbers are between 0 and 15 inclusive.
10623 @item -mno-flush-trap
10624 @opindex mno-flush-trap
10625 Specifies that the cache cannot be flushed by using a trap.
10627 @item -mflush-func=@var{name}
10628 @opindex mflush-func=@var{name}
10629 Specifies the name of the operating system function to call to flush
10630 the cache. The default is @emph{_flush_cache}, but a function call
10631 will only be used if a trap is not available.
10633 @item -mno-flush-func
10634 @opindex mno-flush-func
10635 Indicates that there is no OS function for flushing the cache.
10639 @node M680x0 Options
10640 @subsection M680x0 Options
10641 @cindex M680x0 options
10643 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
10644 The default settings depend on which architecture was selected when
10645 the compiler was configured; the defaults for the most common choices
10649 @item -march=@var{arch}
10651 Generate code for a specific M680x0 or ColdFire instruction set
10652 architecture. Permissible values of @var{arch} for M680x0
10653 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
10654 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
10655 architectures are selected according to Freescale's ISA classification
10656 and the permissible values are: @samp{isaa}, @samp{isaaplus},
10657 @samp{isab} and @samp{isac}.
10659 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
10660 code for a ColdFire target. The @var{arch} in this macro is one of the
10661 @option{-march} arguments given above.
10663 When used together, @option{-march} and @option{-mtune} select code
10664 that runs on a family of similar processors but that is optimized
10665 for a particular microarchitecture.
10667 @item -mcpu=@var{cpu}
10669 Generate code for a specific M680x0 or ColdFire processor.
10670 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
10671 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
10672 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
10673 below, which also classifies the CPUs into families:
10675 @multitable @columnfractions 0.20 0.80
10676 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
10677 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
10678 @item @samp{5206e} @tab @samp{5206e}
10679 @item @samp{5208} @tab @samp{5207} @samp{5208}
10680 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
10681 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
10682 @item @samp{5216} @tab @samp{5214} @samp{5216}
10683 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
10684 @item @samp{5225} @tab @samp{5224} @samp{5225}
10685 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
10686 @item @samp{5249} @tab @samp{5249}
10687 @item @samp{5250} @tab @samp{5250}
10688 @item @samp{5271} @tab @samp{5270} @samp{5271}
10689 @item @samp{5272} @tab @samp{5272}
10690 @item @samp{5275} @tab @samp{5274} @samp{5275}
10691 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
10692 @item @samp{5307} @tab @samp{5307}
10693 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
10694 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
10695 @item @samp{5407} @tab @samp{5407}
10696 @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}
10699 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
10700 @var{arch} is compatible with @var{cpu}. Other combinations of
10701 @option{-mcpu} and @option{-march} are rejected.
10703 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
10704 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
10705 where the value of @var{family} is given by the table above.
10707 @item -mtune=@var{tune}
10709 Tune the code for a particular microarchitecture, within the
10710 constraints set by @option{-march} and @option{-mcpu}.
10711 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
10712 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
10713 and @samp{cpu32}. The ColdFire microarchitectures
10714 are: @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
10716 You can also use @option{-mtune=68020-40} for code that needs
10717 to run relatively well on 68020, 68030 and 68040 targets.
10718 @option{-mtune=68020-60} is similar but includes 68060 targets
10719 as well. These two options select the same tuning decisions as
10720 @option{-m68020-40} and @option{-m68020-60} respectively.
10722 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
10723 when tuning for 680x0 architecture @var{arch}. It also defines
10724 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
10725 option is used. If gcc is tuning for a range of architectures,
10726 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
10727 it defines the macros for every architecture in the range.
10729 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
10730 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
10731 of the arguments given above.
10737 Generate output for a 68000. This is the default
10738 when the compiler is configured for 68000-based systems.
10739 It is equivalent to @option{-march=68000}.
10741 Use this option for microcontrollers with a 68000 or EC000 core,
10742 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10746 Generate output for a 68010. This is the default
10747 when the compiler is configured for 68010-based systems.
10748 It is equivalent to @option{-march=68010}.
10754 Generate output for a 68020. This is the default
10755 when the compiler is configured for 68020-based systems.
10756 It is equivalent to @option{-march=68020}.
10760 Generate output for a 68030. This is the default when the compiler is
10761 configured for 68030-based systems. It is equivalent to
10762 @option{-march=68030}.
10766 Generate output for a 68040. This is the default when the compiler is
10767 configured for 68040-based systems. It is equivalent to
10768 @option{-march=68040}.
10770 This option inhibits the use of 68881/68882 instructions that have to be
10771 emulated by software on the 68040. Use this option if your 68040 does not
10772 have code to emulate those instructions.
10776 Generate output for a 68060. This is the default when the compiler is
10777 configured for 68060-based systems. It is equivalent to
10778 @option{-march=68060}.
10780 This option inhibits the use of 68020 and 68881/68882 instructions that
10781 have to be emulated by software on the 68060. Use this option if your 68060
10782 does not have code to emulate those instructions.
10786 Generate output for a CPU32. This is the default
10787 when the compiler is configured for CPU32-based systems.
10788 It is equivalent to @option{-march=cpu32}.
10790 Use this option for microcontrollers with a
10791 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10792 68336, 68340, 68341, 68349 and 68360.
10796 Generate output for a 520X ColdFire CPU. This is the default
10797 when the compiler is configured for 520X-based systems.
10798 It is equivalent to @option{-mcpu=5206}, and is now deprecated
10799 in favor of that option.
10801 Use this option for microcontroller with a 5200 core, including
10802 the MCF5202, MCF5203, MCF5204 and MCF5206.
10806 Generate output for a 5206e ColdFire CPU. The option is now
10807 deprecated in favor of the equivalent @option{-mcpu=5206e}.
10811 Generate output for a member of the ColdFire 528X family.
10812 The option is now deprecated in favor of the equivalent
10813 @option{-mcpu=528x}.
10817 Generate output for a ColdFire 5307 CPU. The option is now deprecated
10818 in favor of the equivalent @option{-mcpu=5307}.
10822 Generate output for a ColdFire 5407 CPU. The option is now deprecated
10823 in favor of the equivalent @option{-mcpu=5407}.
10827 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
10828 This includes use of hardware floating point instructions.
10829 The option is equivalent to @option{-mcpu=547x}, and is now
10830 deprecated in favor of that option.
10834 Generate output for a 68040, without using any of the new instructions.
10835 This results in code which can run relatively efficiently on either a
10836 68020/68881 or a 68030 or a 68040. The generated code does use the
10837 68881 instructions that are emulated on the 68040.
10839 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
10843 Generate output for a 68060, without using any of the new instructions.
10844 This results in code which can run relatively efficiently on either a
10845 68020/68881 or a 68030 or a 68040. The generated code does use the
10846 68881 instructions that are emulated on the 68060.
10848 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
10852 @opindex mhard-float
10854 Generate floating-point instructions. This is the default for 68020
10855 and above, and for ColdFire devices that have an FPU. It defines the
10856 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
10857 on ColdFire targets.
10860 @opindex msoft-float
10861 Do not generate floating-point instructions; use library calls instead.
10862 This is the default for 68000, 68010, and 68832 targets. It is also
10863 the default for ColdFire devices that have no FPU.
10869 Generate (do not generate) ColdFire hardware divide and remainder
10870 instructions. If @option{-march} is used without @option{-mcpu},
10871 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
10872 architectures. Otherwise, the default is taken from the target CPU
10873 (either the default CPU, or the one specified by @option{-mcpu}). For
10874 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
10875 @option{-mcpu=5206e}.
10877 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
10881 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10882 Additionally, parameters passed on the stack are also aligned to a
10883 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10886 @opindex -mno-short
10887 Do not consider type @code{int} to be 16 bits wide. This is the default.
10890 @itemx -mno-bitfield
10891 @opindex mnobitfield
10892 @opindex mno-bitfield
10893 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10894 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10898 Do use the bit-field instructions. The @option{-m68020} option implies
10899 @option{-mbitfield}. This is the default if you use a configuration
10900 designed for a 68020.
10904 Use a different function-calling convention, in which functions
10905 that take a fixed number of arguments return with the @code{rtd}
10906 instruction, which pops their arguments while returning. This
10907 saves one instruction in the caller since there is no need to pop
10908 the arguments there.
10910 This calling convention is incompatible with the one normally
10911 used on Unix, so you cannot use it if you need to call libraries
10912 compiled with the Unix compiler.
10914 Also, you must provide function prototypes for all functions that
10915 take variable numbers of arguments (including @code{printf});
10916 otherwise incorrect code will be generated for calls to those
10919 In addition, seriously incorrect code will result if you call a
10920 function with too many arguments. (Normally, extra arguments are
10921 harmlessly ignored.)
10923 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10924 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10928 Do not use the calling conventions selected by @option{-mrtd}.
10929 This is the default.
10932 @itemx -mno-align-int
10933 @opindex malign-int
10934 @opindex mno-align-int
10935 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10936 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10937 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10938 Aligning variables on 32-bit boundaries produces code that runs somewhat
10939 faster on processors with 32-bit busses at the expense of more memory.
10941 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10942 align structures containing the above types differently than
10943 most published application binary interface specifications for the m68k.
10947 Use the pc-relative addressing mode of the 68000 directly, instead of
10948 using a global offset table. At present, this option implies @option{-fpic},
10949 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10950 not presently supported with @option{-mpcrel}, though this could be supported for
10951 68020 and higher processors.
10953 @item -mno-strict-align
10954 @itemx -mstrict-align
10955 @opindex mno-strict-align
10956 @opindex mstrict-align
10957 Do not (do) assume that unaligned memory references will be handled by
10961 Generate code that allows the data segment to be located in a different
10962 area of memory from the text segment. This allows for execute in place in
10963 an environment without virtual memory management. This option implies
10966 @item -mno-sep-data
10967 Generate code that assumes that the data segment follows the text segment.
10968 This is the default.
10970 @item -mid-shared-library
10971 Generate code that supports shared libraries via the library ID method.
10972 This allows for execute in place and shared libraries in an environment
10973 without virtual memory management. This option implies @option{-fPIC}.
10975 @item -mno-id-shared-library
10976 Generate code that doesn't assume ID based shared libraries are being used.
10977 This is the default.
10979 @item -mshared-library-id=n
10980 Specified the identification number of the ID based shared library being
10981 compiled. Specifying a value of 0 will generate more compact code, specifying
10982 other values will force the allocation of that number to the current
10983 library but is no more space or time efficient than omitting this option.
10987 @node M68hc1x Options
10988 @subsection M68hc1x Options
10989 @cindex M68hc1x options
10991 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10992 microcontrollers. The default values for these options depends on
10993 which style of microcontroller was selected when the compiler was configured;
10994 the defaults for the most common choices are given below.
11001 Generate output for a 68HC11. This is the default
11002 when the compiler is configured for 68HC11-based systems.
11008 Generate output for a 68HC12. This is the default
11009 when the compiler is configured for 68HC12-based systems.
11015 Generate output for a 68HCS12.
11017 @item -mauto-incdec
11018 @opindex mauto-incdec
11019 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11026 Enable the use of 68HC12 min and max instructions.
11029 @itemx -mno-long-calls
11030 @opindex mlong-calls
11031 @opindex mno-long-calls
11032 Treat all calls as being far away (near). If calls are assumed to be
11033 far away, the compiler will use the @code{call} instruction to
11034 call a function and the @code{rtc} instruction for returning.
11038 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11040 @item -msoft-reg-count=@var{count}
11041 @opindex msoft-reg-count
11042 Specify the number of pseudo-soft registers which are used for the
11043 code generation. The maximum number is 32. Using more pseudo-soft
11044 register may or may not result in better code depending on the program.
11045 The default is 4 for 68HC11 and 2 for 68HC12.
11049 @node MCore Options
11050 @subsection MCore Options
11051 @cindex MCore options
11053 These are the @samp{-m} options defined for the Motorola M*Core
11059 @itemx -mno-hardlit
11061 @opindex mno-hardlit
11062 Inline constants into the code stream if it can be done in two
11063 instructions or less.
11069 Use the divide instruction. (Enabled by default).
11071 @item -mrelax-immediate
11072 @itemx -mno-relax-immediate
11073 @opindex mrelax-immediate
11074 @opindex mno-relax-immediate
11075 Allow arbitrary sized immediates in bit operations.
11077 @item -mwide-bitfields
11078 @itemx -mno-wide-bitfields
11079 @opindex mwide-bitfields
11080 @opindex mno-wide-bitfields
11081 Always treat bit-fields as int-sized.
11083 @item -m4byte-functions
11084 @itemx -mno-4byte-functions
11085 @opindex m4byte-functions
11086 @opindex mno-4byte-functions
11087 Force all functions to be aligned to a four byte boundary.
11089 @item -mcallgraph-data
11090 @itemx -mno-callgraph-data
11091 @opindex mcallgraph-data
11092 @opindex mno-callgraph-data
11093 Emit callgraph information.
11096 @itemx -mno-slow-bytes
11097 @opindex mslow-bytes
11098 @opindex mno-slow-bytes
11099 Prefer word access when reading byte quantities.
11101 @item -mlittle-endian
11102 @itemx -mbig-endian
11103 @opindex mlittle-endian
11104 @opindex mbig-endian
11105 Generate code for a little endian target.
11111 Generate code for the 210 processor.
11115 @subsection MIPS Options
11116 @cindex MIPS options
11122 Generate big-endian code.
11126 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11129 @item -march=@var{arch}
11131 Generate code that will run on @var{arch}, which can be the name of a
11132 generic MIPS ISA, or the name of a particular processor.
11134 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11135 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11136 The processor names are:
11137 @samp{4kc}, @samp{4km}, @samp{4kp},
11138 @samp{4kec}, @samp{4kem}, @samp{4kep},
11139 @samp{5kc}, @samp{5kf},
11141 @samp{24kc}, @samp{24kf}, @samp{24kx},
11142 @samp{24kec}, @samp{24kef}, @samp{24kex},
11143 @samp{34kc}, @samp{34kf}, @samp{34kx},
11146 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11147 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11148 @samp{rm7000}, @samp{rm9000},
11151 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11152 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11153 The special value @samp{from-abi} selects the
11154 most compatible architecture for the selected ABI (that is,
11155 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11157 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11158 (for example, @samp{-march=r2k}). Prefixes are optional, and
11159 @samp{vr} may be written @samp{r}.
11161 GCC defines two macros based on the value of this option. The first
11162 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11163 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11164 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11165 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11166 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11168 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11169 above. In other words, it will have the full prefix and will not
11170 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11171 the macro names the resolved architecture (either @samp{"mips1"} or
11172 @samp{"mips3"}). It names the default architecture when no
11173 @option{-march} option is given.
11175 @item -mtune=@var{arch}
11177 Optimize for @var{arch}. Among other things, this option controls
11178 the way instructions are scheduled, and the perceived cost of arithmetic
11179 operations. The list of @var{arch} values is the same as for
11182 When this option is not used, GCC will optimize for the processor
11183 specified by @option{-march}. By using @option{-march} and
11184 @option{-mtune} together, it is possible to generate code that will
11185 run on a family of processors, but optimize the code for one
11186 particular member of that family.
11188 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11189 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11190 @samp{-march} ones described above.
11194 Equivalent to @samp{-march=mips1}.
11198 Equivalent to @samp{-march=mips2}.
11202 Equivalent to @samp{-march=mips3}.
11206 Equivalent to @samp{-march=mips4}.
11210 Equivalent to @samp{-march=mips32}.
11214 Equivalent to @samp{-march=mips32r2}.
11218 Equivalent to @samp{-march=mips64}.
11223 @opindex mno-mips16
11224 Generate (do not generate) MIPS16 code. If GCC is targetting a
11225 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11237 Generate code for the given ABI@.
11239 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11240 generates 64-bit code when you select a 64-bit architecture, but you
11241 can use @option{-mgp32} to get 32-bit code instead.
11243 For information about the O64 ABI, see
11244 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11246 GCC supports a variant of the o32 ABI in which floating-point registers
11247 are 64 rather than 32 bits wide. You can select this combination with
11248 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11249 and @samp{mfhc1} instructions and is therefore only supported for
11250 MIPS32R2 processors.
11252 The register assignments for arguments and return values remain the
11253 same, but each scalar value is passed in a single 64-bit register
11254 rather than a pair of 32-bit registers. For example, scalar
11255 floating-point values are returned in @samp{$f0} only, not a
11256 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
11257 remains the same, but all 64 bits are saved.
