1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
15 2007 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
151 * Running Protoize:: Automatically adding or removing function prototypes.
157 @section Option Summary
159 Here is a summary of all the options, grouped by type. Explanations are
160 in the following sections.
163 @item Overall Options
164 @xref{Overall Options,,Options Controlling the Kind of Output}.
165 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{]} --target-help @gol
167 --version @@@var{file}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
193 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
194 -fno-default-inline -fvisibility-inlines-hidden @gol
195 -fvisibility-ms-compat @gol
196 -Wabi -Wctor-dtor-privacy @gol
197 -Wnon-virtual-dtor -Wreorder @gol
198 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
199 -Wno-non-template-friend -Wold-style-cast @gol
200 -Woverloaded-virtual -Wno-pmf-conversions @gol
203 @item Objective-C and Objective-C++ Language Options
204 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
205 Objective-C and Objective-C++ Dialects}.
206 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
207 -fgnu-runtime -fnext-runtime @gol
208 -fno-nil-receivers @gol
209 -fobjc-call-cxx-cdtors @gol
210 -fobjc-direct-dispatch @gol
211 -fobjc-exceptions @gol
213 -freplace-objc-classes @gol
216 -Wassign-intercept @gol
217 -Wno-protocol -Wselector @gol
218 -Wstrict-selector-match @gol
219 -Wundeclared-selector}
221 @item Language Independent Options
222 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
223 @gccoptlist{-fmessage-length=@var{n} @gol
224 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
225 -fdiagnostics-show-option}
227 @item Warning Options
228 @xref{Warning Options,,Options to Request or Suppress Warnings}.
229 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
230 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
231 -Wno-attributes -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
232 -Wchar-subscripts -Wclobbered -Wcomment @gol
233 -Wconversion -Wcoverage-mismatch -Wno-deprecated-declarations @gol
234 -Wdisabled-optimization -Wno-div-by-zero @gol
235 -Wempty-body -Wno-endif-labels @gol
236 -Werror -Werror=* @gol
237 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
238 -Wno-format-extra-args -Wformat-nonliteral @gol
239 -Wformat-security -Wformat-y2k @gol
240 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
241 -Wimport -Wno-import -Winit-self -Winline @gol
242 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
243 -Winvalid-pch -Wlarger-than-@var{len} -Wunsafe-loop-optimizations @gol
244 -Wlogical-op -Wlong-long @gol
245 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
246 -Wmissing-format-attribute -Wmissing-include-dirs @gol
247 -Wmissing-noreturn @gol
248 -Wno-multichar -Wnonnull -Wno-overflow @gol
249 -Woverlength-strings -Wpacked -Wpadded @gol
250 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
251 -Wredundant-decls @gol
252 -Wreturn-type -Wsequence-point -Wshadow @gol
253 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
254 -Wstrict-aliasing -Wstrict-aliasing=n @gol
255 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
256 -Wswitch -Wswitch-default -Wswitch-enum @gol
257 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
258 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
259 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
260 -Wunused-value -Wunused-variable @gol
261 -Wvariadic-macros -Wvla @gol
262 -Wvolatile-register-var -Wwrite-strings}
264 @item C-only Warning Options
265 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
266 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
267 -Wold-style-declaration -Wold-style-definition @gol
268 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
269 -Wdeclaration-after-statement -Wpointer-sign}
271 @item Debugging Options
272 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
273 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
274 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
275 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
276 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
277 -fdump-ipa-all -fdump-ipa-cgraph @gol
279 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
284 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
289 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-nrv -fdump-tree-vect @gol
293 -fdump-tree-sink @gol
294 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-salias @gol
296 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
298 -ftree-vectorizer-verbose=@var{n} @gol
299 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
300 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
301 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
302 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
303 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
304 -ftest-coverage -ftime-report -fvar-tracking @gol
305 -g -g@var{level} -gcoff -gdwarf-2 @gol
306 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
307 -fdebug-prefix-map=@var{old}=@var{new} @gol
308 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
309 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
310 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
311 -print-multi-directory -print-multi-lib @gol
312 -print-prog-name=@var{program} -print-search-dirs -Q @gol
313 -print-sysroot-headers-suffix @gol
316 @item Optimization Options
317 @xref{Optimize Options,,Options that Control Optimization}.
318 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
319 -falign-labels=@var{n} -falign-loops=@var{n} @gol
320 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
321 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
322 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
323 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
324 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
325 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
326 -fexpensive-optimizations -ffast-math -ffloat-store @gol
327 -fforce-addr -fforward-propagate -ffunction-sections @gol
328 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
329 -fcrossjumping -fif-conversion -fif-conversion2 @gol
330 -finline-functions -finline-functions-called-once @gol
331 -finline-small-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
332 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
333 -fmodulo-sched -fmodulo-sched-allow-regmoves -fno-branch-count-reg @gol
334 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
335 -fno-function-cse -fno-guess-branch-probability @gol
336 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
337 -funsafe-math-optimizations -funsafe-loop-optimizations @gol
338 -ffinite-math-only -fno-signed-zeros @gol
339 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
340 -fomit-frame-pointer -foptimize-register-move @gol
341 -foptimize-sibling-calls -fpredictive-commoning -fprefetch-loop-arrays @gol
342 -fprofile-generate -fprofile-use @gol
343 -fregmove -frename-registers @gol
344 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
345 -frerun-cse-after-loop @gol
346 -frounding-math -frtl-abstract-sequences @gol
347 -fschedule-insns -fschedule-insns2 @gol
348 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
349 -fsched-spec-load-dangerous @gol
350 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
351 -fsched2-use-superblocks @gol
352 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
353 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
354 -fno-split-wide-types -fstack-protector -fstack-protector-all @gol
355 -fstrict-aliasing -fstrict-overflow -ftracer -fthread-jumps @gol
356 -funroll-all-loops -funroll-loops -fpeel-loops @gol
357 -fsplit-ivs-in-unroller -funswitch-loops @gol
358 -fvariable-expansion-in-unroller -ftree-reassoc @gol
359 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
360 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
361 -fcheck-data-deps @gol
362 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
363 -ftree-ch -ftree-sra -ftree-ter -ftree-fre -ftree-vectorize @gol
364 -ftree-vect-loop-version -fvect-cost-model -ftree-salias -fipa-pta -fweb @gol
365 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
366 --param @var{name}=@var{value}
367 -O -O0 -O1 -O2 -O3 -Os}
369 @item Preprocessor Options
370 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
371 @gccoptlist{-A@var{question}=@var{answer} @gol
372 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
373 -C -dD -dI -dM -dN @gol
374 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
375 -idirafter @var{dir} @gol
376 -include @var{file} -imacros @var{file} @gol
377 -iprefix @var{file} -iwithprefix @var{dir} @gol
378 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
379 -imultilib @var{dir} -isysroot @var{dir} @gol
380 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
381 -P -fworking-directory -remap @gol
382 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
383 -Xpreprocessor @var{option}}
385 @item Assembler Option
386 @xref{Assembler Options,,Passing Options to the Assembler}.
387 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
390 @xref{Link Options,,Options for Linking}.
391 @gccoptlist{@var{object-file-name} -l@var{library} @gol
392 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
393 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
394 -Wl,@var{option} -Xlinker @var{option} @gol
397 @item Directory Options
398 @xref{Directory Options,,Options for Directory Search}.
399 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
400 -specs=@var{file} -I- --sysroot=@var{dir}}
403 @c I wrote this xref this way to avoid overfull hbox. -- rms
404 @xref{Target Options}.
405 @gccoptlist{-V @var{version} -b @var{machine}}
407 @item Machine Dependent Options
408 @xref{Submodel Options,,Hardware Models and Configurations}.
409 @c This list is ordered alphanumerically by subsection name.
410 @c Try and put the significant identifier (CPU or system) first,
411 @c so users have a clue at guessing where the ones they want will be.
414 @gccoptlist{-EB -EL @gol
415 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
416 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
419 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
420 -mabi=@var{name} @gol
421 -mapcs-stack-check -mno-apcs-stack-check @gol
422 -mapcs-float -mno-apcs-float @gol
423 -mapcs-reentrant -mno-apcs-reentrant @gol
424 -msched-prolog -mno-sched-prolog @gol
425 -mlittle-endian -mbig-endian -mwords-little-endian @gol
426 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
427 -mthumb-interwork -mno-thumb-interwork @gol
428 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
429 -mstructure-size-boundary=@var{n} @gol
430 -mabort-on-noreturn @gol
431 -mlong-calls -mno-long-calls @gol
432 -msingle-pic-base -mno-single-pic-base @gol
433 -mpic-register=@var{reg} @gol
434 -mnop-fun-dllimport @gol
435 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
436 -mpoke-function-name @gol
438 -mtpcs-frame -mtpcs-leaf-frame @gol
439 -mcaller-super-interworking -mcallee-super-interworking @gol
443 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
444 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
446 @emph{Blackfin Options}
447 @gccoptlist{-mcpu=@var{cpu} -msim -momit-leaf-frame-pointer @gol
448 -mno-omit-leaf-frame-pointer -mspecld-anomaly -mno-specld-anomaly @gol
449 -mcsync-anomaly -mno-csync-anomaly -mlow-64k -mno-low64k @gol
450 -mstack-check-l1 -mid-shared-library -mno-id-shared-library @gol
451 -mshared-library-id=@var{n} -mleaf-id-shared-library @gol
452 -mno-leaf-id-shared-library -msep-data -mno-sep-data -mlong-calls @gol
453 -mno-long-calls -mfast-fp -minline-plt}
456 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
457 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
458 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
459 -mstack-align -mdata-align -mconst-align @gol
460 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
461 -melf -maout -melinux -mlinux -sim -sim2 @gol
462 -mmul-bug-workaround -mno-mul-bug-workaround}
465 @gccoptlist{-mmac -mpush-args}
467 @emph{Darwin Options}
468 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
469 -arch_only -bind_at_load -bundle -bundle_loader @gol
470 -client_name -compatibility_version -current_version @gol
472 -dependency-file -dylib_file -dylinker_install_name @gol
473 -dynamic -dynamiclib -exported_symbols_list @gol
474 -filelist -flat_namespace -force_cpusubtype_ALL @gol
475 -force_flat_namespace -headerpad_max_install_names @gol
477 -image_base -init -install_name -keep_private_externs @gol
478 -multi_module -multiply_defined -multiply_defined_unused @gol
479 -noall_load -no_dead_strip_inits_and_terms @gol
480 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
481 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
482 -private_bundle -read_only_relocs -sectalign @gol
483 -sectobjectsymbols -whyload -seg1addr @gol
484 -sectcreate -sectobjectsymbols -sectorder @gol
485 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
486 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
487 -segprot -segs_read_only_addr -segs_read_write_addr @gol
488 -single_module -static -sub_library -sub_umbrella @gol
489 -twolevel_namespace -umbrella -undefined @gol
490 -unexported_symbols_list -weak_reference_mismatches @gol
491 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
492 -mkernel -mone-byte-bool}
494 @emph{DEC Alpha Options}
495 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
496 -mieee -mieee-with-inexact -mieee-conformant @gol
497 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
498 -mtrap-precision=@var{mode} -mbuild-constants @gol
499 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
500 -mbwx -mmax -mfix -mcix @gol
501 -mfloat-vax -mfloat-ieee @gol
502 -mexplicit-relocs -msmall-data -mlarge-data @gol
503 -msmall-text -mlarge-text @gol
504 -mmemory-latency=@var{time}}
506 @emph{DEC Alpha/VMS Options}
507 @gccoptlist{-mvms-return-codes}
510 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
511 -mhard-float -msoft-float @gol
512 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
513 -mdouble -mno-double @gol
514 -mmedia -mno-media -mmuladd -mno-muladd @gol
515 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
516 -mlinked-fp -mlong-calls -malign-labels @gol
517 -mlibrary-pic -macc-4 -macc-8 @gol
518 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
519 -moptimize-membar -mno-optimize-membar @gol
520 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
521 -mvliw-branch -mno-vliw-branch @gol
522 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
523 -mno-nested-cond-exec -mtomcat-stats @gol
527 @emph{GNU/Linux Options}
528 @gccoptlist{-muclibc}
530 @emph{H8/300 Options}
531 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
534 @gccoptlist{-march=@var{architecture-type} @gol
535 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
536 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
537 -mfixed-range=@var{register-range} @gol
538 -mjump-in-delay -mlinker-opt -mlong-calls @gol
539 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
540 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
541 -mno-jump-in-delay -mno-long-load-store @gol
542 -mno-portable-runtime -mno-soft-float @gol
543 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
544 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
545 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
546 -munix=@var{unix-std} -nolibdld -static -threads}
548 @emph{i386 and x86-64 Options}
549 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
550 -mfpmath=@var{unit} @gol
551 -masm=@var{dialect} -mno-fancy-math-387 @gol
552 -mno-fp-ret-in-387 -msoft-float @gol
553 -mno-wide-multiply -mrtd -malign-double @gol
554 -mpreferred-stack-boundary=@var{num} -mcx16 -msahf -mrecip @gol
555 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 @gol
556 -msse4a -m3dnow -mpopcnt -mabm @gol
557 -mthreads -mno-align-stringops -minline-all-stringops @gol
558 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
559 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
560 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
561 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
562 -mcmodel=@var{code-model} @gol
563 -m32 -m64 -mlarge-data-threshold=@var{num}}
566 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
567 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
568 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
569 -minline-float-divide-max-throughput @gol
570 -minline-int-divide-min-latency @gol
571 -minline-int-divide-max-throughput @gol
572 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
573 -mno-dwarf2-asm -mearly-stop-bits @gol
574 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
575 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
576 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
577 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
578 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
579 -mno-sched-prefer-non-data-spec-insns @gol
580 -mno-sched-prefer-non-control-spec-insns @gol
581 -mno-sched-count-spec-in-critical-path}
583 @emph{M32R/D Options}
584 @gccoptlist{-m32r2 -m32rx -m32r @gol
586 -malign-loops -mno-align-loops @gol
587 -missue-rate=@var{number} @gol
588 -mbranch-cost=@var{number} @gol
589 -mmodel=@var{code-size-model-type} @gol
590 -msdata=@var{sdata-type} @gol
591 -mno-flush-func -mflush-func=@var{name} @gol
592 -mno-flush-trap -mflush-trap=@var{number} @gol
596 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
598 @emph{M680x0 Options}
599 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
600 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
601 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
602 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
603 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
604 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
605 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
606 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
608 @emph{M68hc1x Options}
609 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
610 -mauto-incdec -minmax -mlong-calls -mshort @gol
611 -msoft-reg-count=@var{count}}
614 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
615 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
616 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
617 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
618 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
621 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
622 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
623 -mips16 -mno-mips16 -mflip-mips16 @gol
624 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
625 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
626 -mfp32 -mfp64 -mhard-float -msoft-float @gol
627 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
628 -msmartmips -mno-smartmips @gol
629 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
630 -mips3d -mno-mips3d -mmt -mno-mt @gol
631 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
632 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
633 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
634 -membedded-data -mno-embedded-data @gol
635 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
636 -mcode-readable=@var{setting} @gol
637 -msplit-addresses -mno-split-addresses @gol
638 -mexplicit-relocs -mno-explicit-relocs @gol
639 -mcheck-zero-division -mno-check-zero-division @gol
640 -mdivide-traps -mdivide-breaks @gol
641 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
642 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
643 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
644 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
645 -mfix-sb1 -mno-fix-sb1 @gol
646 -mflush-func=@var{func} -mno-flush-func @gol
647 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
648 -mfp-exceptions -mno-fp-exceptions @gol
649 -mvr4130-align -mno-vr4130-align}
652 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
653 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
654 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
655 -mno-base-addresses -msingle-exit -mno-single-exit}
657 @emph{MN10300 Options}
658 @gccoptlist{-mmult-bug -mno-mult-bug @gol
659 -mam33 -mno-am33 @gol
660 -mam33-2 -mno-am33-2 @gol
661 -mreturn-pointer-on-d0 @gol
665 @gccoptlist{-mno-crt0 -mbacc -msim @gol
666 -march=@var{cpu-type} }
668 @emph{PDP-11 Options}
669 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
670 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
671 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
672 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
673 -mbranch-expensive -mbranch-cheap @gol
674 -msplit -mno-split -munix-asm -mdec-asm}
676 @emph{PowerPC Options}
677 See RS/6000 and PowerPC Options.
679 @emph{RS/6000 and PowerPC Options}
680 @gccoptlist{-mcpu=@var{cpu-type} @gol
681 -mtune=@var{cpu-type} @gol
682 -mpower -mno-power -mpower2 -mno-power2 @gol
683 -mpowerpc -mpowerpc64 -mno-powerpc @gol
684 -maltivec -mno-altivec @gol
685 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
686 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
687 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
688 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mdfp -mno-dfp @gol
689 -mnew-mnemonics -mold-mnemonics @gol
690 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
691 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
692 -malign-power -malign-natural @gol
693 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
694 -mstring -mno-string -mupdate -mno-update @gol
695 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
696 -mstrict-align -mno-strict-align -mrelocatable @gol
697 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
698 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
699 -mdynamic-no-pic -maltivec -mswdiv @gol
700 -mprioritize-restricted-insns=@var{priority} @gol
701 -msched-costly-dep=@var{dependence_type} @gol
702 -minsert-sched-nops=@var{scheme} @gol
703 -mcall-sysv -mcall-netbsd @gol
704 -maix-struct-return -msvr4-struct-return @gol
705 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
706 -misel -mno-isel @gol
707 -misel=yes -misel=no @gol
709 -mspe=yes -mspe=no @gol
711 -mvrsave -mno-vrsave @gol
712 -mmulhw -mno-mulhw @gol
713 -mdlmzb -mno-dlmzb @gol
714 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
715 -mprototype -mno-prototype @gol
716 -msim -mmvme -mads -myellowknife -memb -msdata @gol
717 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
719 @emph{S/390 and zSeries Options}
720 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
721 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
722 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
723 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
724 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
725 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
726 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
729 @gccoptlist{-meb -mel @gol
733 -mscore5 -mscore5u -mscore7 -mscore7d}
736 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
737 -m4-nofpu -m4-single-only -m4-single -m4 @gol
738 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
739 -m5-64media -m5-64media-nofpu @gol
740 -m5-32media -m5-32media-nofpu @gol
741 -m5-compact -m5-compact-nofpu @gol
742 -mb -ml -mdalign -mrelax @gol
743 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
744 -mieee -misize -minline-ic_invalidate -mpadstruct -mspace @gol
745 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
746 -mdivsi3_libfunc=@var{name} @gol
747 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
751 @gccoptlist{-mcpu=@var{cpu-type} @gol
752 -mtune=@var{cpu-type} @gol
753 -mcmodel=@var{code-model} @gol
754 -m32 -m64 -mapp-regs -mno-app-regs @gol
755 -mfaster-structs -mno-faster-structs @gol
756 -mfpu -mno-fpu -mhard-float -msoft-float @gol
757 -mhard-quad-float -msoft-quad-float @gol
758 -mimpure-text -mno-impure-text -mlittle-endian @gol
759 -mstack-bias -mno-stack-bias @gol
760 -munaligned-doubles -mno-unaligned-doubles @gol
761 -mv8plus -mno-v8plus -mvis -mno-vis
762 -threads -pthreads -pthread}
765 @gccoptlist{-mwarn-reloc -merror-reloc @gol
766 -msafe-dma -munsafe-dma @gol
768 -msmall-mem -mlarge-mem -mstdmain @gol
769 -mfixed-range=@var{register-range}}
771 @emph{System V Options}
772 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
774 @emph{TMS320C3x/C4x Options}
775 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
776 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
777 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
778 -mparallel-insns -mparallel-mpy -mpreserve-float}
781 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
782 -mprolog-function -mno-prolog-function -mspace @gol
783 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
784 -mapp-regs -mno-app-regs @gol
785 -mdisable-callt -mno-disable-callt @gol
791 @gccoptlist{-mg -mgnu -munix}
793 @emph{VxWorks Options}
794 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
795 -Xbind-lazy -Xbind-now}
797 @emph{x86-64 Options}
798 See i386 and x86-64 Options.
800 @emph{Xstormy16 Options}
803 @emph{Xtensa Options}
804 @gccoptlist{-mconst16 -mno-const16 @gol
805 -mfused-madd -mno-fused-madd @gol
806 -mtext-section-literals -mno-text-section-literals @gol
807 -mtarget-align -mno-target-align @gol
808 -mlongcalls -mno-longcalls}
810 @emph{zSeries Options}
811 See S/390 and zSeries Options.
813 @item Code Generation Options
814 @xref{Code Gen Options,,Options for Code Generation Conventions}.
815 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
816 -ffixed-@var{reg} -fexceptions @gol
817 -fnon-call-exceptions -funwind-tables @gol
818 -fasynchronous-unwind-tables @gol
819 -finhibit-size-directive -finstrument-functions @gol
820 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
821 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
822 -fno-common -fno-ident @gol
823 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
824 -fno-jump-tables @gol
825 -frecord-gcc-switches @gol
826 -freg-struct-return -fshort-enums @gol
827 -fshort-double -fshort-wchar @gol
828 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
829 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
830 -fargument-alias -fargument-noalias @gol
831 -fargument-noalias-global -fargument-noalias-anything
832 -fleading-underscore -ftls-model=@var{model} @gol
833 -ftrapv -fwrapv -fbounds-check @gol
838 * Overall Options:: Controlling the kind of output:
839 an executable, object files, assembler files,
840 or preprocessed source.
841 * C Dialect Options:: Controlling the variant of C language compiled.
842 * C++ Dialect Options:: Variations on C++.
843 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
845 * Language Independent Options:: Controlling how diagnostics should be
847 * Warning Options:: How picky should the compiler be?
848 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
849 * Optimize Options:: How much optimization?
850 * Preprocessor Options:: Controlling header files and macro definitions.
851 Also, getting dependency information for Make.
852 * Assembler Options:: Passing options to the assembler.
853 * Link Options:: Specifying libraries and so on.
854 * Directory Options:: Where to find header files and libraries.
855 Where to find the compiler executable files.
856 * Spec Files:: How to pass switches to sub-processes.
857 * Target Options:: Running a cross-compiler, or an old version of GCC.
860 @node Overall Options
861 @section Options Controlling the Kind of Output
863 Compilation can involve up to four stages: preprocessing, compilation
864 proper, assembly and linking, always in that order. GCC is capable of
865 preprocessing and compiling several files either into several
866 assembler input files, or into one assembler input file; then each
867 assembler input file produces an object file, and linking combines all
868 the object files (those newly compiled, and those specified as input)
869 into an executable file.
871 @cindex file name suffix
872 For any given input file, the file name suffix determines what kind of
877 C source code which must be preprocessed.
880 C source code which should not be preprocessed.
883 C++ source code which should not be preprocessed.
886 Objective-C source code. Note that you must link with the @file{libobjc}
887 library to make an Objective-C program work.
890 Objective-C source code which should not be preprocessed.
894 Objective-C++ source code. Note that you must link with the @file{libobjc}
895 library to make an Objective-C++ program work. Note that @samp{.M} refers
896 to a literal capital M@.
899 Objective-C++ source code which should not be preprocessed.
902 C, C++, Objective-C or Objective-C++ header file to be turned into a
907 @itemx @var{file}.cxx
908 @itemx @var{file}.cpp
909 @itemx @var{file}.CPP
910 @itemx @var{file}.c++
912 C++ source code which must be preprocessed. Note that in @samp{.cxx},
913 the last two letters must both be literally @samp{x}. Likewise,
914 @samp{.C} refers to a literal capital C@.
918 Objective-C++ source code which must be preprocessed.
921 Objective-C++ source code which should not be preprocessed.
926 @itemx @var{file}.hxx
927 @itemx @var{file}.hpp
928 @itemx @var{file}.HPP
929 @itemx @var{file}.h++
930 @itemx @var{file}.tcc
931 C++ header file to be turned into a precompiled header.
934 @itemx @var{file}.for
935 @itemx @var{file}.FOR
936 Fixed form Fortran source code which should not be preprocessed.
939 @itemx @var{file}.fpp
940 @itemx @var{file}.FPP
941 Fixed form Fortran source code which must be preprocessed (with the traditional
945 @itemx @var{file}.f95
946 Free form Fortran source code which should not be preprocessed.
949 @itemx @var{file}.F95
950 Free form Fortran source code which must be preprocessed (with the
951 traditional preprocessor).
953 @c FIXME: Descriptions of Java file types.
960 Ada source code file which contains a library unit declaration (a
961 declaration of a package, subprogram, or generic, or a generic
962 instantiation), or a library unit renaming declaration (a package,
963 generic, or subprogram renaming declaration). Such files are also
966 @itemx @var{file}.adb
967 Ada source code file containing a library unit body (a subprogram or
968 package body). Such files are also called @dfn{bodies}.
970 @c GCC also knows about some suffixes for languages not yet included:
981 Assembler code which must be preprocessed.
984 An object file to be fed straight into linking.
985 Any file name with no recognized suffix is treated this way.
989 You can specify the input language explicitly with the @option{-x} option:
992 @item -x @var{language}
993 Specify explicitly the @var{language} for the following input files
994 (rather than letting the compiler choose a default based on the file
995 name suffix). This option applies to all following input files until
996 the next @option{-x} option. Possible values for @var{language} are:
998 c c-header c-cpp-output
999 c++ c++-header c++-cpp-output
1000 objective-c objective-c-header objective-c-cpp-output
1001 objective-c++ objective-c++-header objective-c++-cpp-output
1002 assembler assembler-with-cpp
1010 Turn off any specification of a language, so that subsequent files are
1011 handled according to their file name suffixes (as they are if @option{-x}
1012 has not been used at all).
1014 @item -pass-exit-codes
1015 @opindex pass-exit-codes
1016 Normally the @command{gcc} program will exit with the code of 1 if any
1017 phase of the compiler returns a non-success return code. If you specify
1018 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1019 numerically highest error produced by any phase that returned an error
1020 indication. The C, C++, and Fortran frontends return 4, if an internal
1021 compiler error is encountered.
1024 If you only want some of the stages of compilation, you can use
1025 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1026 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1027 @command{gcc} is to stop. Note that some combinations (for example,
1028 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1033 Compile or assemble the source files, but do not link. The linking
1034 stage simply is not done. The ultimate output is in the form of an
1035 object file for each source file.
1037 By default, the object file name for a source file is made by replacing
1038 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1040 Unrecognized input files, not requiring compilation or assembly, are
1045 Stop after the stage of compilation proper; do not assemble. The output
1046 is in the form of an assembler code file for each non-assembler input
1049 By default, the assembler file name for a source file is made by
1050 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1052 Input files that don't require compilation are ignored.
1056 Stop after the preprocessing stage; do not run the compiler proper. The
1057 output is in the form of preprocessed source code, which is sent to the
1060 Input files which don't require preprocessing are ignored.
1062 @cindex output file option
1065 Place output in file @var{file}. This applies regardless to whatever
1066 sort of output is being produced, whether it be an executable file,
1067 an object file, an assembler file or preprocessed C code.
1069 If @option{-o} is not specified, the default is to put an executable
1070 file in @file{a.out}, the object file for
1071 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1072 assembler file in @file{@var{source}.s}, a precompiled header file in
1073 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1078 Print (on standard error output) the commands executed to run the stages
1079 of compilation. Also print the version number of the compiler driver
1080 program and of the preprocessor and the compiler proper.
1084 Like @option{-v} except the commands are not executed and all command
1085 arguments are quoted. This is useful for shell scripts to capture the
1086 driver-generated command lines.
1090 Use pipes rather than temporary files for communication between the
1091 various stages of compilation. This fails to work on some systems where
1092 the assembler is unable to read from a pipe; but the GNU assembler has
1097 If you are compiling multiple source files, this option tells the driver
1098 to pass all the source files to the compiler at once (for those
1099 languages for which the compiler can handle this). This will allow
1100 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1101 language for which this is supported is C@. If you pass source files for
1102 multiple languages to the driver, using this option, the driver will invoke
1103 the compiler(s) that support IMA once each, passing each compiler all the
1104 source files appropriate for it. For those languages that do not support
1105 IMA this option will be ignored, and the compiler will be invoked once for
1106 each source file in that language. If you use this option in conjunction
1107 with @option{-save-temps}, the compiler will generate multiple
1109 (one for each source file), but only one (combined) @file{.o} or
1114 Print (on the standard output) a description of the command line options
1115 understood by @command{gcc}. If the @option{-v} option is also specified
1116 then @option{--help} will also be passed on to the various processes
1117 invoked by @command{gcc}, so that they can display the command line options
1118 they accept. If the @option{-Wextra} option has also been specified
1119 (prior to the @option{--help} option), then command line options which
1120 have no documentation associated with them will also be displayed.
1123 @opindex target-help
1124 Print (on the standard output) a description of target-specific command
1125 line options for each tool. For some targets extra target-specific
1126 information may also be printed.
1128 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1129 Print (on the standard output) a description of the command line
1130 options understood by the compiler that fit into a specific class.