11260 @itemx -mno-abicalls
11262 @opindex mno-abicalls
11263 Generate (do not generate) code that is suitable for SVR4-style
11264 dynamic objects. @option{-mabicalls} is the default for SVR4-based
11269 Generate (do not generate) code that is fully position-independent,
11270 and that can therefore be linked into shared libraries. This option
11271 only affects @option{-mabicalls}.
11273 All @option{-mabicalls} code has traditionally been position-independent,
11274 regardless of options like @option{-fPIC} and @option{-fpic}. However,
11275 as an extension, the GNU toolchain allows executables to use absolute
11276 accesses for locally-binding symbols. It can also use shorter GP
11277 initialization sequences and generate direct calls to locally-defined
11278 functions. This mode is selected by @option{-mno-shared}.
11280 @option{-mno-shared} depends on binutils 2.16 or higher and generates
11281 objects that can only be linked by the GNU linker. However, the option
11282 does not affect the ABI of the final executable; it only affects the ABI
11283 of relocatable objects. Using @option{-mno-shared} will generally make
11284 executables both smaller and quicker.
11286 @option{-mshared} is the default.
11292 Lift (do not lift) the usual restrictions on the size of the global
11295 GCC normally uses a single instruction to load values from the GOT@.
11296 While this is relatively efficient, it will only work if the GOT
11297 is smaller than about 64k. Anything larger will cause the linker
11298 to report an error such as:
11300 @cindex relocation truncated to fit (MIPS)
11302 relocation truncated to fit: R_MIPS_GOT16 foobar
11305 If this happens, you should recompile your code with @option{-mxgot}.
11306 It should then work with very large GOTs, although it will also be
11307 less efficient, since it will take three instructions to fetch the
11308 value of a global symbol.
11310 Note that some linkers can create multiple GOTs. If you have such a
11311 linker, you should only need to use @option{-mxgot} when a single object
11312 file accesses more than 64k's worth of GOT entries. Very few do.
11314 These options have no effect unless GCC is generating position
11319 Assume that general-purpose registers are 32 bits wide.
11323 Assume that general-purpose registers are 64 bits wide.
11327 Assume that floating-point registers are 32 bits wide.
11331 Assume that floating-point registers are 64 bits wide.
11334 @opindex mhard-float
11335 Use floating-point coprocessor instructions.
11338 @opindex msoft-float
11339 Do not use floating-point coprocessor instructions. Implement
11340 floating-point calculations using library calls instead.
11342 @item -msingle-float
11343 @opindex msingle-float
11344 Assume that the floating-point coprocessor only supports single-precision
11347 @itemx -mdouble-float
11348 @opindex mdouble-float
11349 Assume that the floating-point coprocessor supports double-precision
11350 operations. This is the default.
11356 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
11362 Use (do not use) the MIPS DSP ASE REV 2. @xref{MIPS DSP Built-in Functions}.
11363 The option @option{-mdspr2} implies @option{-mdsp}.
11365 @itemx -mpaired-single
11366 @itemx -mno-paired-single
11367 @opindex mpaired-single
11368 @opindex mno-paired-single
11369 Use (do not use) paired-single floating-point instructions.
11370 @xref{MIPS Paired-Single Support}. This option can only be used
11371 when generating 64-bit code and requires hardware floating-point
11372 support to be enabled.
11377 @opindex mno-mips3d
11378 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
11379 The option @option{-mips3d} implies @option{-mpaired-single}.
11383 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
11384 an explanation of the default and the way that the pointer size is
11389 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
11391 The default size of @code{int}s, @code{long}s and pointers depends on
11392 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
11393 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
11394 32-bit @code{long}s. Pointers are the same size as @code{long}s,
11395 or the same size as integer registers, whichever is smaller.
11401 Assume (do not assume) that all symbols have 32-bit values, regardless
11402 of the selected ABI@. This option is useful in combination with
11403 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
11404 to generate shorter and faster references to symbolic addresses.
11408 @cindex smaller data references (MIPS)
11409 @cindex gp-relative references (MIPS)
11410 Put global and static items less than or equal to @var{num} bytes into
11411 the small data or bss section instead of the normal data or bss section.
11412 This allows the data to be accessed using a single instruction.
11414 All modules should be compiled with the same @option{-G @var{num}}
11417 @item -membedded-data
11418 @itemx -mno-embedded-data
11419 @opindex membedded-data
11420 @opindex mno-embedded-data
11421 Allocate variables to the read-only data section first if possible, then
11422 next in the small data section if possible, otherwise in data. This gives
11423 slightly slower code than the default, but reduces the amount of RAM required
11424 when executing, and thus may be preferred for some embedded systems.
11426 @item -muninit-const-in-rodata
11427 @itemx -mno-uninit-const-in-rodata
11428 @opindex muninit-const-in-rodata
11429 @opindex mno-uninit-const-in-rodata
11430 Put uninitialized @code{const} variables in the read-only data section.
11431 This option is only meaningful in conjunction with @option{-membedded-data}.
11433 @item -msplit-addresses
11434 @itemx -mno-split-addresses
11435 @opindex msplit-addresses
11436 @opindex mno-split-addresses
11437 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
11438 relocation operators. This option has been superseded by
11439 @option{-mexplicit-relocs} but is retained for backwards compatibility.
11441 @item -mexplicit-relocs
11442 @itemx -mno-explicit-relocs
11443 @opindex mexplicit-relocs
11444 @opindex mno-explicit-relocs
11445 Use (do not use) assembler relocation operators when dealing with symbolic
11446 addresses. The alternative, selected by @option{-mno-explicit-relocs},
11447 is to use assembler macros instead.
11449 @option{-mexplicit-relocs} is the default if GCC was configured
11450 to use an assembler that supports relocation operators.
11452 @item -mcheck-zero-division
11453 @itemx -mno-check-zero-division
11454 @opindex mcheck-zero-division
11455 @opindex mno-check-zero-division
11456 Trap (do not trap) on integer division by zero.
11458 The default is @option{-mcheck-zero-division}.
11460 @item -mdivide-traps
11461 @itemx -mdivide-breaks
11462 @opindex mdivide-traps
11463 @opindex mdivide-breaks
11464 MIPS systems check for division by zero by generating either a
11465 conditional trap or a break instruction. Using traps results in
11466 smaller code, but is only supported on MIPS II and later. Also, some
11467 versions of the Linux kernel have a bug that prevents trap from
11468 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
11469 allow conditional traps on architectures that support them and
11470 @option{-mdivide-breaks} to force the use of breaks.
11472 The default is usually @option{-mdivide-traps}, but this can be
11473 overridden at configure time using @option{--with-divide=breaks}.
11474 Divide-by-zero checks can be completely disabled using
11475 @option{-mno-check-zero-division}.
11480 @opindex mno-memcpy
11481 Force (do not force) the use of @code{memcpy()} for non-trivial block
11482 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
11483 most constant-sized copies.
11486 @itemx -mno-long-calls
11487 @opindex mlong-calls
11488 @opindex mno-long-calls
11489 Disable (do not disable) use of the @code{jal} instruction. Calling
11490 functions using @code{jal} is more efficient but requires the caller
11491 and callee to be in the same 256 megabyte segment.
11493 This option has no effect on abicalls code. The default is
11494 @option{-mno-long-calls}.
11500 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
11501 instructions, as provided by the R4650 ISA@.
11504 @itemx -mno-fused-madd
11505 @opindex mfused-madd
11506 @opindex mno-fused-madd
11507 Enable (disable) use of the floating point multiply-accumulate
11508 instructions, when they are available. The default is
11509 @option{-mfused-madd}.
11511 When multiply-accumulate instructions are used, the intermediate
11512 product is calculated to infinite precision and is not subject to
11513 the FCSR Flush to Zero bit. This may be undesirable in some
11518 Tell the MIPS assembler to not run its preprocessor over user
11519 assembler files (with a @samp{.s} suffix) when assembling them.
11522 @itemx -mno-fix-r4000
11523 @opindex mfix-r4000
11524 @opindex mno-fix-r4000
11525 Work around certain R4000 CPU errata:
11528 A double-word or a variable shift may give an incorrect result if executed
11529 immediately after starting an integer division.
11531 A double-word or a variable shift may give an incorrect result if executed
11532 while an integer multiplication is in progress.
11534 An integer division may give an incorrect result if started in a delay slot
11535 of a taken branch or a jump.
11539 @itemx -mno-fix-r4400
11540 @opindex mfix-r4400
11541 @opindex mno-fix-r4400
11542 Work around certain R4400 CPU errata:
11545 A double-word or a variable shift may give an incorrect result if executed
11546 immediately after starting an integer division.
11550 @itemx -mno-fix-vr4120
11551 @opindex mfix-vr4120
11552 Work around certain VR4120 errata:
11555 @code{dmultu} does not always produce the correct result.
11557 @code{div} and @code{ddiv} do not always produce the correct result if one
11558 of the operands is negative.
11560 The workarounds for the division errata rely on special functions in
11561 @file{libgcc.a}. At present, these functions are only provided by
11562 the @code{mips64vr*-elf} configurations.
11564 Other VR4120 errata require a nop to be inserted between certain pairs of
11565 instructions. These errata are handled by the assembler, not by GCC itself.
11568 @opindex mfix-vr4130
11569 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
11570 workarounds are implemented by the assembler rather than by GCC,
11571 although GCC will avoid using @code{mflo} and @code{mfhi} if the
11572 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
11573 instructions are available instead.
11576 @itemx -mno-fix-sb1
11578 Work around certain SB-1 CPU core errata.
11579 (This flag currently works around the SB-1 revision 2
11580 ``F1'' and ``F2'' floating point errata.)
11582 @item -mflush-func=@var{func}
11583 @itemx -mno-flush-func
11584 @opindex mflush-func
11585 Specifies the function to call to flush the I and D caches, or to not
11586 call any such function. If called, the function must take the same
11587 arguments as the common @code{_flush_func()}, that is, the address of the
11588 memory range for which the cache is being flushed, the size of the
11589 memory range, and the number 3 (to flush both caches). The default
11590 depends on the target GCC was configured for, but commonly is either
11591 @samp{_flush_func} or @samp{__cpu_flush}.
11593 @item -mbranch-likely
11594 @itemx -mno-branch-likely
11595 @opindex mbranch-likely
11596 @opindex mno-branch-likely
11597 Enable or disable use of Branch Likely instructions, regardless of the
11598 default for the selected architecture. By default, Branch Likely
11599 instructions may be generated if they are supported by the selected
11600 architecture. An exception is for the MIPS32 and MIPS64 architectures
11601 and processors which implement those architectures; for those, Branch
11602 Likely instructions will not be generated by default because the MIPS32
11603 and MIPS64 architectures specifically deprecate their use.
11605 @item -mfp-exceptions
11606 @itemx -mno-fp-exceptions
11607 @opindex mfp-exceptions
11608 Specifies whether FP exceptions are enabled. This affects how we schedule
11609 FP instructions for some processors. The default is that FP exceptions are
11612 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
11613 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
11616 @item -mvr4130-align
11617 @itemx -mno-vr4130-align
11618 @opindex mvr4130-align
11619 The VR4130 pipeline is two-way superscalar, but can only issue two
11620 instructions together if the first one is 8-byte aligned. When this
11621 option is enabled, GCC will align pairs of instructions that it
11622 thinks should execute in parallel.
11624 This option only has an effect when optimizing for the VR4130.
11625 It normally makes code faster, but at the expense of making it bigger.
11626 It is enabled by default at optimization level @option{-O3}.
11630 @subsection MMIX Options
11631 @cindex MMIX Options
11633 These options are defined for the MMIX:
11637 @itemx -mno-libfuncs
11639 @opindex mno-libfuncs
11640 Specify that intrinsic library functions are being compiled, passing all
11641 values in registers, no matter the size.
11644 @itemx -mno-epsilon
11646 @opindex mno-epsilon
11647 Generate floating-point comparison instructions that compare with respect
11648 to the @code{rE} epsilon register.
11650 @item -mabi=mmixware
11652 @opindex mabi-mmixware
11654 Generate code that passes function parameters and return values that (in
11655 the called function) are seen as registers @code{$0} and up, as opposed to
11656 the GNU ABI which uses global registers @code{$231} and up.
11658 @item -mzero-extend
11659 @itemx -mno-zero-extend
11660 @opindex mzero-extend
11661 @opindex mno-zero-extend
11662 When reading data from memory in sizes shorter than 64 bits, use (do not
11663 use) zero-extending load instructions by default, rather than
11664 sign-extending ones.
11667 @itemx -mno-knuthdiv
11669 @opindex mno-knuthdiv
11670 Make the result of a division yielding a remainder have the same sign as
11671 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
11672 remainder follows the sign of the dividend. Both methods are
11673 arithmetically valid, the latter being almost exclusively used.
11675 @item -mtoplevel-symbols
11676 @itemx -mno-toplevel-symbols
11677 @opindex mtoplevel-symbols
11678 @opindex mno-toplevel-symbols
11679 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
11680 code can be used with the @code{PREFIX} assembly directive.
11684 Generate an executable in the ELF format, rather than the default
11685 @samp{mmo} format used by the @command{mmix} simulator.
11687 @item -mbranch-predict
11688 @itemx -mno-branch-predict
11689 @opindex mbranch-predict
11690 @opindex mno-branch-predict
11691 Use (do not use) the probable-branch instructions, when static branch
11692 prediction indicates a probable branch.
11694 @item -mbase-addresses
11695 @itemx -mno-base-addresses
11696 @opindex mbase-addresses
11697 @opindex mno-base-addresses
11698 Generate (do not generate) code that uses @emph{base addresses}. Using a
11699 base address automatically generates a request (handled by the assembler
11700 and the linker) for a constant to be set up in a global register. The
11701 register is used for one or more base address requests within the range 0
11702 to 255 from the value held in the register. The generally leads to short
11703 and fast code, but the number of different data items that can be
11704 addressed is limited. This means that a program that uses lots of static
11705 data may require @option{-mno-base-addresses}.
11707 @item -msingle-exit
11708 @itemx -mno-single-exit
11709 @opindex msingle-exit
11710 @opindex mno-single-exit
11711 Force (do not force) generated code to have a single exit point in each
11715 @node MN10300 Options
11716 @subsection MN10300 Options
11717 @cindex MN10300 options
11719 These @option{-m} options are defined for Matsushita MN10300 architectures:
11724 Generate code to avoid bugs in the multiply instructions for the MN10300
11725 processors. This is the default.
11727 @item -mno-mult-bug
11728 @opindex mno-mult-bug
11729 Do not generate code to avoid bugs in the multiply instructions for the
11730 MN10300 processors.
11734 Generate code which uses features specific to the AM33 processor.
11738 Do not generate code which uses features specific to the AM33 processor. This
11741 @item -mreturn-pointer-on-d0
11742 @opindex mreturn-pointer-on-d0
11743 When generating a function which returns a pointer, return the pointer
11744 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
11745 only in a0, and attempts to call such functions without a prototype
11746 would result in errors. Note that this option is on by default; use
11747 @option{-mno-return-pointer-on-d0} to disable it.
11751 Do not link in the C run-time initialization object file.
11755 Indicate to the linker that it should perform a relaxation optimization pass
11756 to shorten branches, calls and absolute memory addresses. This option only
11757 has an effect when used on the command line for the final link step.
11759 This option makes symbolic debugging impossible.
11763 @subsection MT Options
11766 These @option{-m} options are defined for Morpho MT architectures:
11770 @item -march=@var{cpu-type}
11772 Generate code that will run on @var{cpu-type}, which is the name of a system
11773 representing a certain processor type. Possible values for
11774 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11775 @samp{ms1-16-003} and @samp{ms2}.
11777 When this option is not used, the default is @option{-march=ms1-16-002}.
11781 Use byte loads and stores when generating code.
11785 Do not use byte loads and stores when generating code.
11789 Use simulator runtime
11793 Do not link in the C run-time initialization object file
11794 @file{crti.o}. Other run-time initialization and termination files
11795 such as @file{startup.o} and @file{exit.o} are still included on the
11796 linker command line.
11800 @node PDP-11 Options
11801 @subsection PDP-11 Options
11802 @cindex PDP-11 Options
11804 These options are defined for the PDP-11:
11809 Use hardware FPP floating point. This is the default. (FIS floating
11810 point on the PDP-11/40 is not supported.)
11813 @opindex msoft-float
11814 Do not use hardware floating point.
11818 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11822 Return floating-point results in memory. This is the default.
11826 Generate code for a PDP-11/40.
11830 Generate code for a PDP-11/45. This is the default.
11834 Generate code for a PDP-11/10.
11836 @item -mbcopy-builtin
11837 @opindex bcopy-builtin
11838 Use inline @code{movmemhi} patterns for copying memory. This is the
11843 Do not use inline @code{movmemhi} patterns for copying memory.
11849 Use 16-bit @code{int}. This is the default.
11855 Use 32-bit @code{int}.
11858 @itemx -mno-float32
11860 @opindex mno-float32
11861 Use 64-bit @code{float}. This is the default.
11864 @itemx -mno-float64
11866 @opindex mno-float64
11867 Use 32-bit @code{float}.
11871 Use @code{abshi2} pattern. This is the default.
11875 Do not use @code{abshi2} pattern.
11877 @item -mbranch-expensive
11878 @opindex mbranch-expensive
11879 Pretend that branches are expensive. This is for experimenting with
11880 code generation only.
11882 @item -mbranch-cheap
11883 @opindex mbranch-cheap
11884 Do not pretend that branches are expensive. This is the default.
11888 Generate code for a system with split I&D@.
11892 Generate code for a system without split I&D@. This is the default.
11896 Use Unix assembler syntax. This is the default when configured for
11897 @samp{pdp11-*-bsd}.
11901 Use DEC assembler syntax. This is the default when configured for any
11902 PDP-11 target other than @samp{pdp11-*-bsd}.
11905 @node PowerPC Options
11906 @subsection PowerPC Options
11907 @cindex PowerPC options
11909 These are listed under @xref{RS/6000 and PowerPC Options}.
11911 @node RS/6000 and PowerPC Options
11912 @subsection IBM RS/6000 and PowerPC Options
11913 @cindex RS/6000 and PowerPC Options
11914 @cindex IBM RS/6000 and PowerPC Options
11916 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11923 @itemx -mno-powerpc
11924 @itemx -mpowerpc-gpopt
11925 @itemx -mno-powerpc-gpopt
11926 @itemx -mpowerpc-gfxopt
11927 @itemx -mno-powerpc-gfxopt
11929 @itemx -mno-powerpc64
11933 @itemx -mno-popcntb
11945 @opindex mno-power2
11947 @opindex mno-powerpc
11948 @opindex mpowerpc-gpopt
11949 @opindex mno-powerpc-gpopt
11950 @opindex mpowerpc-gfxopt
11951 @opindex mno-powerpc-gfxopt
11952 @opindex mpowerpc64
11953 @opindex mno-powerpc64
11957 @opindex mno-popcntb
11963 @opindex mno-mfpgpr
11966 GCC supports two related instruction set architectures for the
11967 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11968 instructions supported by the @samp{rios} chip set used in the original
11969 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11970 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11971 the IBM 4xx, 6xx, and follow-on microprocessors.
11973 Neither architecture is a subset of the other. However there is a
11974 large common subset of instructions supported by both. An MQ
11975 register is included in processors supporting the POWER architecture.
11977 You use these options to specify which instructions are available on the
11978 processor you are using. The default value of these options is
11979 determined when configuring GCC@. Specifying the
11980 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11981 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11982 rather than the options listed above.
11984 The @option{-mpower} option allows GCC to generate instructions that
11985 are found only in the POWER architecture and to use the MQ register.
11986 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11987 to generate instructions that are present in the POWER2 architecture but
11988 not the original POWER architecture.
11990 The @option{-mpowerpc} option allows GCC to generate instructions that
11991 are found only in the 32-bit subset of the PowerPC architecture.
11992 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11993 GCC to use the optional PowerPC architecture instructions in the
11994 General Purpose group, including floating-point square root. Specifying
11995 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11996 use the optional PowerPC architecture instructions in the Graphics
11997 group, including floating-point select.
11999 The @option{-mmfcrf} option allows GCC to generate the move from
12000 condition register field instruction implemented on the POWER4
12001 processor and other processors that support the PowerPC V2.01
12003 The @option{-mpopcntb} option allows GCC to generate the popcount and
12004 double precision FP reciprocal estimate instruction implemented on the
12005 POWER5 processor and other processors that support the PowerPC V2.02
12007 The @option{-mfprnd} option allows GCC to generate the FP round to
12008 integer instructions implemented on the POWER5+ processor and other
12009 processors that support the PowerPC V2.03 architecture.
12010 The @option{-mcmpb} option allows GCC to generate the compare bytes
12011 instruction implemented on the POWER6 processor and other processors
12012 that support the PowerPC V2.05 architecture.
12013 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12014 general purpose register instructions implemented on the POWER6X
12015 processor and other processors that support the extended PowerPC V2.05
12017 The @option{-mdfp} option allows GCC to generate the decimal floating
12018 point instructions implemented on some POWER processors.
12020 The @option{-mpowerpc64} option allows GCC to generate the additional
12021 64-bit instructions that are found in the full PowerPC64 architecture
12022 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12023 @option{-mno-powerpc64}.
12025 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12026 will use only the instructions in the common subset of both
12027 architectures plus some special AIX common-mode calls, and will not use
12028 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12029 permits GCC to use any instruction from either architecture and to
12030 allow use of the MQ register; specify this for the Motorola MPC601.
12032 @item -mnew-mnemonics
12033 @itemx -mold-mnemonics
12034 @opindex mnew-mnemonics
12035 @opindex mold-mnemonics
12036 Select which mnemonics to use in the generated assembler code. With
12037 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12038 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12039 assembler mnemonics defined for the POWER architecture. Instructions
12040 defined in only one architecture have only one mnemonic; GCC uses that
12041 mnemonic irrespective of which of these options is specified.
12043 GCC defaults to the mnemonics appropriate for the architecture in
12044 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
12045 value of these option. Unless you are building a cross-compiler, you
12046 should normally not specify either @option{-mnew-mnemonics} or
12047 @option{-mold-mnemonics}, but should instead accept the default.
12049 @item -mcpu=@var{cpu_type}
12051 Set architecture type, register usage, choice of mnemonics, and
12052 instruction scheduling parameters for machine type @var{cpu_type}.
12053 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
12054 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
12055 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
12056 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
12057 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
12058 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
12059 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
12060 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
12061 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
12062 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
12064 @option{-mcpu=common} selects a completely generic processor. Code
12065 generated under this option will run on any POWER or PowerPC processor.
12066 GCC will use only the instructions in the common subset of both
12067 architectures, and will not use the MQ register. GCC assumes a generic
12068 processor model for scheduling purposes.
12070 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
12071 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
12072 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
12073 types, with an appropriate, generic processor model assumed for
12074 scheduling purposes.
12076 The other options specify a specific processor. Code generated under
12077 those options will run best on that processor, and may not run at all on
12080 The @option{-mcpu} options automatically enable or disable the
12083 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
12084 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
12085 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
12087 The particular options set for any particular CPU will vary between
12088 compiler versions, depending on what setting seems to produce optimal
12089 code for that CPU; it doesn't necessarily reflect the actual hardware's
12090 capabilities. If you wish to set an individual option to a particular
12091 value, you may specify it after the @option{-mcpu} option, like
12092 @samp{-mcpu=970 -mno-altivec}.
12094 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
12095 not enabled or disabled by the @option{-mcpu} option at present because
12096 AIX does not have full support for these options. You may still
12097 enable or disable them individually if you're sure it'll work in your
12100 @item -mtune=@var{cpu_type}
12102 Set the instruction scheduling parameters for machine type
12103 @var{cpu_type}, but do not set the architecture type, register usage, or
12104 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
12105 values for @var{cpu_type} are used for @option{-mtune} as for
12106 @option{-mcpu}. If both are specified, the code generated will use the
12107 architecture, registers, and mnemonics set by @option{-mcpu}, but the
12108 scheduling parameters set by @option{-mtune}.
12114 Generate code to compute division as reciprocal estimate and iterative
12115 refinement, creating opportunities for increased throughput. This
12116 feature requires: optional PowerPC Graphics instruction set for single
12117 precision and FRE instruction for double precision, assuming divides
12118 cannot generate user-visible traps, and the domain values not include
12119 Infinities, denormals or zero denominator.
12122 @itemx -mno-altivec
12124 @opindex mno-altivec
12125 Generate code that uses (does not use) AltiVec instructions, and also
12126 enable the use of built-in functions that allow more direct access to
12127 the AltiVec instruction set. You may also need to set
12128 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
12134 @opindex mno-vrsave
12135 Generate VRSAVE instructions when generating AltiVec code.
12138 @opindex msecure-plt
12139 Generate code that allows ld and ld.so to build executables and shared
12140 libraries with non-exec .plt and .got sections. This is a PowerPC
12141 32-bit SYSV ABI option.
12145 Generate code that uses a BSS .plt section that ld.so fills in, and
12146 requires .plt and .got sections that are both writable and executable.
12147 This is a PowerPC 32-bit SYSV ABI option.
12153 This switch enables or disables the generation of ISEL instructions.
12155 @item -misel=@var{yes/no}
12156 This switch has been deprecated. Use @option{-misel} and
12157 @option{-mno-isel} instead.
12163 This switch enables or disables the generation of SPE simd
12166 @item -mspe=@var{yes/no}
12167 This option has been deprecated. Use @option{-mspe} and
12168 @option{-mno-spe} instead.
12170 @item -mfloat-gprs=@var{yes/single/double/no}
12171 @itemx -mfloat-gprs
12172 @opindex mfloat-gprs
12173 This switch enables or disables the generation of floating point
12174 operations on the general purpose registers for architectures that
12177 The argument @var{yes} or @var{single} enables the use of
12178 single-precision floating point operations.
12180 The argument @var{double} enables the use of single and
12181 double-precision floating point operations.
12183 The argument @var{no} disables floating point operations on the
12184 general purpose registers.
12186 This option is currently only available on the MPC854x.
12192 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
12193 targets (including GNU/Linux). The 32-bit environment sets int, long
12194 and pointer to 32 bits and generates code that runs on any PowerPC
12195 variant. The 64-bit environment sets int to 32 bits and long and
12196 pointer to 64 bits, and generates code for PowerPC64, as for
12197 @option{-mpowerpc64}.
12200 @itemx -mno-fp-in-toc
12201 @itemx -mno-sum-in-toc
12202 @itemx -mminimal-toc
12204 @opindex mno-fp-in-toc
12205 @opindex mno-sum-in-toc
12206 @opindex mminimal-toc
12207 Modify generation of the TOC (Table Of Contents), which is created for
12208 every executable file. The @option{-mfull-toc} option is selected by
12209 default. In that case, GCC will allocate at least one TOC entry for
12210 each unique non-automatic variable reference in your program. GCC
12211 will also place floating-point constants in the TOC@. However, only
12212 16,384 entries are available in the TOC@.
12214 If you receive a linker error message that saying you have overflowed
12215 the available TOC space, you can reduce the amount of TOC space used
12216 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
12217 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
12218 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
12219 generate code to calculate the sum of an address and a constant at
12220 run-time instead of putting that sum into the TOC@. You may specify one
12221 or both of these options. Each causes GCC to produce very slightly
12222 slower and larger code at the expense of conserving TOC space.
12224 If you still run out of space in the TOC even when you specify both of
12225 these options, specify @option{-mminimal-toc} instead. This option causes
12226 GCC to make only one TOC entry for every file. When you specify this
12227 option, GCC will produce code that is slower and larger but which
12228 uses extremely little TOC space. You may wish to use this option
12229 only on files that contain less frequently executed code.
12235 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
12236 @code{long} type, and the infrastructure needed to support them.
12237 Specifying @option{-maix64} implies @option{-mpowerpc64} and
12238 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
12239 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
12242 @itemx -mno-xl-compat
12243 @opindex mxl-compat
12244 @opindex mno-xl-compat
12245 Produce code that conforms more closely to IBM XL compiler semantics
12246 when using AIX-compatible ABI. Pass floating-point arguments to
12247 prototyped functions beyond the register save area (RSA) on the stack
12248 in addition to argument FPRs. Do not assume that most significant
12249 double in 128-bit long double value is properly rounded when comparing
12250 values and converting to double. Use XL symbol names for long double
12253 The AIX calling convention was extended but not initially documented to
12254 handle an obscure K&R C case of calling a function that takes the
12255 address of its arguments with fewer arguments than declared. IBM XL
12256 compilers access floating point arguments which do not fit in the
12257 RSA from the stack when a subroutine is compiled without
12258 optimization. Because always storing floating-point arguments on the
12259 stack is inefficient and rarely needed, this option is not enabled by
12260 default and only is necessary when calling subroutines compiled by IBM
12261 XL compilers without optimization.
12265 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
12266 application written to use message passing with special startup code to
12267 enable the application to run. The system must have PE installed in the
12268 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
12269 must be overridden with the @option{-specs=} option to specify the
12270 appropriate directory location. The Parallel Environment does not
12271 support threads, so the @option{-mpe} option and the @option{-pthread}
12272 option are incompatible.
12274 @item -malign-natural
12275 @itemx -malign-power
12276 @opindex malign-natural
12277 @opindex malign-power
12278 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
12279 @option{-malign-natural} overrides the ABI-defined alignment of larger
12280 types, such as floating-point doubles, on their natural size-based boundary.
12281 The option @option{-malign-power} instructs GCC to follow the ABI-specified
12282 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
12284 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
12288 @itemx -mhard-float
12289 @opindex msoft-float
12290 @opindex mhard-float
12291 Generate code that does not use (uses) the floating-point register set.
12292 Software floating point emulation is provided if you use the
12293 @option{-msoft-float} option, and pass the option to GCC when linking.
12296 @itemx -mno-multiple
12298 @opindex mno-multiple
12299 Generate code that uses (does not use) the load multiple word
12300 instructions and the store multiple word instructions. These
12301 instructions are generated by default on POWER systems, and not
12302 generated on PowerPC systems. Do not use @option{-mmultiple} on little
12303 endian PowerPC systems, since those instructions do not work when the
12304 processor is in little endian mode. The exceptions are PPC740 and
12305 PPC750 which permit the instructions usage in little endian mode.
12310 @opindex mno-string
12311 Generate code that uses (does not use) the load string instructions
12312 and the store string word instructions to save multiple registers and
12313 do small block moves. These instructions are generated by default on
12314 POWER systems, and not generated on PowerPC systems. Do not use
12315 @option{-mstring} on little endian PowerPC systems, since those
12316 instructions do not work when the processor is in little endian mode.
12317 The exceptions are PPC740 and PPC750 which permit the instructions
12318 usage in little endian mode.
12323 @opindex mno-update
12324 Generate code that uses (does not use) the load or store instructions
12325 that update the base register to the address of the calculated memory
12326 location. These instructions are generated by default. If you use
12327 @option{-mno-update}, there is a small window between the time that the
12328 stack pointer is updated and the address of the previous frame is
12329 stored, which means code that walks the stack frame across interrupts or
12330 signals may get corrupted data.
12333 @itemx -mno-fused-madd
12334 @opindex mfused-madd
12335 @opindex mno-fused-madd
12336 Generate code that uses (does not use) the floating point multiply and
12337 accumulate instructions. These instructions are generated by default if
12338 hardware floating is used.
12344 Generate code that uses (does not use) the half-word multiply and
12345 multiply-accumulate instructions on the IBM 405 and 440 processors.
12346 These instructions are generated by default when targetting those
12353 Generate code that uses (does not use) the string-search @samp{dlmzb}
12354 instruction on the IBM 405 and 440 processors. This instruction is
12355 generated by default when targetting those processors.
12357 @item -mno-bit-align
12359 @opindex mno-bit-align
12360 @opindex mbit-align
12361 On System V.4 and embedded PowerPC systems do not (do) force structures
12362 and unions that contain bit-fields to be aligned to the base type of the
12365 For example, by default a structure containing nothing but 8
12366 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
12367 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
12368 the structure would be aligned to a 1 byte boundary and be one byte in
12371 @item -mno-strict-align
12372 @itemx -mstrict-align
12373 @opindex mno-strict-align
12374 @opindex mstrict-align
12375 On System V.4 and embedded PowerPC systems do not (do) assume that
12376 unaligned memory references will be handled by the system.
12378 @item -mrelocatable
12379 @itemx -mno-relocatable
12380 @opindex mrelocatable
12381 @opindex mno-relocatable
12382 On embedded PowerPC systems generate code that allows (does not allow)
12383 the program to be relocated to a different address at runtime. If you
12384 use @option{-mrelocatable} on any module, all objects linked together must
12385 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
12387 @item -mrelocatable-lib
12388 @itemx -mno-relocatable-lib
12389 @opindex mrelocatable-lib
12390 @opindex mno-relocatable-lib
12391 On embedded PowerPC systems generate code that allows (does not allow)
12392 the program to be relocated to a different address at runtime. Modules
12393 compiled with @option{-mrelocatable-lib} can be linked with either modules
12394 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
12395 with modules compiled with the @option{-mrelocatable} options.