1131 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1132 @samp{params}, or @var{language}:
1135 @item @samp{optimizers}
1136 This will display all of the optimization options supported by the
1139 @item @samp{warnings}
1140 This will display all of the options controlling warning messages
1141 produced by the compiler.
1144 This will display target-specific options. Unlike the
1145 @option{--target-help} option however, target-specific options of the
1146 linker and assembler will not be displayed. This is because those
1147 tools do not currently support the extended @option{--help=} syntax.
1150 This will display the values recognized by the @option{--param}
1153 @item @var{language}
1154 This will display the options supported for @var{language}, where
1155 @var{language} is the name of one of the languages supported in this
1159 This will display the options that are common to all languages.
1162 It is possible to further refine the output of the @option{--help=}
1163 option by adding a comma separated list of qualifiers after the
1164 class. These can be any from the following list:
1167 @item @samp{undocumented}
1168 Display only those options which are undocumented.
1171 Display options which take an argument that appears after an equal
1172 sign in the same continuous piece of text, such as:
1173 @samp{--help=target}.
1175 @item @samp{separate}
1176 Display options which take an argument that appears as a separate word
1177 following the original option, such as: @samp{-o output-file}.
1180 Thus for example to display all the undocumented target-specific
1181 switches supported by the compiler the following can be used:
1184 --help=target,undocumented
1187 The sense of a qualifier can be inverted by prefixing it with the
1188 @var{^} character, so for example to display all binary warning
1189 options (i.e. ones that are either on or off and that do not take an
1190 argument), which have a description the following can be used:
1193 --help=warnings,^joined,^undocumented
1196 A class can also be used as a qualifier, although this usually
1197 restricts the output by so much that there is nothing to display. One
1198 case where it does work however is when one of the classes is
1199 @var{target}. So for example to display all the target-specific
1200 optimization options the following can be used:
1203 --help=target,optimizers
1206 The @option{--help=} option can be repeated on the command line. Each
1207 successive use will display its requested class of options, skipping
1208 those that have already been displayed.
1210 If the @option{-Q} option appears on the command line before the
1211 @option{--help=} option, then the descriptive text displayed by
1212 @option{--help=} is changed. Instead of describing the displayed
1213 options, an indication is given as to whether the option is enabled,
1214 disabled or set to a specific value (assuming that the compiler
1215 knows this at the point where the @option{--help=} option is used).
1217 Here is a truncated example from the ARM port of @command{gcc}:
1220 % gcc -Q -mabi=2 --help=target -c
1221 The following options are target specific:
1223 -mabort-on-noreturn [disabled]
1227 The output is sensitive to the effects of previous command line
1228 options, so for example it is possible to find out which optimizations
1229 are enabled at @option{-O2} by using:
1232 -O2 --help=optimizers
1235 Alternatively you can discover which binary optimizations are enabled
1236 by @option{-O3} by using:
1239 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1240 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1241 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1246 Display the version number and copyrights of the invoked GCC@.
1248 @include @value{srcdir}/../libiberty/at-file.texi
1252 @section Compiling C++ Programs
1254 @cindex suffixes for C++ source
1255 @cindex C++ source file suffixes
1256 C++ source files conventionally use one of the suffixes @samp{.C},
1257 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1258 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1259 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1260 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1261 files with these names and compiles them as C++ programs even if you
1262 call the compiler the same way as for compiling C programs (usually
1263 with the name @command{gcc}).
1267 However, the use of @command{gcc} does not add the C++ library.
1268 @command{g++} is a program that calls GCC and treats @samp{.c},
1269 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1270 files unless @option{-x} is used, and automatically specifies linking
1271 against the C++ library. This program is also useful when
1272 precompiling a C header file with a @samp{.h} extension for use in C++
1273 compilations. On many systems, @command{g++} is also installed with
1274 the name @command{c++}.
1276 @cindex invoking @command{g++}
1277 When you compile C++ programs, you may specify many of the same
1278 command-line options that you use for compiling programs in any
1279 language; or command-line options meaningful for C and related
1280 languages; or options that are meaningful only for C++ programs.
1281 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1282 explanations of options for languages related to C@.
1283 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1284 explanations of options that are meaningful only for C++ programs.
1286 @node C Dialect Options
1287 @section Options Controlling C Dialect
1288 @cindex dialect options
1289 @cindex language dialect options
1290 @cindex options, dialect
1292 The following options control the dialect of C (or languages derived
1293 from C, such as C++, Objective-C and Objective-C++) that the compiler
1297 @cindex ANSI support
1301 In C mode, support all ISO C90 programs. In C++ mode,
1302 remove GNU extensions that conflict with ISO C++.
1304 This turns off certain features of GCC that are incompatible with ISO
1305 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1306 such as the @code{asm} and @code{typeof} keywords, and
1307 predefined macros such as @code{unix} and @code{vax} that identify the
1308 type of system you are using. It also enables the undesirable and
1309 rarely used ISO trigraph feature. For the C compiler,
1310 it disables recognition of C++ style @samp{//} comments as well as
1311 the @code{inline} keyword.
1313 The alternate keywords @code{__asm__}, @code{__extension__},
1314 @code{__inline__} and @code{__typeof__} continue to work despite
1315 @option{-ansi}. You would not want to use them in an ISO C program, of
1316 course, but it is useful to put them in header files that might be included
1317 in compilations done with @option{-ansi}. Alternate predefined macros
1318 such as @code{__unix__} and @code{__vax__} are also available, with or
1319 without @option{-ansi}.
1321 The @option{-ansi} option does not cause non-ISO programs to be
1322 rejected gratuitously. For that, @option{-pedantic} is required in
1323 addition to @option{-ansi}. @xref{Warning Options}.
1325 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1326 option is used. Some header files may notice this macro and refrain
1327 from declaring certain functions or defining certain macros that the
1328 ISO standard doesn't call for; this is to avoid interfering with any
1329 programs that might use these names for other things.
1331 Functions which would normally be built in but do not have semantics
1332 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1333 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1334 built-in functions provided by GCC}, for details of the functions
1339 Determine the language standard. This option is currently only
1340 supported when compiling C or C++. A value for this option must be
1341 provided; possible values are
1346 ISO C90 (same as @option{-ansi}).
1348 @item iso9899:199409
1349 ISO C90 as modified in amendment 1.
1355 ISO C99. Note that this standard is not yet fully supported; see
1356 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1357 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1360 Default, ISO C90 plus GNU extensions (including some C99 features).
1364 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1365 this will become the default. The name @samp{gnu9x} is deprecated.
1368 The 1998 ISO C++ standard plus amendments.
1371 The same as @option{-std=c++98} plus GNU extensions. This is the
1372 default for C++ code.
1375 The working draft of the upcoming ISO C++0x standard. This option
1376 enables experimental features that are likely to be included in
1377 C++0x. The working draft is constantly changing, and any feature that is
1378 enabled by this flag may be removed from future versions of GCC if it is
1379 not part of the C++0x standard.
1382 The same as @option{-std=c++0x} plus GNU extensions. As with
1383 @option{-std=c++0x}, this option enables experimental features that may
1384 be removed in future versions of GCC.
1387 Even when this option is not specified, you can still use some of the
1388 features of newer standards in so far as they do not conflict with
1389 previous C standards. For example, you may use @code{__restrict__} even
1390 when @option{-std=c99} is not specified.
1392 The @option{-std} options specifying some version of ISO C have the same
1393 effects as @option{-ansi}, except that features that were not in ISO C90
1394 but are in the specified version (for example, @samp{//} comments and
1395 the @code{inline} keyword in ISO C99) are not disabled.
1397 @xref{Standards,,Language Standards Supported by GCC}, for details of
1398 these standard versions.
1400 @item -fgnu89-inline
1401 @opindex fgnu89-inline
1402 The option @option{-fgnu89-inline} tells GCC to use the traditional
1403 GNU semantics for @code{inline} functions when in C99 mode.
1404 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1405 is accepted and ignored by GCC versions 4.1.3 up to but not including
1406 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1407 C99 mode. Using this option is roughly equivalent to adding the
1408 @code{gnu_inline} function attribute to all inline functions
1409 (@pxref{Function Attributes}).
1411 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1412 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1413 specifies the default behavior). This option was first supported in
1414 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1416 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1417 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1418 in effect for @code{inline} functions. @xref{Common Predefined
1419 Macros,,,cpp,The C Preprocessor}.
1421 @item -aux-info @var{filename}
1423 Output to the given filename prototyped declarations for all functions
1424 declared and/or defined in a translation unit, including those in header
1425 files. This option is silently ignored in any language other than C@.
1427 Besides declarations, the file indicates, in comments, the origin of
1428 each declaration (source file and line), whether the declaration was
1429 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1430 @samp{O} for old, respectively, in the first character after the line
1431 number and the colon), and whether it came from a declaration or a
1432 definition (@samp{C} or @samp{F}, respectively, in the following
1433 character). In the case of function definitions, a K&R-style list of
1434 arguments followed by their declarations is also provided, inside
1435 comments, after the declaration.
1439 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1440 keyword, so that code can use these words as identifiers. You can use
1441 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1442 instead. @option{-ansi} implies @option{-fno-asm}.
1444 In C++, this switch only affects the @code{typeof} keyword, since
1445 @code{asm} and @code{inline} are standard keywords. You may want to
1446 use the @option{-fno-gnu-keywords} flag instead, which has the same
1447 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1448 switch only affects the @code{asm} and @code{typeof} keywords, since
1449 @code{inline} is a standard keyword in ISO C99.
1452 @itemx -fno-builtin-@var{function}
1453 @opindex fno-builtin
1454 @cindex built-in functions
1455 Don't recognize built-in functions that do not begin with
1456 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1457 functions provided by GCC}, for details of the functions affected,
1458 including those which are not built-in functions when @option{-ansi} or
1459 @option{-std} options for strict ISO C conformance are used because they
1460 do not have an ISO standard meaning.
1462 GCC normally generates special code to handle certain built-in functions
1463 more efficiently; for instance, calls to @code{alloca} may become single
1464 instructions that adjust the stack directly, and calls to @code{memcpy}
1465 may become inline copy loops. The resulting code is often both smaller
1466 and faster, but since the function calls no longer appear as such, you
1467 cannot set a breakpoint on those calls, nor can you change the behavior
1468 of the functions by linking with a different library. In addition,
1469 when a function is recognized as a built-in function, GCC may use
1470 information about that function to warn about problems with calls to
1471 that function, or to generate more efficient code, even if the
1472 resulting code still contains calls to that function. For example,
1473 warnings are given with @option{-Wformat} for bad calls to
1474 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1475 known not to modify global memory.
1477 With the @option{-fno-builtin-@var{function}} option
1478 only the built-in function @var{function} is
1479 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1480 function is named this is not built-in in this version of GCC, this
1481 option is ignored. There is no corresponding
1482 @option{-fbuiltin-@var{function}} option; if you wish to enable
1483 built-in functions selectively when using @option{-fno-builtin} or
1484 @option{-ffreestanding}, you may define macros such as:
1487 #define abs(n) __builtin_abs ((n))
1488 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1493 @cindex hosted environment
1495 Assert that compilation takes place in a hosted environment. This implies
1496 @option{-fbuiltin}. A hosted environment is one in which the
1497 entire standard library is available, and in which @code{main} has a return
1498 type of @code{int}. Examples are nearly everything except a kernel.
1499 This is equivalent to @option{-fno-freestanding}.
1501 @item -ffreestanding
1502 @opindex ffreestanding
1503 @cindex hosted environment
1505 Assert that compilation takes place in a freestanding environment. This
1506 implies @option{-fno-builtin}. A freestanding environment
1507 is one in which the standard library may not exist, and program startup may
1508 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1509 This is equivalent to @option{-fno-hosted}.
1511 @xref{Standards,,Language Standards Supported by GCC}, for details of
1512 freestanding and hosted environments.
1516 @cindex openmp parallel
1517 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1518 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1519 compiler generates parallel code according to the OpenMP Application
1520 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1522 @item -fms-extensions
1523 @opindex fms-extensions
1524 Accept some non-standard constructs used in Microsoft header files.
1526 Some cases of unnamed fields in structures and unions are only
1527 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1528 fields within structs/unions}, for details.
1532 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1533 options for strict ISO C conformance) implies @option{-trigraphs}.
1535 @item -no-integrated-cpp
1536 @opindex no-integrated-cpp
1537 Performs a compilation in two passes: preprocessing and compiling. This
1538 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1539 @option{-B} option. The user supplied compilation step can then add in
1540 an additional preprocessing step after normal preprocessing but before
1541 compiling. The default is to use the integrated cpp (internal cpp)
1543 The semantics of this option will change if "cc1", "cc1plus", and
1544 "cc1obj" are merged.
1546 @cindex traditional C language
1547 @cindex C language, traditional
1549 @itemx -traditional-cpp
1550 @opindex traditional-cpp
1551 @opindex traditional
1552 Formerly, these options caused GCC to attempt to emulate a pre-standard
1553 C compiler. They are now only supported with the @option{-E} switch.
1554 The preprocessor continues to support a pre-standard mode. See the GNU
1555 CPP manual for details.
1557 @item -fcond-mismatch
1558 @opindex fcond-mismatch
1559 Allow conditional expressions with mismatched types in the second and
1560 third arguments. The value of such an expression is void. This option
1561 is not supported for C++.
1563 @item -flax-vector-conversions
1564 @opindex flax-vector-conversions
1565 Allow implicit conversions between vectors with differing numbers of
1566 elements and/or incompatible element types. This option should not be
1569 @item -funsigned-char
1570 @opindex funsigned-char
1571 Let the type @code{char} be unsigned, like @code{unsigned char}.
1573 Each kind of machine has a default for what @code{char} should
1574 be. It is either like @code{unsigned char} by default or like
1575 @code{signed char} by default.
1577 Ideally, a portable program should always use @code{signed char} or
1578 @code{unsigned char} when it depends on the signedness of an object.
1579 But many programs have been written to use plain @code{char} and
1580 expect it to be signed, or expect it to be unsigned, depending on the
1581 machines they were written for. This option, and its inverse, let you
1582 make such a program work with the opposite default.
1584 The type @code{char} is always a distinct type from each of
1585 @code{signed char} or @code{unsigned char}, even though its behavior
1586 is always just like one of those two.
1589 @opindex fsigned-char
1590 Let the type @code{char} be signed, like @code{signed char}.
1592 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1593 the negative form of @option{-funsigned-char}. Likewise, the option
1594 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1596 @item -fsigned-bitfields
1597 @itemx -funsigned-bitfields
1598 @itemx -fno-signed-bitfields
1599 @itemx -fno-unsigned-bitfields
1600 @opindex fsigned-bitfields
1601 @opindex funsigned-bitfields
1602 @opindex fno-signed-bitfields
1603 @opindex fno-unsigned-bitfields
1604 These options control whether a bit-field is signed or unsigned, when the
1605 declaration does not use either @code{signed} or @code{unsigned}. By
1606 default, such a bit-field is signed, because this is consistent: the
1607 basic integer types such as @code{int} are signed types.
1610 @node C++ Dialect Options
1611 @section Options Controlling C++ Dialect
1613 @cindex compiler options, C++
1614 @cindex C++ options, command line
1615 @cindex options, C++
1616 This section describes the command-line options that are only meaningful
1617 for C++ programs; but you can also use most of the GNU compiler options
1618 regardless of what language your program is in. For example, you
1619 might compile a file @code{firstClass.C} like this:
1622 g++ -g -frepo -O -c firstClass.C
1626 In this example, only @option{-frepo} is an option meant
1627 only for C++ programs; you can use the other options with any
1628 language supported by GCC@.
1630 Here is a list of options that are @emph{only} for compiling C++ programs:
1634 @item -fabi-version=@var{n}
1635 @opindex fabi-version
1636 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1637 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1638 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1639 the version that conforms most closely to the C++ ABI specification.
1640 Therefore, the ABI obtained using version 0 will change as ABI bugs
1643 The default is version 2.
1645 @item -fno-access-control
1646 @opindex fno-access-control
1647 Turn off all access checking. This switch is mainly useful for working
1648 around bugs in the access control code.
1652 Check that the pointer returned by @code{operator new} is non-null
1653 before attempting to modify the storage allocated. This check is
1654 normally unnecessary because the C++ standard specifies that
1655 @code{operator new} will only return @code{0} if it is declared
1656 @samp{throw()}, in which case the compiler will always check the
1657 return value even without this option. In all other cases, when
1658 @code{operator new} has a non-empty exception specification, memory
1659 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1660 @samp{new (nothrow)}.
1662 @item -fconserve-space
1663 @opindex fconserve-space
1664 Put uninitialized or runtime-initialized global variables into the
1665 common segment, as C does. This saves space in the executable at the
1666 cost of not diagnosing duplicate definitions. If you compile with this
1667 flag and your program mysteriously crashes after @code{main()} has
1668 completed, you may have an object that is being destroyed twice because
1669 two definitions were merged.
1671 This option is no longer useful on most targets, now that support has
1672 been added for putting variables into BSS without making them common.
1674 @item -ffriend-injection
1675 @opindex ffriend-injection
1676 Inject friend functions into the enclosing namespace, so that they are
1677 visible outside the scope of the class in which they are declared.
1678 Friend functions were documented to work this way in the old Annotated
1679 C++ Reference Manual, and versions of G++ before 4.1 always worked
1680 that way. However, in ISO C++ a friend function which is not declared
1681 in an enclosing scope can only be found using argument dependent
1682 lookup. This option causes friends to be injected as they were in
1685 This option is for compatibility, and may be removed in a future
1688 @item -fno-elide-constructors
1689 @opindex fno-elide-constructors
1690 The C++ standard allows an implementation to omit creating a temporary
1691 which is only used to initialize another object of the same type.
1692 Specifying this option disables that optimization, and forces G++ to
1693 call the copy constructor in all cases.
1695 @item -fno-enforce-eh-specs
1696 @opindex fno-enforce-eh-specs
1697 Don't generate code to check for violation of exception specifications
1698 at runtime. This option violates the C++ standard, but may be useful
1699 for reducing code size in production builds, much like defining
1700 @samp{NDEBUG}. This does not give user code permission to throw
1701 exceptions in violation of the exception specifications; the compiler
1702 will still optimize based on the specifications, so throwing an
1703 unexpected exception will result in undefined behavior.
1706 @itemx -fno-for-scope
1708 @opindex fno-for-scope
1709 If @option{-ffor-scope} is specified, the scope of variables declared in
1710 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1711 as specified by the C++ standard.
1712 If @option{-fno-for-scope} is specified, the scope of variables declared in
1713 a @i{for-init-statement} extends to the end of the enclosing scope,
1714 as was the case in old versions of G++, and other (traditional)
1715 implementations of C++.
1717 The default if neither flag is given to follow the standard,
1718 but to allow and give a warning for old-style code that would
1719 otherwise be invalid, or have different behavior.
1721 @item -fno-gnu-keywords
1722 @opindex fno-gnu-keywords
1723 Do not recognize @code{typeof} as a keyword, so that code can use this
1724 word as an identifier. You can use the keyword @code{__typeof__} instead.
1725 @option{-ansi} implies @option{-fno-gnu-keywords}.
1727 @item -fno-implicit-templates
1728 @opindex fno-implicit-templates
1729 Never emit code for non-inline templates which are instantiated
1730 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1731 @xref{Template Instantiation}, for more information.
1733 @item -fno-implicit-inline-templates
1734 @opindex fno-implicit-inline-templates
1735 Don't emit code for implicit instantiations of inline templates, either.
1736 The default is to handle inlines differently so that compiles with and
1737 without optimization will need the same set of explicit instantiations.
1739 @item -fno-implement-inlines
1740 @opindex fno-implement-inlines
1741 To save space, do not emit out-of-line copies of inline functions
1742 controlled by @samp{#pragma implementation}. This will cause linker
1743 errors if these functions are not inlined everywhere they are called.
1745 @item -fms-extensions
1746 @opindex fms-extensions
1747 Disable pedantic warnings about constructs used in MFC, such as implicit
1748 int and getting a pointer to member function via non-standard syntax.
1750 @item -fno-nonansi-builtins
1751 @opindex fno-nonansi-builtins
1752 Disable built-in declarations of functions that are not mandated by
1753 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1754 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1756 @item -fno-operator-names
1757 @opindex fno-operator-names
1758 Do not treat the operator name keywords @code{and}, @code{bitand},
1759 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1760 synonyms as keywords.
1762 @item -fno-optional-diags
1763 @opindex fno-optional-diags
1764 Disable diagnostics that the standard says a compiler does not need to
1765 issue. Currently, the only such diagnostic issued by G++ is the one for
1766 a name having multiple meanings within a class.
1769 @opindex fpermissive
1770 Downgrade some diagnostics about nonconformant code from errors to
1771 warnings. Thus, using @option{-fpermissive} will allow some
1772 nonconforming code to compile.
1776 Enable automatic template instantiation at link time. This option also
1777 implies @option{-fno-implicit-templates}. @xref{Template
1778 Instantiation}, for more information.
1782 Disable generation of information about every class with virtual
1783 functions for use by the C++ runtime type identification features
1784 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1785 of the language, you can save some space by using this flag. Note that
1786 exception handling uses the same information, but it will generate it as
1787 needed. The @samp{dynamic_cast} operator can still be used for casts that
1788 do not require runtime type information, i.e. casts to @code{void *} or to
1789 unambiguous base classes.
1793 Emit statistics about front-end processing at the end of the compilation.
1794 This information is generally only useful to the G++ development team.
1796 @item -ftemplate-depth-@var{n}
1797 @opindex ftemplate-depth
1798 Set the maximum instantiation depth for template classes to @var{n}.
1799 A limit on the template instantiation depth is needed to detect
1800 endless recursions during template class instantiation. ANSI/ISO C++
1801 conforming programs must not rely on a maximum depth greater than 17.
1803 @item -fno-threadsafe-statics
1804 @opindex fno-threadsafe-statics
1805 Do not emit the extra code to use the routines specified in the C++
1806 ABI for thread-safe initialization of local statics. You can use this
1807 option to reduce code size slightly in code that doesn't need to be
1810 @item -fuse-cxa-atexit
1811 @opindex fuse-cxa-atexit
1812 Register destructors for objects with static storage duration with the
1813 @code{__cxa_atexit} function rather than the @code{atexit} function.
1814 This option is required for fully standards-compliant handling of static
1815 destructors, but will only work if your C library supports
1816 @code{__cxa_atexit}.
1818 @item -fno-use-cxa-get-exception-ptr
1819 @opindex fno-use-cxa-get-exception-ptr
1820 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1821 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1822 if the runtime routine is not available.
1824 @item -fvisibility-inlines-hidden
1825 @opindex fvisibility-inlines-hidden
1826 This switch declares that the user does not attempt to compare
1827 pointers to inline methods where the addresses of the two functions
1828 were taken in different shared objects.
1830 The effect of this is that GCC may, effectively, mark inline methods with
1831 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1832 appear in the export table of a DSO and do not require a PLT indirection
1833 when used within the DSO@. Enabling this option can have a dramatic effect
1834 on load and link times of a DSO as it massively reduces the size of the
1835 dynamic export table when the library makes heavy use of templates.
1837 The behavior of this switch is not quite the same as marking the
1838 methods as hidden directly, because it does not affect static variables
1839 local to the function or cause the compiler to deduce that
1840 the function is defined in only one shared object.
1842 You may mark a method as having a visibility explicitly to negate the
1843 effect of the switch for that method. For example, if you do want to
1844 compare pointers to a particular inline method, you might mark it as
1845 having default visibility. Marking the enclosing class with explicit
1846 visibility will have no effect.
1848 Explicitly instantiated inline methods are unaffected by this option
1849 as their linkage might otherwise cross a shared library boundary.
1850 @xref{Template Instantiation}.
1852 @item -fvisibility-ms-compat
1853 @opindex fvisibility-ms-compat
1854 This flag attempts to use visibility settings to make GCC's C++
1855 linkage model compatible with that of Microsoft Visual Studio.
1857 The flag makes these changes to GCC's linkage model:
1861 It sets the default visibility to @code{hidden}, like
1862 @option{-fvisibility=hidden}.
1865 Types, but not their members, are not hidden by default.
1868 The One Definition Rule is relaxed for types without explicit
1869 visibility specifications which are defined in more than one different
1870 shared object: those declarations are permitted if they would have
1871 been permitted when this option was not used.
1874 In new code it is better to use @option{-fvisibility=hidden} and
1875 export those classes which are intended to be externally visible.
1876 Unfortunately it is possible for code to rely, perhaps accidentally,
1877 on the Visual Studio behavior.
1879 Among the consequences of these changes are that static data members
1880 of the same type with the same name but defined in different shared
1881 objects will be different, so changing one will not change the other;
1882 and that pointers to function members defined in different shared
1883 objects may not compare equal. When this flag is given, it is a
1884 violation of the ODR to define types with the same name differently.
1888 Do not use weak symbol support, even if it is provided by the linker.
1889 By default, G++ will use weak symbols if they are available. This
1890 option exists only for testing, and should not be used by end-users;
1891 it will result in inferior code and has no benefits. This option may
1892 be removed in a future release of G++.
1896 Do not search for header files in the standard directories specific to
1897 C++, but do still search the other standard directories. (This option
1898 is used when building the C++ library.)
1901 In addition, these optimization, warning, and code generation options
1902 have meanings only for C++ programs:
1905 @item -fno-default-inline
1906 @opindex fno-default-inline
1907 Do not assume @samp{inline} for functions defined inside a class scope.
1908 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1909 functions will have linkage like inline functions; they just won't be
1912 @item -Wabi @r{(C++ only)}
1914 Warn when G++ generates code that is probably not compatible with the
1915 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1916 all such cases, there are probably some cases that are not warned about,
1917 even though G++ is generating incompatible code. There may also be
1918 cases where warnings are emitted even though the code that is generated
1921 You should rewrite your code to avoid these warnings if you are
1922 concerned about the fact that code generated by G++ may not be binary
1923 compatible with code generated by other compilers.
1925 The known incompatibilities at this point include:
1930 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1931 pack data into the same byte as a base class. For example:
1934 struct A @{ virtual void f(); int f1 : 1; @};
1935 struct B : public A @{ int f2 : 1; @};
1939 In this case, G++ will place @code{B::f2} into the same byte
1940 as@code{A::f1}; other compilers will not. You can avoid this problem
1941 by explicitly padding @code{A} so that its size is a multiple of the
1942 byte size on your platform; that will cause G++ and other compilers to
1943 layout @code{B} identically.
1946 Incorrect handling of tail-padding for virtual bases. G++ does not use
1947 tail padding when laying out virtual bases. For example:
1950 struct A @{ virtual void f(); char c1; @};
1951 struct B @{ B(); char c2; @};
1952 struct C : public A, public virtual B @{@};
1956 In this case, G++ will not place @code{B} into the tail-padding for
1957 @code{A}; other compilers will. You can avoid this problem by
1958 explicitly padding @code{A} so that its size is a multiple of its
1959 alignment (ignoring virtual base classes); that will cause G++ and other
1960 compilers to layout @code{C} identically.
1963 Incorrect handling of bit-fields with declared widths greater than that
1964 of their underlying types, when the bit-fields appear in a union. For
1968 union U @{ int i : 4096; @};
1972 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1973 union too small by the number of bits in an @code{int}.
1976 Empty classes can be placed at incorrect offsets. For example:
1986 struct C : public B, public A @{@};
1990 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1991 it should be placed at offset zero. G++ mistakenly believes that the
1992 @code{A} data member of @code{B} is already at offset zero.
1995 Names of template functions whose types involve @code{typename} or
1996 template template parameters can be mangled incorrectly.
1999 template <typename Q>
2000 void f(typename Q::X) @{@}
2002 template <template <typename> class Q>
2003 void f(typename Q<int>::X) @{@}
2007 Instantiations of these templates may be mangled incorrectly.
2011 @item -Wctor-dtor-privacy @r{(C++ only)}
2012 @opindex Wctor-dtor-privacy
2013 Warn when a class seems unusable because all the constructors or
2014 destructors in that class are private, and it has neither friends nor
2015 public static member functions.
2017 @item -Wnon-virtual-dtor @r{(C++ only)}
2018 @opindex Wnon-virtual-dtor
2019 Warn when a class has virtual functions and accessible non-virtual
2020 destructor, in which case it would be possible but unsafe to delete
2021 an instance of a derived class through a pointer to the base class.
2022 This warning is also enabled if -Weffc++ is specified.
2024 @item -Wreorder @r{(C++ only)}
2026 @cindex reordering, warning
2027 @cindex warning for reordering of member initializers
2028 Warn when the order of member initializers given in the code does not
2029 match the order in which they must be executed. For instance:
2035 A(): j (0), i (1) @{ @}
2039 The compiler will rearrange the member initializers for @samp{i}
2040 and @samp{j} to match the declaration order of the members, emitting
2041 a warning to that effect. This warning is enabled by @option{-Wall}.
2044 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2047 @item -Weffc++ @r{(C++ only)}
2049 Warn about violations of the following style guidelines from Scott Meyers'
2050 @cite{Effective C++} book:
2054 Item 11: Define a copy constructor and an assignment operator for classes
2055 with dynamically allocated memory.
2058 Item 12: Prefer initialization to assignment in constructors.
2061 Item 14: Make destructors virtual in base classes.
2064 Item 15: Have @code{operator=} return a reference to @code{*this}.
2067 Item 23: Don't try to return a reference when you must return an object.
2071 Also warn about violations of the following style guidelines from
2072 Scott Meyers' @cite{More Effective C++} book:
2076 Item 6: Distinguish between prefix and postfix forms of increment and
2077 decrement operators.
2080 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2084 When selecting this option, be aware that the standard library
2085 headers do not obey all of these guidelines; use @samp{grep -v}
2086 to filter out those warnings.
2088 @item -Wno-deprecated @r{(C++ only)}
2089 @opindex Wno-deprecated
2090 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
2092 @item -Wstrict-null-sentinel @r{(C++ only)}
2093 @opindex Wstrict-null-sentinel
2094 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2095 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2096 to @code{__null}. Although it is a null pointer constant not a null pointer,
2097 it is guaranteed to of the same size as a pointer. But this use is
2098 not portable across different compilers.
2100 @item -Wno-non-template-friend @r{(C++ only)}
2101 @opindex Wno-non-template-friend
2102 Disable warnings when non-templatized friend functions are declared
2103 within a template. Since the advent of explicit template specification
2104 support in G++, if the name of the friend is an unqualified-id (i.e.,
2105 @samp{friend foo(int)}), the C++ language specification demands that the
2106 friend declare or define an ordinary, nontemplate function. (Section
2107 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2108 could be interpreted as a particular specialization of a templatized
2109 function. Because this non-conforming behavior is no longer the default
2110 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2111 check existing code for potential trouble spots and is on by default.
2112 This new compiler behavior can be turned off with
2113 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2114 but disables the helpful warning.
2116 @item -Wold-style-cast @r{(C++ only)}
2117 @opindex Wold-style-cast
2118 Warn if an old-style (C-style) cast to a non-void type is used within
2119 a C++ program. The new-style casts (@samp{dynamic_cast},
2120 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2121 less vulnerable to unintended effects and much easier to search for.
2123 @item -Woverloaded-virtual @r{(C++ only)}
2124 @opindex Woverloaded-virtual
2125 @cindex overloaded virtual fn, warning
2126 @cindex warning for overloaded virtual fn
2127 Warn when a function declaration hides virtual functions from a
2128 base class. For example, in:
2135 struct B: public A @{
2140 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2148 will fail to compile.
2150 @item -Wno-pmf-conversions @r{(C++ only)}
2151 @opindex Wno-pmf-conversions
2152 Disable the diagnostic for converting a bound pointer to member function
2155 @item -Wsign-promo @r{(C++ only)}
2156 @opindex Wsign-promo
2157 Warn when overload resolution chooses a promotion from unsigned or
2158 enumerated type to a signed type, over a conversion to an unsigned type of
2159 the same size. Previous versions of G++ would try to preserve
2160 unsignedness, but the standard mandates the current behavior.
2165 A& operator = (int);
2175 In this example, G++ will synthesize a default @samp{A& operator =
2176 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2179 @node Objective-C and Objective-C++ Dialect Options
2180 @section Options Controlling Objective-C and Objective-C++ Dialects
2182 @cindex compiler options, Objective-C and Objective-C++
2183 @cindex Objective-C and Objective-C++ options, command line
2184 @cindex options, Objective-C and Objective-C++
2185 (NOTE: This manual does not describe the Objective-C and Objective-C++
2186 languages themselves. See @xref{Standards,,Language Standards
2187 Supported by GCC}, for references.)
2189 This section describes the command-line options that are only meaningful
2190 for Objective-C and Objective-C++ programs, but you can also use most of
2191 the language-independent GNU compiler options.
2192 For example, you might compile a file @code{some_class.m} like this:
2195 gcc -g -fgnu-runtime -O -c some_class.m
2199 In this example, @option{-fgnu-runtime} is an option meant only for
2200 Objective-C and Objective-C++ programs; you can use the other options with
2201 any language supported by GCC@.
2203 Note that since Objective-C is an extension of the C language, Objective-C
2204 compilations may also use options specific to the C front-end (e.g.,
2205 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2206 C++-specific options (e.g., @option{-Wabi}).
2208 Here is a list of options that are @emph{only} for compiling Objective-C
2209 and Objective-C++ programs:
2212 @item -fconstant-string-class=@var{class-name}
2213 @opindex fconstant-string-class
2214 Use @var{class-name} as the name of the class to instantiate for each
2215 literal string specified with the syntax @code{@@"@dots{}"}. The default
2216 class name is @code{NXConstantString} if the GNU runtime is being used, and
2217 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2218 @option{-fconstant-cfstrings} option, if also present, will override the
2219 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2220 to be laid out as constant CoreFoundation strings.
2223 @opindex fgnu-runtime
2224 Generate object code compatible with the standard GNU Objective-C
2225 runtime. This is the default for most types of systems.
2227 @item -fnext-runtime
2228 @opindex fnext-runtime
2229 Generate output compatible with the NeXT runtime. This is the default
2230 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2231 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2234 @item -fno-nil-receivers
2235 @opindex fno-nil-receivers
2236 Assume that all Objective-C message dispatches (e.g.,
2237 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2238 is not @code{nil}. This allows for more efficient entry points in the runtime
2239 to be used. Currently, this option is only available in conjunction with
2240 the NeXT runtime on Mac OS X 10.3 and later.
2242 @item -fobjc-call-cxx-cdtors
2243 @opindex fobjc-call-cxx-cdtors
2244 For each Objective-C class, check if any of its instance variables is a
2245 C++ object with a non-trivial default constructor. If so, synthesize a
2246 special @code{- (id) .cxx_construct} instance method that will run
2247 non-trivial default constructors on any such instance variables, in order,
2248 and then return @code{self}. Similarly, check if any instance variable
2249 is a C++ object with a non-trivial destructor, and if so, synthesize a
2250 special @code{- (void) .cxx_destruct} method that will run
2251 all such default destructors, in reverse order.
2253 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2254 thusly generated will only operate on instance variables declared in the
2255 current Objective-C class, and not those inherited from superclasses. It
2256 is the responsibility of the Objective-C runtime to invoke all such methods
2257 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2258 will be invoked by the runtime immediately after a new object
2259 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2260 be invoked immediately before the runtime deallocates an object instance.
2262 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2263 support for invoking the @code{- (id) .cxx_construct} and
2264 @code{- (void) .cxx_destruct} methods.
2266 @item -fobjc-direct-dispatch
2267 @opindex fobjc-direct-dispatch
2268 Allow fast jumps to the message dispatcher. On Darwin this is
2269 accomplished via the comm page.
2271 @item -fobjc-exceptions
2272 @opindex fobjc-exceptions
2273 Enable syntactic support for structured exception handling in Objective-C,
2274 similar to what is offered by C++ and Java. This option is
2275 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2284 @@catch (AnObjCClass *exc) @{
2291 @@catch (AnotherClass *exc) @{
2294 @@catch (id allOthers) @{
2304 The @code{@@throw} statement may appear anywhere in an Objective-C or
2305 Objective-C++ program; when used inside of a @code{@@catch} block, the
2306 @code{@@throw} may appear without an argument (as shown above), in which case
2307 the object caught by the @code{@@catch} will be rethrown.
2309 Note that only (pointers to) Objective-C objects may be thrown and
2310 caught using this scheme. When an object is thrown, it will be caught
2311 by the nearest @code{@@catch} clause capable of handling objects of that type,
2312 analogously to how @code{catch} blocks work in C++ and Java. A
2313 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2314 any and all Objective-C exceptions not caught by previous @code{@@catch}
2317 The @code{@@finally} clause, if present, will be executed upon exit from the
2318 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2319 regardless of whether any exceptions are thrown, caught or rethrown
2320 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2321 of the @code{finally} clause in Java.
2323 There are several caveats to using the new exception mechanism:
2327 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2328 idioms provided by the @code{NSException} class, the new
2329 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2330 systems, due to additional functionality needed in the (NeXT) Objective-C
2334 As mentioned above, the new exceptions do not support handling
2335 types other than Objective-C objects. Furthermore, when used from
2336 Objective-C++, the Objective-C exception model does not interoperate with C++
2337 exceptions at this time. This means you cannot @code{@@throw} an exception
2338 from Objective-C and @code{catch} it in C++, or vice versa
2339 (i.e., @code{throw @dots{} @@catch}).
2342 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2343 blocks for thread-safe execution:
2346 @@synchronized (ObjCClass *guard) @{
2351 Upon entering the @code{@@synchronized} block, a thread of execution shall
2352 first check whether a lock has been placed on the corresponding @code{guard}
2353 object by another thread. If it has, the current thread shall wait until
2354 the other thread relinquishes its lock. Once @code{guard} becomes available,
2355 the current thread will place its own lock on it, execute the code contained in
2356 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2357 making @code{guard} available to other threads).
2359 Unlike Java, Objective-C does not allow for entire methods to be marked
2360 @code{@@synchronized}. Note that throwing exceptions out of
2361 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2362 to be unlocked properly.
2366 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2368 @item -freplace-objc-classes
2369 @opindex freplace-objc-classes
2370 Emit a special marker instructing @command{ld(1)} not to statically link in
2371 the resulting object file, and allow @command{dyld(1)} to load it in at
2372 run time instead. This is used in conjunction with the Fix-and-Continue
2373 debugging mode, where the object file in question may be recompiled and
2374 dynamically reloaded in the course of program execution, without the need
2375 to restart the program itself. Currently, Fix-and-Continue functionality
2376 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2381 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2382 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2383 compile time) with static class references that get initialized at load time,
2384 which improves run-time performance. Specifying the @option{-fzero-link} flag
2385 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2386 to be retained. This is useful in Zero-Link debugging mode, since it allows
2387 for individual class implementations to be modified during program execution.
2391 Dump interface declarations for all classes seen in the source file to a
2392 file named @file{@var{sourcename}.decl}.
2394 @item -Wassign-intercept
2395 @opindex Wassign-intercept
2396 Warn whenever an Objective-C assignment is being intercepted by the
2400 @opindex Wno-protocol
2401 If a class is declared to implement a protocol, a warning is issued for
2402 every method in the protocol that is not implemented by the class. The
2403 default behavior is to issue a warning for every method not explicitly
2404 implemented in the class, even if a method implementation is inherited
2405 from the superclass. If you use the @option{-Wno-protocol} option, then
2406 methods inherited from the superclass are considered to be implemented,
2407 and no warning is issued for them.
2411 Warn if multiple methods of different types for the same selector are
2412 found during compilation. The check is performed on the list of methods
2413 in the final stage of compilation. Additionally, a check is performed
2414 for each selector appearing in a @code{@@selector(@dots{})}
2415 expression, and a corresponding method for that selector has been found
2416 during compilation. Because these checks scan the method table only at
2417 the end of compilation, these warnings are not produced if the final
2418 stage of compilation is not reached, for example because an error is
2419 found during compilation, or because the @option{-fsyntax-only} option is
2422 @item -Wstrict-selector-match
2423 @opindex Wstrict-selector-match
2424 Warn if multiple methods with differing argument and/or return types are
2425 found for a given selector when attempting to send a message using this
2426 selector to a receiver of type @code{id} or @code{Class}. When this flag
2427 is off (which is the default behavior), the compiler will omit such warnings
2428 if any differences found are confined to types which share the same size
2431 @item -Wundeclared-selector
2432 @opindex Wundeclared-selector
2433 Warn if a @code{@@selector(@dots{})} expression referring to an
2434 undeclared selector is found. A selector is considered undeclared if no
2435 method with that name has been declared before the
2436 @code{@@selector(@dots{})} expression, either explicitly in an
2437 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2438 an @code{@@implementation} section. This option always performs its
2439 checks as soon as a @code{@@selector(@dots{})} expression is found,
2440 while @option{-Wselector} only performs its checks in the final stage of
2441 compilation. This also enforces the coding style convention
2442 that methods and selectors must be declared before being used.
2444 @item -print-objc-runtime-info
2445 @opindex print-objc-runtime-info
2446 Generate C header describing the largest structure that is passed by
2451 @node Language Independent Options
2452 @section Options to Control Diagnostic Messages Formatting
2453 @cindex options to control diagnostics formatting
2454 @cindex diagnostic messages
2455 @cindex message formatting
2457 Traditionally, diagnostic messages have been formatted irrespective of
2458 the output device's aspect (e.g.@: its width, @dots{}). The options described
2459 below can be used to control the diagnostic messages formatting
2460 algorithm, e.g.@: how many characters per line, how often source location
2461 information should be reported. Right now, only the C++ front end can
2462 honor these options. However it is expected, in the near future, that
2463 the remaining front ends would be able to digest them correctly.
2466 @item -fmessage-length=@var{n}
2467 @opindex fmessage-length
2468 Try to format error messages so that they fit on lines of about @var{n}
2469 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2470 the front ends supported by GCC@. If @var{n} is zero, then no
2471 line-wrapping will be done; each error message will appear on a single
2474 @opindex fdiagnostics-show-location
2475 @item -fdiagnostics-show-location=once
2476 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2477 reporter to emit @emph{once} source location information; that is, in
2478 case the message is too long to fit on a single physical line and has to
2479 be wrapped, the source location won't be emitted (as prefix) again,
2480 over and over, in subsequent continuation lines. This is the default
2483 @item -fdiagnostics-show-location=every-line
2484 Only meaningful in line-wrapping mode. Instructs the diagnostic
2485 messages reporter to emit the same source location information (as
2486 prefix) for physical lines that result from the process of breaking
2487 a message which is too long to fit on a single line.
2489 @item -fdiagnostics-show-option
2490 @opindex fdiagnostics-show-option
2491 This option instructs the diagnostic machinery to add text to each
2492 diagnostic emitted, which indicates which command line option directly
2493 controls that diagnostic, when such an option is known to the
2494 diagnostic machinery.
2496 @item -Wcoverage-mismatch
2497 @opindex Wcoverage-mismatch
2498 Warn if feedback profiles do not match when using the
2499 @option{-fprofile-use} option.
2500 If a source file was changed between @option{-fprofile-gen} and
2501 @option{-fprofile-use}, the files with the profile feedback can fail
2502 to match the source file and GCC can not use the profile feedback
2503 information. By default, GCC emits an error message in this case.
2504 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2505 error. GCC does not use appropriate feedback profiles, so using this
2506 option can result in poorly optimized code. This option is useful
2507 only in the case of very minor changes such as bug fixes to an
2512 @node Warning Options
2513 @section Options to Request or Suppress Warnings
2514 @cindex options to control warnings
2515 @cindex warning messages
2516 @cindex messages, warning
2517 @cindex suppressing warnings
2519 Warnings are diagnostic messages that report constructions which
2520 are not inherently erroneous but which are risky or suggest there
2521 may have been an error.
2523 You can request many specific warnings with options beginning @samp{-W},
2524 for example @option{-Wimplicit} to request warnings on implicit
2525 declarations. Each of these specific warning options also has a
2526 negative form beginning @samp{-Wno-} to turn off warnings;
2527 for example, @option{-Wno-implicit}. This manual lists only one of the
2528 two forms, whichever is not the default.
2530 The following options control the amount and kinds of warnings produced
2531 by GCC; for further, language-specific options also refer to
2532 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2536 @cindex syntax checking
2538 @opindex fsyntax-only
2539 Check the code for syntax errors, but don't do anything beyond that.
2543 Issue all the warnings demanded by strict ISO C and ISO C++;
2544 reject all programs that use forbidden extensions, and some other
2545 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2546 version of the ISO C standard specified by any @option{-std} option used.
2548 Valid ISO C and ISO C++ programs should compile properly with or without
2549 this option (though a rare few will require @option{-ansi} or a
2550 @option{-std} option specifying the required version of ISO C)@. However,
2551 without this option, certain GNU extensions and traditional C and C++
2552 features are supported as well. With this option, they are rejected.
2554 @option{-pedantic} does not cause warning messages for use of the
2555 alternate keywords whose names begin and end with @samp{__}. Pedantic
2556 warnings are also disabled in the expression that follows
2557 @code{__extension__}. However, only system header files should use
2558 these escape routes; application programs should avoid them.
2559 @xref{Alternate Keywords}.
2561 Some users try to use @option{-pedantic} to check programs for strict ISO
2562 C conformance. They soon find that it does not do quite what they want:
2563 it finds some non-ISO practices, but not all---only those for which
2564 ISO C @emph{requires} a diagnostic, and some others for which
2565 diagnostics have been added.
2567 A feature to report any failure to conform to ISO C might be useful in
2568 some instances, but would require considerable additional work and would
2569 be quite different from @option{-pedantic}. We don't have plans to
2570 support such a feature in the near future.
2572 Where the standard specified with @option{-std} represents a GNU
2573 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2574 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2575 extended dialect is based. Warnings from @option{-pedantic} are given
2576 where they are required by the base standard. (It would not make sense
2577 for such warnings to be given only for features not in the specified GNU
2578 C dialect, since by definition the GNU dialects of C include all
2579 features the compiler supports with the given option, and there would be
2580 nothing to warn about.)
2582 @item -pedantic-errors
2583 @opindex pedantic-errors
2584 Like @option{-pedantic}, except that errors are produced rather than
2589 Inhibit all warning messages.
2593 Inhibit warning messages about the use of @samp{#import}.
2595 @item -Wchar-subscripts
2596 @opindex Wchar-subscripts
2597 Warn if an array subscript has type @code{char}. This is a common cause
2598 of error, as programmers often forget that this type is signed on some
2600 This warning is enabled by @option{-Wall}.
2604 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2605 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2606 This warning is enabled by @option{-Wall}.
2608 @item -Wfatal-errors
2609 @opindex Wfatal-errors
2610 This option causes the compiler to abort compilation on the first error
2611 occurred rather than trying to keep going and printing further error
2616 @opindex ffreestanding
2617 @opindex fno-builtin
2618 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2619 the arguments supplied have types appropriate to the format string
2620 specified, and that the conversions specified in the format string make
2621 sense. This includes standard functions, and others specified by format
2622 attributes (@pxref{Function Attributes}), in the @code{printf},
2623 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2624 not in the C standard) families (or other target-specific families).
2625 Which functions are checked without format attributes having been
2626 specified depends on the standard version selected, and such checks of
2627 functions without the attribute specified are disabled by
2628 @option{-ffreestanding} or @option{-fno-builtin}.
2630 The formats are checked against the format features supported by GNU
2631 libc version 2.2. These include all ISO C90 and C99 features, as well
2632 as features from the Single Unix Specification and some BSD and GNU
2633 extensions. Other library implementations may not support all these
2634 features; GCC does not support warning about features that go beyond a
2635 particular library's limitations. However, if @option{-pedantic} is used
2636 with @option{-Wformat}, warnings will be given about format features not
2637 in the selected standard version (but not for @code{strfmon} formats,
2638 since those are not in any version of the C standard). @xref{C Dialect
2639 Options,,Options Controlling C Dialect}.
2641 Since @option{-Wformat} also checks for null format arguments for
2642 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2644 @option{-Wformat} is included in @option{-Wall}. For more control over some
2645 aspects of format checking, the options @option{-Wformat-y2k},
2646 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2647 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2648 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2651 @opindex Wformat-y2k
2652 If @option{-Wformat} is specified, also warn about @code{strftime}
2653 formats which may yield only a two-digit year.
2655 @item -Wno-format-extra-args
2656 @opindex Wno-format-extra-args
2657 If @option{-Wformat} is specified, do not warn about excess arguments to a
2658 @code{printf} or @code{scanf} format function. The C standard specifies
2659 that such arguments are ignored.
2661 Where the unused arguments lie between used arguments that are
2662 specified with @samp{$} operand number specifications, normally
2663 warnings are still given, since the implementation could not know what
2664 type to pass to @code{va_arg} to skip the unused arguments. However,
2665 in the case of @code{scanf} formats, this option will suppress the
2666 warning if the unused arguments are all pointers, since the Single
2667 Unix Specification says that such unused arguments are allowed.
2669 @item -Wno-format-zero-length
2670 @opindex Wno-format-zero-length
2671 If @option{-Wformat} is specified, do not warn about zero-length formats.
2672 The C standard specifies that zero-length formats are allowed.
2674 @item -Wformat-nonliteral
2675 @opindex Wformat-nonliteral
2676 If @option{-Wformat} is specified, also warn if the format string is not a
2677 string literal and so cannot be checked, unless the format function
2678 takes its format arguments as a @code{va_list}.
2680 @item -Wformat-security
2681 @opindex Wformat-security
2682 If @option{-Wformat} is specified, also warn about uses of format
2683 functions that represent possible security problems. At present, this
2684 warns about calls to @code{printf} and @code{scanf} functions where the
2685 format string is not a string literal and there are no format arguments,
2686 as in @code{printf (foo);}. This may be a security hole if the format
2687 string came from untrusted input and contains @samp{%n}. (This is
2688 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2689 in future warnings may be added to @option{-Wformat-security} that are not
2690 included in @option{-Wformat-nonliteral}.)
2694 Enable @option{-Wformat} plus format checks not included in
2695 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2696 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2700 Warn about passing a null pointer for arguments marked as
2701 requiring a non-null value by the @code{nonnull} function attribute.
2703 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2704 can be disabled with the @option{-Wno-nonnull} option.
2706 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2708 Warn about uninitialized variables which are initialized with themselves.
2709 Note this option can only be used with the @option{-Wuninitialized} option,
2710 which in turn only works with @option{-O1} and above.
2712 For example, GCC will warn about @code{i} being uninitialized in the
2713 following snippet only when @option{-Winit-self} has been specified:
2724 @item -Wimplicit-int
2725 @opindex Wimplicit-int
2726 Warn when a declaration does not specify a type.
2727 This warning is enabled by @option{-Wall}.
2729 @item -Wimplicit-function-declaration
2730 @opindex Wimplicit-function-declaration
2731 @opindex Wno-implicit-function-declaration
2732 Give a warning whenever a function is used before being declared. In
2733 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2734 enabled by default and it is made into an error by
2735 @option{-pedantic-errors}. This warning is also enabled by
2740 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2741 This warning is enabled by @option{-Wall}.
2745 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2746 function with external linkage, returning int, taking either zero
2747 arguments, two, or three arguments of appropriate types.
2748 This warning is enabled by @option{-Wall}.
2750 @item -Wmissing-braces
2751 @opindex Wmissing-braces
2752 Warn if an aggregate or union initializer is not fully bracketed. In
2753 the following example, the initializer for @samp{a} is not fully
2754 bracketed, but that for @samp{b} is fully bracketed.
2757 int a[2][2] = @{ 0, 1, 2, 3 @};
2758 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2761 This warning is enabled by @option{-Wall}.
2763 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2764 @opindex Wmissing-include-dirs
2765 Warn if a user-supplied include directory does not exist.
2768 @opindex Wparentheses
2769 Warn if parentheses are omitted in certain contexts, such
2770 as when there is an assignment in a context where a truth value
2771 is expected, or when operators are nested whose precedence people
2772 often get confused about.
2774 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2775 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2776 interpretation from that of ordinary mathematical notation.
2778 Also warn about constructions where there may be confusion to which
2779 @code{if} statement an @code{else} branch belongs. Here is an example of
2794 In C/C++, every @code{else} branch belongs to the innermost possible
2795 @code{if} statement, which in this example is @code{if (b)}. This is
2796 often not what the programmer expected, as illustrated in the above
2797 example by indentation the programmer chose. When there is the
2798 potential for this confusion, GCC will issue a warning when this flag
2799 is specified. To eliminate the warning, add explicit braces around
2800 the innermost @code{if} statement so there is no way the @code{else}
2801 could belong to the enclosing @code{if}. The resulting code would
2818 This warning is enabled by @option{-Wall}.
2820 @item -Wsequence-point
2821 @opindex Wsequence-point
2822 Warn about code that may have undefined semantics because of violations
2823 of sequence point rules in the C and C++ standards.
2825 The C and C++ standards defines the order in which expressions in a C/C++
2826 program are evaluated in terms of @dfn{sequence points}, which represent
2827 a partial ordering between the execution of parts of the program: those
2828 executed before the sequence point, and those executed after it. These
2829 occur after the evaluation of a full expression (one which is not part
2830 of a larger expression), after the evaluation of the first operand of a
2831 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2832 function is called (but after the evaluation of its arguments and the
2833 expression denoting the called function), and in certain other places.
2834 Other than as expressed by the sequence point rules, the order of
2835 evaluation of subexpressions of an expression is not specified. All
2836 these rules describe only a partial order rather than a total order,
2837 since, for example, if two functions are called within one expression
2838 with no sequence point between them, the order in which the functions
2839 are called is not specified. However, the standards committee have
2840 ruled that function calls do not overlap.
2842 It is not specified when between sequence points modifications to the
2843 values of objects take effect. Programs whose behavior depends on this
2844 have undefined behavior; the C and C++ standards specify that ``Between
2845 the previous and next sequence point an object shall have its stored
2846 value modified at most once by the evaluation of an expression.
2847 Furthermore, the prior value shall be read only to determine the value
2848 to be stored.''. If a program breaks these rules, the results on any
2849 particular implementation are entirely unpredictable.
2851 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2852 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2853 diagnosed by this option, and it may give an occasional false positive
2854 result, but in general it has been found fairly effective at detecting
2855 this sort of problem in programs.
2857 The standard is worded confusingly, therefore there is some debate
2858 over the precise meaning of the sequence point rules in subtle cases.
2859 Links to discussions of the problem, including proposed formal
2860 definitions, may be found on the GCC readings page, at
2861 @w{@uref{http://gcc.gnu.org/readings.html}}.
2863 This warning is enabled by @option{-Wall} for C and C++.
2866 @opindex Wreturn-type
2867 @opindex Wno-return-type
2868 Warn whenever a function is defined with a return-type that defaults
2869 to @code{int}. Also warn about any @code{return} statement with no
2870 return-value in a function whose return-type is not @code{void}
2871 (falling off the end of the function body is considered returning
2872 without a value), and about a @code{return} statement with a
2873 expression in a function whose return-type is @code{void}.
2875 Also warn if the return type of a function has a type qualifier
2876 such as @code{const}. For ISO C such a type qualifier has no effect,
2877 since the value returned by a function is not an lvalue.
2878 For C++, the warning is only emitted for scalar types or @code{void}.
2879 ISO C prohibits qualified @code{void} return types on function
2880 definitions, so such return types always receive a warning
2881 even without this option.
2883 For C++, a function without return type always produces a diagnostic
2884 message, even when @option{-Wno-return-type} is specified. The only
2885 exceptions are @samp{main} and functions defined in system headers.
2887 This warning is enabled by @option{-Wall}.
2891 Warn whenever a @code{switch} statement has an index of enumerated type
2892 and lacks a @code{case} for one or more of the named codes of that
2893 enumeration. (The presence of a @code{default} label prevents this
2894 warning.) @code{case} labels outside the enumeration range also
2895 provoke warnings when this option is used.
2896 This warning is enabled by @option{-Wall}.
2898 @item -Wswitch-default
2899 @opindex Wswitch-switch
2900 Warn whenever a @code{switch} statement does not have a @code{default}
2904 @opindex Wswitch-enum
2905 Warn whenever a @code{switch} statement has an index of enumerated type
2906 and lacks a @code{case} for one or more of the named codes of that
2907 enumeration. @code{case} labels outside the enumeration range also
2908 provoke warnings when this option is used.
2912 Warn if any trigraphs are encountered that might change the meaning of
2913 the program (trigraphs within comments are not warned about).
2914 This warning is enabled by @option{-Wall}.
2916 @item -Wunused-function
2917 @opindex Wunused-function
2918 Warn whenever a static function is declared but not defined or a
2919 non-inline static function is unused.
2920 This warning is enabled by @option{-Wall}.
2922 @item -Wunused-label
2923 @opindex Wunused-label
2924 Warn whenever a label is declared but not used.
2925 This warning is enabled by @option{-Wall}.
2927 To suppress this warning use the @samp{unused} attribute
2928 (@pxref{Variable Attributes}).
2930 @item -Wunused-parameter
2931 @opindex Wunused-parameter
2932 Warn whenever a function parameter is unused aside from its declaration.
2934 To suppress this warning use the @samp{unused} attribute
2935 (@pxref{Variable Attributes}).
2937 @item -Wunused-variable
2938 @opindex Wunused-variable
2939 Warn whenever a local variable or non-constant static variable is unused
2940 aside from its declaration.
2941 This warning is enabled by @option{-Wall}.
2943 To suppress this warning use the @samp{unused} attribute
2944 (@pxref{Variable Attributes}).
2946 @item -Wunused-value
2947 @opindex Wunused-value
2948 Warn whenever a statement computes a result that is explicitly not
2949 used. To suppress this warning cast the unused expression to
2950 @samp{void}. This includes an expression-statement or the left-hand
2951 side of a comma expression that contains no side effects. For example,
2952 an expression such as @samp{x[i,j]} will cause a warning, while
2953 @samp{x[(void)i,j]} will not.