12401 On System V.4 and embedded PowerPC systems do not (do) assume that
12402 register 2 contains a pointer to a global area pointing to the addresses
12403 used in the program.
12406 @itemx -mlittle-endian
12408 @opindex mlittle-endian
12409 On System V.4 and embedded PowerPC systems compile code for the
12410 processor in little endian mode. The @option{-mlittle-endian} option is
12411 the same as @option{-mlittle}.
12414 @itemx -mbig-endian
12416 @opindex mbig-endian
12417 On System V.4 and embedded PowerPC systems compile code for the
12418 processor in big endian mode. The @option{-mbig-endian} option is
12419 the same as @option{-mbig}.
12421 @item -mdynamic-no-pic
12422 @opindex mdynamic-no-pic
12423 On Darwin and Mac OS X systems, compile code so that it is not
12424 relocatable, but that its external references are relocatable. The
12425 resulting code is suitable for applications, but not shared
12428 @item -mprioritize-restricted-insns=@var{priority}
12429 @opindex mprioritize-restricted-insns
12430 This option controls the priority that is assigned to
12431 dispatch-slot restricted instructions during the second scheduling
12432 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
12433 @var{no/highest/second-highest} priority to dispatch slot restricted
12436 @item -msched-costly-dep=@var{dependence_type}
12437 @opindex msched-costly-dep
12438 This option controls which dependences are considered costly
12439 by the target during instruction scheduling. The argument
12440 @var{dependence_type} takes one of the following values:
12441 @var{no}: no dependence is costly,
12442 @var{all}: all dependences are costly,
12443 @var{true_store_to_load}: a true dependence from store to load is costly,
12444 @var{store_to_load}: any dependence from store to load is costly,
12445 @var{number}: any dependence which latency >= @var{number} is costly.
12447 @item -minsert-sched-nops=@var{scheme}
12448 @opindex minsert-sched-nops
12449 This option controls which nop insertion scheme will be used during
12450 the second scheduling pass. The argument @var{scheme} takes one of the
12452 @var{no}: Don't insert nops.
12453 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
12454 according to the scheduler's grouping.
12455 @var{regroup_exact}: Insert nops to force costly dependent insns into
12456 separate groups. Insert exactly as many nops as needed to force an insn
12457 to a new group, according to the estimated processor grouping.
12458 @var{number}: Insert nops to force costly dependent insns into
12459 separate groups. Insert @var{number} nops to force an insn to a new group.
12462 @opindex mcall-sysv
12463 On System V.4 and embedded PowerPC systems compile code using calling
12464 conventions that adheres to the March 1995 draft of the System V
12465 Application Binary Interface, PowerPC processor supplement. This is the
12466 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
12468 @item -mcall-sysv-eabi
12469 @opindex mcall-sysv-eabi
12470 Specify both @option{-mcall-sysv} and @option{-meabi} options.
12472 @item -mcall-sysv-noeabi
12473 @opindex mcall-sysv-noeabi
12474 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
12476 @item -mcall-solaris
12477 @opindex mcall-solaris
12478 On System V.4 and embedded PowerPC systems compile code for the Solaris
12482 @opindex mcall-linux
12483 On System V.4 and embedded PowerPC systems compile code for the
12484 Linux-based GNU system.
12488 On System V.4 and embedded PowerPC systems compile code for the
12489 Hurd-based GNU system.
12491 @item -mcall-netbsd
12492 @opindex mcall-netbsd
12493 On System V.4 and embedded PowerPC systems compile code for the
12494 NetBSD operating system.
12496 @item -maix-struct-return
12497 @opindex maix-struct-return
12498 Return all structures in memory (as specified by the AIX ABI)@.
12500 @item -msvr4-struct-return
12501 @opindex msvr4-struct-return
12502 Return structures smaller than 8 bytes in registers (as specified by the
12505 @item -mabi=@var{abi-type}
12507 Extend the current ABI with a particular extension, or remove such extension.
12508 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
12509 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
12513 Extend the current ABI with SPE ABI extensions. This does not change
12514 the default ABI, instead it adds the SPE ABI extensions to the current
12518 @opindex mabi=no-spe
12519 Disable Booke SPE ABI extensions for the current ABI@.
12521 @item -mabi=ibmlongdouble
12522 @opindex mabi=ibmlongdouble
12523 Change the current ABI to use IBM extended precision long double.
12524 This is a PowerPC 32-bit SYSV ABI option.
12526 @item -mabi=ieeelongdouble
12527 @opindex mabi=ieeelongdouble
12528 Change the current ABI to use IEEE extended precision long double.
12529 This is a PowerPC 32-bit Linux ABI option.
12532 @itemx -mno-prototype
12533 @opindex mprototype
12534 @opindex mno-prototype
12535 On System V.4 and embedded PowerPC systems assume that all calls to
12536 variable argument functions are properly prototyped. Otherwise, the
12537 compiler must insert an instruction before every non prototyped call to
12538 set or clear bit 6 of the condition code register (@var{CR}) to
12539 indicate whether floating point values were passed in the floating point
12540 registers in case the function takes a variable arguments. With
12541 @option{-mprototype}, only calls to prototyped variable argument functions
12542 will set or clear the bit.
12546 On embedded PowerPC systems, assume that the startup module is called
12547 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
12548 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
12553 On embedded PowerPC systems, assume that the startup module is called
12554 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
12559 On embedded PowerPC systems, assume that the startup module is called
12560 @file{crt0.o} and the standard C libraries are @file{libads.a} and
12563 @item -myellowknife
12564 @opindex myellowknife
12565 On embedded PowerPC systems, assume that the startup module is called
12566 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
12571 On System V.4 and embedded PowerPC systems, specify that you are
12572 compiling for a VxWorks system.
12576 Specify that you are compiling for the WindISS simulation environment.
12580 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
12581 header to indicate that @samp{eabi} extended relocations are used.
12587 On System V.4 and embedded PowerPC systems do (do not) adhere to the
12588 Embedded Applications Binary Interface (eabi) which is a set of
12589 modifications to the System V.4 specifications. Selecting @option{-meabi}
12590 means that the stack is aligned to an 8 byte boundary, a function
12591 @code{__eabi} is called to from @code{main} to set up the eabi
12592 environment, and the @option{-msdata} option can use both @code{r2} and
12593 @code{r13} to point to two separate small data areas. Selecting
12594 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
12595 do not call an initialization function from @code{main}, and the
12596 @option{-msdata} option will only use @code{r13} to point to a single
12597 small data area. The @option{-meabi} option is on by default if you
12598 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
12601 @opindex msdata=eabi
12602 On System V.4 and embedded PowerPC systems, put small initialized
12603 @code{const} global and static data in the @samp{.sdata2} section, which
12604 is pointed to by register @code{r2}. Put small initialized
12605 non-@code{const} global and static data in the @samp{.sdata} section,
12606 which is pointed to by register @code{r13}. Put small uninitialized
12607 global and static data in the @samp{.sbss} section, which is adjacent to
12608 the @samp{.sdata} section. The @option{-msdata=eabi} option is
12609 incompatible with the @option{-mrelocatable} option. The
12610 @option{-msdata=eabi} option also sets the @option{-memb} option.
12613 @opindex msdata=sysv
12614 On System V.4 and embedded PowerPC systems, put small global and static
12615 data in the @samp{.sdata} section, which is pointed to by register
12616 @code{r13}. Put small uninitialized global and static data in the
12617 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
12618 The @option{-msdata=sysv} option is incompatible with the
12619 @option{-mrelocatable} option.
12621 @item -msdata=default
12623 @opindex msdata=default
12625 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
12626 compile code the same as @option{-msdata=eabi}, otherwise compile code the
12627 same as @option{-msdata=sysv}.
12630 @opindex msdata-data
12631 On System V.4 and embedded PowerPC systems, put small global
12632 data in the @samp{.sdata} section. Put small uninitialized global
12633 data in the @samp{.sbss} section. Do not use register @code{r13}
12634 to address small data however. This is the default behavior unless
12635 other @option{-msdata} options are used.
12639 @opindex msdata=none
12641 On embedded PowerPC systems, put all initialized global and static data
12642 in the @samp{.data} section, and all uninitialized data in the
12643 @samp{.bss} section.
12647 @cindex smaller data references (PowerPC)
12648 @cindex .sdata/.sdata2 references (PowerPC)
12649 On embedded PowerPC systems, put global and static items less than or
12650 equal to @var{num} bytes into the small data or bss sections instead of
12651 the normal data or bss section. By default, @var{num} is 8. The
12652 @option{-G @var{num}} switch is also passed to the linker.
12653 All modules should be compiled with the same @option{-G @var{num}} value.
12656 @itemx -mno-regnames
12658 @opindex mno-regnames
12659 On System V.4 and embedded PowerPC systems do (do not) emit register
12660 names in the assembly language output using symbolic forms.
12663 @itemx -mno-longcall
12665 @opindex mno-longcall
12666 By default assume that all calls are far away so that a longer more
12667 expensive calling sequence is required. This is required for calls
12668 further than 32 megabytes (33,554,432 bytes) from the current location.
12669 A short call will be generated if the compiler knows
12670 the call cannot be that far away. This setting can be overridden by
12671 the @code{shortcall} function attribute, or by @code{#pragma
12674 Some linkers are capable of detecting out-of-range calls and generating
12675 glue code on the fly. On these systems, long calls are unnecessary and
12676 generate slower code. As of this writing, the AIX linker can do this,
12677 as can the GNU linker for PowerPC/64. It is planned to add this feature
12678 to the GNU linker for 32-bit PowerPC systems as well.
12680 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
12681 callee, L42'', plus a ``branch island'' (glue code). The two target
12682 addresses represent the callee and the ``branch island''. The
12683 Darwin/PPC linker will prefer the first address and generate a ``bl
12684 callee'' if the PPC ``bl'' instruction will reach the callee directly;
12685 otherwise, the linker will generate ``bl L42'' to call the ``branch
12686 island''. The ``branch island'' is appended to the body of the
12687 calling function; it computes the full 32-bit address of the callee
12690 On Mach-O (Darwin) systems, this option directs the compiler emit to
12691 the glue for every direct call, and the Darwin linker decides whether
12692 to use or discard it.
12694 In the future, we may cause GCC to ignore all longcall specifications
12695 when the linker is known to generate glue.
12699 Adds support for multithreading with the @dfn{pthreads} library.
12700 This option sets flags for both the preprocessor and linker.
12704 @node S/390 and zSeries Options
12705 @subsection S/390 and zSeries Options
12706 @cindex S/390 and zSeries Options
12708 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
12712 @itemx -msoft-float
12713 @opindex mhard-float
12714 @opindex msoft-float
12715 Use (do not use) the hardware floating-point instructions and registers
12716 for floating-point operations. When @option{-msoft-float} is specified,
12717 functions in @file{libgcc.a} will be used to perform floating-point
12718 operations. When @option{-mhard-float} is specified, the compiler
12719 generates IEEE floating-point instructions. This is the default.
12721 @item -mlong-double-64
12722 @itemx -mlong-double-128
12723 @opindex mlong-double-64
12724 @opindex mlong-double-128
12725 These switches control the size of @code{long double} type. A size
12726 of 64bit makes the @code{long double} type equivalent to the @code{double}
12727 type. This is the default.
12730 @itemx -mno-backchain
12731 @opindex mbackchain
12732 @opindex mno-backchain
12733 Store (do not store) the address of the caller's frame as backchain pointer
12734 into the callee's stack frame.
12735 A backchain may be needed to allow debugging using tools that do not understand
12736 DWARF-2 call frame information.
12737 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
12738 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
12739 the backchain is placed into the topmost word of the 96/160 byte register
12742 In general, code compiled with @option{-mbackchain} is call-compatible with
12743 code compiled with @option{-mmo-backchain}; however, use of the backchain
12744 for debugging purposes usually requires that the whole binary is built with
12745 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
12746 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12747 to build a linux kernel use @option{-msoft-float}.
12749 The default is to not maintain the backchain.
12751 @item -mpacked-stack
12752 @item -mno-packed-stack
12753 @opindex mpacked-stack
12754 @opindex mno-packed-stack
12755 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
12756 specified, the compiler uses the all fields of the 96/160 byte register save
12757 area only for their default purpose; unused fields still take up stack space.
12758 When @option{-mpacked-stack} is specified, register save slots are densely
12759 packed at the top of the register save area; unused space is reused for other
12760 purposes, allowing for more efficient use of the available stack space.
12761 However, when @option{-mbackchain} is also in effect, the topmost word of
12762 the save area is always used to store the backchain, and the return address
12763 register is always saved two words below the backchain.
12765 As long as the stack frame backchain is not used, code generated with
12766 @option{-mpacked-stack} is call-compatible with code generated with
12767 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
12768 S/390 or zSeries generated code that uses the stack frame backchain at run
12769 time, not just for debugging purposes. Such code is not call-compatible
12770 with code compiled with @option{-mpacked-stack}. Also, note that the
12771 combination of @option{-mbackchain},
12772 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12773 to build a linux kernel use @option{-msoft-float}.
12775 The default is to not use the packed stack layout.
12778 @itemx -mno-small-exec
12779 @opindex msmall-exec
12780 @opindex mno-small-exec
12781 Generate (or do not generate) code using the @code{bras} instruction
12782 to do subroutine calls.
12783 This only works reliably if the total executable size does not
12784 exceed 64k. The default is to use the @code{basr} instruction instead,
12785 which does not have this limitation.
12791 When @option{-m31} is specified, generate code compliant to the
12792 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12793 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12794 particular to generate 64-bit instructions. For the @samp{s390}
12795 targets, the default is @option{-m31}, while the @samp{s390x}
12796 targets default to @option{-m64}.
12802 When @option{-mzarch} is specified, generate code using the
12803 instructions available on z/Architecture.
12804 When @option{-mesa} is specified, generate code using the
12805 instructions available on ESA/390. Note that @option{-mesa} is
12806 not possible with @option{-m64}.
12807 When generating code compliant to the GNU/Linux for S/390 ABI,
12808 the default is @option{-mesa}. When generating code compliant
12809 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12815 Generate (or do not generate) code using the @code{mvcle} instruction
12816 to perform block moves. When @option{-mno-mvcle} is specified,
12817 use a @code{mvc} loop instead. This is the default unless optimizing for
12824 Print (or do not print) additional debug information when compiling.
12825 The default is to not print debug information.
12827 @item -march=@var{cpu-type}
12829 Generate code that will run on @var{cpu-type}, which is the name of a system
12830 representing a certain processor type. Possible values for
12831 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12832 When generating code using the instructions available on z/Architecture,
12833 the default is @option{-march=z900}. Otherwise, the default is
12834 @option{-march=g5}.
12836 @item -mtune=@var{cpu-type}
12838 Tune to @var{cpu-type} everything applicable about the generated code,
12839 except for the ABI and the set of available instructions.
12840 The list of @var{cpu-type} values is the same as for @option{-march}.
12841 The default is the value used for @option{-march}.
12844 @itemx -mno-tpf-trace
12845 @opindex mtpf-trace
12846 @opindex mno-tpf-trace
12847 Generate code that adds (does not add) in TPF OS specific branches to trace
12848 routines in the operating system. This option is off by default, even
12849 when compiling for the TPF OS@.
12852 @itemx -mno-fused-madd
12853 @opindex mfused-madd
12854 @opindex mno-fused-madd
12855 Generate code that uses (does not use) the floating point multiply and
12856 accumulate instructions. These instructions are generated by default if
12857 hardware floating point is used.
12859 @item -mwarn-framesize=@var{framesize}
12860 @opindex mwarn-framesize
12861 Emit a warning if the current function exceeds the given frame size. Because
12862 this is a compile time check it doesn't need to be a real problem when the program
12863 runs. It is intended to identify functions which most probably cause
12864 a stack overflow. It is useful to be used in an environment with limited stack
12865 size e.g.@: the linux kernel.
12867 @item -mwarn-dynamicstack
12868 @opindex mwarn-dynamicstack
12869 Emit a warning if the function calls alloca or uses dynamically
12870 sized arrays. This is generally a bad idea with a limited stack size.
12872 @item -mstack-guard=@var{stack-guard}
12873 @item -mstack-size=@var{stack-size}
12874 @opindex mstack-guard
12875 @opindex mstack-size
12876 If these options are provided the s390 back end emits additional instructions in
12877 the function prologue which trigger a trap if the stack size is @var{stack-guard}
12878 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
12879 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
12880 the frame size of the compiled function is chosen.
12881 These options are intended to be used to help debugging stack overflow problems.
12882 The additionally emitted code causes only little overhead and hence can also be
12883 used in production like systems without greater performance degradation. The given
12884 values have to be exact powers of 2 and @var{stack-size} has to be greater than
12885 @var{stack-guard} without exceeding 64k.
12886 In order to be efficient the extra code makes the assumption that the stack starts
12887 at an address aligned to the value given by @var{stack-size}.
12888 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
12891 @node Score Options
12892 @subsection Score Options
12893 @cindex Score Options
12895 These options are defined for Score implementations:
12900 Compile code for little endian mode.
12904 Compile code for big endian mode. This is the default.
12908 Enable the use of multiply-accumulate instructions. Disabled by default.
12912 Specify the SCORE5U of the target architecture.
12916 Specify the SCORE7 of the target architecture. This is the default.
12920 @subsection SH Options
12922 These @samp{-m} options are defined for the SH implementations:
12927 Generate code for the SH1.
12931 Generate code for the SH2.
12934 Generate code for the SH2e.
12938 Generate code for the SH3.
12942 Generate code for the SH3e.
12946 Generate code for the SH4 without a floating-point unit.
12948 @item -m4-single-only
12949 @opindex m4-single-only
12950 Generate code for the SH4 with a floating-point unit that only
12951 supports single-precision arithmetic.
12955 Generate code for the SH4 assuming the floating-point unit is in
12956 single-precision mode by default.
12960 Generate code for the SH4.
12964 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12965 floating-point unit is not used.
12967 @item -m4a-single-only
12968 @opindex m4a-single-only
12969 Generate code for the SH4a, in such a way that no double-precision
12970 floating point operations are used.
12973 @opindex m4a-single
12974 Generate code for the SH4a assuming the floating-point unit is in
12975 single-precision mode by default.
12979 Generate code for the SH4a.
12983 Same as @option{-m4a-nofpu}, except that it implicitly passes
12984 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12985 instructions at the moment.
12989 Compile code for the processor in big endian mode.
12993 Compile code for the processor in little endian mode.
12997 Align doubles at 64-bit boundaries. Note that this changes the calling
12998 conventions, and thus some functions from the standard C library will
12999 not work unless you recompile it first with @option{-mdalign}.
13003 Shorten some address references at link time, when possible; uses the
13004 linker option @option{-relax}.
13008 Use 32-bit offsets in @code{switch} tables. The default is to use
13013 Enable the use of the instruction @code{fmovd}.
13017 Comply with the calling conventions defined by Renesas.
13021 Comply with the calling conventions defined by Renesas.
13025 Comply with the calling conventions defined for GCC before the Renesas
13026 conventions were available. This option is the default for all
13027 targets of the SH toolchain except for @samp{sh-symbianelf}.
13030 @opindex mnomacsave
13031 Mark the @code{MAC} register as call-clobbered, even if
13032 @option{-mhitachi} is given.
13036 Increase IEEE-compliance of floating-point code.
13037 At the moment, this is equivalent to @option{-fno-finite-math-only}.
13038 When generating 16 bit SH opcodes, getting IEEE-conforming results for
13039 comparisons of NANs / infinities incurs extra overhead in every
13040 floating point comparison, therefore the default is set to
13041 @option{-ffinite-math-only}.
13043 @item -minline-ic_invalidate
13044 @opindex minline-ic_invalidate
13045 Inline code to invalidate instruction cache entries after setting up
13046 nested function trampolines.
13047 This option has no effect if -musermode is in effect and the selected
13048 code generation option (e.g. -m4) does not allow the use of the icbi
13050 If the selected code generation option does not allow the use of the icbi
13051 instruction, and -musermode is not in effect, the inlined code will
13052 manipulate the instruction cache address array directly with an associative
13053 write. This not only requires privileged mode, but it will also
13054 fail if the cache line had been mapped via the TLB and has become unmapped.
13058 Dump instruction size and location in the assembly code.
13061 @opindex mpadstruct
13062 This option is deprecated. It pads structures to multiple of 4 bytes,
13063 which is incompatible with the SH ABI@.
13067 Optimize for space instead of speed. Implied by @option{-Os}.
13070 @opindex mprefergot
13071 When generating position-independent code, emit function calls using
13072 the Global Offset Table instead of the Procedure Linkage Table.
13076 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
13077 if the inlined code would not work in user mode.
13078 This is the default when the target is @code{sh-*-linux*}.
13080 @item -multcost=@var{number}
13081 @opindex multcost=@var{number}
13082 Set the cost to assume for a multiply insn.
13084 @item -mdiv=@var{strategy}
13085 @opindex mdiv=@var{strategy}
13086 Set the division strategy to use for SHmedia code. @var{strategy} must be
13087 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
13088 inv:call2, inv:fp .
13089 "fp" performs the operation in floating point. This has a very high latency,
13090 but needs only a few instructions, so it might be a good choice if
13091 your code has enough easily exploitable ILP to allow the compiler to
13092 schedule the floating point instructions together with other instructions.
13093 Division by zero causes a floating point exception.
13094 "inv" uses integer operations to calculate the inverse of the divisor,
13095 and then multiplies the dividend with the inverse. This strategy allows
13096 cse and hoisting of the inverse calculation. Division by zero calculates
13097 an unspecified result, but does not trap.
13098 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
13099 have been found, or if the entire operation has been hoisted to the same
13100 place, the last stages of the inverse calculation are intertwined with the
13101 final multiply to reduce the overall latency, at the expense of using a few
13102 more instructions, and thus offering fewer scheduling opportunities with
13104 "call" calls a library function that usually implements the inv:minlat
13106 This gives high code density for m5-*media-nofpu compilations.
13107 "call2" uses a different entry point of the same library function, where it
13108 assumes that a pointer to a lookup table has already been set up, which
13109 exposes the pointer load to cse / code hoisting optimizations.
13110 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
13111 code generation, but if the code stays unoptimized, revert to the "call",
13112 "call2", or "fp" strategies, respectively. Note that the
13113 potentially-trapping side effect of division by zero is carried by a
13114 separate instruction, so it is possible that all the integer instructions
13115 are hoisted out, but the marker for the side effect stays where it is.
13116 A recombination to fp operations or a call is not possible in that case.
13117 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
13118 that the inverse calculation was nor separated from the multiply, they speed
13119 up division where the dividend fits into 20 bits (plus sign where applicable),
13120 by inserting a test to skip a number of operations in this case; this test
13121 slows down the case of larger dividends. inv20u assumes the case of a such
13122 a small dividend to be unlikely, and inv20l assumes it to be likely.
13124 @item -mdivsi3_libfunc=@var{name}
13125 @opindex mdivsi3_libfunc=@var{name}
13126 Set the name of the library function used for 32 bit signed division to
13127 @var{name}. This only affect the name used in the call and inv:call
13128 division strategies, and the compiler will still expect the same
13129 sets of input/output/clobbered registers as if this option was not present.
13131 @item -madjust-unroll
13132 @opindex madjust-unroll
13133 Throttle unrolling to avoid thrashing target registers.
13134 This option only has an effect if the gcc code base supports the
13135 TARGET_ADJUST_UNROLL_MAX target hook.
13137 @item -mindexed-addressing
13138 @opindex mindexed-addressing
13139 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
13140 This is only safe if the hardware and/or OS implement 32 bit wrap-around
13141 semantics for the indexed addressing mode. The architecture allows the
13142 implementation of processors with 64 bit MMU, which the OS could use to
13143 get 32 bit addressing, but since no current hardware implementation supports
13144 this or any other way to make the indexed addressing mode safe to use in
13145 the 32 bit ABI, the default is -mno-indexed-addressing.
13147 @item -mgettrcost=@var{number}
13148 @opindex mgettrcost=@var{number}
13149 Set the cost assumed for the gettr instruction to @var{number}.
13150 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
13154 Assume pt* instructions won't trap. This will generally generate better
13155 scheduled code, but is unsafe on current hardware. The current architecture
13156 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
13157 This has the unintentional effect of making it unsafe to schedule ptabs /
13158 ptrel before a branch, or hoist it out of a loop. For example,
13159 __do_global_ctors, a part of libgcc that runs constructors at program
13160 startup, calls functions in a list which is delimited by @minus{}1. With the
13161 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
13162 That means that all the constructors will be run a bit quicker, but when
13163 the loop comes to the end of the list, the program crashes because ptabs
13164 loads @minus{}1 into a target register. Since this option is unsafe for any
13165 hardware implementing the current architecture specification, the default
13166 is -mno-pt-fixed. Unless the user specifies a specific cost with
13167 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
13168 this deters register allocation using target registers for storing
13171 @item -minvalid-symbols
13172 @opindex minvalid-symbols
13173 Assume symbols might be invalid. Ordinary function symbols generated by
13174 the compiler will always be valid to load with movi/shori/ptabs or
13175 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
13176 to generate symbols that will cause ptabs / ptrel to trap.
13177 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
13178 It will then prevent cross-basic-block cse, hoisting and most scheduling
13179 of symbol loads. The default is @option{-mno-invalid-symbols}.
13182 @node SPARC Options
13183 @subsection SPARC Options
13184 @cindex SPARC options
13186 These @samp{-m} options are supported on the SPARC:
13189 @item -mno-app-regs
13191 @opindex mno-app-regs
13193 Specify @option{-mapp-regs} to generate output using the global registers
13194 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
13197 To be fully SVR4 ABI compliant at the cost of some performance loss,
13198 specify @option{-mno-app-regs}. You should compile libraries and system
13199 software with this option.
13202 @itemx -mhard-float
13204 @opindex mhard-float
13205 Generate output containing floating point instructions. This is the
13209 @itemx -msoft-float
13211 @opindex msoft-float
13212 Generate output containing library calls for floating point.
13213 @strong{Warning:} the requisite libraries are not available for all SPARC
13214 targets. Normally the facilities of the machine's usual C compiler are
13215 used, but this cannot be done directly in cross-compilation. You must make
13216 your own arrangements to provide suitable library functions for
13217 cross-compilation. The embedded targets @samp{sparc-*-aout} and
13218 @samp{sparclite-*-*} do provide software floating point support.
13220 @option{-msoft-float} changes the calling convention in the output file;
13221 therefore, it is only useful if you compile @emph{all} of a program with
13222 this option. In particular, you need to compile @file{libgcc.a}, the
13223 library that comes with GCC, with @option{-msoft-float} in order for
13226 @item -mhard-quad-float
13227 @opindex mhard-quad-float
13228 Generate output containing quad-word (long double) floating point
13231 @item -msoft-quad-float
13232 @opindex msoft-quad-float
13233 Generate output containing library calls for quad-word (long double)
13234 floating point instructions. The functions called are those specified
13235 in the SPARC ABI@. This is the default.
13237 As of this writing, there are no SPARC implementations that have hardware
13238 support for the quad-word floating point instructions. They all invoke
13239 a trap handler for one of these instructions, and then the trap handler
13240 emulates the effect of the instruction. Because of the trap handler overhead,
13241 this is much slower than calling the ABI library routines. Thus the
13242 @option{-msoft-quad-float} option is the default.
13244 @item -mno-unaligned-doubles
13245 @itemx -munaligned-doubles
13246 @opindex mno-unaligned-doubles
13247 @opindex munaligned-doubles
13248 Assume that doubles have 8 byte alignment. This is the default.
13250 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
13251 alignment only if they are contained in another type, or if they have an
13252 absolute address. Otherwise, it assumes they have 4 byte alignment.
13253 Specifying this option avoids some rare compatibility problems with code
13254 generated by other compilers. It is not the default because it results
13255 in a performance loss, especially for floating point code.
13257 @item -mno-faster-structs
13258 @itemx -mfaster-structs
13259 @opindex mno-faster-structs
13260 @opindex mfaster-structs
13261 With @option{-mfaster-structs}, the compiler assumes that structures
13262 should have 8 byte alignment. This enables the use of pairs of
13263 @code{ldd} and @code{std} instructions for copies in structure
13264 assignment, in place of twice as many @code{ld} and @code{st} pairs.
13265 However, the use of this changed alignment directly violates the SPARC
13266 ABI@. Thus, it's intended only for use on targets where the developer
13267 acknowledges that their resulting code will not be directly in line with
13268 the rules of the ABI@.
13270 @item -mimpure-text
13271 @opindex mimpure-text
13272 @option{-mimpure-text}, used in addition to @option{-shared}, tells
13273 the compiler to not pass @option{-z text} to the linker when linking a
13274 shared object. Using this option, you can link position-dependent
13275 code into a shared object.
13277 @option{-mimpure-text} suppresses the ``relocations remain against
13278 allocatable but non-writable sections'' linker error message.
13279 However, the necessary relocations will trigger copy-on-write, and the
13280 shared object is not actually shared across processes. Instead of
13281 using @option{-mimpure-text}, you should compile all source code with
13282 @option{-fpic} or @option{-fPIC}.
13284 This option is only available on SunOS and Solaris.
13286 @item -mcpu=@var{cpu_type}
13288 Set the instruction set, register set, and instruction scheduling parameters
13289 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
13290 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
13291 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
13292 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
13293 @samp{ultrasparc3}, and @samp{niagara}.
13295 Default instruction scheduling parameters are used for values that select
13296 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
13297 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
13299 Here is a list of each supported architecture and their supported
13304 v8: supersparc, hypersparc
13305 sparclite: f930, f934, sparclite86x
13307 v9: ultrasparc, ultrasparc3, niagara
13310 By default (unless configured otherwise), GCC generates code for the V7
13311 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
13312 additionally optimizes it for the Cypress CY7C602 chip, as used in the
13313 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
13314 SPARCStation 1, 2, IPX etc.
13316 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
13317 architecture. The only difference from V7 code is that the compiler emits
13318 the integer multiply and integer divide instructions which exist in SPARC-V8
13319 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
13320 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
13323 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
13324 the SPARC architecture. This adds the integer multiply, integer divide step
13325 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
13326 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
13327 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
13328 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
13329 MB86934 chip, which is the more recent SPARClite with FPU@.
13331 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
13332 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
13333 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
13334 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
13335 optimizes it for the TEMIC SPARClet chip.
13337 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
13338 architecture. This adds 64-bit integer and floating-point move instructions,
13339 3 additional floating-point condition code registers and conditional move
13340 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
13341 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
13342 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
13343 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
13344 @option{-mcpu=niagara}, the compiler additionally optimizes it for
13345 Sun UltraSPARC T1 chips.
13347 @item -mtune=@var{cpu_type}
13349 Set the instruction scheduling parameters for machine type
13350 @var{cpu_type}, but do not set the instruction set or register set that the
13351 option @option{-mcpu=@var{cpu_type}} would.
13353 The same values for @option{-mcpu=@var{cpu_type}} can be used for
13354 @option{-mtune=@var{cpu_type}}, but the only useful values are those
13355 that select a particular cpu implementation. Those are @samp{cypress},
13356 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
13357 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
13358 @samp{ultrasparc3}, and @samp{niagara}.
13363 @opindex mno-v8plus
13364 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
13365 difference from the V8 ABI is that the global and out registers are
13366 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
13367 mode for all SPARC-V9 processors.
13373 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
13374 Visual Instruction Set extensions. The default is @option{-mno-vis}.
13377 These @samp{-m} options are supported in addition to the above
13378 on SPARC-V9 processors in 64-bit environments:
13381 @item -mlittle-endian
13382 @opindex mlittle-endian
13383 Generate code for a processor running in little-endian mode. It is only
13384 available for a few configurations and most notably not on Solaris and Linux.
13390 Generate code for a 32-bit or 64-bit environment.
13391 The 32-bit environment sets int, long and pointer to 32 bits.
13392 The 64-bit environment sets int to 32 bits and long and pointer
13395 @item -mcmodel=medlow
13396 @opindex mcmodel=medlow
13397 Generate code for the Medium/Low code model: 64-bit addresses, programs
13398 must be linked in the low 32 bits of memory. Programs can be statically
13399 or dynamically linked.