2955 This warning is enabled by @option{-Wall}.
2959 All the above @option{-Wunused} options combined.
2961 In order to get a warning about an unused function parameter, you must
2962 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2963 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2965 @item -Wuninitialized
2966 @opindex Wuninitialized
2967 Warn if an automatic variable is used without first being initialized or
2968 if a variable may be clobbered by a @code{setjmp} call.
2970 These warnings are possible only in optimizing compilation,
2971 because they require data flow information that is computed only
2972 when optimizing. If you do not specify @option{-O}, you will not get
2973 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2974 requiring @option{-O}.
2976 If you want to warn about code which uses the uninitialized value of the
2977 variable in its own initializer, use the @option{-Winit-self} option.
2979 These warnings occur for individual uninitialized or clobbered
2980 elements of structure, union or array variables as well as for
2981 variables which are uninitialized or clobbered as a whole. They do
2982 not occur for variables or elements declared @code{volatile}. Because
2983 these warnings depend on optimization, the exact variables or elements
2984 for which there are warnings will depend on the precise optimization
2985 options and version of GCC used.
2987 Note that there may be no warning about a variable that is used only
2988 to compute a value that itself is never used, because such
2989 computations may be deleted by data flow analysis before the warnings
2992 These warnings are made optional because GCC is not smart
2993 enough to see all the reasons why the code might be correct
2994 despite appearing to have an error. Here is one example of how
3015 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3016 always initialized, but GCC doesn't know this. Here is
3017 another common case:
3022 if (change_y) save_y = y, y = new_y;
3024 if (change_y) y = save_y;
3029 This has no bug because @code{save_y} is used only if it is set.
3031 @cindex @code{longjmp} warnings
3032 This option also warns when a non-volatile automatic variable might be
3033 changed by a call to @code{longjmp}. These warnings as well are possible
3034 only in optimizing compilation.
3036 The compiler sees only the calls to @code{setjmp}. It cannot know
3037 where @code{longjmp} will be called; in fact, a signal handler could
3038 call it at any point in the code. As a result, you may get a warning
3039 even when there is in fact no problem because @code{longjmp} cannot
3040 in fact be called at the place which would cause a problem.
3042 Some spurious warnings can be avoided if you declare all the functions
3043 you use that never return as @code{noreturn}. @xref{Function
3046 This warning is enabled by @option{-Wall}.
3048 @item -Wunknown-pragmas
3049 @opindex Wunknown-pragmas
3050 @cindex warning for unknown pragmas
3051 @cindex unknown pragmas, warning
3052 @cindex pragmas, warning of unknown
3053 Warn when a #pragma directive is encountered which is not understood by
3054 GCC@. If this command line option is used, warnings will even be issued
3055 for unknown pragmas in system header files. This is not the case if
3056 the warnings were only enabled by the @option{-Wall} command line option.
3059 @opindex Wno-pragmas
3061 Do not warn about misuses of pragmas, such as incorrect parameters,
3062 invalid syntax, or conflicts between pragmas. See also
3063 @samp{-Wunknown-pragmas}.
3065 @item -Wstrict-aliasing
3066 @opindex Wstrict-aliasing
3067 This option is only active when @option{-fstrict-aliasing} is active.
3068 It warns about code which might break the strict aliasing rules that the
3069 compiler is using for optimization. The warning does not catch all
3070 cases, but does attempt to catch the more common pitfalls. It is
3071 included in @option{-Wall}.
3072 It is equivalent to -Wstrict-aliasing=3
3074 @item -Wstrict-aliasing=n
3075 @opindex Wstrict-aliasing=n
3076 This option is only active when @option{-fstrict-aliasing} is active.
3077 It warns about code which might break the strict aliasing rules that the
3078 compiler is using for optimization.
3079 Higher levels correspond to higher accuracy (fewer false positives).
3080 Higher levels also correspond to more effort, similar to the way -O works.
3081 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3084 Level 1: Most aggressive, quick, least accurate.
3085 Possibly useful when higher levels
3086 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3087 false negatives. However, it has many false positives.
3088 Warns for all pointer conversions between possibly incompatible types,
3089 even if never dereferenced. Runs in the frontend only.
3091 Level 2: Aggressive, quick, not too precise.
3092 May still have many false positives (not as many as level 1 though),
3093 and few false negatives (but possibly more than level 1).
3094 Unlike level 1, it only warns when an address is taken. Warns about
3095 incomplete types. Runs in the frontend only.
3097 Level 3 (default for @option{-Wstrict-aliasing}):
3098 Should have very few false positives and few false
3099 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3100 Takes care of the common punn+dereference pattern in the frontend:
3101 @code{*(int*)&some_float}.
3102 If optimization is enabled, it also runs in the backend, where it deals
3103 with multiple statement cases using flow-sensitive points-to information.
3104 Only warns when the converted pointer is dereferenced.
3105 Does not warn about incomplete types.
3107 @item -Wstrict-overflow
3108 @item -Wstrict-overflow=@var{n}
3109 @opindex Wstrict-overflow
3110 This option is only active when @option{-fstrict-overflow} is active.
3111 It warns about cases where the compiler optimizes based on the
3112 assumption that signed overflow does not occur. Note that it does not
3113 warn about all cases where the code might overflow: it only warns
3114 about cases where the compiler implements some optimization. Thus
3115 this warning depends on the optimization level.
3117 An optimization which assumes that signed overflow does not occur is
3118 perfectly safe if the values of the variables involved are such that
3119 overflow never does, in fact, occur. Therefore this warning can
3120 easily give a false positive: a warning about code which is not
3121 actually a problem. To help focus on important issues, several
3122 warning levels are defined. No warnings are issued for the use of
3123 undefined signed overflow when estimating how many iterations a loop
3124 will require, in particular when determining whether a loop will be
3128 @item -Wstrict-overflow=1
3129 Warn about cases which are both questionable and easy to avoid. For
3130 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3131 compiler will simplify this to @code{1}. This level of
3132 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3133 are not, and must be explicitly requested.
3135 @item -Wstrict-overflow=2
3136 Also warn about other cases where a comparison is simplified to a
3137 constant. For example: @code{abs (x) >= 0}. This can only be
3138 simplified when @option{-fstrict-overflow} is in effect, because
3139 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3140 zero. @option{-Wstrict-overflow} (with no level) is the same as
3141 @option{-Wstrict-overflow=2}.
3143 @item -Wstrict-overflow=3
3144 Also warn about other cases where a comparison is simplified. For
3145 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3147 @item -Wstrict-overflow=4
3148 Also warn about other simplifications not covered by the above cases.
3149 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3151 @item -Wstrict-overflow=5
3152 Also warn about cases where the compiler reduces the magnitude of a
3153 constant involved in a comparison. For example: @code{x + 2 > y} will
3154 be simplified to @code{x + 1 >= y}. This is reported only at the
3155 highest warning level because this simplification applies to many
3156 comparisons, so this warning level will give a very large number of
3160 @item -Warray-bounds
3161 @opindex Wno-array-bounds
3162 @opindex Warray-bounds
3163 This option is only active when @option{-ftree-vrp} is active
3164 (default for -O2 and above). It warns about subscripts to arrays
3165 that are always out of bounds. This warning is enabled by @option{-Wall}.
3169 All of the above @samp{-W} options combined. This enables all the
3170 warnings about constructions that some users consider questionable, and
3171 that are easy to avoid (or modify to prevent the warning), even in
3172 conjunction with macros. This also enables some language-specific
3173 warnings described in @ref{C++ Dialect Options} and
3174 @ref{Objective-C and Objective-C++ Dialect Options}.
3177 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
3178 Some of them warn about constructions that users generally do not
3179 consider questionable, but which occasionally you might wish to check
3180 for; others warn about constructions that are necessary or hard to avoid
3181 in some cases, and there is no simple way to modify the code to suppress
3188 (This option used to be called @option{-W}. The older name is still
3189 supported, but the newer name is more descriptive.) Print extra warning
3190 messages for these events:
3194 Warn if a comparison is always true or always false due to the limited
3195 range of the data type, but do not warn for constant expressions. For
3196 example, warn if an unsigned variable is compared against zero with
3197 @samp{<} or @samp{>=}. This warning can be independently controlled
3198 by @option{-Wtype-limits}.
3201 Storage-class specifiers like @code{static} are not the first things
3202 in a declaration. According to the C Standard, this usage is
3203 obsolescent. This warning can be independently controlled by
3204 @option{-Wold-style-declaration}.
3207 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
3211 A comparison between signed and unsigned values could produce an
3212 incorrect result when the signed value is converted to unsigned.
3213 (But don't warn if @option{-Wno-sign-compare} is also specified.)
3216 An aggregate has an initializer which does not initialize all members.
3217 This warning can be independently controlled by
3218 @option{-Wmissing-field-initializers}.
3221 An initialized field without side effects is overridden when using
3222 designated initializers (@pxref{Designated Inits, , Designated
3223 Initializers}). This warning can be independently controlled by
3224 @option{-Woverride-init}.
3227 A function parameter is declared without a type specifier in K&R-style
3228 functions. This warning can be independently controlled by
3229 @option{-Wmissing-parameter-type}.
3232 An empty body occurs in an @samp{if}, @samp{else} or
3233 @samp{do while} statement. This warning can be independently
3234 controlled by @option{-Wempty-body}.
3236 @item @r{(C++ only)}
3237 An empty body occurs in a @samp{while} or @samp{for} statement with no
3238 whitespacing before the semicolon. This warning can be independently
3239 controlled by @option{-Wempty-body}.
3242 A pointer is compared against integer zero with @samp{<}, @samp{<=},
3243 @samp{>}, or @samp{>=}.
3246 A variable might be changed by @samp{longjmp} or @samp{vfork}.
3247 This warning can be independently controlled by @option{-Wclobbered}.
3249 @item @r{(C++ only)}
3250 An enumerator and a non-enumerator both appear in a conditional expression.
3252 @item @r{(C++ only)}
3253 A non-static reference or non-static @samp{const} member appears in a
3254 class without constructors.
3256 @item @r{(C++ only)}
3257 Ambiguous virtual bases.
3259 @item @r{(C++ only)}
3260 Subscripting an array which has been declared @samp{register}.
3262 @item @r{(C++ only)}
3263 Taking the address of a variable which has been declared @samp{register}.
3265 @item @r{(C++ only)}
3266 A base class is not initialized in a derived class' copy constructor.
3269 @item -Wno-div-by-zero
3270 @opindex Wno-div-by-zero
3271 @opindex Wdiv-by-zero
3272 Do not warn about compile-time integer division by zero. Floating point
3273 division by zero is not warned about, as it can be a legitimate way of
3274 obtaining infinities and NaNs.
3276 @item -Wsystem-headers
3277 @opindex Wsystem-headers
3278 @cindex warnings from system headers
3279 @cindex system headers, warnings from
3280 Print warning messages for constructs found in system header files.
3281 Warnings from system headers are normally suppressed, on the assumption
3282 that they usually do not indicate real problems and would only make the
3283 compiler output harder to read. Using this command line option tells
3284 GCC to emit warnings from system headers as if they occurred in user
3285 code. However, note that using @option{-Wall} in conjunction with this
3286 option will @emph{not} warn about unknown pragmas in system
3287 headers---for that, @option{-Wunknown-pragmas} must also be used.
3290 @opindex Wfloat-equal
3291 Warn if floating point values are used in equality comparisons.
3293 The idea behind this is that sometimes it is convenient (for the
3294 programmer) to consider floating-point values as approximations to
3295 infinitely precise real numbers. If you are doing this, then you need
3296 to compute (by analyzing the code, or in some other way) the maximum or
3297 likely maximum error that the computation introduces, and allow for it
3298 when performing comparisons (and when producing output, but that's a
3299 different problem). In particular, instead of testing for equality, you
3300 would check to see whether the two values have ranges that overlap; and
3301 this is done with the relational operators, so equality comparisons are
3304 @item -Wtraditional @r{(C only)}
3305 @opindex Wtraditional
3306 Warn about certain constructs that behave differently in traditional and
3307 ISO C@. Also warn about ISO C constructs that have no traditional C
3308 equivalent, and/or problematic constructs which should be avoided.
3312 Macro parameters that appear within string literals in the macro body.
3313 In traditional C macro replacement takes place within string literals,
3314 but does not in ISO C@.
3317 In traditional C, some preprocessor directives did not exist.
3318 Traditional preprocessors would only consider a line to be a directive
3319 if the @samp{#} appeared in column 1 on the line. Therefore
3320 @option{-Wtraditional} warns about directives that traditional C
3321 understands but would ignore because the @samp{#} does not appear as the
3322 first character on the line. It also suggests you hide directives like
3323 @samp{#pragma} not understood by traditional C by indenting them. Some
3324 traditional implementations would not recognize @samp{#elif}, so it
3325 suggests avoiding it altogether.
3328 A function-like macro that appears without arguments.
3331 The unary plus operator.
3334 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3335 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3336 constants.) Note, these suffixes appear in macros defined in the system
3337 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3338 Use of these macros in user code might normally lead to spurious
3339 warnings, however GCC's integrated preprocessor has enough context to
3340 avoid warning in these cases.
3343 A function declared external in one block and then used after the end of
3347 A @code{switch} statement has an operand of type @code{long}.
3350 A non-@code{static} function declaration follows a @code{static} one.
3351 This construct is not accepted by some traditional C compilers.
3354 The ISO type of an integer constant has a different width or
3355 signedness from its traditional type. This warning is only issued if
3356 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3357 typically represent bit patterns, are not warned about.
3360 Usage of ISO string concatenation is detected.
3363 Initialization of automatic aggregates.
3366 Identifier conflicts with labels. Traditional C lacks a separate
3367 namespace for labels.
3370 Initialization of unions. If the initializer is zero, the warning is
3371 omitted. This is done under the assumption that the zero initializer in
3372 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3373 initializer warnings and relies on default initialization to zero in the
3377 Conversions by prototypes between fixed/floating point values and vice
3378 versa. The absence of these prototypes when compiling with traditional
3379 C would cause serious problems. This is a subset of the possible
3380 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3383 Use of ISO C style function definitions. This warning intentionally is
3384 @emph{not} issued for prototype declarations or variadic functions
3385 because these ISO C features will appear in your code when using
3386 libiberty's traditional C compatibility macros, @code{PARAMS} and
3387 @code{VPARAMS}. This warning is also bypassed for nested functions
3388 because that feature is already a GCC extension and thus not relevant to
3389 traditional C compatibility.
3392 @item -Wtraditional-conversion @r{(C only)}
3393 @opindex Wtraditional-conversion
3394 Warn if a prototype causes a type conversion that is different from what
3395 would happen to the same argument in the absence of a prototype. This
3396 includes conversions of fixed point to floating and vice versa, and
3397 conversions changing the width or signedness of a fixed point argument
3398 except when the same as the default promotion.
3400 @item -Wdeclaration-after-statement @r{(C only)}
3401 @opindex Wdeclaration-after-statement
3402 Warn when a declaration is found after a statement in a block. This
3403 construct, known from C++, was introduced with ISO C99 and is by default
3404 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3405 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3409 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3411 @item -Wno-endif-labels
3412 @opindex Wno-endif-labels
3413 @opindex Wendif-labels
3414 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3418 Warn whenever a local variable shadows another local variable, parameter or
3419 global variable or whenever a built-in function is shadowed.
3421 @item -Wlarger-than-@var{len}
3422 @opindex Wlarger-than
3423 Warn whenever an object of larger than @var{len} bytes is defined.
3425 @item -Wunsafe-loop-optimizations
3426 @opindex Wunsafe-loop-optimizations
3427 Warn if the loop cannot be optimized because the compiler could not
3428 assume anything on the bounds of the loop indices. With
3429 @option{-funsafe-loop-optimizations} warn if the compiler made
3432 @item -Wpointer-arith
3433 @opindex Wpointer-arith
3434 Warn about anything that depends on the ``size of'' a function type or
3435 of @code{void}. GNU C assigns these types a size of 1, for
3436 convenience in calculations with @code{void *} pointers and pointers
3437 to functions. In C++, warn also when an arithmetic operation involves
3438 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3441 @opindex Wtype-limits
3442 @opindex Wno-type-limits
3443 Warn if a comparison is always true or always false due to the limited
3444 range of the data type, but do not warn for constant expressions. For
3445 example, warn if an unsigned variable is compared against zero with
3446 @samp{<} or @samp{>=}. This warning is also enabled by
3449 @item -Wbad-function-cast @r{(C only)}
3450 @opindex Wbad-function-cast
3451 Warn whenever a function call is cast to a non-matching type.
3452 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3455 Warn about ISO C constructs that are outside of the common subset of
3456 ISO C and ISO C++, e.g.@: request for implicit conversion from
3457 @code{void *} to a pointer to non-@code{void} type.
3459 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3460 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3461 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3462 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3466 Warn whenever a pointer is cast so as to remove a type qualifier from
3467 the target type. For example, warn if a @code{const char *} is cast
3468 to an ordinary @code{char *}.
3471 @opindex Wcast-align
3472 Warn whenever a pointer is cast such that the required alignment of the
3473 target is increased. For example, warn if a @code{char *} is cast to
3474 an @code{int *} on machines where integers can only be accessed at
3475 two- or four-byte boundaries.
3477 @item -Wwrite-strings
3478 @opindex Wwrite-strings
3479 When compiling C, give string constants the type @code{const
3480 char[@var{length}]} so that
3481 copying the address of one into a non-@code{const} @code{char *}
3482 pointer will get a warning; when compiling C++, warn about the
3483 deprecated conversion from string literals to @code{char *}. This
3484 warning, by default, is enabled for C++ programs.
3485 These warnings will help you find at
3486 compile time code that can try to write into a string constant, but
3487 only if you have been very careful about using @code{const} in
3488 declarations and prototypes. Otherwise, it will just be a nuisance;
3489 this is why we did not make @option{-Wall} request these warnings.
3493 Warn for variables that might be changed by @samp{longjmp} or
3494 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3497 @opindex Wconversion
3498 @opindex Wno-conversion
3499 Warn for implicit conversions that may alter a value. This includes
3500 conversions between real and integer, like @code{abs (x)} when
3501 @code{x} is @code{double}; conversions between signed and unsigned,
3502 like @code{unsigned ui = -1}; and conversions to smaller types, like
3503 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3504 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3505 changed by the conversion like in @code{abs (2.0)}. Warnings about
3506 conversions between signed and unsigned integers can be disabled by
3507 using @option{-Wno-sign-conversion}.
3509 For C++, also warn for conversions between @code{NULL} and non-pointer
3510 types; confusing overload resolution for user-defined conversions; and
3511 conversions that will never use a type conversion operator:
3512 conversions to @code{void}, the same type, a base class or a reference
3513 to them. Warnings about conversions between signed and unsigned
3514 integers are disabled by default in C++ unless
3515 @option{-Wsign-conversion} is explicitly enabled.
3518 @opindex Wempty-body
3519 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3520 while} statement. Additionally, in C++, warn when an empty body occurs
3521 in a @samp{while} or @samp{for} statement with no whitespacing before
3522 the semicolon. This warning is also enabled by @option{-Wextra}.
3524 @item -Wsign-compare
3525 @opindex Wsign-compare
3526 @cindex warning for comparison of signed and unsigned values
3527 @cindex comparison of signed and unsigned values, warning
3528 @cindex signed and unsigned values, comparison warning
3529 Warn when a comparison between signed and unsigned values could produce
3530 an incorrect result when the signed value is converted to unsigned.
3531 This warning is also enabled by @option{-Wextra}; to get the other warnings
3532 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3534 @item -Wsign-conversion
3535 @opindex Wsign-conversion
3536 @opindex Wno-sign-conversion
3537 Warn for implicit conversions that may change the sign of an integer
3538 value, like assigning a signed integer expression to an unsigned
3539 integer variable. An explicit cast silences the warning. In C, this
3540 option is enabled also by @option{-Wconversion}.
3544 @opindex Wno-address
3545 Warn about suspicious uses of memory addresses. These include using
3546 the address of a function in a conditional expression, such as
3547 @code{void func(void); if (func)}, and comparisons against the memory
3548 address of a string literal, such as @code{if (x == "abc")}. Such
3549 uses typically indicate a programmer error: the address of a function
3550 always evaluates to true, so their use in a conditional usually
3551 indicate that the programmer forgot the parentheses in a function
3552 call; and comparisons against string literals result in unspecified
3553 behavior and are not portable in C, so they usually indicate that the
3554 programmer intended to use @code{strcmp}. This warning is enabled by
3558 @opindex Wlogical-op
3559 @opindex Wno-logical-op
3560 Warn about suspicious uses of logical operators in expressions.
3561 This includes using logical operators in contexts where a
3562 bit-wise operator is likely to be expected.
3564 @item -Waggregate-return
3565 @opindex Waggregate-return
3566 Warn if any functions that return structures or unions are defined or
3567 called. (In languages where you can return an array, this also elicits
3570 @item -Wno-attributes
3571 @opindex Wno-attributes
3572 @opindex Wattributes
3573 Do not warn if an unexpected @code{__attribute__} is used, such as
3574 unrecognized attributes, function attributes applied to variables,
3575 etc. This will not stop errors for incorrect use of supported
3578 @item -Wstrict-prototypes @r{(C only)}
3579 @opindex Wstrict-prototypes
3580 Warn if a function is declared or defined without specifying the
3581 argument types. (An old-style function definition is permitted without
3582 a warning if preceded by a declaration which specifies the argument
3585 @item -Wold-style-declaration @r{(C only)}
3586 @opindex Wold-style-declaration
3587 Warn for obsolescent usages, according to the C Standard, in a
3588 declaration. For example, warn if storage-class specifiers like
3589 @code{static} are not the first things in a declaration. This warning
3590 is also enabled by @option{-Wextra}.
3592 @item -Wold-style-definition @r{(C only)}
3593 @opindex Wold-style-definition
3594 Warn if an old-style function definition is used. A warning is given
3595 even if there is a previous prototype.
3597 @item -Wmissing-parameter-type @r{(C only)}
3598 @opindex Wmissing-parameter-type
3599 A function parameter is declared without a type specifier in K&R-style
3606 This warning is also enabled by @option{-Wextra}.
3608 @item -Wmissing-prototypes @r{(C only)}
3609 @opindex Wmissing-prototypes
3610 Warn if a global function is defined without a previous prototype
3611 declaration. This warning is issued even if the definition itself
3612 provides a prototype. The aim is to detect global functions that fail
3613 to be declared in header files.
3615 @item -Wmissing-declarations @r{(C and C++ only)}
3616 @opindex Wmissing-declarations
3617 Warn if a global function is defined without a previous declaration.
3618 Do so even if the definition itself provides a prototype.
3619 Use this option to detect global functions that are not declared in
3620 header files. In C++, no warnings are issued for function templates,
3621 or for inline functions, or for functions in anonymous namespaces.
3623 @item -Wmissing-field-initializers
3624 @opindex Wmissing-field-initializers
3627 Warn if a structure's initializer has some fields missing. For
3628 example, the following code would cause such a warning, because
3629 @code{x.h} is implicitly zero:
3632 struct s @{ int f, g, h; @};
3633 struct s x = @{ 3, 4 @};
3636 This option does not warn about designated initializers, so the following
3637 modification would not trigger a warning:
3640 struct s @{ int f, g, h; @};
3641 struct s x = @{ .f = 3, .g = 4 @};
3644 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3645 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3647 @item -Wmissing-noreturn
3648 @opindex Wmissing-noreturn
3649 Warn about functions which might be candidates for attribute @code{noreturn}.
3650 Note these are only possible candidates, not absolute ones. Care should
3651 be taken to manually verify functions actually do not ever return before
3652 adding the @code{noreturn} attribute, otherwise subtle code generation
3653 bugs could be introduced. You will not get a warning for @code{main} in
3654 hosted C environments.
3656 @item -Wmissing-format-attribute
3657 @opindex Wmissing-format-attribute
3659 Warn about function pointers which might be candidates for @code{format}
3660 attributes. Note these are only possible candidates, not absolute ones.
3661 GCC will guess that function pointers with @code{format} attributes that
3662 are used in assignment, initialization, parameter passing or return
3663 statements should have a corresponding @code{format} attribute in the
3664 resulting type. I.e.@: the left-hand side of the assignment or
3665 initialization, the type of the parameter variable, or the return type
3666 of the containing function respectively should also have a @code{format}
3667 attribute to avoid the warning.
3669 GCC will also warn about function definitions which might be
3670 candidates for @code{format} attributes. Again, these are only
3671 possible candidates. GCC will guess that @code{format} attributes
3672 might be appropriate for any function that calls a function like
3673 @code{vprintf} or @code{vscanf}, but this might not always be the
3674 case, and some functions for which @code{format} attributes are
3675 appropriate may not be detected.
3677 @item -Wno-multichar
3678 @opindex Wno-multichar
3680 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3681 Usually they indicate a typo in the user's code, as they have
3682 implementation-defined values, and should not be used in portable code.
3684 @item -Wnormalized=<none|id|nfc|nfkc>
3685 @opindex Wnormalized
3688 @cindex character set, input normalization
3689 In ISO C and ISO C++, two identifiers are different if they are
3690 different sequences of characters. However, sometimes when characters
3691 outside the basic ASCII character set are used, you can have two
3692 different character sequences that look the same. To avoid confusion,
3693 the ISO 10646 standard sets out some @dfn{normalization rules} which
3694 when applied ensure that two sequences that look the same are turned into
3695 the same sequence. GCC can warn you if you are using identifiers which
3696 have not been normalized; this option controls that warning.
3698 There are four levels of warning that GCC supports. The default is
3699 @option{-Wnormalized=nfc}, which warns about any identifier which is
3700 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3701 recommended form for most uses.
3703 Unfortunately, there are some characters which ISO C and ISO C++ allow
3704 in identifiers that when turned into NFC aren't allowable as
3705 identifiers. That is, there's no way to use these symbols in portable
3706 ISO C or C++ and have all your identifiers in NFC.
3707 @option{-Wnormalized=id} suppresses the warning for these characters.
3708 It is hoped that future versions of the standards involved will correct
3709 this, which is why this option is not the default.
3711 You can switch the warning off for all characters by writing
3712 @option{-Wnormalized=none}. You would only want to do this if you
3713 were using some other normalization scheme (like ``D''), because
3714 otherwise you can easily create bugs that are literally impossible to see.
3716 Some characters in ISO 10646 have distinct meanings but look identical
3717 in some fonts or display methodologies, especially once formatting has
3718 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3719 LETTER N'', will display just like a regular @code{n} which has been
3720 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3721 normalization scheme to convert all these into a standard form as
3722 well, and GCC will warn if your code is not in NFKC if you use
3723 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3724 about every identifier that contains the letter O because it might be
3725 confused with the digit 0, and so is not the default, but may be
3726 useful as a local coding convention if the programming environment is
3727 unable to be fixed to display these characters distinctly.
3729 @item -Wno-deprecated-declarations
3730 @opindex Wno-deprecated-declarations
3731 Do not warn about uses of functions (@pxref{Function Attributes}),
3732 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3733 Attributes}) marked as deprecated by using the @code{deprecated}
3737 @opindex Wno-overflow
3738 Do not warn about compile-time overflow in constant expressions.
3740 @item -Woverride-init
3741 @opindex Woverride-init
3744 Warn if an initialized field without side effects is overridden when
3745 using designated initializers (@pxref{Designated Inits, , Designated
3748 This warning is included in @option{-Wextra}. To get other
3749 @option{-Wextra} warnings without this one, use @samp{-Wextra
3750 -Wno-override-init}.
3754 Warn if a structure is given the packed attribute, but the packed
3755 attribute has no effect on the layout or size of the structure.
3756 Such structures may be mis-aligned for little benefit. For
3757 instance, in this code, the variable @code{f.x} in @code{struct bar}
3758 will be misaligned even though @code{struct bar} does not itself
3759 have the packed attribute:
3766 @} __attribute__((packed));
3776 Warn if padding is included in a structure, either to align an element
3777 of the structure or to align the whole structure. Sometimes when this
3778 happens it is possible to rearrange the fields of the structure to
3779 reduce the padding and so make the structure smaller.
3781 @item -Wredundant-decls
3782 @opindex Wredundant-decls
3783 Warn if anything is declared more than once in the same scope, even in
3784 cases where multiple declaration is valid and changes nothing.
3786 @item -Wnested-externs @r{(C only)}
3787 @opindex Wnested-externs
3788 Warn if an @code{extern} declaration is encountered within a function.
3790 @item -Wunreachable-code
3791 @opindex Wunreachable-code
3792 Warn if the compiler detects that code will never be executed.
3794 This option is intended to warn when the compiler detects that at
3795 least a whole line of source code will never be executed, because
3796 some condition is never satisfied or because it is after a
3797 procedure that never returns.
3799 It is possible for this option to produce a warning even though there
3800 are circumstances under which part of the affected line can be executed,
3801 so care should be taken when removing apparently-unreachable code.
3803 For instance, when a function is inlined, a warning may mean that the
3804 line is unreachable in only one inlined copy of the function.
3806 This option is not made part of @option{-Wall} because in a debugging
3807 version of a program there is often substantial code which checks
3808 correct functioning of the program and is, hopefully, unreachable
3809 because the program does work. Another common use of unreachable
3810 code is to provide behavior which is selectable at compile-time.
3814 Warn if a function can not be inlined and it was declared as inline.
3815 Even with this option, the compiler will not warn about failures to
3816 inline functions declared in system headers.
3818 The compiler uses a variety of heuristics to determine whether or not
3819 to inline a function. For example, the compiler takes into account
3820 the size of the function being inlined and the amount of inlining
3821 that has already been done in the current function. Therefore,
3822 seemingly insignificant changes in the source program can cause the
3823 warnings produced by @option{-Winline} to appear or disappear.
3825 @item -Wno-invalid-offsetof @r{(C++ only)}
3826 @opindex Wno-invalid-offsetof
3827 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3828 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3829 to a non-POD type is undefined. In existing C++ implementations,
3830 however, @samp{offsetof} typically gives meaningful results even when
3831 applied to certain kinds of non-POD types. (Such as a simple
3832 @samp{struct} that fails to be a POD type only by virtue of having a
3833 constructor.) This flag is for users who are aware that they are
3834 writing nonportable code and who have deliberately chosen to ignore the
3837 The restrictions on @samp{offsetof} may be relaxed in a future version
3838 of the C++ standard.
3840 @item -Wno-int-to-pointer-cast @r{(C only)}
3841 @opindex Wno-int-to-pointer-cast
3842 Suppress warnings from casts to pointer type of an integer of a
3845 @item -Wno-pointer-to-int-cast @r{(C only)}
3846 @opindex Wno-pointer-to-int-cast
3847 Suppress warnings from casts from a pointer to an integer type of a
3851 @opindex Winvalid-pch
3852 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3853 the search path but can't be used.
3857 @opindex Wno-long-long
3858 Warn if @samp{long long} type is used. This is default. To inhibit
3859 the warning messages, use @option{-Wno-long-long}. Flags
3860 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3861 only when @option{-pedantic} flag is used.
3863 @item -Wvariadic-macros
3864 @opindex Wvariadic-macros
3865 @opindex Wno-variadic-macros
3866 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3867 alternate syntax when in pedantic ISO C99 mode. This is default.
3868 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3873 Warn if variable length array is used in the code.
3874 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
3875 the variable length array.
3877 @item -Wvolatile-register-var
3878 @opindex Wvolatile-register-var
3879 @opindex Wno-volatile-register-var
3880 Warn if a register variable is declared volatile. The volatile
3881 modifier does not inhibit all optimizations that may eliminate reads
3882 and/or writes to register variables.
3884 @item -Wdisabled-optimization
3885 @opindex Wdisabled-optimization
3886 Warn if a requested optimization pass is disabled. This warning does
3887 not generally indicate that there is anything wrong with your code; it
3888 merely indicates that GCC's optimizers were unable to handle the code
3889 effectively. Often, the problem is that your code is too big or too
3890 complex; GCC will refuse to optimize programs when the optimization
3891 itself is likely to take inordinate amounts of time.
3893 @item -Wpointer-sign
3894 @opindex Wpointer-sign
3895 @opindex Wno-pointer-sign
3896 Warn for pointer argument passing or assignment with different signedness.
3897 This option is only supported for C and Objective-C@. It is implied by
3898 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3899 @option{-Wno-pointer-sign}.
3903 Make all warnings into errors.
3907 Make the specified warning into an errors. The specifier for a
3908 warning is appended, for example @option{-Werror=switch} turns the
3909 warnings controlled by @option{-Wswitch} into errors. This switch
3910 takes a negative form, to be used to negate @option{-Werror} for
3911 specific warnings, for example @option{-Wno-error=switch} makes
3912 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3913 is in effect. You can use the @option{-fdiagnostics-show-option}
3914 option to have each controllable warning amended with the option which
3915 controls it, to determine what to use with this option.
3917 Note that specifying @option{-Werror=}@var{foo} automatically implies
3918 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3921 @item -Wstack-protector
3922 @opindex Wstack-protector
3923 This option is only active when @option{-fstack-protector} is active. It
3924 warns about functions that will not be protected against stack smashing.
3926 @item -Woverlength-strings
3927 @opindex Woverlength-strings
3928 Warn about string constants which are longer than the ``minimum
3929 maximum'' length specified in the C standard. Modern compilers
3930 generally allow string constants which are much longer than the
3931 standard's minimum limit, but very portable programs should avoid
3932 using longer strings.
3934 The limit applies @emph{after} string constant concatenation, and does
3935 not count the trailing NUL@. In C89, the limit was 509 characters; in
3936 C99, it was raised to 4095. C++98 does not specify a normative
3937 minimum maximum, so we do not diagnose overlength strings in C++@.
3939 This option is implied by @option{-pedantic}, and can be disabled with
3940 @option{-Wno-overlength-strings}.
3943 @node Debugging Options
3944 @section Options for Debugging Your Program or GCC
3945 @cindex options, debugging
3946 @cindex debugging information options
3948 GCC has various special options that are used for debugging
3949 either your program or GCC:
3954 Produce debugging information in the operating system's native format
3955 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3958 On most systems that use stabs format, @option{-g} enables use of extra
3959 debugging information that only GDB can use; this extra information
3960 makes debugging work better in GDB but will probably make other debuggers
3962 refuse to read the program. If you want to control for certain whether
3963 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3964 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3966 GCC allows you to use @option{-g} with
3967 @option{-O}. The shortcuts taken by optimized code may occasionally
3968 produce surprising results: some variables you declared may not exist
3969 at all; flow of control may briefly move where you did not expect it;
3970 some statements may not be executed because they compute constant
3971 results or their values were already at hand; some statements may
3972 execute in different places because they were moved out of loops.
3974 Nevertheless it proves possible to debug optimized output. This makes
3975 it reasonable to use the optimizer for programs that might have bugs.
3977 The following options are useful when GCC is generated with the
3978 capability for more than one debugging format.
3982 Produce debugging information for use by GDB@. This means to use the
3983 most expressive format available (DWARF 2, stabs, or the native format
3984 if neither of those are supported), including GDB extensions if at all
3989 Produce debugging information in stabs format (if that is supported),
3990 without GDB extensions. This is the format used by DBX on most BSD
3991 systems. On MIPS, Alpha and System V Release 4 systems this option
3992 produces stabs debugging output which is not understood by DBX or SDB@.
3993 On System V Release 4 systems this option requires the GNU assembler.
3995 @item -feliminate-unused-debug-symbols
3996 @opindex feliminate-unused-debug-symbols
3997 Produce debugging information in stabs format (if that is supported),
3998 for only symbols that are actually used.
4000 @item -femit-class-debug-always
4001 Instead of emitting debugging information for a C++ class in only one
4002 object file, emit it in all object files using the class. This option
4003 should be used only with debuggers that are unable to handle the way GCC
4004 normally emits debugging information for classes because using this
4005 option will increase the size of debugging information by as much as a
4010 Produce debugging information in stabs format (if that is supported),
4011 using GNU extensions understood only by the GNU debugger (GDB)@. The
4012 use of these extensions is likely to make other debuggers crash or
4013 refuse to read the program.
4017 Produce debugging information in COFF format (if that is supported).
4018 This is the format used by SDB on most System V systems prior to
4023 Produce debugging information in XCOFF format (if that is supported).
4024 This is the format used by the DBX debugger on IBM RS/6000 systems.
4028 Produce debugging information in XCOFF format (if that is supported),
4029 using GNU extensions understood only by the GNU debugger (GDB)@. The
4030 use of these extensions is likely to make other debuggers crash or
4031 refuse to read the program, and may cause assemblers other than the GNU
4032 assembler (GAS) to fail with an error.
4036 Produce debugging information in DWARF version 2 format (if that is
4037 supported). This is the format used by DBX on IRIX 6. With this
4038 option, GCC uses features of DWARF version 3 when they are useful;
4039 version 3 is upward compatible with version 2, but may still cause
4040 problems for older debuggers.
4044 Produce debugging information in VMS debug format (if that is
4045 supported). This is the format used by DEBUG on VMS systems.
4048 @itemx -ggdb@var{level}
4049 @itemx -gstabs@var{level}
4050 @itemx -gcoff@var{level}
4051 @itemx -gxcoff@var{level}
4052 @itemx -gvms@var{level}
4053 Request debugging information and also use @var{level} to specify how
4054 much information. The default level is 2.
4056 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4059 Level 1 produces minimal information, enough for making backtraces in
4060 parts of the program that you don't plan to debug. This includes
4061 descriptions of functions and external variables, but no information
4062 about local variables and no line numbers.
4064 Level 3 includes extra information, such as all the macro definitions
4065 present in the program. Some debuggers support macro expansion when
4066 you use @option{-g3}.
4068 @option{-gdwarf-2} does not accept a concatenated debug level, because
4069 GCC used to support an option @option{-gdwarf} that meant to generate
4070 debug information in version 1 of the DWARF format (which is very
4071 different from version 2), and it would have been too confusing. That
4072 debug format is long obsolete, but the option cannot be changed now.
4073 Instead use an additional @option{-g@var{level}} option to change the
4074 debug level for DWARF2.
4076 @item -feliminate-dwarf2-dups
4077 @opindex feliminate-dwarf2-dups
4078 Compress DWARF2 debugging information by eliminating duplicated
4079 information about each symbol. This option only makes sense when
4080 generating DWARF2 debugging information with @option{-gdwarf-2}.
4082 @item -femit-struct-debug-baseonly
4083 Emit debug information for struct-like types
4084 only when the base name of the compilation source file
4085 matches the base name of file in which the struct was defined.
4087 This option substantially reduces the size of debugging information,
4088 but at significant potential loss in type information to the debugger.
4089 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4090 See @option{-femit-struct-debug-detailed} for more detailed control.
4092 This option works only with DWARF 2.
4094 @item -femit-struct-debug-reduced
4095 Emit debug information for struct-like types
4096 only when the base name of the compilation source file
4097 matches the base name of file in which the type was defined,
4098 unless the struct is a template or defined in a system header.
4100 This option significantly reduces the size of debugging information,
4101 with some potential loss in type information to the debugger.
4102 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4103 See @option{-femit-struct-debug-detailed} for more detailed control.
4105 This option works only with DWARF 2.
4107 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4108 Specify the struct-like types
4109 for which the compiler will generate debug information.
4110 The intent is to reduce duplicate struct debug information
4111 between different object files within the same program.
4113 This option is a detailed version of
4114 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4115 which will serve for most needs.
4117 A specification has the syntax
4118 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4120 The optional first word limits the specification to
4121 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4122 A struct type is used directly when it is the type of a variable, member.
4123 Indirect uses arise through pointers to structs.
4124 That is, when use of an incomplete struct would be legal, the use is indirect.
4126 @samp{struct one direct; struct two * indirect;}.
4128 The optional second word limits the specification to
4129 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4130 Generic structs are a bit complicated to explain.
4131 For C++, these are non-explicit specializations of template classes,
4132 or non-template classes within the above.
4133 Other programming languages have generics,
4134 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4136 The third word specifies the source files for those
4137 structs for which the compiler will emit debug information.
4138 The values @samp{none} and @samp{any} have the normal meaning.
4139 The value @samp{base} means that
4140 the base of name of the file in which the type declaration appears
4141 must match the base of the name of the main compilation file.
4142 In practice, this means that
4143 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4144 but types declared in other header will not.
4145 The value @samp{sys} means those types satisfying @samp{base}
4146 or declared in system or compiler headers.
4148 You may need to experiment to determine the best settings for your application.
4150 The default is @samp{-femit-struct-debug-detailed=all}.
4152 This option works only with DWARF 2.
4154 @item -fdebug-prefix-map=@var{old}=@var{new}
4155 @opindex fdebug-prefix-map
4156 When compiling files in directory @file{@var{old}}, record debugging
4157 information describing them as in @file{@var{new}} instead.
4159 @cindex @command{prof}
4162 Generate extra code to write profile information suitable for the
4163 analysis program @command{prof}. You must use this option when compiling
4164 the source files you want data about, and you must also use it when
4167 @cindex @command{gprof}
4170 Generate extra code to write profile information suitable for the
4171 analysis program @command{gprof}. You must use this option when compiling
4172 the source files you want data about, and you must also use it when
4177 Makes the compiler print out each function name as it is compiled, and
4178 print some statistics about each pass when it finishes.
4181 @opindex ftime-report
4182 Makes the compiler print some statistics about the time consumed by each
4183 pass when it finishes.
4186 @opindex fmem-report
4187 Makes the compiler print some statistics about permanent memory
4188 allocation when it finishes.
4190 @item -fpre-ipa-mem-report
4191 @opindex fpre-ipa-mem-report
4192 @item -fpost-ipa-mem-report
4193 @opindex fpost-ipa-mem-report
4194 Makes the compiler print some statistics about permanent memory
4195 allocation before or after interprocedural optimization.
4197 @item -fprofile-arcs
4198 @opindex fprofile-arcs
4199 Add code so that program flow @dfn{arcs} are instrumented. During
4200 execution the program records how many times each branch and call is
4201 executed and how many times it is taken or returns. When the compiled
4202 program exits it saves this data to a file called
4203 @file{@var{auxname}.gcda} for each source file. The data may be used for
4204 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4205 test coverage analysis (@option{-ftest-coverage}). Each object file's
4206 @var{auxname} is generated from the name of the output file, if
4207 explicitly specified and it is not the final executable, otherwise it is
4208 the basename of the source file. In both cases any suffix is removed
4209 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4210 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4211 @xref{Cross-profiling}.
4213 @cindex @command{gcov}
4217 This option is used to compile and link code instrumented for coverage
4218 analysis. The option is a synonym for @option{-fprofile-arcs}
4219 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4220 linking). See the documentation for those options for more details.
4225 Compile the source files with @option{-fprofile-arcs} plus optimization
4226 and code generation options. For test coverage analysis, use the
4227 additional @option{-ftest-coverage} option. You do not need to profile
4228 every source file in a program.
4231 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4232 (the latter implies the former).
4235 Run the program on a representative workload to generate the arc profile
4236 information. This may be repeated any number of times. You can run
4237 concurrent instances of your program, and provided that the file system
4238 supports locking, the data files will be correctly updated. Also
4239 @code{fork} calls are detected and correctly handled (double counting
4243 For profile-directed optimizations, compile the source files again with
4244 the same optimization and code generation options plus
4245 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4246 Control Optimization}).
4249 For test coverage analysis, use @command{gcov} to produce human readable
4250 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4251 @command{gcov} documentation for further information.
4255 With @option{-fprofile-arcs}, for each function of your program GCC
4256 creates a program flow graph, then finds a spanning tree for the graph.
4257 Only arcs that are not on the spanning tree have to be instrumented: the
4258 compiler adds code to count the number of times that these arcs are
4259 executed. When an arc is the only exit or only entrance to a block, the
4260 instrumentation code can be added to the block; otherwise, a new basic
4261 block must be created to hold the instrumentation code.
4264 @item -ftest-coverage
4265 @opindex ftest-coverage
4266 Produce a notes file that the @command{gcov} code-coverage utility
4267 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4268 show program coverage. Each source file's note file is called
4269 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4270 above for a description of @var{auxname} and instructions on how to
4271 generate test coverage data. Coverage data will match the source files
4272 more closely, if you do not optimize.
4274 @item -fdbg-cnt-list
4275 @opindex fdbg-cnt-list
4276 Print the name and the counter upperbound for all debug counters.
4278 @item -fdbg-cnt=@var{counter-value-list}
4280 Set the internal debug counter upperbound. @var{counter-value-list}
4281 is a comma-separated list of @var{name}:@var{value} pairs
4282 which sets the upperbound of each debug counter @var{name} to @var{value}.
4283 All debug counters have the initial upperbound of @var{UINT_MAX},
4284 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4285 e.g. With -fdbg-cnt=dce:10,tail_call:0
4286 dbg_cnt(dce) will return true only for first 10 invocations
4287 and dbg_cnt(tail_call) will return false always.
4289 @item -d@var{letters}
4290 @item -fdump-rtl-@var{pass}
4292 Says to make debugging dumps during compilation at times specified by
4293 @var{letters}. This is used for debugging the RTL-based passes of the
4294 compiler. The file names for most of the dumps are made by appending a
4295 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4296 from the name of the output file, if explicitly specified and it is not
4297 an executable, otherwise it is the basename of the source file. These
4298 switches may have different effects when @option{-E} is used for
4301 Most debug dumps can be enabled either passing a letter to the @option{-d}
4302 option, or with a long @option{-fdump-rtl} switch; here are the possible
4303 letters for use in @var{letters} and @var{pass}, and their meanings:
4308 Annotate the assembler output with miscellaneous debugging information.
4311 @itemx -fdump-rtl-bbro
4313 @opindex fdump-rtl-bbro
4314 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4317 @itemx -fdump-rtl-combine
4319 @opindex fdump-rtl-combine
4320 Dump after the RTL instruction combination pass, to the file
4321 @file{@var{file}.129r.combine}.
4324 @itemx -fdump-rtl-ce1
4325 @itemx -fdump-rtl-ce2
4327 @opindex fdump-rtl-ce1
4328 @opindex fdump-rtl-ce2
4329 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
4330 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
4331 and @option{-fdump-rtl-ce2} enable dumping after the second if
4332 conversion, to the file @file{@var{file}.130r.ce2}.
4335 @itemx -fdump-rtl-btl
4336 @itemx -fdump-rtl-dbr
4338 @opindex fdump-rtl-btl
4339 @opindex fdump-rtl-dbr
4340 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
4341 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
4342 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
4343 scheduling, to @file{@var{file}.36.dbr}.
4347 Dump all macro definitions, at the end of preprocessing, in addition to
4351 @itemx -fdump-rtl-ce3
4353 @opindex fdump-rtl-ce3
4354 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4357 @itemx -fdump-rtl-cfg
4358 @itemx -fdump-rtl-life
4360 @opindex fdump-rtl-cfg
4361 @opindex fdump-rtl-life
4362 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
4363 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
4364 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4365 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4368 @itemx -fdump-rtl-greg
4370 @opindex fdump-rtl-greg
4371 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4374 @itemx -fdump-rtl-gcse
4375 @itemx -fdump-rtl-bypass
4377 @opindex fdump-rtl-gcse
4378 @opindex fdump-rtl-bypass
4379 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4380 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
4381 enable dumping after jump bypassing and control flow optimizations, to
4382 @file{@var{file}.115r.bypass}.
4385 @itemx -fdump-rtl-eh
4387 @opindex fdump-rtl-eh
4388 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4391 @itemx -fdump-rtl-sibling
4393 @opindex fdump-rtl-sibling
4394 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4397 @itemx -fdump-rtl-jump
4399 @opindex fdump-rtl-jump
4400 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4403 @itemx -fdump-rtl-stack
4405 @opindex fdump-rtl-stack
4406 Dump after conversion from GCC's "flat register file" registers to the
4407 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4410 @itemx -fdump-rtl-lreg
4412 @opindex fdump-rtl-lreg
4413 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4416 @itemx -fdump-rtl-loop2
4418 @opindex fdump-rtl-loop2
4419 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4420 loop optimization pass, to @file{@var{file}.119r.loop2},
4421 @file{@var{file}.120r.loop2_init},
4422 @file{@var{file}.121r.loop2_invariant}, and
4423 @file{@var{file}.125r.loop2_done}.
4426 @itemx -fdump-rtl-sms
4428 @opindex fdump-rtl-sms
4429 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4432 @itemx -fdump-rtl-mach
4434 @opindex fdump-rtl-mach
4435 Dump after performing the machine dependent reorganization pass, to
4436 @file{@var{file}.155r.mach} if that pass exists.
4439 @itemx -fdump-rtl-rnreg
4441 @opindex fdump-rtl-rnreg
4442 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4445 @itemx -fdump-rtl-regmove
4447 @opindex fdump-rtl-regmove
4448 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4451 @itemx -fdump-rtl-postreload
4453 @opindex fdump-rtl-postreload
4454 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4457 @itemx -fdump-rtl-expand
4459 @opindex fdump-rtl-expand
4460 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4463 @itemx -fdump-rtl-sched2
4465 @opindex fdump-rtl-sched2
4466 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4469 @itemx -fdump-rtl-cse
4471 @opindex fdump-rtl-cse
4472 Dump after CSE (including the jump optimization that sometimes follows
4473 CSE), to @file{@var{file}.113r.cse}.
4476 @itemx -fdump-rtl-sched1
4478 @opindex fdump-rtl-sched1
4479 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4482 @itemx -fdump-rtl-cse2
4484 @opindex fdump-rtl-cse2
4485 Dump after the second CSE pass (including the jump optimization that
4486 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4489 @itemx -fdump-rtl-tracer
4491 @opindex fdump-rtl-tracer
4492 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4495 @itemx -fdump-rtl-vpt
4496 @itemx -fdump-rtl-vartrack
4498 @opindex fdump-rtl-vpt
4499 @opindex fdump-rtl-vartrack
4500 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
4501 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
4502 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4503 to @file{@var{file}.154r.vartrack}.
4506 @itemx -fdump-rtl-flow2
4508 @opindex fdump-rtl-flow2
4509 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4512 @itemx -fdump-rtl-peephole2
4514 @opindex fdump-rtl-peephole2
4515 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4518 @itemx -fdump-rtl-web
4520 @opindex fdump-rtl-web
4521 Dump after live range splitting, to @file{@var{file}.126r.web}.
4524 @itemx -fdump-rtl-all
4526 @opindex fdump-rtl-all
4527 Produce all the dumps listed above.
4531 Produce a core dump whenever an error occurs.
4535 Print statistics on memory usage, at the end of the run, to
4540 Annotate the assembler output with a comment indicating which
4541 pattern and alternative was used. The length of each instruction is
4546 Dump the RTL in the assembler output as a comment before each instruction.
4547 Also turns on @option{-dp} annotation.
4551 For each of the other indicated dump files (either with @option{-d} or
4552 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4553 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4557 Just generate RTL for a function instead of compiling it. Usually used
4558 with @samp{r} (@option{-fdump-rtl-expand}).
4562 Dump debugging information during parsing, to standard error.
4566 @opindex fdump-noaddr
4567 When doing debugging dumps (see @option{-d} option above), suppress
4568 address output. This makes it more feasible to use diff on debugging
4569 dumps for compiler invocations with different compiler binaries and/or
4570 different text / bss / data / heap / stack / dso start locations.
4572 @item -fdump-unnumbered
4573 @opindex fdump-unnumbered
4574 When doing debugging dumps (see @option{-d} option above), suppress instruction
4575 numbers and address output. This makes it more feasible to
4576 use diff on debugging dumps for compiler invocations with different
4577 options, in particular with and without @option{-g}.
4579 @item -fdump-translation-unit @r{(C++ only)}
4580 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4581 @opindex fdump-translation-unit
4582 Dump a representation of the tree structure for the entire translation
4583 unit to a file. The file name is made by appending @file{.tu} to the
4584 source file name. If the @samp{-@var{options}} form is used, @var{options}
4585 controls the details of the dump as described for the
4586 @option{-fdump-tree} options.
4588 @item -fdump-class-hierarchy @r{(C++ only)}
4589 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4590 @opindex fdump-class-hierarchy
4591 Dump a representation of each class's hierarchy and virtual function
4592 table layout to a file. The file name is made by appending @file{.class}
4593 to the source file name. If the @samp{-@var{options}} form is used,
4594 @var{options} controls the details of the dump as described for the
4595 @option{-fdump-tree} options.
4597 @item -fdump-ipa-@var{switch}
4599 Control the dumping at various stages of inter-procedural analysis
4600 language tree to a file. The file name is generated by appending a switch
4601 specific suffix to the source file name. The following dumps are possible:
4605 Enables all inter-procedural analysis dumps; currently the only produced
4606 dump is the @samp{cgraph} dump.
4609 Dumps information about call-graph optimization, unused function removal,
4610 and inlining decisions.
4613 @item -fdump-tree-@var{switch}
4614 @itemx -fdump-tree-@var{switch}-@var{options}
4616 Control the dumping at various stages of processing the intermediate
4617 language tree to a file. The file name is generated by appending a switch
4618 specific suffix to the source file name. If the @samp{-@var{options}}
4619 form is used, @var{options} is a list of @samp{-} separated options that
4620 control the details of the dump. Not all options are applicable to all
4621 dumps, those which are not meaningful will be ignored. The following
4622 options are available
4626 Print the address of each node. Usually this is not meaningful as it
4627 changes according to the environment and source file. Its primary use
4628 is for tying up a dump file with a debug environment.
4630 Inhibit dumping of members of a scope or body of a function merely
4631 because that scope has been reached. Only dump such items when they
4632 are directly reachable by some other path. When dumping pretty-printed
4633 trees, this option inhibits dumping the bodies of control structures.
4635 Print a raw representation of the tree. By default, trees are
4636 pretty-printed into a C-like representation.
4638 Enable more detailed dumps (not honored by every dump option).
4640 Enable dumping various statistics about the pass (not honored by every dump
4643 Enable showing basic block boundaries (disabled in raw dumps).
4645 Enable showing virtual operands for every statement.
4647 Enable showing line numbers for statements.
4649 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4651 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4654 The following tree dumps are possible:
4658 Dump before any tree based optimization, to @file{@var{file}.original}.
4661 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4664 Dump after function inlining, to @file{@var{file}.inlined}.
4667 @opindex fdump-tree-gimple
4668 Dump each function before and after the gimplification pass to a file. The
4669 file name is made by appending @file{.gimple} to the source file name.
4672 @opindex fdump-tree-cfg
4673 Dump the control flow graph of each function to a file. The file name is
4674 made by appending @file{.cfg} to the source file name.
4677 @opindex fdump-tree-vcg
4678 Dump the control flow graph of each function to a file in VCG format. The
4679 file name is made by appending @file{.vcg} to the source file name. Note
4680 that if the file contains more than one function, the generated file cannot
4681 be used directly by VCG@. You will need to cut and paste each function's
4682 graph into its own separate file first.
4685 @opindex fdump-tree-ch
4686 Dump each function after copying loop headers. The file name is made by
4687 appending @file{.ch} to the source file name.
4690 @opindex fdump-tree-ssa
4691 Dump SSA related information to a file. The file name is made by appending
4692 @file{.ssa} to the source file name.
4695 @opindex fdump-tree-salias
4696 Dump structure aliasing variable information to a file. This file name
4697 is made by appending @file{.salias} to the source file name.
4700 @opindex fdump-tree-alias
4701 Dump aliasing information for each function. The file name is made by
4702 appending @file{.alias} to the source file name.
4705 @opindex fdump-tree-ccp
4706 Dump each function after CCP@. The file name is made by appending
4707 @file{.ccp} to the source file name.
4710 @opindex fdump-tree-storeccp
4711 Dump each function after STORE-CCP. The file name is made by appending
4712 @file{.storeccp} to the source file name.
4715 @opindex fdump-tree-pre
4716 Dump trees after partial redundancy elimination. The file name is made
4717 by appending @file{.pre} to the source file name.
4720 @opindex fdump-tree-fre
4721 Dump trees after full redundancy elimination. The file name is made
4722 by appending @file{.fre} to the source file name.
4725 @opindex fdump-tree-copyprop
4726 Dump trees after copy propagation. The file name is made
4727 by appending @file{.copyprop} to the source file name.
4729 @item store_copyprop
4730 @opindex fdump-tree-store_copyprop
4731 Dump trees after store copy-propagation. The file name is made
4732 by appending @file{.store_copyprop} to the source file name.
4735 @opindex fdump-tree-dce
4736 Dump each function after dead code elimination. The file name is made by
4737 appending @file{.dce} to the source file name.
4740 @opindex fdump-tree-mudflap
4741 Dump each function after adding mudflap instrumentation. The file name is
4742 made by appending @file{.mudflap} to the source file name.
4745 @opindex fdump-tree-sra
4746 Dump each function after performing scalar replacement of aggregates. The
4747 file name is made by appending @file{.sra} to the source file name.
4750 @opindex fdump-tree-sink
4751 Dump each function after performing code sinking. The file name is made
4752 by appending @file{.sink} to the source file name.
4755 @opindex fdump-tree-dom
4756 Dump each function after applying dominator tree optimizations. The file
4757 name is made by appending @file{.dom} to the source file name.
4760 @opindex fdump-tree-dse
4761 Dump each function after applying dead store elimination. The file
4762 name is made by appending @file{.dse} to the source file name.
4765 @opindex fdump-tree-phiopt
4766 Dump each function after optimizing PHI nodes into straightline code. The file
4767 name is made by appending @file{.phiopt} to the source file name.
4770 @opindex fdump-tree-forwprop
4771 Dump each function after forward propagating single use variables. The file
4772 name is made by appending @file{.forwprop} to the source file name.