13401 @item -mcmodel=medmid
13402 @opindex mcmodel=medmid
13403 Generate code for the Medium/Middle code model: 64-bit addresses, programs
13404 must be linked in the low 44 bits of memory, the text and data segments must
13405 be less than 2GB in size and the data segment must be located within 2GB of
13408 @item -mcmodel=medany
13409 @opindex mcmodel=medany
13410 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
13411 may be linked anywhere in memory, the text and data segments must be less
13412 than 2GB in size and the data segment must be located within 2GB of the
13415 @item -mcmodel=embmedany
13416 @opindex mcmodel=embmedany
13417 Generate code for the Medium/Anywhere code model for embedded systems:
13418 64-bit addresses, the text and data segments must be less than 2GB in
13419 size, both starting anywhere in memory (determined at link time). The
13420 global register %g4 points to the base of the data segment. Programs
13421 are statically linked and PIC is not supported.
13424 @itemx -mno-stack-bias
13425 @opindex mstack-bias
13426 @opindex mno-stack-bias
13427 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
13428 frame pointer if present, are offset by @minus{}2047 which must be added back
13429 when making stack frame references. This is the default in 64-bit mode.
13430 Otherwise, assume no such offset is present.
13433 These switches are supported in addition to the above on Solaris:
13438 Add support for multithreading using the Solaris threads library. This
13439 option sets flags for both the preprocessor and linker. This option does
13440 not affect the thread safety of object code produced by the compiler or
13441 that of libraries supplied with it.
13445 Add support for multithreading using the POSIX threads library. This
13446 option sets flags for both the preprocessor and linker. This option does
13447 not affect the thread safety of object code produced by the compiler or
13448 that of libraries supplied with it.
13452 This is a synonym for @option{-pthreads}.
13456 @subsection SPU Options
13457 @cindex SPU options
13459 These @samp{-m} options are supported on the SPU:
13463 @itemx -merror-reloc
13464 @opindex mwarn-reloc
13465 @opindex merror-reloc
13467 The loader for SPU does not handle dynamic relocations. By default, GCC
13468 will give an error when it generates code that requires a dynamic
13469 relocation. @option{-mno-error-reloc} disables the error,
13470 @option{-mwarn-reloc} will generate a warning instead.
13473 @itemx -munsafe-dma
13475 @opindex munsafe-dma
13477 Instructions which initiate or test completion of DMA must not be
13478 reordered with respect to loads and stores of the memory which is being
13479 accessed. Users typically address this problem using the volatile
13480 keyword, but that can lead to inefficient code in places where the
13481 memory is known to not change. Rather than mark the memory as volatile
13482 we treat the DMA instructions as potentially effecting all memory. With
13483 @option{-munsafe-dma} users must use the volatile keyword to protect
13486 @item -mbranch-hints
13487 @opindex mbranch-hints
13489 By default, GCC will generate a branch hint instruction to avoid
13490 pipeline stalls for always taken or probably taken branches. A hint
13491 will not be generated closer than 8 instructions away from its branch.
13492 There is little reason to disable them, except for debugging purposes,
13493 or to make an object a little bit smaller.
13497 @opindex msmall-mem
13498 @opindex mlarge-mem
13500 By default, GCC generates code assuming that addresses are never larger
13501 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
13502 a full 32 bit address.
13507 By default, GCC links against startup code that assumes the SPU-style
13508 main function interface (which has an unconventional parameter list).
13509 With @option{-mstdmain}, GCC will link your program against startup
13510 code that assumes a C99-style interface to @code{main}, including a
13511 local copy of @code{argv} strings.
13513 @item -mfixed-range=@var{register-range}
13514 @opindex mfixed-range
13515 Generate code treating the given register range as fixed registers.
13516 A fixed register is one that the register allocator can not use. This is
13517 useful when compiling kernel code. A register range is specified as
13518 two registers separated by a dash. Multiple register ranges can be
13519 specified separated by a comma.
13523 @node System V Options
13524 @subsection Options for System V
13526 These additional options are available on System V Release 4 for
13527 compatibility with other compilers on those systems:
13532 Create a shared object.
13533 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
13537 Identify the versions of each tool used by the compiler, in a
13538 @code{.ident} assembler directive in the output.
13542 Refrain from adding @code{.ident} directives to the output file (this is
13545 @item -YP,@var{dirs}
13547 Search the directories @var{dirs}, and no others, for libraries
13548 specified with @option{-l}.
13550 @item -Ym,@var{dir}
13552 Look in the directory @var{dir} to find the M4 preprocessor.
13553 The assembler uses this option.
13554 @c This is supposed to go with a -Yd for predefined M4 macro files, but
13555 @c the generic assembler that comes with Solaris takes just -Ym.
13558 @node TMS320C3x/C4x Options
13559 @subsection TMS320C3x/C4x Options
13560 @cindex TMS320C3x/C4x Options
13562 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
13566 @item -mcpu=@var{cpu_type}
13568 Set the instruction set, register set, and instruction scheduling
13569 parameters for machine type @var{cpu_type}. Supported values for
13570 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
13571 @samp{c44}. The default is @samp{c40} to generate code for the
13576 @itemx -msmall-memory
13578 @opindex mbig-memory
13580 @opindex msmall-memory
13582 Generates code for the big or small memory model. The small memory
13583 model assumed that all data fits into one 64K word page. At run-time
13584 the data page (DP) register must be set to point to the 64K page
13585 containing the .bss and .data program sections. The big memory model is
13586 the default and requires reloading of the DP register for every direct
13593 Allow (disallow) allocation of general integer operands into the block
13594 count register BK@.
13600 Enable (disable) generation of code using decrement and branch,
13601 DBcond(D), instructions. This is enabled by default for the C4x. To be
13602 on the safe side, this is disabled for the C3x, since the maximum
13603 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
13604 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
13605 that it can utilize the decrement and branch instruction, but will give
13606 up if there is more than one memory reference in the loop. Thus a loop
13607 where the loop counter is decremented can generate slightly more
13608 efficient code, in cases where the RPTB instruction cannot be utilized.
13610 @item -mdp-isr-reload
13612 @opindex mdp-isr-reload
13614 Force the DP register to be saved on entry to an interrupt service
13615 routine (ISR), reloaded to point to the data section, and restored on
13616 exit from the ISR@. This should not be required unless someone has
13617 violated the small memory model by modifying the DP register, say within
13624 For the C3x use the 24-bit MPYI instruction for integer multiplies
13625 instead of a library call to guarantee 32-bit results. Note that if one
13626 of the operands is a constant, then the multiplication will be performed
13627 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
13628 then squaring operations are performed inline instead of a library call.
13631 @itemx -mno-fast-fix
13633 @opindex mno-fast-fix
13634 The C3x/C4x FIX instruction to convert a floating point value to an
13635 integer value chooses the nearest integer less than or equal to the
13636 floating point value rather than to the nearest integer. Thus if the
13637 floating point number is negative, the result will be incorrectly
13638 truncated an additional code is necessary to detect and correct this
13639 case. This option can be used to disable generation of the additional
13640 code required to correct the result.
13646 Enable (disable) generation of repeat block sequences using the RPTB
13647 instruction for zero overhead looping. The RPTB construct is only used
13648 for innermost loops that do not call functions or jump across the loop
13649 boundaries. There is no advantage having nested RPTB loops due to the
13650 overhead required to save and restore the RC, RS, and RE registers.
13651 This is enabled by default with @option{-O2}.
13653 @item -mrpts=@var{count}
13657 Enable (disable) the use of the single instruction repeat instruction
13658 RPTS@. If a repeat block contains a single instruction, and the loop
13659 count can be guaranteed to be less than the value @var{count}, GCC will
13660 emit a RPTS instruction instead of a RPTB@. If no value is specified,
13661 then a RPTS will be emitted even if the loop count cannot be determined
13662 at compile time. Note that the repeated instruction following RPTS does
13663 not have to be reloaded from memory each iteration, thus freeing up the
13664 CPU buses for operands. However, since interrupts are blocked by this
13665 instruction, it is disabled by default.
13667 @item -mloop-unsigned
13668 @itemx -mno-loop-unsigned
13669 @opindex mloop-unsigned
13670 @opindex mno-loop-unsigned
13671 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
13672 is @math{2^{31} + 1} since these instructions test if the iteration count is
13673 negative to terminate the loop. If the iteration count is unsigned
13674 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
13675 exceeded. This switch allows an unsigned iteration count.
13679 Try to emit an assembler syntax that the TI assembler (asm30) is happy
13680 with. This also enforces compatibility with the API employed by the TI
13681 C3x C compiler. For example, long doubles are passed as structures
13682 rather than in floating point registers.
13688 Generate code that uses registers (stack) for passing arguments to functions.
13689 By default, arguments are passed in registers where possible rather
13690 than by pushing arguments on to the stack.
13692 @item -mparallel-insns
13693 @itemx -mno-parallel-insns
13694 @opindex mparallel-insns
13695 @opindex mno-parallel-insns
13696 Allow the generation of parallel instructions. This is enabled by
13697 default with @option{-O2}.
13699 @item -mparallel-mpy
13700 @itemx -mno-parallel-mpy
13701 @opindex mparallel-mpy
13702 @opindex mno-parallel-mpy
13703 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
13704 provided @option{-mparallel-insns} is also specified. These instructions have
13705 tight register constraints which can pessimize the code generation
13706 of large functions.
13711 @subsection V850 Options
13712 @cindex V850 Options
13714 These @samp{-m} options are defined for V850 implementations:
13718 @itemx -mno-long-calls
13719 @opindex mlong-calls
13720 @opindex mno-long-calls
13721 Treat all calls as being far away (near). If calls are assumed to be
13722 far away, the compiler will always load the functions address up into a
13723 register, and call indirect through the pointer.
13729 Do not optimize (do optimize) basic blocks that use the same index
13730 pointer 4 or more times to copy pointer into the @code{ep} register, and
13731 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
13732 option is on by default if you optimize.
13734 @item -mno-prolog-function
13735 @itemx -mprolog-function
13736 @opindex mno-prolog-function
13737 @opindex mprolog-function
13738 Do not use (do use) external functions to save and restore registers
13739 at the prologue and epilogue of a function. The external functions
13740 are slower, but use less code space if more than one function saves
13741 the same number of registers. The @option{-mprolog-function} option
13742 is on by default if you optimize.
13746 Try to make the code as small as possible. At present, this just turns
13747 on the @option{-mep} and @option{-mprolog-function} options.
13749 @item -mtda=@var{n}
13751 Put static or global variables whose size is @var{n} bytes or less into
13752 the tiny data area that register @code{ep} points to. The tiny data
13753 area can hold up to 256 bytes in total (128 bytes for byte references).
13755 @item -msda=@var{n}
13757 Put static or global variables whose size is @var{n} bytes or less into
13758 the small data area that register @code{gp} points to. The small data
13759 area can hold up to 64 kilobytes.
13761 @item -mzda=@var{n}
13763 Put static or global variables whose size is @var{n} bytes or less into
13764 the first 32 kilobytes of memory.
13768 Specify that the target processor is the V850.
13771 @opindex mbig-switch
13772 Generate code suitable for big switch tables. Use this option only if
13773 the assembler/linker complain about out of range branches within a switch
13778 This option will cause r2 and r5 to be used in the code generated by
13779 the compiler. This setting is the default.
13781 @item -mno-app-regs
13782 @opindex mno-app-regs
13783 This option will cause r2 and r5 to be treated as fixed registers.
13787 Specify that the target processor is the V850E1. The preprocessor
13788 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
13789 this option is used.
13793 Specify that the target processor is the V850E@. The preprocessor
13794 constant @samp{__v850e__} will be defined if this option is used.
13796 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
13797 are defined then a default target processor will be chosen and the
13798 relevant @samp{__v850*__} preprocessor constant will be defined.
13800 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
13801 defined, regardless of which processor variant is the target.
13803 @item -mdisable-callt
13804 @opindex mdisable-callt
13805 This option will suppress generation of the CALLT instruction for the
13806 v850e and v850e1 flavors of the v850 architecture. The default is
13807 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
13812 @subsection VAX Options
13813 @cindex VAX options
13815 These @samp{-m} options are defined for the VAX:
13820 Do not output certain jump instructions (@code{aobleq} and so on)
13821 that the Unix assembler for the VAX cannot handle across long
13826 Do output those jump instructions, on the assumption that you
13827 will assemble with the GNU assembler.
13831 Output code for g-format floating point numbers instead of d-format.
13834 @node VxWorks Options
13835 @subsection VxWorks Options
13836 @cindex VxWorks Options
13838 The options in this section are defined for all VxWorks targets.
13839 Options specific to the target hardware are listed with the other
13840 options for that target.
13845 GCC can generate code for both VxWorks kernels and real time processes
13846 (RTPs). This option switches from the former to the latter. It also
13847 defines the preprocessor macro @code{__RTP__}.
13850 @opindex non-static
13851 Link an RTP executable against shared libraries rather than static
13852 libraries. The options @option{-static} and @option{-shared} can
13853 also be used for RTPs (@pxref{Link Options}); @option{-static}
13860 These options are passed down to the linker. They are defined for
13861 compatibility with Diab.
13864 @opindex Xbind-lazy
13865 Enable lazy binding of function calls. This option is equivalent to
13866 @option{-Wl,-z,now} and is defined for compatibility with Diab.
13870 Disable lazy binding of function calls. This option is the default and
13871 is defined for compatibility with Diab.
13874 @node x86-64 Options
13875 @subsection x86-64 Options
13876 @cindex x86-64 options
13878 These are listed under @xref{i386 and x86-64 Options}.
13880 @node Xstormy16 Options
13881 @subsection Xstormy16 Options
13882 @cindex Xstormy16 Options
13884 These options are defined for Xstormy16:
13889 Choose startup files and linker script suitable for the simulator.
13892 @node Xtensa Options
13893 @subsection Xtensa Options
13894 @cindex Xtensa Options
13896 These options are supported for Xtensa targets:
13900 @itemx -mno-const16
13902 @opindex mno-const16
13903 Enable or disable use of @code{CONST16} instructions for loading
13904 constant values. The @code{CONST16} instruction is currently not a
13905 standard option from Tensilica. When enabled, @code{CONST16}
13906 instructions are always used in place of the standard @code{L32R}
13907 instructions. The use of @code{CONST16} is enabled by default only if
13908 the @code{L32R} instruction is not available.
13911 @itemx -mno-fused-madd
13912 @opindex mfused-madd
13913 @opindex mno-fused-madd
13914 Enable or disable use of fused multiply/add and multiply/subtract
13915 instructions in the floating-point option. This has no effect if the
13916 floating-point option is not also enabled. Disabling fused multiply/add
13917 and multiply/subtract instructions forces the compiler to use separate
13918 instructions for the multiply and add/subtract operations. This may be
13919 desirable in some cases where strict IEEE 754-compliant results are
13920 required: the fused multiply add/subtract instructions do not round the
13921 intermediate result, thereby producing results with @emph{more} bits of
13922 precision than specified by the IEEE standard. Disabling fused multiply
13923 add/subtract instructions also ensures that the program output is not
13924 sensitive to the compiler's ability to combine multiply and add/subtract
13927 @item -mtext-section-literals
13928 @itemx -mno-text-section-literals
13929 @opindex mtext-section-literals
13930 @opindex mno-text-section-literals
13931 Control the treatment of literal pools. The default is
13932 @option{-mno-text-section-literals}, which places literals in a separate
13933 section in the output file. This allows the literal pool to be placed
13934 in a data RAM/ROM, and it also allows the linker to combine literal
13935 pools from separate object files to remove redundant literals and
13936 improve code size. With @option{-mtext-section-literals}, the literals
13937 are interspersed in the text section in order to keep them as close as
13938 possible to their references. This may be necessary for large assembly
13941 @item -mtarget-align
13942 @itemx -mno-target-align
13943 @opindex mtarget-align
13944 @opindex mno-target-align
13945 When this option is enabled, GCC instructs the assembler to
13946 automatically align instructions to reduce branch penalties at the
13947 expense of some code density. The assembler attempts to widen density
13948 instructions to align branch targets and the instructions following call
13949 instructions. If there are not enough preceding safe density
13950 instructions to align a target, no widening will be performed. The
13951 default is @option{-mtarget-align}. These options do not affect the
13952 treatment of auto-aligned instructions like @code{LOOP}, which the
13953 assembler will always align, either by widening density instructions or
13954 by inserting no-op instructions.