4775 @opindex fdump-tree-copyrename
4776 Dump each function after applying the copy rename optimization. The file
4777 name is made by appending @file{.copyrename} to the source file name.
4780 @opindex fdump-tree-nrv
4781 Dump each function after applying the named return value optimization on
4782 generic trees. The file name is made by appending @file{.nrv} to the source
4786 @opindex fdump-tree-vect
4787 Dump each function after applying vectorization of loops. The file name is
4788 made by appending @file{.vect} to the source file name.
4791 @opindex fdump-tree-vrp
4792 Dump each function after Value Range Propagation (VRP). The file name
4793 is made by appending @file{.vrp} to the source file name.
4796 @opindex fdump-tree-all
4797 Enable all the available tree dumps with the flags provided in this option.
4800 @item -ftree-vectorizer-verbose=@var{n}
4801 @opindex ftree-vectorizer-verbose
4802 This option controls the amount of debugging output the vectorizer prints.
4803 This information is written to standard error, unless
4804 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4805 in which case it is output to the usual dump listing file, @file{.vect}.
4806 For @var{n}=0 no diagnostic information is reported.
4807 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4808 and the total number of loops that got vectorized.
4809 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4810 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4811 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4812 level that @option{-fdump-tree-vect-stats} uses.
4813 Higher verbosity levels mean either more information dumped for each
4814 reported loop, or same amount of information reported for more loops:
4815 If @var{n}=3, alignment related information is added to the reports.
4816 If @var{n}=4, data-references related information (e.g. memory dependences,
4817 memory access-patterns) is added to the reports.
4818 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4819 that did not pass the first analysis phase (i.e. may not be countable, or
4820 may have complicated control-flow).
4821 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4822 For @var{n}=7, all the information the vectorizer generates during its
4823 analysis and transformation is reported. This is the same verbosity level
4824 that @option{-fdump-tree-vect-details} uses.
4826 @item -frandom-seed=@var{string}
4827 @opindex frandom-string
4828 This option provides a seed that GCC uses when it would otherwise use
4829 random numbers. It is used to generate certain symbol names
4830 that have to be different in every compiled file. It is also used to
4831 place unique stamps in coverage data files and the object files that
4832 produce them. You can use the @option{-frandom-seed} option to produce
4833 reproducibly identical object files.
4835 The @var{string} should be different for every file you compile.
4837 @item -fsched-verbose=@var{n}
4838 @opindex fsched-verbose
4839 On targets that use instruction scheduling, this option controls the
4840 amount of debugging output the scheduler prints. This information is
4841 written to standard error, unless @option{-dS} or @option{-dR} is
4842 specified, in which case it is output to the usual dump
4843 listing file, @file{.sched} or @file{.sched2} respectively. However
4844 for @var{n} greater than nine, the output is always printed to standard
4847 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4848 same information as @option{-dRS}. For @var{n} greater than one, it
4849 also output basic block probabilities, detailed ready list information
4850 and unit/insn info. For @var{n} greater than two, it includes RTL
4851 at abort point, control-flow and regions info. And for @var{n} over
4852 four, @option{-fsched-verbose} also includes dependence info.
4856 Store the usual ``temporary'' intermediate files permanently; place them
4857 in the current directory and name them based on the source file. Thus,
4858 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4859 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4860 preprocessed @file{foo.i} output file even though the compiler now
4861 normally uses an integrated preprocessor.
4863 When used in combination with the @option{-x} command line option,
4864 @option{-save-temps} is sensible enough to avoid over writing an
4865 input source file with the same extension as an intermediate file.
4866 The corresponding intermediate file may be obtained by renaming the
4867 source file before using @option{-save-temps}.
4871 Report the CPU time taken by each subprocess in the compilation
4872 sequence. For C source files, this is the compiler proper and assembler
4873 (plus the linker if linking is done). The output looks like this:
4880 The first number on each line is the ``user time'', that is time spent
4881 executing the program itself. The second number is ``system time'',
4882 time spent executing operating system routines on behalf of the program.
4883 Both numbers are in seconds.
4885 @item -fvar-tracking
4886 @opindex fvar-tracking
4887 Run variable tracking pass. It computes where variables are stored at each
4888 position in code. Better debugging information is then generated
4889 (if the debugging information format supports this information).
4891 It is enabled by default when compiling with optimization (@option{-Os},
4892 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4893 the debug info format supports it.
4895 @item -print-file-name=@var{library}
4896 @opindex print-file-name
4897 Print the full absolute name of the library file @var{library} that
4898 would be used when linking---and don't do anything else. With this
4899 option, GCC does not compile or link anything; it just prints the
4902 @item -print-multi-directory
4903 @opindex print-multi-directory
4904 Print the directory name corresponding to the multilib selected by any
4905 other switches present in the command line. This directory is supposed
4906 to exist in @env{GCC_EXEC_PREFIX}.
4908 @item -print-multi-lib
4909 @opindex print-multi-lib
4910 Print the mapping from multilib directory names to compiler switches
4911 that enable them. The directory name is separated from the switches by
4912 @samp{;}, and each switch starts with an @samp{@@} instead of the
4913 @samp{-}, without spaces between multiple switches. This is supposed to
4914 ease shell-processing.
4916 @item -print-prog-name=@var{program}
4917 @opindex print-prog-name
4918 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4920 @item -print-libgcc-file-name
4921 @opindex print-libgcc-file-name
4922 Same as @option{-print-file-name=libgcc.a}.
4924 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4925 but you do want to link with @file{libgcc.a}. You can do
4928 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4931 @item -print-search-dirs
4932 @opindex print-search-dirs
4933 Print the name of the configured installation directory and a list of
4934 program and library directories @command{gcc} will search---and don't do anything else.
4936 This is useful when @command{gcc} prints the error message
4937 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4938 To resolve this you either need to put @file{cpp0} and the other compiler
4939 components where @command{gcc} expects to find them, or you can set the environment
4940 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4941 Don't forget the trailing @samp{/}.
4942 @xref{Environment Variables}.
4944 @item -print-sysroot-headers-suffix
4945 @opindex print-sysroot-headers-suffix
4946 Print the suffix added to the target sysroot when searching for
4947 headers, or give an error if the compiler is not configured with such
4948 a suffix---and don't do anything else.
4951 @opindex dumpmachine
4952 Print the compiler's target machine (for example,
4953 @samp{i686-pc-linux-gnu})---and don't do anything else.
4956 @opindex dumpversion
4957 Print the compiler version (for example, @samp{3.0})---and don't do
4962 Print the compiler's built-in specs---and don't do anything else. (This
4963 is used when GCC itself is being built.) @xref{Spec Files}.
4965 @item -feliminate-unused-debug-types
4966 @opindex feliminate-unused-debug-types
4967 Normally, when producing DWARF2 output, GCC will emit debugging
4968 information for all types declared in a compilation
4969 unit, regardless of whether or not they are actually used
4970 in that compilation unit. Sometimes this is useful, such as
4971 if, in the debugger, you want to cast a value to a type that is
4972 not actually used in your program (but is declared). More often,
4973 however, this results in a significant amount of wasted space.
4974 With this option, GCC will avoid producing debug symbol output
4975 for types that are nowhere used in the source file being compiled.
4978 @node Optimize Options
4979 @section Options That Control Optimization
4980 @cindex optimize options
4981 @cindex options, optimization
4983 These options control various sorts of optimizations.
4985 Without any optimization option, the compiler's goal is to reduce the
4986 cost of compilation and to make debugging produce the expected
4987 results. Statements are independent: if you stop the program with a
4988 breakpoint between statements, you can then assign a new value to any
4989 variable or change the program counter to any other statement in the
4990 function and get exactly the results you would expect from the source
4993 Turning on optimization flags makes the compiler attempt to improve
4994 the performance and/or code size at the expense of compilation time
4995 and possibly the ability to debug the program.
4997 The compiler performs optimization based on the knowledge it has of
4998 the program. Optimization levels @option{-O} and above, in
4999 particular, enable @emph{unit-at-a-time} mode, which allows the
5000 compiler to consider information gained from later functions in
5001 the file when compiling a function. Compiling multiple files at
5002 once to a single output file in @emph{unit-at-a-time} mode allows
5003 the compiler to use information gained from all of the files when
5004 compiling each of them.
5006 Not all optimizations are controlled directly by a flag. Only
5007 optimizations that have a flag are listed.
5014 Optimize. Optimizing compilation takes somewhat more time, and a lot
5015 more memory for a large function.
5017 With @option{-O}, the compiler tries to reduce code size and execution
5018 time, without performing any optimizations that take a great deal of
5021 @option{-O} turns on the following optimization flags:
5022 @gccoptlist{-fdefer-pop @gol
5023 -fdelayed-branch @gol
5024 -fguess-branch-probability @gol
5025 -fcprop-registers @gol
5026 -fif-conversion @gol
5027 -fif-conversion2 @gol
5028 -fsplit-wide-types @gol
5031 -ftree-dominator-opts @gol
5035 -ftree-copyrename @gol
5038 -funit-at-a-time @gol
5039 -finline-small-functions @gol
5042 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5043 where doing so does not interfere with debugging.
5047 Optimize even more. GCC performs nearly all supported optimizations
5048 that do not involve a space-speed tradeoff. The compiler does not
5049 perform loop unrolling or function inlining when you specify @option{-O2}.
5050 As compared to @option{-O}, this option increases both compilation time
5051 and the performance of the generated code.
5053 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5054 also turns on the following optimization flags:
5055 @gccoptlist{-fthread-jumps @gol
5057 -foptimize-sibling-calls @gol
5058 -fcse-follow-jumps -fcse-skip-blocks @gol
5059 -fgcse -fgcse-lm @gol
5060 -fexpensive-optimizations @gol
5061 -frerun-cse-after-loop @gol
5064 -fschedule-insns -fschedule-insns2 @gol
5065 -fsched-interblock -fsched-spec @gol
5067 -fstrict-aliasing -fstrict-overflow @gol
5068 -fdelete-null-pointer-checks @gol
5069 -freorder-blocks -freorder-functions @gol
5070 -falign-functions -falign-jumps @gol
5071 -falign-loops -falign-labels @gol
5075 Please note the warning under @option{-fgcse} about
5076 invoking @option{-O2} on programs that use computed gotos.
5080 Optimize yet more. @option{-O3} turns on all optimizations specified by
5081 @option{-O2} and also turns on the @option{-finline-functions},
5082 @option{-funswitch-loops}, @option{-fpredictive-commoning} and
5083 @option{-fgcse-after-reload} options.
5087 Reduce compilation time and make debugging produce the expected
5088 results. This is the default.
5092 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5093 do not typically increase code size. It also performs further
5094 optimizations designed to reduce code size.
5096 @option{-Os} disables the following optimization flags:
5097 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5098 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5099 -fprefetch-loop-arrays -ftree-vect-loop-version}
5101 If you use multiple @option{-O} options, with or without level numbers,
5102 the last such option is the one that is effective.
5105 Options of the form @option{-f@var{flag}} specify machine-independent
5106 flags. Most flags have both positive and negative forms; the negative
5107 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5108 below, only one of the forms is listed---the one you typically will
5109 use. You can figure out the other form by either removing @samp{no-}
5112 The following options control specific optimizations. They are either
5113 activated by @option{-O} options or are related to ones that are. You
5114 can use the following flags in the rare cases when ``fine-tuning'' of
5115 optimizations to be performed is desired.
5118 @item -fno-default-inline
5119 @opindex fno-default-inline
5120 Do not make member functions inline by default merely because they are
5121 defined inside the class scope (C++ only). Otherwise, when you specify
5122 @w{@option{-O}}, member functions defined inside class scope are compiled
5123 inline by default; i.e., you don't need to add @samp{inline} in front of
5124 the member function name.
5126 @item -fno-defer-pop
5127 @opindex fno-defer-pop
5128 Always pop the arguments to each function call as soon as that function
5129 returns. For machines which must pop arguments after a function call,
5130 the compiler normally lets arguments accumulate on the stack for several
5131 function calls and pops them all at once.
5133 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5136 @opindex fforce-addr
5137 Force memory address constants to be copied into registers before
5138 doing arithmetic on them.
5140 @item -fforward-propagate
5141 @opindex fforward-propagate
5142 Perform a forward propagation pass on RTL. The pass tries to combine two
5143 instructions and checks if the result can be simplified. If loop unrolling
5144 is active, two passes are performed and the second is scheduled after
5147 This option is enabled by default at optimization levels @option{-O2},
5148 @option{-O3}, @option{-Os}.
5150 @item -fomit-frame-pointer
5151 @opindex fomit-frame-pointer
5152 Don't keep the frame pointer in a register for functions that
5153 don't need one. This avoids the instructions to save, set up and
5154 restore frame pointers; it also makes an extra register available
5155 in many functions. @strong{It also makes debugging impossible on
5158 On some machines, such as the VAX, this flag has no effect, because
5159 the standard calling sequence automatically handles the frame pointer
5160 and nothing is saved by pretending it doesn't exist. The
5161 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5162 whether a target machine supports this flag. @xref{Registers,,Register
5163 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5165 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5167 @item -foptimize-sibling-calls
5168 @opindex foptimize-sibling-calls
5169 Optimize sibling and tail recursive calls.
5171 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5175 Don't pay attention to the @code{inline} keyword. Normally this option
5176 is used to keep the compiler from expanding any functions inline.
5177 Note that if you are not optimizing, no functions can be expanded inline.
5179 @item -finline-small-functions
5180 @opindex finline-small-functions
5181 Integrate functions into their callers when their body is smaller than expected
5182 function call code (so overall size of program gets smaller). The compiler
5183 heuristically decides which functions are simple enough to be worth integrating
5186 Enabled at level @option{-O2}.
5188 @item -finline-functions
5189 @opindex finline-functions
5190 Integrate all simple functions into their callers. The compiler
5191 heuristically decides which functions are simple enough to be worth
5192 integrating in this way.
5194 If all calls to a given function are integrated, and the function is
5195 declared @code{static}, then the function is normally not output as
5196 assembler code in its own right.
5198 Enabled at level @option{-O3}.
5200 @item -finline-functions-called-once
5201 @opindex finline-functions-called-once
5202 Consider all @code{static} functions called once for inlining into their
5203 caller even if they are not marked @code{inline}. If a call to a given
5204 function is integrated, then the function is not output as assembler code
5207 Enabled if @option{-funit-at-a-time} is enabled.
5209 @item -fearly-inlining
5210 @opindex fearly-inlining
5211 Inline functions marked by @code{always_inline} and functions whose body seems
5212 smaller than the function call overhead early before doing
5213 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5214 makes profiling significantly cheaper and usually inlining faster on programs
5215 having large chains of nested wrapper functions.
5219 @item -finline-limit=@var{n}
5220 @opindex finline-limit
5221 By default, GCC limits the size of functions that can be inlined. This flag
5222 allows the control of this limit for functions that are explicitly marked as
5223 inline (i.e., marked with the inline keyword or defined within the class
5224 definition in c++). @var{n} is the size of functions that can be inlined in
5225 number of pseudo instructions (not counting parameter handling). The default
5226 value of @var{n} is 600.
5227 Increasing this value can result in more inlined code at
5228 the cost of compilation time and memory consumption. Decreasing usually makes
5229 the compilation faster and less code will be inlined (which presumably
5230 means slower programs). This option is particularly useful for programs that
5231 use inlining heavily such as those based on recursive templates with C++.
5233 Inlining is actually controlled by a number of parameters, which may be
5234 specified individually by using @option{--param @var{name}=@var{value}}.
5235 The @option{-finline-limit=@var{n}} option sets some of these parameters
5239 @item max-inline-insns-single
5240 is set to @var{n}/2.
5241 @item max-inline-insns-auto
5242 is set to @var{n}/2.
5243 @item min-inline-insns
5244 is set to 130 or @var{n}/4, whichever is smaller.
5245 @item max-inline-insns-rtl
5249 See below for a documentation of the individual
5250 parameters controlling inlining.
5252 @emph{Note:} pseudo instruction represents, in this particular context, an
5253 abstract measurement of function's size. In no way does it represent a count
5254 of assembly instructions and as such its exact meaning might change from one
5255 release to an another.
5257 @item -fkeep-inline-functions
5258 @opindex fkeep-inline-functions
5259 In C, emit @code{static} functions that are declared @code{inline}
5260 into the object file, even if the function has been inlined into all
5261 of its callers. This switch does not affect functions using the
5262 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5263 inline functions into the object file.
5265 @item -fkeep-static-consts
5266 @opindex fkeep-static-consts
5267 Emit variables declared @code{static const} when optimization isn't turned
5268 on, even if the variables aren't referenced.
5270 GCC enables this option by default. If you want to force the compiler to
5271 check if the variable was referenced, regardless of whether or not
5272 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5274 @item -fmerge-constants
5275 Attempt to merge identical constants (string constants and floating point
5276 constants) across compilation units.
5278 This option is the default for optimized compilation if the assembler and
5279 linker support it. Use @option{-fno-merge-constants} to inhibit this
5282 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5284 @item -fmerge-all-constants
5285 Attempt to merge identical constants and identical variables.
5287 This option implies @option{-fmerge-constants}. In addition to
5288 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5289 arrays or initialized constant variables with integral or floating point
5290 types. Languages like C or C++ require each non-automatic variable to
5291 have distinct location, so using this option will result in non-conforming
5294 @item -fmodulo-sched
5295 @opindex fmodulo-sched
5296 Perform swing modulo scheduling immediately before the first scheduling
5297 pass. This pass looks at innermost loops and reorders their
5298 instructions by overlapping different iterations.
5300 @item -fmodulo-sched-allow-regmoves
5301 @opindex fmodulo-sched-allow-regmoves
5302 Perform more aggressive SMS based modulo scheduling with register moves
5303 allowed. By setting this flag certain anti-dependences edges will be
5304 deleted which will trigger the generation of reg-moves based on the
5305 life-range analysis. This option is effective only with
5306 @option{-fmodulo-sched} enabled.
5308 @item -fno-branch-count-reg
5309 @opindex fno-branch-count-reg
5310 Do not use ``decrement and branch'' instructions on a count register,
5311 but instead generate a sequence of instructions that decrement a
5312 register, compare it against zero, then branch based upon the result.
5313 This option is only meaningful on architectures that support such
5314 instructions, which include x86, PowerPC, IA-64 and S/390.
5316 The default is @option{-fbranch-count-reg}.
5318 @item -fno-function-cse
5319 @opindex fno-function-cse
5320 Do not put function addresses in registers; make each instruction that
5321 calls a constant function contain the function's address explicitly.
5323 This option results in less efficient code, but some strange hacks
5324 that alter the assembler output may be confused by the optimizations
5325 performed when this option is not used.
5327 The default is @option{-ffunction-cse}
5329 @item -fno-zero-initialized-in-bss
5330 @opindex fno-zero-initialized-in-bss
5331 If the target supports a BSS section, GCC by default puts variables that
5332 are initialized to zero into BSS@. This can save space in the resulting
5335 This option turns off this behavior because some programs explicitly
5336 rely on variables going to the data section. E.g., so that the
5337 resulting executable can find the beginning of that section and/or make
5338 assumptions based on that.
5340 The default is @option{-fzero-initialized-in-bss}.
5342 @item -fbounds-check
5343 @opindex fbounds-check
5344 For front-ends that support it, generate additional code to check that
5345 indices used to access arrays are within the declared range. This is
5346 currently only supported by the Java and Fortran front-ends, where
5347 this option defaults to true and false respectively.
5349 @item -fmudflap -fmudflapth -fmudflapir
5353 @cindex bounds checking
5355 For front-ends that support it (C and C++), instrument all risky
5356 pointer/array dereferencing operations, some standard library
5357 string/heap functions, and some other associated constructs with
5358 range/validity tests. Modules so instrumented should be immune to
5359 buffer overflows, invalid heap use, and some other classes of C/C++
5360 programming errors. The instrumentation relies on a separate runtime
5361 library (@file{libmudflap}), which will be linked into a program if
5362 @option{-fmudflap} is given at link time. Run-time behavior of the
5363 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5364 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5367 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5368 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5369 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5370 instrumentation should ignore pointer reads. This produces less
5371 instrumentation (and therefore faster execution) and still provides
5372 some protection against outright memory corrupting writes, but allows
5373 erroneously read data to propagate within a program.
5375 @item -fthread-jumps
5376 @opindex fthread-jumps
5377 Perform optimizations where we check to see if a jump branches to a
5378 location where another comparison subsumed by the first is found. If
5379 so, the first branch is redirected to either the destination of the
5380 second branch or a point immediately following it, depending on whether
5381 the condition is known to be true or false.
5383 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5385 @item -fsplit-wide-types
5386 @opindex fsplit-wide-types
5387 When using a type that occupies multiple registers, such as @code{long
5388 long} on a 32-bit system, split the registers apart and allocate them
5389 independently. This normally generates better code for those types,
5390 but may make debugging more difficult.
5392 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5395 @item -fcse-follow-jumps
5396 @opindex fcse-follow-jumps
5397 In common subexpression elimination, scan through jump instructions
5398 when the target of the jump is not reached by any other path. For
5399 example, when CSE encounters an @code{if} statement with an
5400 @code{else} clause, CSE will follow the jump when the condition
5403 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5405 @item -fcse-skip-blocks
5406 @opindex fcse-skip-blocks
5407 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5408 follow jumps which conditionally skip over blocks. When CSE
5409 encounters a simple @code{if} statement with no else clause,
5410 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5411 body of the @code{if}.
5413 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5415 @item -frerun-cse-after-loop
5416 @opindex frerun-cse-after-loop
5417 Re-run common subexpression elimination after loop optimizations has been
5420 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5424 Perform a global common subexpression elimination pass.
5425 This pass also performs global constant and copy propagation.
5427 @emph{Note:} When compiling a program using computed gotos, a GCC
5428 extension, you may get better runtime performance if you disable
5429 the global common subexpression elimination pass by adding
5430 @option{-fno-gcse} to the command line.
5432 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5436 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5437 attempt to move loads which are only killed by stores into themselves. This
5438 allows a loop containing a load/store sequence to be changed to a load outside
5439 the loop, and a copy/store within the loop.
5441 Enabled by default when gcse is enabled.
5445 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5446 global common subexpression elimination. This pass will attempt to move
5447 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5448 loops containing a load/store sequence can be changed to a load before
5449 the loop and a store after the loop.
5451 Not enabled at any optimization level.
5455 When @option{-fgcse-las} is enabled, the global common subexpression
5456 elimination pass eliminates redundant loads that come after stores to the
5457 same memory location (both partial and full redundancies).
5459 Not enabled at any optimization level.
5461 @item -fgcse-after-reload
5462 @opindex fgcse-after-reload
5463 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5464 pass is performed after reload. The purpose of this pass is to cleanup
5467 @item -funsafe-loop-optimizations
5468 @opindex funsafe-loop-optimizations
5469 If given, the loop optimizer will assume that loop indices do not
5470 overflow, and that the loops with nontrivial exit condition are not
5471 infinite. This enables a wider range of loop optimizations even if
5472 the loop optimizer itself cannot prove that these assumptions are valid.
5473 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5474 if it finds this kind of loop.
5476 @item -fcrossjumping
5477 @opindex crossjumping
5478 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5479 resulting code may or may not perform better than without cross-jumping.
5481 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5483 @item -fif-conversion
5484 @opindex if-conversion
5485 Attempt to transform conditional jumps into branch-less equivalents. This
5486 include use of conditional moves, min, max, set flags and abs instructions, and
5487 some tricks doable by standard arithmetics. The use of conditional execution
5488 on chips where it is available is controlled by @code{if-conversion2}.
5490 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5492 @item -fif-conversion2
5493 @opindex if-conversion2
5494 Use conditional execution (where available) to transform conditional jumps into
5495 branch-less equivalents.
5497 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5499 @item -fdelete-null-pointer-checks
5500 @opindex fdelete-null-pointer-checks
5501 Use global dataflow analysis to identify and eliminate useless checks
5502 for null pointers. The compiler assumes that dereferencing a null
5503 pointer would have halted the program. If a pointer is checked after
5504 it has already been dereferenced, it cannot be null.
5506 In some environments, this assumption is not true, and programs can
5507 safely dereference null pointers. Use
5508 @option{-fno-delete-null-pointer-checks} to disable this optimization
5509 for programs which depend on that behavior.
5511 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5513 @item -fexpensive-optimizations
5514 @opindex fexpensive-optimizations
5515 Perform a number of minor optimizations that are relatively expensive.
5517 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5519 @item -foptimize-register-move
5521 @opindex foptimize-register-move
5523 Attempt to reassign register numbers in move instructions and as
5524 operands of other simple instructions in order to maximize the amount of
5525 register tying. This is especially helpful on machines with two-operand
5528 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5531 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5533 @item -fdelayed-branch
5534 @opindex fdelayed-branch
5535 If supported for the target machine, attempt to reorder instructions
5536 to exploit instruction slots available after delayed branch
5539 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5541 @item -fschedule-insns
5542 @opindex fschedule-insns
5543 If supported for the target machine, attempt to reorder instructions to
5544 eliminate execution stalls due to required data being unavailable. This
5545 helps machines that have slow floating point or memory load instructions
5546 by allowing other instructions to be issued until the result of the load
5547 or floating point instruction is required.
5549 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5551 @item -fschedule-insns2
5552 @opindex fschedule-insns2
5553 Similar to @option{-fschedule-insns}, but requests an additional pass of
5554 instruction scheduling after register allocation has been done. This is
5555 especially useful on machines with a relatively small number of
5556 registers and where memory load instructions take more than one cycle.
5558 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5560 @item -fno-sched-interblock
5561 @opindex fno-sched-interblock
5562 Don't schedule instructions across basic blocks. This is normally
5563 enabled by default when scheduling before register allocation, i.e.@:
5564 with @option{-fschedule-insns} or at @option{-O2} or higher.
5566 @item -fno-sched-spec
5567 @opindex fno-sched-spec
5568 Don't allow speculative motion of non-load instructions. This is normally
5569 enabled by default when scheduling before register allocation, i.e.@:
5570 with @option{-fschedule-insns} or at @option{-O2} or higher.
5572 @item -fsched-spec-load
5573 @opindex fsched-spec-load
5574 Allow speculative motion of some load instructions. This only makes
5575 sense when scheduling before register allocation, i.e.@: with
5576 @option{-fschedule-insns} or at @option{-O2} or higher.
5578 @item -fsched-spec-load-dangerous
5579 @opindex fsched-spec-load-dangerous
5580 Allow speculative motion of more load instructions. This only makes
5581 sense when scheduling before register allocation, i.e.@: with
5582 @option{-fschedule-insns} or at @option{-O2} or higher.
5584 @item -fsched-stalled-insns=@var{n}
5585 @opindex fsched-stalled-insns
5586 Define how many insns (if any) can be moved prematurely from the queue
5587 of stalled insns into the ready list, during the second scheduling pass.
5589 @item -fsched-stalled-insns-dep=@var{n}
5590 @opindex fsched-stalled-insns-dep
5591 Define how many insn groups (cycles) will be examined for a dependency
5592 on a stalled insn that is candidate for premature removal from the queue
5593 of stalled insns. Has an effect only during the second scheduling pass,
5594 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5596 @item -fsched2-use-superblocks
5597 @opindex fsched2-use-superblocks
5598 When scheduling after register allocation, do use superblock scheduling
5599 algorithm. Superblock scheduling allows motion across basic block boundaries
5600 resulting on faster schedules. This option is experimental, as not all machine
5601 descriptions used by GCC model the CPU closely enough to avoid unreliable
5602 results from the algorithm.
5604 This only makes sense when scheduling after register allocation, i.e.@: with
5605 @option{-fschedule-insns2} or at @option{-O2} or higher.
5607 @item -fsched2-use-traces
5608 @opindex fsched2-use-traces
5609 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5610 allocation and additionally perform code duplication in order to increase the
5611 size of superblocks using tracer pass. See @option{-ftracer} for details on
5614 This mode should produce faster but significantly longer programs. Also
5615 without @option{-fbranch-probabilities} the traces constructed may not
5616 match the reality and hurt the performance. This only makes
5617 sense when scheduling after register allocation, i.e.@: with
5618 @option{-fschedule-insns2} or at @option{-O2} or higher.
5622 Eliminates redundant extension instructions and move the non redundant
5623 ones to optimal placement using LCM.
5625 @item -freschedule-modulo-scheduled-loops
5626 @opindex fscheduling-in-modulo-scheduled-loops
5627 The modulo scheduling comes before the traditional scheduling, if a loop
5628 was modulo scheduled we may want to prevent the later scheduling passes
5629 from changing its schedule, we use this option to control that.
5631 @item -fcaller-saves
5632 @opindex fcaller-saves
5633 Enable values to be allocated in registers that will be clobbered by
5634 function calls, by emitting extra instructions to save and restore the
5635 registers around such calls. Such allocation is done only when it
5636 seems to result in better code than would otherwise be produced.
5638 This option is always enabled by default on certain machines, usually
5639 those which have no call-preserved registers to use instead.
5641 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5643 @item -ftree-reassoc
5644 Perform Reassociation on trees This flag is enabled by default
5645 at @option{-O} and higher.
5648 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5649 enabled by default at @option{-O2} and @option{-O3}.
5652 Perform Full Redundancy Elimination (FRE) on trees. The difference
5653 between FRE and PRE is that FRE only considers expressions
5654 that are computed on all paths leading to the redundant computation.
5655 This analysis is faster than PRE, though it exposes fewer redundancies.
5656 This flag is enabled by default at @option{-O} and higher.
5658 @item -ftree-copy-prop
5659 Perform copy propagation on trees. This pass eliminates unnecessary
5660 copy operations. This flag is enabled by default at @option{-O} and
5663 @item -ftree-store-copy-prop
5664 Perform copy propagation of memory loads and stores. This pass
5665 eliminates unnecessary copy operations in memory references
5666 (structures, global variables, arrays, etc). This flag is enabled by
5667 default at @option{-O2} and higher.
5670 Perform structural alias analysis on trees. This flag
5671 is enabled by default at @option{-O} and higher.
5674 Perform interprocedural pointer analysis.
5677 Perform forward store motion on trees. This flag is
5678 enabled by default at @option{-O} and higher.
5681 Perform sparse conditional constant propagation (CCP) on trees. This
5682 pass only operates on local scalar variables and is enabled by default
5683 at @option{-O} and higher.
5685 @item -ftree-store-ccp
5686 Perform sparse conditional constant propagation (CCP) on trees. This
5687 pass operates on both local scalar variables and memory stores and
5688 loads (global variables, structures, arrays, etc). This flag is
5689 enabled by default at @option{-O2} and higher.
5692 Perform dead code elimination (DCE) on trees. This flag is enabled by
5693 default at @option{-O} and higher.
5695 @item -ftree-dominator-opts
5696 Perform a variety of simple scalar cleanups (constant/copy
5697 propagation, redundancy elimination, range propagation and expression
5698 simplification) based on a dominator tree traversal. This also
5699 performs jump threading (to reduce jumps to jumps). This flag is
5700 enabled by default at @option{-O} and higher.
5704 Perform dead store elimination (DSE) on trees. A dead store is a store into
5705 a memory location which will later be overwritten by another store without
5706 any intervening loads. In this case the earlier store can be deleted. This
5707 flag is enabled by default at @option{-O} and higher.
5710 Perform loop header copying on trees. This is beneficial since it increases
5711 effectiveness of code motion optimizations. It also saves one jump. This flag
5712 is enabled by default at @option{-O} and higher. It is not enabled
5713 for @option{-Os}, since it usually increases code size.
5715 @item -ftree-loop-optimize
5716 Perform loop optimizations on trees. This flag is enabled by default
5717 at @option{-O} and higher.
5719 @item -ftree-loop-linear
5720 Perform linear loop transformations on tree. This flag can improve cache
5721 performance and allow further loop optimizations to take place.
5723 @item -fcheck-data-deps
5724 Compare the results of several data dependence analyzers. This option
5725 is used for debugging the data dependence analyzers.
5727 @item -ftree-loop-im
5728 Perform loop invariant motion on trees. This pass moves only invariants that
5729 would be hard to handle at RTL level (function calls, operations that expand to
5730 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5731 operands of conditions that are invariant out of the loop, so that we can use
5732 just trivial invariantness analysis in loop unswitching. The pass also includes
5735 @item -ftree-loop-ivcanon
5736 Create a canonical counter for number of iterations in the loop for that
5737 determining number of iterations requires complicated analysis. Later
5738 optimizations then may determine the number easily. Useful especially
5739 in connection with unrolling.
5742 Perform induction variable optimizations (strength reduction, induction
5743 variable merging and induction variable elimination) on trees.
5746 Perform scalar replacement of aggregates. This pass replaces structure
5747 references with scalars to prevent committing structures to memory too
5748 early. This flag is enabled by default at @option{-O} and higher.
5750 @item -ftree-copyrename
5751 Perform copy renaming on trees. This pass attempts to rename compiler
5752 temporaries to other variables at copy locations, usually resulting in
5753 variable names which more closely resemble the original variables. This flag
5754 is enabled by default at @option{-O} and higher.
5757 Perform temporary expression replacement during the SSA->normal phase. Single
5758 use/single def temporaries are replaced at their use location with their
5759 defining expression. This results in non-GIMPLE code, but gives the expanders
5760 much more complex trees to work on resulting in better RTL generation. This is
5761 enabled by default at @option{-O} and higher.
5763 @item -ftree-vectorize
5764 Perform loop vectorization on trees.
5766 @item -ftree-vect-loop-version
5767 @opindex ftree-vect-loop-version
5768 Perform loop versioning when doing loop vectorization on trees. When a loop
5769 appears to be vectorizable except that data alignment or data dependence cannot
5770 be determined at compile time then vectorized and non-vectorized versions of
5771 the loop are generated along with runtime checks for alignment or dependence
5772 to control which version is executed. This option is enabled by default
5773 except at level @option{-Os} where it is disabled.
5775 @item -fvect-cost-model
5776 Enable cost model for vectorization.
5779 Perform Value Range Propagation on trees. This is similar to the
5780 constant propagation pass, but instead of values, ranges of values are
5781 propagated. This allows the optimizers to remove unnecessary range
5782 checks like array bound checks and null pointer checks. This is
5783 enabled by default at @option{-O2} and higher. Null pointer check
5784 elimination is only done if @option{-fdelete-null-pointer-checks} is
5789 Perform tail duplication to enlarge superblock size. This transformation
5790 simplifies the control flow of the function allowing other optimizations to do
5793 @item -funroll-loops
5794 @opindex funroll-loops
5795 Unroll loops whose number of iterations can be determined at compile
5796 time or upon entry to the loop. @option{-funroll-loops} implies
5797 @option{-frerun-cse-after-loop}. This option makes code larger,
5798 and may or may not make it run faster.
5800 @item -funroll-all-loops
5801 @opindex funroll-all-loops
5802 Unroll all loops, even if their number of iterations is uncertain when
5803 the loop is entered. This usually makes programs run more slowly.
5804 @option{-funroll-all-loops} implies the same options as
5805 @option{-funroll-loops},
5807 @item -fsplit-ivs-in-unroller
5808 @opindex fsplit-ivs-in-unroller
5809 Enables expressing of values of induction variables in later iterations
5810 of the unrolled loop using the value in the first iteration. This breaks
5811 long dependency chains, thus improving efficiency of the scheduling passes.
5813 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5814 same effect. However in cases the loop body is more complicated than
5815 a single basic block, this is not reliable. It also does not work at all
5816 on some of the architectures due to restrictions in the CSE pass.
5818 This optimization is enabled by default.
5820 @item -fvariable-expansion-in-unroller
5821 @opindex fvariable-expansion-in-unroller
5822 With this option, the compiler will create multiple copies of some
5823 local variables when unrolling a loop which can result in superior code.
5825 @item -fpredictive-commoning
5826 @opindex fpredictive-commoning
5827 Perform predictive commoning optimization, i.e., reusing computations
5828 (especially memory loads and stores) performed in previous
5829 iterations of loops.
5831 This option is enabled at level @option{-O3}.
5833 @item -fprefetch-loop-arrays
5834 @opindex fprefetch-loop-arrays
5835 If supported by the target machine, generate instructions to prefetch
5836 memory to improve the performance of loops that access large arrays.
5838 This option may generate better or worse code; results are highly
5839 dependent on the structure of loops within the source code.
5841 Disabled at level @option{-Os}.
5844 @itemx -fno-peephole2
5845 @opindex fno-peephole
5846 @opindex fno-peephole2
5847 Disable any machine-specific peephole optimizations. The difference
5848 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5849 are implemented in the compiler; some targets use one, some use the
5850 other, a few use both.
5852 @option{-fpeephole} is enabled by default.
5853 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5855 @item -fno-guess-branch-probability
5856 @opindex fno-guess-branch-probability
5857 Do not guess branch probabilities using heuristics.
5859 GCC will use heuristics to guess branch probabilities if they are
5860 not provided by profiling feedback (@option{-fprofile-arcs}). These
5861 heuristics are based on the control flow graph. If some branch probabilities
5862 are specified by @samp{__builtin_expect}, then the heuristics will be
5863 used to guess branch probabilities for the rest of the control flow graph,
5864 taking the @samp{__builtin_expect} info into account. The interactions
5865 between the heuristics and @samp{__builtin_expect} can be complex, and in
5866 some cases, it may be useful to disable the heuristics so that the effects
5867 of @samp{__builtin_expect} are easier to understand.
5869 The default is @option{-fguess-branch-probability} at levels
5870 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5872 @item -freorder-blocks
5873 @opindex freorder-blocks
5874 Reorder basic blocks in the compiled function in order to reduce number of
5875 taken branches and improve code locality.
5877 Enabled at levels @option{-O2}, @option{-O3}.
5879 @item -freorder-blocks-and-partition
5880 @opindex freorder-blocks-and-partition
5881 In addition to reordering basic blocks in the compiled function, in order
5882 to reduce number of taken branches, partitions hot and cold basic blocks
5883 into separate sections of the assembly and .o files, to improve
5884 paging and cache locality performance.
5886 This optimization is automatically turned off in the presence of
5887 exception handling, for linkonce sections, for functions with a user-defined
5888 section attribute and on any architecture that does not support named
5891 @item -freorder-functions
5892 @opindex freorder-functions
5893 Reorder functions in the object file in order to
5894 improve code locality. This is implemented by using special
5895 subsections @code{.text.hot} for most frequently executed functions and
5896 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5897 the linker so object file format must support named sections and linker must
5898 place them in a reasonable way.
5900 Also profile feedback must be available in to make this option effective. See
5901 @option{-fprofile-arcs} for details.
5903 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5905 @item -fstrict-aliasing
5906 @opindex fstrict-aliasing
5907 Allows the compiler to assume the strictest aliasing rules applicable to
5908 the language being compiled. For C (and C++), this activates
5909 optimizations based on the type of expressions. In particular, an
5910 object of one type is assumed never to reside at the same address as an
5911 object of a different type, unless the types are almost the same. For
5912 example, an @code{unsigned int} can alias an @code{int}, but not a
5913 @code{void*} or a @code{double}. A character type may alias any other
5916 Pay special attention to code like this:
5929 The practice of reading from a different union member than the one most
5930 recently written to (called ``type-punning'') is common. Even with
5931 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5932 is accessed through the union type. So, the code above will work as
5933 expected. However, this code might not:
5944 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5946 @item -fstrict-overflow
5947 @opindex fstrict-overflow
5948 Allow the compiler to assume strict signed overflow rules, depending
5949 on the language being compiled. For C (and C++) this means that
5950 overflow when doing arithmetic with signed numbers is undefined, which
5951 means that the compiler may assume that it will not happen. This
5952 permits various optimizations. For example, the compiler will assume
5953 that an expression like @code{i + 10 > i} will always be true for
5954 signed @code{i}. This assumption is only valid if signed overflow is
5955 undefined, as the expression is false if @code{i + 10} overflows when
5956 using twos complement arithmetic. When this option is in effect any
5957 attempt to determine whether an operation on signed numbers will
5958 overflow must be written carefully to not actually involve overflow.
5960 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
5961 that signed overflow is fully defined: it wraps. When
5962 @option{-fwrapv} is used, there is no difference between
5963 @option{-fstrict-overflow} and @option{-fno-strict-overflow}. With
5964 @option{-fwrapv} certain types of overflow are permitted. For
5965 example, if the compiler gets an overflow when doing arithmetic on
5966 constants, the overflowed value can still be used with
5967 @option{-fwrapv}, but not otherwise.
5969 The @option{-fstrict-overflow} option is enabled at levels
5970 @option{-O2}, @option{-O3}, @option{-Os}.
5972 @item -falign-functions
5973 @itemx -falign-functions=@var{n}
5974 @opindex falign-functions
5975 Align the start of functions to the next power-of-two greater than
5976 @var{n}, skipping up to @var{n} bytes. For instance,
5977 @option{-falign-functions=32} aligns functions to the next 32-byte
5978 boundary, but @option{-falign-functions=24} would align to the next
5979 32-byte boundary only if this can be done by skipping 23 bytes or less.
5981 @option{-fno-align-functions} and @option{-falign-functions=1} are
5982 equivalent and mean that functions will not be aligned.
5984 Some assemblers only support this flag when @var{n} is a power of two;
5985 in that case, it is rounded up.
5987 If @var{n} is not specified or is zero, use a machine-dependent default.
5989 Enabled at levels @option{-O2}, @option{-O3}.
5991 @item -falign-labels
5992 @itemx -falign-labels=@var{n}
5993 @opindex falign-labels
5994 Align all branch targets to a power-of-two boundary, skipping up to
5995 @var{n} bytes like @option{-falign-functions}. This option can easily
5996 make code slower, because it must insert dummy operations for when the
5997 branch target is reached in the usual flow of the code.
5999 @option{-fno-align-labels} and @option{-falign-labels=1} are
6000 equivalent and mean that labels will not be aligned.
6002 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6003 are greater than this value, then their values are used instead.
6005 If @var{n} is not specified or is zero, use a machine-dependent default
6006 which is very likely to be @samp{1}, meaning no alignment.
6008 Enabled at levels @option{-O2}, @option{-O3}.
6011 @itemx -falign-loops=@var{n}
6012 @opindex falign-loops
6013 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6014 like @option{-falign-functions}. The hope is that the loop will be
6015 executed many times, which will make up for any execution of the dummy
6018 @option{-fno-align-loops} and @option{-falign-loops=1} are
6019 equivalent and mean that loops will not be aligned.
6021 If @var{n} is not specified or is zero, use a machine-dependent default.
6023 Enabled at levels @option{-O2}, @option{-O3}.
6026 @itemx -falign-jumps=@var{n}
6027 @opindex falign-jumps
6028 Align branch targets to a power-of-two boundary, for branch targets
6029 where the targets can only be reached by jumping, skipping up to @var{n}
6030 bytes like @option{-falign-functions}. In this case, no dummy operations
6033 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6034 equivalent and mean that loops will not be aligned.
6036 If @var{n} is not specified or is zero, use a machine-dependent default.
6038 Enabled at levels @option{-O2}, @option{-O3}.
6040 @item -funit-at-a-time
6041 @opindex funit-at-a-time
6042 Parse the whole compilation unit before starting to produce code.
6043 This allows some extra optimizations to take place but consumes
6044 more memory (in general). There are some compatibility issues
6045 with @emph{unit-at-a-time} mode:
6048 enabling @emph{unit-at-a-time} mode may change the order
6049 in which functions, variables, and top-level @code{asm} statements
6050 are emitted, and will likely break code relying on some particular
6051 ordering. The majority of such top-level @code{asm} statements,
6052 though, can be replaced by @code{section} attributes. The
6053 @option{fno-toplevel-reorder} option may be used to keep the ordering
6054 used in the input file, at the cost of some optimizations.
6057 @emph{unit-at-a-time} mode removes unreferenced static variables
6058 and functions. This may result in undefined references
6059 when an @code{asm} statement refers directly to variables or functions
6060 that are otherwise unused. In that case either the variable/function
6061 shall be listed as an operand of the @code{asm} statement operand or,
6062 in the case of top-level @code{asm} statements the attribute @code{used}
6063 shall be used on the declaration.
6066 Static functions now can use non-standard passing conventions that
6067 may break @code{asm} statements calling functions directly. Again,
6068 attribute @code{used} will prevent this behavior.
6071 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
6072 but this scheme may not be supported by future releases of GCC@.
6074 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6076 @item -fno-toplevel-reorder
6077 Do not reorder top-level functions, variables, and @code{asm}
6078 statements. Output them in the same order that they appear in the
6079 input file. When this option is used, unreferenced static variables
6080 will not be removed. This option is intended to support existing code
6081 which relies on a particular ordering. For new code, it is better to
6086 Constructs webs as commonly used for register allocation purposes and assign
6087 each web individual pseudo register. This allows the register allocation pass
6088 to operate on pseudos directly, but also strengthens several other optimization
6089 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6090 however, make debugging impossible, since variables will no longer stay in a
6093 Enabled by default with @option{-funroll-loops}.
6095 @item -fwhole-program
6096 @opindex fwhole-program
6097 Assume that the current compilation unit represents whole program being
6098 compiled. All public functions and variables with the exception of @code{main}
6099 and those merged by attribute @code{externally_visible} become static functions
6100 and in a affect gets more aggressively optimized by interprocedural optimizers.
6101 While this option is equivalent to proper use of @code{static} keyword for
6102 programs consisting of single file, in combination with option
6103 @option{--combine} this flag can be used to compile most of smaller scale C
6104 programs since the functions and variables become local for the whole combined
6105 compilation unit, not for the single source file itself.
6108 @item -fno-cprop-registers
6109 @opindex fno-cprop-registers
6110 After register allocation and post-register allocation instruction splitting,
6111 we perform a copy-propagation pass to try to reduce scheduling dependencies
6112 and occasionally eliminate the copy.
6114 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6116 @item -fprofile-generate
6117 @opindex fprofile-generate
6119 Enable options usually used for instrumenting application to produce
6120 profile useful for later recompilation with profile feedback based
6121 optimization. You must use @option{-fprofile-generate} both when
6122 compiling and when linking your program.
6124 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6127 @opindex fprofile-use
6128 Enable profile feedback directed optimizations, and optimizations
6129 generally profitable only with profile feedback available.
6131 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6132 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6134 By default, GCC emits an error message if the feedback profiles do not
6135 match the source code. This error can be turned into a warning by using
6136 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6140 The following options control compiler behavior regarding floating
6141 point arithmetic. These options trade off between speed and
6142 correctness. All must be specifically enabled.
6146 @opindex ffloat-store
6147 Do not store floating point variables in registers, and inhibit other
6148 options that might change whether a floating point value is taken from a
6151 @cindex floating point precision
6152 This option prevents undesirable excess precision on machines such as
6153 the 68000 where the floating registers (of the 68881) keep more
6154 precision than a @code{double} is supposed to have. Similarly for the
6155 x86 architecture. For most programs, the excess precision does only
6156 good, but a few programs rely on the precise definition of IEEE floating
6157 point. Use @option{-ffloat-store} for such programs, after modifying
6158 them to store all pertinent intermediate computations into variables.
6162 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
6163 @option{-fno-trapping-math}, @option{-ffinite-math-only},
6164 @option{-fno-rounding-math}, @option{-fno-signaling-nans},
6165 @option{-fno-signed-zeros} and @option{fcx-limited-range}.
6167 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6169 This option is not turned on by any @option{-O} option since
6170 it can result in incorrect output for programs which depend on
6171 an exact implementation of IEEE or ISO rules/specifications for
6172 math functions. It may, however, yield faster code for programs
6173 that do not require the guarantees of these specifications.
6175 @item -fno-math-errno
6176 @opindex fno-math-errno
6177 Do not set ERRNO after calling math functions that are executed
6178 with a single instruction, e.g., sqrt. A program that relies on
6179 IEEE exceptions for math error handling may want to use this flag
6180 for speed while maintaining IEEE arithmetic compatibility.
6182 This option is not turned on by any @option{-O} option since
6183 it can result in incorrect output for programs which depend on
6184 an exact implementation of IEEE or ISO rules/specifications for
6185 math functions. It may, however, yield faster code for programs
6186 that do not require the guarantees of these specifications.
6188 The default is @option{-fmath-errno}.
6190 On Darwin systems, the math library never sets @code{errno}. There is
6191 therefore no reason for the compiler to consider the possibility that
6192 it might, and @option{-fno-math-errno} is the default.
6194 @item -funsafe-math-optimizations
6195 @opindex funsafe-math-optimizations
6197 Allow optimizations for floating-point arithmetic that (a) assume
6198 that arguments and results are valid and (b) may violate IEEE or
6199 ANSI standards. When used at link-time, it may include libraries
6200 or startup files that change the default FPU control word or other
6201 similar optimizations.
6203 This option is not turned on by any @option{-O} option since
6204 it can result in incorrect output for programs which depend on
6205 an exact implementation of IEEE or ISO rules/specifications for
6206 math functions. It may, however, yield faster code for programs
6207 that do not require the guarantees of these specifications.
6208 Enables @option{-freciprocal-math} and @option{-fassociative-math}.
6210 The default is @option{-fno-unsafe-math-optimizations}.
6212 @item -fassociative-math
6213 @opindex -fassociative-math
6215 Allow re-association of operands in series of floating-point operations.
6216 This violates the ISO C and C++ language standard by possibly changing
6217 computation result. NOTE: re-ordering may change the sign of zero as
6218 well as ignore NaNs and inhibit or create underflow or overflow (and
6219 thus cannot be used on a code which relies on rounding behavior like
6220 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6221 and thus may not be used when ordered comparisons are required.
6222 This flag doesn't make much sense without @option{-fno-signed-zeros}
6223 or @option{-fno-trapping-math} or with @option{-frounding-math}.
6225 The default is @option{-fno-associative-math}.
6227 @item -freciprocal-math
6228 @opindex -freciprocal-math
6230 Allow the reciprocal of a value to be used instead of dividing by
6231 the value if this enables optimizations. For example @code{x / y}
6232 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6233 is subject to common subexpression elimination. Note that this loses
6234 precision and increases the number of flops operating on the value.
6236 The default is @option{-fno-reciprocal-math}.
6238 @item -ffinite-math-only
6239 @opindex ffinite-math-only
6240 Allow optimizations for floating-point arithmetic that assume
6241 that arguments and results are not NaNs or +-Infs.
6243 This option is not turned on by any @option{-O} option since
6244 it can result in incorrect output for programs which depend on
6245 an exact implementation of IEEE or ISO rules/specifications for
6246 math functions. It may, however, yield faster code for programs
6247 that do not require the guarantees of these specifications.
6249 The default is @option{-fno-finite-math-only}.
6251 @item -fno-signed-zeros
6252 @opindex fno-signed-zeros
6253 Allow optimizations for floating point arithmetic that ignore the
6254 signedness of zero. IEEE arithmetic specifies the behavior of
6255 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6256 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6257 This option implies that the sign of a zero result isn't significant.
6259 The default is @option{-fsigned-zeros}.
6261 @item -fno-trapping-math
6262 @opindex fno-trapping-math
6263 Compile code assuming that floating-point operations cannot generate
6264 user-visible traps. These traps include division by zero, overflow,
6265 underflow, inexact result and invalid operation. This option implies
6266 @option{-fno-signaling-nans}. Setting this option may allow faster
6267 code if one relies on ``non-stop'' IEEE arithmetic, for example.
6269 This option should never be turned on by any @option{-O} option since
6270 it can result in incorrect output for programs which depend on
6271 an exact implementation of IEEE or ISO rules/specifications for
6274 The default is @option{-ftrapping-math}.
6276 @item -frounding-math
6277 @opindex frounding-math
6278 Disable transformations and optimizations that assume default floating
6279 point rounding behavior. This is round-to-zero for all floating point
6280 to integer conversions, and round-to-nearest for all other arithmetic
6281 truncations. This option should be specified for programs that change
6282 the FP rounding mode dynamically, or that may be executed with a
6283 non-default rounding mode. This option disables constant folding of
6284 floating point expressions at compile-time (which may be affected by
6285 rounding mode) and arithmetic transformations that are unsafe in the
6286 presence of sign-dependent rounding modes.
6288 The default is @option{-fno-rounding-math}.
6290 This option is experimental and does not currently guarantee to
6291 disable all GCC optimizations that are affected by rounding mode.
6292 Future versions of GCC may provide finer control of this setting
6293 using C99's @code{FENV_ACCESS} pragma. This command line option
6294 will be used to specify the default state for @code{FENV_ACCESS}.
6296 @item -frtl-abstract-sequences
6297 @opindex frtl-abstract-sequences
6298 It is a size optimization method. This option is to find identical
6299 sequences of code, which can be turned into pseudo-procedures and
6300 then replace all occurrences with calls to the newly created
6301 subroutine. It is kind of an opposite of @option{-finline-functions}.
6302 This optimization runs at RTL level.
6304 @item -fsignaling-nans
6305 @opindex fsignaling-nans
6306 Compile code assuming that IEEE signaling NaNs may generate user-visible
6307 traps during floating-point operations. Setting this option disables
6308 optimizations that may change the number of exceptions visible with
6309 signaling NaNs. This option implies @option{-ftrapping-math}.
6311 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6314 The default is @option{-fno-signaling-nans}.
6316 This option is experimental and does not currently guarantee to
6317 disable all GCC optimizations that affect signaling NaN behavior.
6319 @item -fsingle-precision-constant
6320 @opindex fsingle-precision-constant
6321 Treat floating point constant as single precision constant instead of
6322 implicitly converting it to double precision constant.
6324 @item -fcx-limited-range
6325 @itemx -fno-cx-limited-range
6326 @opindex fcx-limited-range
6327 @opindex fno-cx-limited-range
6328 When enabled, this option states that a range reduction step is not
6329 needed when performing complex division. The default is
6330 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
6332 This option controls the default setting of the ISO C99
6333 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6338 The following options control optimizations that may improve
6339 performance, but are not enabled by any @option{-O} options. This
6340 section includes experimental options that may produce broken code.
6343 @item -fbranch-probabilities
6344 @opindex fbranch-probabilities
6345 After running a program compiled with @option{-fprofile-arcs}
6346 (@pxref{Debugging Options,, Options for Debugging Your Program or
6347 @command{gcc}}), you can compile it a second time using
6348 @option{-fbranch-probabilities}, to improve optimizations based on
6349 the number of times each branch was taken. When the program
6350 compiled with @option{-fprofile-arcs} exits it saves arc execution
6351 counts to a file called @file{@var{sourcename}.gcda} for each source
6352 file. The information in this data file is very dependent on the
6353 structure of the generated code, so you must use the same source code
6354 and the same optimization options for both compilations.
6356 With @option{-fbranch-probabilities}, GCC puts a
6357 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6358 These can be used to improve optimization. Currently, they are only
6359 used in one place: in @file{reorg.c}, instead of guessing which path a
6360 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6361 exactly determine which path is taken more often.
6363 @item -fprofile-values
6364 @opindex fprofile-values
6365 If combined with @option{-fprofile-arcs}, it adds code so that some
6366 data about values of expressions in the program is gathered.
6368 With @option{-fbranch-probabilities}, it reads back the data gathered
6369 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6370 notes to instructions for their later usage in optimizations.
6372 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6376 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6377 a code to gather information about values of expressions.
6379 With @option{-fbranch-probabilities}, it reads back the data gathered
6380 and actually performs the optimizations based on them.
6381 Currently the optimizations include specialization of division operation
6382 using the knowledge about the value of the denominator.
6384 @item -frename-registers
6385 @opindex frename-registers
6386 Attempt to avoid false dependencies in scheduled code by making use
6387 of registers left over after register allocation. This optimization
6388 will most benefit processors with lots of registers. Depending on the
6389 debug information format adopted by the target, however, it can
6390 make debugging impossible, since variables will no longer stay in
6391 a ``home register''.
6393 Enabled by default with @option{-funroll-loops}.
6397 Perform tail duplication to enlarge superblock size. This transformation
6398 simplifies the control flow of the function allowing other optimizations to do
6401 Enabled with @option{-fprofile-use}.
6403 @item -funroll-loops
6404 @opindex funroll-loops
6405 Unroll loops whose number of iterations can be determined at compile time or
6406 upon entry to the loop. @option{-funroll-loops} implies
6407 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6408 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6409 small constant number of iterations). This option makes code larger, and may
6410 or may not make it run faster.
6412 Enabled with @option{-fprofile-use}.
6414 @item -funroll-all-loops
6415 @opindex funroll-all-loops
6416 Unroll all loops, even if their number of iterations is uncertain when
6417 the loop is entered. This usually makes programs run more slowly.
6418 @option{-funroll-all-loops} implies the same options as
6419 @option{-funroll-loops}.
6422 @opindex fpeel-loops
6423 Peels the loops for that there is enough information that they do not
6424 roll much (from profile feedback). It also turns on complete loop peeling
6425 (i.e.@: complete removal of loops with small constant number of iterations).
6427 Enabled with @option{-fprofile-use}.
6429 @item -fmove-loop-invariants
6430 @opindex fmove-loop-invariants
6431 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6432 at level @option{-O1}
6434 @item -funswitch-loops
6435 @opindex funswitch-loops
6436 Move branches with loop invariant conditions out of the loop, with duplicates
6437 of the loop on both branches (modified according to result of the condition).
6439 @item -ffunction-sections
6440 @itemx -fdata-sections
6441 @opindex ffunction-sections
6442 @opindex fdata-sections
6443 Place each function or data item into its own section in the output
6444 file if the target supports arbitrary sections. The name of the
6445 function or the name of the data item determines the section's name
6448 Use these options on systems where the linker can perform optimizations
6449 to improve locality of reference in the instruction space. Most systems
6450 using the ELF object format and SPARC processors running Solaris 2 have
6451 linkers with such optimizations. AIX may have these optimizations in
6454 Only use these options when there are significant benefits from doing
6455 so. When you specify these options, the assembler and linker will
6456 create larger object and executable files and will also be slower.