13957 @itemx -mno-longcalls
13958 @opindex mlongcalls
13959 @opindex mno-longcalls
13960 When this option is enabled, GCC instructs the assembler to translate
13961 direct calls to indirect calls unless it can determine that the target
13962 of a direct call is in the range allowed by the call instruction. This
13963 translation typically occurs for calls to functions in other source
13964 files. Specifically, the assembler translates a direct @code{CALL}
13965 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13966 The default is @option{-mno-longcalls}. This option should be used in
13967 programs where the call target can potentially be out of range. This
13968 option is implemented in the assembler, not the compiler, so the
13969 assembly code generated by GCC will still show direct call
13970 instructions---look at the disassembled object code to see the actual
13971 instructions. Note that the assembler will use an indirect call for
13972 every cross-file call, not just those that really will be out of range.
13975 @node zSeries Options
13976 @subsection zSeries Options
13977 @cindex zSeries options
13979 These are listed under @xref{S/390 and zSeries Options}.
13981 @node Code Gen Options
13982 @section Options for Code Generation Conventions
13983 @cindex code generation conventions
13984 @cindex options, code generation
13985 @cindex run-time options
13987 These machine-independent options control the interface conventions
13988 used in code generation.
13990 Most of them have both positive and negative forms; the negative form
13991 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13992 one of the forms is listed---the one which is not the default. You
13993 can figure out the other form by either removing @samp{no-} or adding
13997 @item -fbounds-check
13998 @opindex fbounds-check
13999 For front-ends that support it, generate additional code to check that
14000 indices used to access arrays are within the declared range. This is
14001 currently only supported by the Java and Fortran front-ends, where
14002 this option defaults to true and false respectively.
14006 This option generates traps for signed overflow on addition, subtraction,
14007 multiplication operations.
14011 This option instructs the compiler to assume that signed arithmetic
14012 overflow of addition, subtraction and multiplication wraps around
14013 using twos-complement representation. This flag enables some optimizations
14014 and disables others. This option is enabled by default for the Java
14015 front-end, as required by the Java language specification.
14018 @opindex fexceptions
14019 Enable exception handling. Generates extra code needed to propagate
14020 exceptions. For some targets, this implies GCC will generate frame
14021 unwind information for all functions, which can produce significant data
14022 size overhead, although it does not affect execution. If you do not
14023 specify this option, GCC will enable it by default for languages like
14024 C++ which normally require exception handling, and disable it for
14025 languages like C that do not normally require it. However, you may need
14026 to enable this option when compiling C code that needs to interoperate
14027 properly with exception handlers written in C++. You may also wish to
14028 disable this option if you are compiling older C++ programs that don't
14029 use exception handling.
14031 @item -fnon-call-exceptions
14032 @opindex fnon-call-exceptions
14033 Generate code that allows trapping instructions to throw exceptions.
14034 Note that this requires platform-specific runtime support that does
14035 not exist everywhere. Moreover, it only allows @emph{trapping}
14036 instructions to throw exceptions, i.e.@: memory references or floating
14037 point instructions. It does not allow exceptions to be thrown from
14038 arbitrary signal handlers such as @code{SIGALRM}.
14040 @item -funwind-tables
14041 @opindex funwind-tables
14042 Similar to @option{-fexceptions}, except that it will just generate any needed
14043 static data, but will not affect the generated code in any other way.
14044 You will normally not enable this option; instead, a language processor
14045 that needs this handling would enable it on your behalf.
14047 @item -fasynchronous-unwind-tables
14048 @opindex fasynchronous-unwind-tables
14049 Generate unwind table in dwarf2 format, if supported by target machine. The
14050 table is exact at each instruction boundary, so it can be used for stack
14051 unwinding from asynchronous events (such as debugger or garbage collector).
14053 @item -fpcc-struct-return
14054 @opindex fpcc-struct-return
14055 Return ``short'' @code{struct} and @code{union} values in memory like
14056 longer ones, rather than in registers. This convention is less
14057 efficient, but it has the advantage of allowing intercallability between
14058 GCC-compiled files and files compiled with other compilers, particularly
14059 the Portable C Compiler (pcc).
14061 The precise convention for returning structures in memory depends
14062 on the target configuration macros.
14064 Short structures and unions are those whose size and alignment match
14065 that of some integer type.
14067 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14068 switch is not binary compatible with code compiled with the
14069 @option{-freg-struct-return} switch.
14070 Use it to conform to a non-default application binary interface.
14072 @item -freg-struct-return
14073 @opindex freg-struct-return
14074 Return @code{struct} and @code{union} values in registers when possible.
14075 This is more efficient for small structures than
14076 @option{-fpcc-struct-return}.
14078 If you specify neither @option{-fpcc-struct-return} nor
14079 @option{-freg-struct-return}, GCC defaults to whichever convention is
14080 standard for the target. If there is no standard convention, GCC
14081 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14082 the principal compiler. In those cases, we can choose the standard, and
14083 we chose the more efficient register return alternative.
14085 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14086 switch is not binary compatible with code compiled with the
14087 @option{-fpcc-struct-return} switch.
14088 Use it to conform to a non-default application binary interface.
14090 @item -fshort-enums
14091 @opindex fshort-enums
14092 Allocate to an @code{enum} type only as many bytes as it needs for the
14093 declared range of possible values. Specifically, the @code{enum} type
14094 will be equivalent to the smallest integer type which has enough room.
14096 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14097 code that is not binary compatible with code generated without that switch.
14098 Use it to conform to a non-default application binary interface.
14100 @item -fshort-double
14101 @opindex fshort-double
14102 Use the same size for @code{double} as for @code{float}.
14104 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14105 code that is not binary compatible with code generated without that switch.
14106 Use it to conform to a non-default application binary interface.
14108 @item -fshort-wchar
14109 @opindex fshort-wchar
14110 Override the underlying type for @samp{wchar_t} to be @samp{short
14111 unsigned int} instead of the default for the target. This option is
14112 useful for building programs to run under WINE@.
14114 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14115 code that is not binary compatible with code generated without that switch.
14116 Use it to conform to a non-default application binary interface.
14119 @opindex fno-common
14120 In C, allocate even uninitialized global variables in the data section of the
14121 object file, rather than generating them as common blocks. This has the
14122 effect that if the same variable is declared (without @code{extern}) in
14123 two different compilations, you will get an error when you link them.
14124 The only reason this might be useful is if you wish to verify that the
14125 program will work on other systems which always work this way.
14129 Ignore the @samp{#ident} directive.
14131 @item -finhibit-size-directive
14132 @opindex finhibit-size-directive
14133 Don't output a @code{.size} assembler directive, or anything else that
14134 would cause trouble if the function is split in the middle, and the
14135 two halves are placed at locations far apart in memory. This option is
14136 used when compiling @file{crtstuff.c}; you should not need to use it
14139 @item -fverbose-asm
14140 @opindex fverbose-asm
14141 Put extra commentary information in the generated assembly code to
14142 make it more readable. This option is generally only of use to those
14143 who actually need to read the generated assembly code (perhaps while
14144 debugging the compiler itself).
14146 @option{-fno-verbose-asm}, the default, causes the
14147 extra information to be omitted and is useful when comparing two assembler
14150 @item -frecord-gcc-switches
14151 @opindex frecord-gcc-switches
14152 This switch causes the command line that was used to invoke the
14153 compiler to be recorded into the object file that is being created.
14154 This switch is only implemented on some targets and the exact format
14155 of the recording is target and binary file format dependent, but it
14156 usually takes the form of a section containing ASCII text. This
14157 switch is related to the @option{-fverbose-asm} switch, but that
14158 switch only records information in the assembler output file as
14159 comments, so it never reaches the object file.
14163 @cindex global offset table
14165 Generate position-independent code (PIC) suitable for use in a shared
14166 library, if supported for the target machine. Such code accesses all
14167 constant addresses through a global offset table (GOT)@. The dynamic
14168 loader resolves the GOT entries when the program starts (the dynamic
14169 loader is not part of GCC; it is part of the operating system). If
14170 the GOT size for the linked executable exceeds a machine-specific
14171 maximum size, you get an error message from the linker indicating that
14172 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14173 instead. (These maximums are 8k on the SPARC and 32k
14174 on the m68k and RS/6000. The 386 has no such limit.)
14176 Position-independent code requires special support, and therefore works
14177 only on certain machines. For the 386, GCC supports PIC for System V
14178 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14179 position-independent.
14181 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14186 If supported for the target machine, emit position-independent code,
14187 suitable for dynamic linking and avoiding any limit on the size of the
14188 global offset table. This option makes a difference on the m68k,
14189 PowerPC and SPARC@.
14191 Position-independent code requires special support, and therefore works
14192 only on certain machines.
14194 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14201 These options are similar to @option{-fpic} and @option{-fPIC}, but
14202 generated position independent code can be only linked into executables.
14203 Usually these options are used when @option{-pie} GCC option will be
14204 used during linking.
14206 @item -fno-jump-tables
14207 @opindex fno-jump-tables
14208 Do not use jump tables for switch statements even where it would be
14209 more efficient than other code generation strategies. This option is
14210 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14211 building code which forms part of a dynamic linker and cannot
14212 reference the address of a jump table. On some targets, jump tables
14213 do not require a GOT and this option is not needed.
14215 @item -ffixed-@var{reg}
14217 Treat the register named @var{reg} as a fixed register; generated code
14218 should never refer to it (except perhaps as a stack pointer, frame
14219 pointer or in some other fixed role).
14221 @var{reg} must be the name of a register. The register names accepted
14222 are machine-specific and are defined in the @code{REGISTER_NAMES}
14223 macro in the machine description macro file.
14225 This flag does not have a negative form, because it specifies a
14228 @item -fcall-used-@var{reg}
14229 @opindex fcall-used
14230 Treat the register named @var{reg} as an allocable register that is
14231 clobbered by function calls. It may be allocated for temporaries or
14232 variables that do not live across a call. Functions compiled this way
14233 will not save and restore the register @var{reg}.
14235 It is an error to used this flag with the frame pointer or stack pointer.
14236 Use of this flag for other registers that have fixed pervasive roles in
14237 the machine's execution model will produce disastrous results.
14239 This flag does not have a negative form, because it specifies a
14242 @item -fcall-saved-@var{reg}
14243 @opindex fcall-saved
14244 Treat the register named @var{reg} as an allocable register saved by
14245 functions. It may be allocated even for temporaries or variables that
14246 live across a call. Functions compiled this way will save and restore
14247 the register @var{reg} if they use it.
14249 It is an error to used this flag with the frame pointer or stack pointer.
14250 Use of this flag for other registers that have fixed pervasive roles in
14251 the machine's execution model will produce disastrous results.
14253 A different sort of disaster will result from the use of this flag for
14254 a register in which function values may be returned.
14256 This flag does not have a negative form, because it specifies a
14259 @item -fpack-struct[=@var{n}]
14260 @opindex fpack-struct
14261 Without a value specified, pack all structure members together without
14262 holes. When a value is specified (which must be a small power of two), pack
14263 structure members according to this value, representing the maximum
14264 alignment (that is, objects with default alignment requirements larger than
14265 this will be output potentially unaligned at the next fitting location.
14267 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14268 code that is not binary compatible with code generated without that switch.
14269 Additionally, it makes the code suboptimal.
14270 Use it to conform to a non-default application binary interface.
14272 @item -finstrument-functions
14273 @opindex finstrument-functions
14274 Generate instrumentation calls for entry and exit to functions. Just
14275 after function entry and just before function exit, the following
14276 profiling functions will be called with the address of the current
14277 function and its call site. (On some platforms,
14278 @code{__builtin_return_address} does not work beyond the current
14279 function, so the call site information may not be available to the
14280 profiling functions otherwise.)
14283 void __cyg_profile_func_enter (void *this_fn,
14285 void __cyg_profile_func_exit (void *this_fn,
14289 The first argument is the address of the start of the current function,
14290 which may be looked up exactly in the symbol table.
14292 This instrumentation is also done for functions expanded inline in other
14293 functions. The profiling calls will indicate where, conceptually, the
14294 inline function is entered and exited. This means that addressable
14295 versions of such functions must be available. If all your uses of a
14296 function are expanded inline, this may mean an additional expansion of
14297 code size. If you use @samp{extern inline} in your C code, an
14298 addressable version of such functions must be provided. (This is
14299 normally the case anyways, but if you get lucky and the optimizer always
14300 expands the functions inline, you might have gotten away without
14301 providing static copies.)
14303 A function may be given the attribute @code{no_instrument_function}, in
14304 which case this instrumentation will not be done. This can be used, for
14305 example, for the profiling functions listed above, high-priority
14306 interrupt routines, and any functions from which the profiling functions
14307 cannot safely be called (perhaps signal handlers, if the profiling
14308 routines generate output or allocate memory).
14310 @item -fstack-check
14311 @opindex fstack-check
14312 Generate code to verify that you do not go beyond the boundary of the
14313 stack. You should specify this flag if you are running in an
14314 environment with multiple threads, but only rarely need to specify it in
14315 a single-threaded environment since stack overflow is automatically
14316 detected on nearly all systems if there is only one stack.
14318 Note that this switch does not actually cause checking to be done; the
14319 operating system must do that. The switch causes generation of code
14320 to ensure that the operating system sees the stack being extended.
14322 @item -fstack-limit-register=@var{reg}
14323 @itemx -fstack-limit-symbol=@var{sym}
14324 @itemx -fno-stack-limit
14325 @opindex fstack-limit-register
14326 @opindex fstack-limit-symbol
14327 @opindex fno-stack-limit
14328 Generate code to ensure that the stack does not grow beyond a certain value,
14329 either the value of a register or the address of a symbol. If the stack
14330 would grow beyond the value, a signal is raised. For most targets,
14331 the signal is raised before the stack overruns the boundary, so
14332 it is possible to catch the signal without taking special precautions.
14334 For instance, if the stack starts at absolute address @samp{0x80000000}
14335 and grows downwards, you can use the flags
14336 @option{-fstack-limit-symbol=__stack_limit} and
14337 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
14338 of 128KB@. Note that this may only work with the GNU linker.
14340 @cindex aliasing of parameters
14341 @cindex parameters, aliased
14342 @item -fargument-alias
14343 @itemx -fargument-noalias
14344 @itemx -fargument-noalias-global
14345 @itemx -fargument-noalias-anything
14346 @opindex fargument-alias
14347 @opindex fargument-noalias
14348 @opindex fargument-noalias-global
14349 @opindex fargument-noalias-anything
14350 Specify the possible relationships among parameters and between
14351 parameters and global data.
14353 @option{-fargument-alias} specifies that arguments (parameters) may
14354 alias each other and may alias global storage.@*
14355 @option{-fargument-noalias} specifies that arguments do not alias
14356 each other, but may alias global storage.@*
14357 @option{-fargument-noalias-global} specifies that arguments do not
14358 alias each other and do not alias global storage.
14359 @option{-fargument-noalias-anything} specifies that arguments do not
14360 alias any other storage.
14362 Each language will automatically use whatever option is required by
14363 the language standard. You should not need to use these options yourself.
14365 @item -fleading-underscore
14366 @opindex fleading-underscore
14367 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
14368 change the way C symbols are represented in the object file. One use
14369 is to help link with legacy assembly code.
14371 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
14372 generate code that is not binary compatible with code generated without that
14373 switch. Use it to conform to a non-default application binary interface.
14374 Not all targets provide complete support for this switch.
14376 @item -ftls-model=@var{model}
14377 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
14378 The @var{model} argument should be one of @code{global-dynamic},
14379 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
14381 The default without @option{-fpic} is @code{initial-exec}; with
14382 @option{-fpic} the default is @code{global-dynamic}.
14384 @item -fvisibility=@var{default|internal|hidden|protected}
14385 @opindex fvisibility
14386 Set the default ELF image symbol visibility to the specified option---all
14387 symbols will be marked with this unless overridden within the code.
14388 Using this feature can very substantially improve linking and
14389 load times of shared object libraries, produce more optimized
14390 code, provide near-perfect API export and prevent symbol clashes.
14391 It is @strong{strongly} recommended that you use this in any shared objects
14394 Despite the nomenclature, @code{default} always means public ie;
14395 available to be linked against from outside the shared object.
14396 @code{protected} and @code{internal} are pretty useless in real-world
14397 usage so the only other commonly used option will be @code{hidden}.
14398 The default if @option{-fvisibility} isn't specified is
14399 @code{default}, i.e., make every
14400 symbol public---this causes the same behavior as previous versions of
14403 A good explanation of the benefits offered by ensuring ELF
14404 symbols have the correct visibility is given by ``How To Write
14405 Shared Libraries'' by Ulrich Drepper (which can be found at
14406 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
14407 solution made possible by this option to marking things hidden when
14408 the default is public is to make the default hidden and mark things
14409 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
14410 and @code{__attribute__ ((visibility("default")))} instead of
14411 @code{__declspec(dllexport)} you get almost identical semantics with
14412 identical syntax. This is a great boon to those working with
14413 cross-platform projects.
14415 For those adding visibility support to existing code, you may find
14416 @samp{#pragma GCC visibility} of use. This works by you enclosing
14417 the declarations you wish to set visibility for with (for example)
14418 @samp{#pragma GCC visibility push(hidden)} and
14419 @samp{#pragma GCC visibility pop}.
14420 Bear in mind that symbol visibility should be viewed @strong{as
14421 part of the API interface contract} and thus all new code should
14422 always specify visibility when it is not the default ie; declarations
14423 only for use within the local DSO should @strong{always} be marked explicitly
14424 as hidden as so to avoid PLT indirection overheads---making this
14425 abundantly clear also aids readability and self-documentation of the code.
14426 Note that due to ISO C++ specification requirements, operator new and
14427 operator delete must always be of default visibility.
14429 Be aware that headers from outside your project, in particular system
14430 headers and headers from any other library you use, may not be
14431 expecting to be compiled with visibility other than the default. You
14432 may need to explicitly say @samp{#pragma GCC visibility push(default)}
14433 before including any such headers.
14435 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
14436 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
14437 no modifications. However, this means that calls to @samp{extern}
14438 functions with no explicit visibility will use the PLT, so it is more
14439 effective to use @samp{__attribute ((visibility))} and/or
14440 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
14441 declarations should be treated as hidden.
14443 Note that @samp{-fvisibility} does affect C++ vague linkage
14444 entities. This means that, for instance, an exception class that will
14445 be thrown between DSOs must be explicitly marked with default
14446 visibility so that the @samp{type_info} nodes will be unified between
14449 An overview of these techniques, their benefits and how to use them
14450 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
14456 @node Environment Variables
14457 @section Environment Variables Affecting GCC
14458 @cindex environment variables
14460 @c man begin ENVIRONMENT
14461 This section describes several environment variables that affect how GCC
14462 operates. Some of them work by specifying directories or prefixes to use
14463 when searching for various kinds of files. Some are used to specify other
14464 aspects of the compilation environment.
14466 Note that you can also specify places to search using options such as
14467 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
14468 take precedence over places specified using environment variables, which
14469 in turn take precedence over those specified by the configuration of GCC@.
14470 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
14471 GNU Compiler Collection (GCC) Internals}.
14476 @c @itemx LC_COLLATE
14478 @c @itemx LC_MONETARY
14479 @c @itemx LC_NUMERIC
14484 @c @findex LC_COLLATE
14485 @findex LC_MESSAGES
14486 @c @findex LC_MONETARY
14487 @c @findex LC_NUMERIC
14491 These environment variables control the way that GCC uses
14492 localization information that allow GCC to work with different
14493 national conventions. GCC inspects the locale categories
14494 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
14495 so. These locale categories can be set to any value supported by your
14496 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
14497 Kingdom encoded in UTF-8.
14499 The @env{LC_CTYPE} environment variable specifies character
14500 classification. GCC uses it to determine the character boundaries in
14501 a string; this is needed for some multibyte encodings that contain quote
14502 and escape characters that would otherwise be interpreted as a string
14505 The @env{LC_MESSAGES} environment variable specifies the language to
14506 use in diagnostic messages.
14508 If the @env{LC_ALL} environment variable is set, it overrides the value
14509 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
14510 and @env{LC_MESSAGES} default to the value of the @env{LANG}
14511 environment variable. If none of these variables are set, GCC
14512 defaults to traditional C English behavior.
14516 If @env{TMPDIR} is set, it specifies the directory to use for temporary
14517 files. GCC uses temporary files to hold the output of one stage of
14518 compilation which is to be used as input to the next stage: for example,
14519 the output of the preprocessor, which is the input to the compiler
14522 @item GCC_EXEC_PREFIX
14523 @findex GCC_EXEC_PREFIX
14524 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
14525 names of the subprograms executed by the compiler. No slash is added
14526 when this prefix is combined with the name of a subprogram, but you can
14527 specify a prefix that ends with a slash if you wish.
14529 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
14530 an appropriate prefix to use based on the pathname it was invoked with.
14532 If GCC cannot find the subprogram using the specified prefix, it
14533 tries looking in the usual places for the subprogram.
14535 The default value of @env{GCC_EXEC_PREFIX} is
14536 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
14537 the installed compiler. In many cases @var{prefix} is the value
14538 of @code{prefix} when you ran the @file{configure} script.
14540 Other prefixes specified with @option{-B} take precedence over this prefix.
14542 This prefix is also used for finding files such as @file{crt0.o} that are
14545 In addition, the prefix is used in an unusual way in finding the
14546 directories to search for header files. For each of the standard
14547 directories whose name normally begins with @samp{/usr/local/lib/gcc}
14548 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
14549 replacing that beginning with the specified prefix to produce an
14550 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
14551 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
14552 These alternate directories are searched first; the standard directories
14553 come next. If a standard directory begins with the configured
14554 @var{prefix} then the value of @var{prefix} is replaced by
14555 @env{GCC_EXEC_PREFIX} when looking for header files.
14557 @item COMPILER_PATH
14558 @findex COMPILER_PATH
14559 The value of @env{COMPILER_PATH} is a colon-separated list of
14560 directories, much like @env{PATH}. GCC tries the directories thus
14561 specified when searching for subprograms, if it can't find the
14562 subprograms using @env{GCC_EXEC_PREFIX}.
14565 @findex LIBRARY_PATH
14566 The value of @env{LIBRARY_PATH} is a colon-separated list of
14567 directories, much like @env{PATH}. When configured as a native compiler,
14568 GCC tries the directories thus specified when searching for special
14569 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
14570 using GCC also uses these directories when searching for ordinary
14571 libraries for the @option{-l} option (but directories specified with
14572 @option{-L} come first).
14576 @cindex locale definition
14577 This variable is used to pass locale information to the compiler. One way in
14578 which this information is used is to determine the character set to be used
14579 when character literals, string literals and comments are parsed in C and C++.
14580 When the compiler is configured to allow multibyte characters,
14581 the following values for @env{LANG} are recognized:
14585 Recognize JIS characters.
14587 Recognize SJIS characters.
14589 Recognize EUCJP characters.
14592 If @env{LANG} is not defined, or if it has some other value, then the
14593 compiler will use mblen and mbtowc as defined by the default locale to
14594 recognize and translate multibyte characters.
14598 Some additional environments variables affect the behavior of the
14601 @include cppenv.texi
14605 @node Precompiled Headers
14606 @section Using Precompiled Headers
14607 @cindex precompiled headers
14608 @cindex speed of compilation
14610 Often large projects have many header files that are included in every
14611 source file. The time the compiler takes to process these header files
14612 over and over again can account for nearly all of the time required to
14613 build the project. To make builds faster, GCC allows users to
14614 `precompile' a header file; then, if builds can use the precompiled
14615 header file they will be much faster.
14617 To create a precompiled header file, simply compile it as you would any
14618 other file, if necessary using the @option{-x} option to make the driver
14619 treat it as a C or C++ header file. You will probably want to use a
14620 tool like @command{make} to keep the precompiled header up-to-date when
14621 the headers it contains change.
14623 A precompiled header file will be searched for when @code{#include} is
14624 seen in the compilation. As it searches for the included file
14625 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
14626 compiler looks for a precompiled header in each directory just before it
14627 looks for the include file in that directory. The name searched for is
14628 the name specified in the @code{#include} with @samp{.gch} appended. If
14629 the precompiled header file can't be used, it is ignored.
14631 For instance, if you have @code{#include "all.h"}, and you have
14632 @file{all.h.gch} in the same directory as @file{all.h}, then the
14633 precompiled header file will be used if possible, and the original
14634 header will be used otherwise.
14636 Alternatively, you might decide to put the precompiled header file in a
14637 directory and use @option{-I} to ensure that directory is searched
14638 before (or instead of) the directory containing the original header.
14639 Then, if you want to check that the precompiled header file is always
14640 used, you can put a file of the same name as the original header in this
14641 directory containing an @code{#error} command.
14643 This also works with @option{-include}. So yet another way to use
14644 precompiled headers, good for projects not designed with precompiled
14645 header files in mind, is to simply take most of the header files used by
14646 a project, include them from another header file, precompile that header
14647 file, and @option{-include} the precompiled header. If the header files
14648 have guards against multiple inclusion, they will be skipped because
14649 they've already been included (in the precompiled header).
14651 If you need to precompile the same header file for different
14652 languages, targets, or compiler options, you can instead make a
14653 @emph{directory} named like @file{all.h.gch}, and put each precompiled
14654 header in the directory, perhaps using @option{-o}. It doesn't matter
14655 what you call the files in the directory, every precompiled header in
14656 the directory will be considered. The first precompiled header
14657 encountered in the directory that is valid for this compilation will
14658 be used; they're searched in no particular order.
14660 There are many other possibilities, limited only by your imagination,
14661 good sense, and the constraints of your build system.
14663 A precompiled header file can be used only when these conditions apply:
14667 Only one precompiled header can be used in a particular compilation.
14670 A precompiled header can't be used once the first C token is seen. You
14671 can have preprocessor directives before a precompiled header; you can
14672 even include a precompiled header from inside another header, so long as
14673 there are no C tokens before the @code{#include}.
14676 The precompiled header file must be produced for the same language as
14677 the current compilation. You can't use a C precompiled header for a C++
14681 The precompiled header file must have been produced by the same compiler
14682 binary as the current compilation is using.
14685 Any macros defined before the precompiled header is included must
14686 either be defined in the same way as when the precompiled header was
14687 generated, or must not affect the precompiled header, which usually
14688 means that they don't appear in the precompiled header at all.
14690 The @option{-D} option is one way to define a macro before a
14691 precompiled header is included; using a @code{#define} can also do it.
14692 There are also some options that define macros implicitly, like
14693 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
14696 @item If debugging information is output when using the precompiled
14697 header, using @option{-g} or similar, the same kind of debugging information
14698 must have been output when building the precompiled header. However,
14699 a precompiled header built using @option{-g} can be used in a compilation
14700 when no debugging information is being output.
14702 @item The same @option{-m} options must generally be used when building
14703 and using the precompiled header. @xref{Submodel Options},
14704 for any cases where this rule is relaxed.
14706 @item Each of the following options must be the same when building and using
14707 the precompiled header:
14709 @gccoptlist{-fexceptions -funit-at-a-time}
14712 Some other command-line options starting with @option{-f},
14713 @option{-p}, or @option{-O} must be defined in the same way as when
14714 the precompiled header was generated. At present, it's not clear
14715 which options are safe to change and which are not; the safest choice
14716 is to use exactly the same options when generating and using the
14717 precompiled header. The following are known to be safe:
14719 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
14720 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
14721 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
14726 For all of these except the last, the compiler will automatically
14727 ignore the precompiled header if the conditions aren't met. If you
14728 find an option combination that doesn't work and doesn't cause the
14729 precompiled header to be ignored, please consider filing a bug report,
14732 If you do use differing options when generating and using the
14733 precompiled header, the actual behavior will be a mixture of the
14734 behavior for the options. For instance, if you use @option{-g} to
14735 generate the precompiled header but not when using it, you may or may
14736 not get debugging information for routines in the precompiled header.
14738 @node Running Protoize
14739 @section Running Protoize
14741 The program @code{protoize} is an optional part of GCC@. You can use
14742 it to add prototypes to a program, thus converting the program to ISO
14743 C in one respect. The companion program @code{unprotoize} does the
14744 reverse: it removes argument types from any prototypes that are found.
14746 When you run these programs, you must specify a set of source files as
14747 command line arguments. The conversion programs start out by compiling
14748 these files to see what functions they define. The information gathered
14749 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
14751 After scanning comes actual conversion. The specified files are all
14752 eligible to be converted; any files they include (whether sources or
14753 just headers) are eligible as well.
14755 But not all the eligible files are converted. By default,
14756 @code{protoize} and @code{unprotoize} convert only source and header
14757 files in the current directory. You can specify additional directories
14758 whose files should be converted with the @option{-d @var{directory}}
14759 option. You can also specify particular files to exclude with the
14760 @option{-x @var{file}} option. A file is converted if it is eligible, its
14761 directory name matches one of the specified directory names, and its
14762 name within the directory has not been excluded.
14764 Basic conversion with @code{protoize} consists of rewriting most
14765 function definitions and function declarations to specify the types of
14766 the arguments. The only ones not rewritten are those for varargs
14769 @code{protoize} optionally inserts prototype declarations at the
14770 beginning of the source file, to make them available for any calls that
14771 precede the function's definition. Or it can insert prototype
14772 declarations with block scope in the blocks where undeclared functions
14775 Basic conversion with @code{unprotoize} consists of rewriting most
14776 function declarations to remove any argument types, and rewriting
14777 function definitions to the old-style pre-ISO form.
14779 Both conversion programs print a warning for any function declaration or
14780 definition that they can't convert. You can suppress these warnings
14783 The output from @code{protoize} or @code{unprotoize} replaces the
14784 original source file. The original file is renamed to a name ending
14785 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
14786 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
14787 for DOS) file already exists, then the source file is simply discarded.
14789 @code{protoize} and @code{unprotoize} both depend on GCC itself to
14790 scan the program and collect information about the functions it uses.
14791 So neither of these programs will work until GCC is installed.
14793 Here is a table of the options you can use with @code{protoize} and
14794 @code{unprotoize}. Each option works with both programs unless
14798 @item -B @var{directory}
14799 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
14800 usual directory (normally @file{/usr/local/lib}). This file contains
14801 prototype information about standard system functions. This option
14802 applies only to @code{protoize}.
14804 @item -c @var{compilation-options}
14805 Use @var{compilation-options} as the options when running @command{gcc} to
14806 produce the @samp{.X} files. The special option @option{-aux-info} is
14807 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
14809 Note that the compilation options must be given as a single argument to
14810 @code{protoize} or @code{unprotoize}. If you want to specify several
14811 @command{gcc} options, you must quote the entire set of compilation options
14812 to make them a single word in the shell.
14814 There are certain @command{gcc} arguments that you cannot use, because they
14815 would produce the wrong kind of output. These include @option{-g},
14816 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
14817 the @var{compilation-options}, they are ignored.
14820 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
14821 systems) instead of @samp{.c}. This is convenient if you are converting
14822 a C program to C++. This option applies only to @code{protoize}.
14825 Add explicit global declarations. This means inserting explicit
14826 declarations at the beginning of each source file for each function
14827 that is called in the file and was not declared. These declarations
14828 precede the first function definition that contains a call to an
14829 undeclared function. This option applies only to @code{protoize}.
14831 @item -i @var{string}
14832 Indent old-style parameter declarations with the string @var{string}.
14833 This option applies only to @code{protoize}.
14835 @code{unprotoize} converts prototyped function definitions to old-style
14836 function definitions, where the arguments are declared between the
14837 argument list and the initial @samp{@{}. By default, @code{unprotoize}
14838 uses five spaces as the indentation. If you want to indent with just
14839 one space instead, use @option{-i " "}.
14842 Keep the @samp{.X} files. Normally, they are deleted after conversion
14846 Add explicit local declarations. @code{protoize} with @option{-l} inserts
14847 a prototype declaration for each function in each block which calls the
14848 function without any declaration. This option applies only to
14852 Make no real changes. This mode just prints information about the conversions
14853 that would have been done without @option{-n}.
14856 Make no @samp{.save} files. The original files are simply deleted.
14857 Use this option with caution.
14859 @item -p @var{program}
14860 Use the program @var{program} as the compiler. Normally, the name
14861 @file{gcc} is used.
14864 Work quietly. Most warnings are suppressed.
14867 Print the version number, just like @option{-v} for @command{gcc}.
14870 If you need special compiler options to compile one of your program's
14871 source files, then you should generate that file's @samp{.X} file
14872 specially, by running @command{gcc} on that source file with the
14873 appropriate options and the option @option{-aux-info}. Then run
14874 @code{protoize} on the entire set of files. @code{protoize} will use
14875 the existing @samp{.X} file because it is newer than the source file.
14879 gcc -Dfoo=bar file1.c -aux-info file1.X
14884 You need to include the special files along with the rest in the
14885 @code{protoize} command, even though their @samp{.X} files already
14886 exist, because otherwise they won't get converted.
14888 @xref{Protoize Caveats}, for more information on how to use
14889 @code{protoize} successfully.