6457 You will not be able to use @code{gprof} on all systems if you
6458 specify this option and you may have problems with debugging if
6459 you specify both this option and @option{-g}.
6461 @item -fbranch-target-load-optimize
6462 @opindex fbranch-target-load-optimize
6463 Perform branch target register load optimization before prologue / epilogue
6465 The use of target registers can typically be exposed only during reload,
6466 thus hoisting loads out of loops and doing inter-block scheduling needs
6467 a separate optimization pass.
6469 @item -fbranch-target-load-optimize2
6470 @opindex fbranch-target-load-optimize2
6471 Perform branch target register load optimization after prologue / epilogue
6474 @item -fbtr-bb-exclusive
6475 @opindex fbtr-bb-exclusive
6476 When performing branch target register load optimization, don't reuse
6477 branch target registers in within any basic block.
6479 @item -fstack-protector
6480 Emit extra code to check for buffer overflows, such as stack smashing
6481 attacks. This is done by adding a guard variable to functions with
6482 vulnerable objects. This includes functions that call alloca, and
6483 functions with buffers larger than 8 bytes. The guards are initialized
6484 when a function is entered and then checked when the function exits.
6485 If a guard check fails, an error message is printed and the program exits.
6487 @item -fstack-protector-all
6488 Like @option{-fstack-protector} except that all functions are protected.
6490 @item -fsection-anchors
6491 @opindex fsection-anchors
6492 Try to reduce the number of symbolic address calculations by using
6493 shared ``anchor'' symbols to address nearby objects. This transformation
6494 can help to reduce the number of GOT entries and GOT accesses on some
6497 For example, the implementation of the following function @code{foo}:
6501 int foo (void) @{ return a + b + c; @}
6504 would usually calculate the addresses of all three variables, but if you
6505 compile it with @option{-fsection-anchors}, it will access the variables
6506 from a common anchor point instead. The effect is similar to the
6507 following pseudocode (which isn't valid C):
6512 register int *xr = &x;
6513 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6517 Not all targets support this option.
6519 @item --param @var{name}=@var{value}
6521 In some places, GCC uses various constants to control the amount of
6522 optimization that is done. For example, GCC will not inline functions
6523 that contain more that a certain number of instructions. You can
6524 control some of these constants on the command-line using the
6525 @option{--param} option.
6527 The names of specific parameters, and the meaning of the values, are
6528 tied to the internals of the compiler, and are subject to change
6529 without notice in future releases.
6531 In each case, the @var{value} is an integer. The allowable choices for
6532 @var{name} are given in the following table:
6535 @item salias-max-implicit-fields
6536 The maximum number of fields in a variable without direct
6537 structure accesses for which structure aliasing will consider trying
6538 to track each field. The default is 5
6540 @item salias-max-array-elements
6541 The maximum number of elements an array can have and its elements
6542 still be tracked individually by structure aliasing. The default is 4
6544 @item sra-max-structure-size
6545 The maximum structure size, in bytes, at which the scalar replacement
6546 of aggregates (SRA) optimization will perform block copies. The
6547 default value, 0, implies that GCC will select the most appropriate
6550 @item sra-field-structure-ratio
6551 The threshold ratio (as a percentage) between instantiated fields and
6552 the complete structure size. We say that if the ratio of the number
6553 of bytes in instantiated fields to the number of bytes in the complete
6554 structure exceeds this parameter, then block copies are not used. The
6557 @item max-crossjump-edges
6558 The maximum number of incoming edges to consider for crossjumping.
6559 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6560 the number of edges incoming to each block. Increasing values mean
6561 more aggressive optimization, making the compile time increase with
6562 probably small improvement in executable size.
6564 @item min-crossjump-insns
6565 The minimum number of instructions which must be matched at the end
6566 of two blocks before crossjumping will be performed on them. This
6567 value is ignored in the case where all instructions in the block being
6568 crossjumped from are matched. The default value is 5.
6570 @item max-grow-copy-bb-insns
6571 The maximum code size expansion factor when copying basic blocks
6572 instead of jumping. The expansion is relative to a jump instruction.
6573 The default value is 8.
6575 @item max-goto-duplication-insns
6576 The maximum number of instructions to duplicate to a block that jumps
6577 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6578 passes, GCC factors computed gotos early in the compilation process,
6579 and unfactors them as late as possible. Only computed jumps at the
6580 end of a basic blocks with no more than max-goto-duplication-insns are
6581 unfactored. The default value is 8.
6583 @item max-delay-slot-insn-search
6584 The maximum number of instructions to consider when looking for an
6585 instruction to fill a delay slot. If more than this arbitrary number of
6586 instructions is searched, the time savings from filling the delay slot
6587 will be minimal so stop searching. Increasing values mean more
6588 aggressive optimization, making the compile time increase with probably
6589 small improvement in executable run time.
6591 @item max-delay-slot-live-search
6592 When trying to fill delay slots, the maximum number of instructions to
6593 consider when searching for a block with valid live register
6594 information. Increasing this arbitrarily chosen value means more
6595 aggressive optimization, increasing the compile time. This parameter
6596 should be removed when the delay slot code is rewritten to maintain the
6599 @item max-gcse-memory
6600 The approximate maximum amount of memory that will be allocated in
6601 order to perform the global common subexpression elimination
6602 optimization. If more memory than specified is required, the
6603 optimization will not be done.
6605 @item max-gcse-passes
6606 The maximum number of passes of GCSE to run. The default is 1.
6608 @item max-pending-list-length
6609 The maximum number of pending dependencies scheduling will allow
6610 before flushing the current state and starting over. Large functions
6611 with few branches or calls can create excessively large lists which
6612 needlessly consume memory and resources.
6614 @item max-inline-insns-single
6615 Several parameters control the tree inliner used in gcc.
6616 This number sets the maximum number of instructions (counted in GCC's
6617 internal representation) in a single function that the tree inliner
6618 will consider for inlining. This only affects functions declared
6619 inline and methods implemented in a class declaration (C++).
6620 The default value is 450.
6622 @item max-inline-insns-auto
6623 When you use @option{-finline-functions} (included in @option{-O3}),
6624 a lot of functions that would otherwise not be considered for inlining
6625 by the compiler will be investigated. To those functions, a different
6626 (more restrictive) limit compared to functions declared inline can
6628 The default value is 90.
6630 @item large-function-insns
6631 The limit specifying really large functions. For functions larger than this
6632 limit after inlining inlining is constrained by
6633 @option{--param large-function-growth}. This parameter is useful primarily
6634 to avoid extreme compilation time caused by non-linear algorithms used by the
6636 This parameter is ignored when @option{-funit-at-a-time} is not used.
6637 The default value is 2700.
6639 @item large-function-growth
6640 Specifies maximal growth of large function caused by inlining in percents.
6641 This parameter is ignored when @option{-funit-at-a-time} is not used.
6642 The default value is 100 which limits large function growth to 2.0 times
6645 @item large-unit-insns
6646 The limit specifying large translation unit. Growth caused by inlining of
6647 units larger than this limit is limited by @option{--param inline-unit-growth}.
6648 For small units this might be too tight (consider unit consisting of function A
6649 that is inline and B that just calls A three time. If B is small relative to
6650 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6651 large units consisting of small inlininable functions however the overall unit
6652 growth limit is needed to avoid exponential explosion of code size. Thus for
6653 smaller units, the size is increased to @option{--param large-unit-insns}
6654 before applying @option{--param inline-unit-growth}. The default is 10000
6656 @item inline-unit-growth
6657 Specifies maximal overall growth of the compilation unit caused by inlining.
6658 This parameter is ignored when @option{-funit-at-a-time} is not used.
6659 The default value is 30 which limits unit growth to 1.3 times the original
6662 @item large-stack-frame
6663 The limit specifying large stack frames. While inlining the algorithm is trying
6664 to not grow past this limit too much. Default value is 256 bytes.
6666 @item large-stack-frame-growth
6667 Specifies maximal growth of large stack frames caused by inlining in percents.
6668 The default value is 1000 which limits large stack frame growth to 11 times
6671 @item max-inline-insns-recursive
6672 @itemx max-inline-insns-recursive-auto
6673 Specifies maximum number of instructions out-of-line copy of self recursive inline
6674 function can grow into by performing recursive inlining.
6676 For functions declared inline @option{--param max-inline-insns-recursive} is
6677 taken into account. For function not declared inline, recursive inlining
6678 happens only when @option{-finline-functions} (included in @option{-O3}) is
6679 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6680 default value is 450.
6682 @item max-inline-recursive-depth
6683 @itemx max-inline-recursive-depth-auto
6684 Specifies maximum recursion depth used by the recursive inlining.
6686 For functions declared inline @option{--param max-inline-recursive-depth} is
6687 taken into account. For function not declared inline, recursive inlining
6688 happens only when @option{-finline-functions} (included in @option{-O3}) is
6689 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6690 default value is 450.
6692 @item min-inline-recursive-probability
6693 Recursive inlining is profitable only for function having deep recursion
6694 in average and can hurt for function having little recursion depth by
6695 increasing the prologue size or complexity of function body to other
6698 When profile feedback is available (see @option{-fprofile-generate}) the actual
6699 recursion depth can be guessed from probability that function will recurse via
6700 given call expression. This parameter limits inlining only to call expression
6701 whose probability exceeds given threshold (in percents). The default value is
6704 @item inline-call-cost
6705 Specify cost of call instruction relative to simple arithmetics operations
6706 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6707 functions and at the same time increases size of leaf function that is believed to
6708 reduce function size by being inlined. In effect it increases amount of
6709 inlining for code having large abstraction penalty (many functions that just
6710 pass the arguments to other functions) and decrease inlining for code with low
6711 abstraction penalty. The default value is 16.
6713 @item min-vect-loop-bound
6714 The minimum number of iterations under which a loop will not get vectorized
6715 when @option{-ftree-vectorize} is used. The number of iterations after
6716 vectorization needs to be greater than the value specified by this option
6717 to allow vectorization. The default value is 0.
6719 @item max-unrolled-insns
6720 The maximum number of instructions that a loop should have if that loop
6721 is unrolled, and if the loop is unrolled, it determines how many times
6722 the loop code is unrolled.
6724 @item max-average-unrolled-insns
6725 The maximum number of instructions biased by probabilities of their execution
6726 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6727 it determines how many times the loop code is unrolled.
6729 @item max-unroll-times
6730 The maximum number of unrollings of a single loop.
6732 @item max-peeled-insns
6733 The maximum number of instructions that a loop should have if that loop
6734 is peeled, and if the loop is peeled, it determines how many times
6735 the loop code is peeled.
6737 @item max-peel-times
6738 The maximum number of peelings of a single loop.
6740 @item max-completely-peeled-insns
6741 The maximum number of insns of a completely peeled loop.
6743 @item max-completely-peel-times
6744 The maximum number of iterations of a loop to be suitable for complete peeling.
6746 @item max-unswitch-insns
6747 The maximum number of insns of an unswitched loop.
6749 @item max-unswitch-level
6750 The maximum number of branches unswitched in a single loop.
6753 The minimum cost of an expensive expression in the loop invariant motion.
6755 @item iv-consider-all-candidates-bound
6756 Bound on number of candidates for induction variables below that
6757 all candidates are considered for each use in induction variable
6758 optimizations. Only the most relevant candidates are considered
6759 if there are more candidates, to avoid quadratic time complexity.
6761 @item iv-max-considered-uses
6762 The induction variable optimizations give up on loops that contain more
6763 induction variable uses.
6765 @item iv-always-prune-cand-set-bound
6766 If number of candidates in the set is smaller than this value,
6767 we always try to remove unnecessary ivs from the set during its
6768 optimization when a new iv is added to the set.
6770 @item scev-max-expr-size
6771 Bound on size of expressions used in the scalar evolutions analyzer.
6772 Large expressions slow the analyzer.
6774 @item omega-max-vars
6775 The maximum number of variables in an Omega constraint system.
6776 The default value is 128.
6778 @item omega-max-geqs
6779 The maximum number of inequalities in an Omega constraint system.
6780 The default value is 256.
6783 The maximum number of equalities in an Omega constraint system.
6784 The default value is 128.
6786 @item omega-max-wild-cards
6787 The maximum number of wildcard variables that the Omega solver will
6788 be able to insert. The default value is 18.
6790 @item omega-hash-table-size
6791 The size of the hash table in the Omega solver. The default value is
6794 @item omega-max-keys
6795 The maximal number of keys used by the Omega solver. The default
6798 @item omega-eliminate-redundant-constraints
6799 When set to 1, use expensive methods to eliminate all redundant
6800 constraints. The default value is 0.
6802 @item vect-max-version-for-alignment-checks
6803 The maximum number of runtime checks that can be performed when
6804 doing loop versioning for alignment in the vectorizer. See option
6805 ftree-vect-loop-version for more information.
6807 @item vect-max-version-for-alias-checks
6808 The maximum number of runtime checks that can be performed when
6809 doing loop versioning for alias in the vectorizer. See option
6810 ftree-vect-loop-version for more information.
6812 @item max-iterations-to-track
6814 The maximum number of iterations of a loop the brute force algorithm
6815 for analysis of # of iterations of the loop tries to evaluate.
6817 @item hot-bb-count-fraction
6818 Select fraction of the maximal count of repetitions of basic block in program
6819 given basic block needs to have to be considered hot.
6821 @item hot-bb-frequency-fraction
6822 Select fraction of the maximal frequency of executions of basic block in
6823 function given basic block needs to have to be considered hot
6825 @item max-predicted-iterations
6826 The maximum number of loop iterations we predict statically. This is useful
6827 in cases where function contain single loop with known bound and other loop
6828 with unknown. We predict the known number of iterations correctly, while
6829 the unknown number of iterations average to roughly 10. This means that the
6830 loop without bounds would appear artificially cold relative to the other one.
6832 @item tracer-dynamic-coverage
6833 @itemx tracer-dynamic-coverage-feedback
6835 This value is used to limit superblock formation once the given percentage of
6836 executed instructions is covered. This limits unnecessary code size
6839 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6840 feedback is available. The real profiles (as opposed to statically estimated
6841 ones) are much less balanced allowing the threshold to be larger value.
6843 @item tracer-max-code-growth
6844 Stop tail duplication once code growth has reached given percentage. This is
6845 rather hokey argument, as most of the duplicates will be eliminated later in
6846 cross jumping, so it may be set to much higher values than is the desired code
6849 @item tracer-min-branch-ratio
6851 Stop reverse growth when the reverse probability of best edge is less than this
6852 threshold (in percent).
6854 @item tracer-min-branch-ratio
6855 @itemx tracer-min-branch-ratio-feedback
6857 Stop forward growth if the best edge do have probability lower than this
6860 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6861 compilation for profile feedback and one for compilation without. The value
6862 for compilation with profile feedback needs to be more conservative (higher) in
6863 order to make tracer effective.
6865 @item max-cse-path-length
6867 Maximum number of basic blocks on path that cse considers. The default is 10.
6870 The maximum instructions CSE process before flushing. The default is 1000.
6872 @item max-aliased-vops
6874 Maximum number of virtual operands per function allowed to represent
6875 aliases before triggering the alias partitioning heuristic. Alias
6876 partitioning reduces compile times and memory consumption needed for
6877 aliasing at the expense of precision loss in alias information. The
6878 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
6881 Notice that if a function contains more memory statements than the
6882 value of this parameter, it is not really possible to achieve this
6883 reduction. In this case, the compiler will use the number of memory
6884 statements as the value for @option{max-aliased-vops}.
6886 @item avg-aliased-vops
6888 Average number of virtual operands per statement allowed to represent
6889 aliases before triggering the alias partitioning heuristic. This
6890 works in conjunction with @option{max-aliased-vops}. If a function
6891 contains more than @option{max-aliased-vops} virtual operators, then
6892 memory symbols will be grouped into memory partitions until either the
6893 total number of virtual operators is below @option{max-aliased-vops}
6894 or the average number of virtual operators per memory statement is
6895 below @option{avg-aliased-vops}. The default value for this parameter
6896 is 1 for -O1 and -O2, and 3 for -O3.
6898 @item ggc-min-expand
6900 GCC uses a garbage collector to manage its own memory allocation. This
6901 parameter specifies the minimum percentage by which the garbage
6902 collector's heap should be allowed to expand between collections.
6903 Tuning this may improve compilation speed; it has no effect on code
6906 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6907 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6908 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6909 GCC is not able to calculate RAM on a particular platform, the lower
6910 bound of 30% is used. Setting this parameter and
6911 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6912 every opportunity. This is extremely slow, but can be useful for
6915 @item ggc-min-heapsize
6917 Minimum size of the garbage collector's heap before it begins bothering
6918 to collect garbage. The first collection occurs after the heap expands
6919 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6920 tuning this may improve compilation speed, and has no effect on code
6923 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6924 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6925 with a lower bound of 4096 (four megabytes) and an upper bound of
6926 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6927 particular platform, the lower bound is used. Setting this parameter
6928 very large effectively disables garbage collection. Setting this
6929 parameter and @option{ggc-min-expand} to zero causes a full collection
6930 to occur at every opportunity.
6932 @item max-reload-search-insns
6933 The maximum number of instruction reload should look backward for equivalent
6934 register. Increasing values mean more aggressive optimization, making the
6935 compile time increase with probably slightly better performance. The default
6938 @item max-cselib-memory-locations
6939 The maximum number of memory locations cselib should take into account.
6940 Increasing values mean more aggressive optimization, making the compile time
6941 increase with probably slightly better performance. The default value is 500.
6943 @item max-flow-memory-locations
6944 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6945 The default value is 100.
6947 @item reorder-blocks-duplicate
6948 @itemx reorder-blocks-duplicate-feedback
6950 Used by basic block reordering pass to decide whether to use unconditional
6951 branch or duplicate the code on its destination. Code is duplicated when its
6952 estimated size is smaller than this value multiplied by the estimated size of
6953 unconditional jump in the hot spots of the program.
6955 The @option{reorder-block-duplicate-feedback} is used only when profile
6956 feedback is available and may be set to higher values than
6957 @option{reorder-block-duplicate} since information about the hot spots is more
6960 @item max-sched-ready-insns
6961 The maximum number of instructions ready to be issued the scheduler should
6962 consider at any given time during the first scheduling pass. Increasing
6963 values mean more thorough searches, making the compilation time increase
6964 with probably little benefit. The default value is 100.
6966 @item max-sched-region-blocks
6967 The maximum number of blocks in a region to be considered for
6968 interblock scheduling. The default value is 10.
6970 @item max-sched-region-insns
6971 The maximum number of insns in a region to be considered for
6972 interblock scheduling. The default value is 100.
6975 The minimum probability (in percents) of reaching a source block
6976 for interblock speculative scheduling. The default value is 40.
6978 @item max-sched-extend-regions-iters
6979 The maximum number of iterations through CFG to extend regions.
6980 0 - disable region extension,
6981 N - do at most N iterations.
6982 The default value is 0.
6984 @item max-sched-insn-conflict-delay
6985 The maximum conflict delay for an insn to be considered for speculative motion.
6986 The default value is 3.
6988 @item sched-spec-prob-cutoff
6989 The minimal probability of speculation success (in percents), so that
6990 speculative insn will be scheduled.
6991 The default value is 40.
6993 @item max-last-value-rtl
6995 The maximum size measured as number of RTLs that can be recorded in an expression
6996 in combiner for a pseudo register as last known value of that register. The default
6999 @item integer-share-limit
7000 Small integer constants can use a shared data structure, reducing the
7001 compiler's memory usage and increasing its speed. This sets the maximum
7002 value of a shared integer constant's. The default value is 256.
7004 @item min-virtual-mappings
7005 Specifies the minimum number of virtual mappings in the incremental
7006 SSA updater that should be registered to trigger the virtual mappings
7007 heuristic defined by virtual-mappings-ratio. The default value is
7010 @item virtual-mappings-ratio
7011 If the number of virtual mappings is virtual-mappings-ratio bigger
7012 than the number of virtual symbols to be updated, then the incremental
7013 SSA updater switches to a full update for those symbols. The default
7016 @item ssp-buffer-size
7017 The minimum size of buffers (i.e. arrays) that will receive stack smashing
7018 protection when @option{-fstack-protection} is used.
7020 @item max-jump-thread-duplication-stmts
7021 Maximum number of statements allowed in a block that needs to be
7022 duplicated when threading jumps.
7024 @item max-fields-for-field-sensitive
7025 Maximum number of fields in a structure we will treat in
7026 a field sensitive manner during pointer analysis.
7028 @item prefetch-latency
7029 Estimate on average number of instructions that are executed before
7030 prefetch finishes. The distance we prefetch ahead is proportional
7031 to this constant. Increasing this number may also lead to less
7032 streams being prefetched (see @option{simultaneous-prefetches}).
7034 @item simultaneous-prefetches
7035 Maximum number of prefetches that can run at the same time.
7037 @item l1-cache-line-size
7038 The size of cache line in L1 cache, in bytes.
7041 The size of L1 cache, in kilobytes.
7044 The size of L2 cache, in kilobytes.
7046 @item use-canonical-types
7047 Whether the compiler should use the ``canonical'' type system. By
7048 default, this should always be 1, which uses a more efficient internal
7049 mechanism for comparing types in C++ and Objective-C++. However, if
7050 bugs in the canonical type system are causing compilation failures,
7051 set this value to 0 to disable canonical types.
7056 @node Preprocessor Options
7057 @section Options Controlling the Preprocessor
7058 @cindex preprocessor options
7059 @cindex options, preprocessor
7061 These options control the C preprocessor, which is run on each C source
7062 file before actual compilation.
7064 If you use the @option{-E} option, nothing is done except preprocessing.
7065 Some of these options make sense only together with @option{-E} because
7066 they cause the preprocessor output to be unsuitable for actual
7071 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7072 and pass @var{option} directly through to the preprocessor. If
7073 @var{option} contains commas, it is split into multiple options at the
7074 commas. However, many options are modified, translated or interpreted
7075 by the compiler driver before being passed to the preprocessor, and
7076 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7077 interface is undocumented and subject to change, so whenever possible
7078 you should avoid using @option{-Wp} and let the driver handle the
7081 @item -Xpreprocessor @var{option}
7082 @opindex preprocessor
7083 Pass @var{option} as an option to the preprocessor. You can use this to
7084 supply system-specific preprocessor options which GCC does not know how to
7087 If you want to pass an option that takes an argument, you must use
7088 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7091 @include cppopts.texi
7093 @node Assembler Options
7094 @section Passing Options to the Assembler
7096 @c prevent bad page break with this line
7097 You can pass options to the assembler.
7100 @item -Wa,@var{option}
7102 Pass @var{option} as an option to the assembler. If @var{option}
7103 contains commas, it is split into multiple options at the commas.
7105 @item -Xassembler @var{option}
7107 Pass @var{option} as an option to the assembler. You can use this to
7108 supply system-specific assembler options which GCC does not know how to
7111 If you want to pass an option that takes an argument, you must use
7112 @option{-Xassembler} twice, once for the option and once for the argument.
7117 @section Options for Linking
7118 @cindex link options
7119 @cindex options, linking
7121 These options come into play when the compiler links object files into
7122 an executable output file. They are meaningless if the compiler is
7123 not doing a link step.
7127 @item @var{object-file-name}
7128 A file name that does not end in a special recognized suffix is
7129 considered to name an object file or library. (Object files are
7130 distinguished from libraries by the linker according to the file
7131 contents.) If linking is done, these object files are used as input
7140 If any of these options is used, then the linker is not run, and
7141 object file names should not be used as arguments. @xref{Overall
7145 @item -l@var{library}
7146 @itemx -l @var{library}
7148 Search the library named @var{library} when linking. (The second
7149 alternative with the library as a separate argument is only for
7150 POSIX compliance and is not recommended.)
7152 It makes a difference where in the command you write this option; the
7153 linker searches and processes libraries and object files in the order they
7154 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7155 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7156 to functions in @samp{z}, those functions may not be loaded.
7158 The linker searches a standard list of directories for the library,
7159 which is actually a file named @file{lib@var{library}.a}. The linker
7160 then uses this file as if it had been specified precisely by name.
7162 The directories searched include several standard system directories
7163 plus any that you specify with @option{-L}.
7165 Normally the files found this way are library files---archive files
7166 whose members are object files. The linker handles an archive file by
7167 scanning through it for members which define symbols that have so far
7168 been referenced but not defined. But if the file that is found is an
7169 ordinary object file, it is linked in the usual fashion. The only
7170 difference between using an @option{-l} option and specifying a file name
7171 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7172 and searches several directories.
7176 You need this special case of the @option{-l} option in order to
7177 link an Objective-C or Objective-C++ program.
7180 @opindex nostartfiles
7181 Do not use the standard system startup files when linking.
7182 The standard system libraries are used normally, unless @option{-nostdlib}
7183 or @option{-nodefaultlibs} is used.
7185 @item -nodefaultlibs
7186 @opindex nodefaultlibs
7187 Do not use the standard system libraries when linking.
7188 Only the libraries you specify will be passed to the linker.
7189 The standard startup files are used normally, unless @option{-nostartfiles}
7190 is used. The compiler may generate calls to @code{memcmp},
7191 @code{memset}, @code{memcpy} and @code{memmove}.
7192 These entries are usually resolved by entries in
7193 libc. These entry points should be supplied through some other
7194 mechanism when this option is specified.
7198 Do not use the standard system startup files or libraries when linking.
7199 No startup files and only the libraries you specify will be passed to
7200 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7201 @code{memcpy} and @code{memmove}.
7202 These entries are usually resolved by entries in
7203 libc. These entry points should be supplied through some other
7204 mechanism when this option is specified.
7206 @cindex @option{-lgcc}, use with @option{-nostdlib}
7207 @cindex @option{-nostdlib} and unresolved references
7208 @cindex unresolved references and @option{-nostdlib}
7209 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7210 @cindex @option{-nodefaultlibs} and unresolved references
7211 @cindex unresolved references and @option{-nodefaultlibs}
7212 One of the standard libraries bypassed by @option{-nostdlib} and
7213 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7214 that GCC uses to overcome shortcomings of particular machines, or special
7215 needs for some languages.
7216 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7217 Collection (GCC) Internals},
7218 for more discussion of @file{libgcc.a}.)
7219 In most cases, you need @file{libgcc.a} even when you want to avoid
7220 other standard libraries. In other words, when you specify @option{-nostdlib}
7221 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7222 This ensures that you have no unresolved references to internal GCC
7223 library subroutines. (For example, @samp{__main}, used to ensure C++
7224 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7225 GNU Compiler Collection (GCC) Internals}.)
7229 Produce a position independent executable on targets which support it.
7230 For predictable results, you must also specify the same set of options
7231 that were used to generate code (@option{-fpie}, @option{-fPIE},
7232 or model suboptions) when you specify this option.
7236 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7237 that support it. This instructs the linker to add all symbols, not
7238 only used ones, to the dynamic symbol table. This option is needed
7239 for some uses of @code{dlopen} or to allow obtaining backtraces
7240 from within a program.
7244 Remove all symbol table and relocation information from the executable.
7248 On systems that support dynamic linking, this prevents linking with the shared
7249 libraries. On other systems, this option has no effect.
7253 Produce a shared object which can then be linked with other objects to
7254 form an executable. Not all systems support this option. For predictable
7255 results, you must also specify the same set of options that were used to
7256 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7257 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7258 needs to build supplementary stub code for constructors to work. On
7259 multi-libbed systems, @samp{gcc -shared} must select the correct support
7260 libraries to link against. Failing to supply the correct flags may lead
7261 to subtle defects. Supplying them in cases where they are not necessary
7264 @item -shared-libgcc
7265 @itemx -static-libgcc
7266 @opindex shared-libgcc
7267 @opindex static-libgcc
7268 On systems that provide @file{libgcc} as a shared library, these options
7269 force the use of either the shared or static version respectively.
7270 If no shared version of @file{libgcc} was built when the compiler was
7271 configured, these options have no effect.
7273 There are several situations in which an application should use the
7274 shared @file{libgcc} instead of the static version. The most common
7275 of these is when the application wishes to throw and catch exceptions
7276 across different shared libraries. In that case, each of the libraries
7277 as well as the application itself should use the shared @file{libgcc}.
7279 Therefore, the G++ and GCJ drivers automatically add
7280 @option{-shared-libgcc} whenever you build a shared library or a main
7281 executable, because C++ and Java programs typically use exceptions, so
7282 this is the right thing to do.
7284 If, instead, you use the GCC driver to create shared libraries, you may
7285 find that they will not always be linked with the shared @file{libgcc}.
7286 If GCC finds, at its configuration time, that you have a non-GNU linker
7287 or a GNU linker that does not support option @option{--eh-frame-hdr},
7288 it will link the shared version of @file{libgcc} into shared libraries
7289 by default. Otherwise, it will take advantage of the linker and optimize
7290 away the linking with the shared version of @file{libgcc}, linking with
7291 the static version of libgcc by default. This allows exceptions to
7292 propagate through such shared libraries, without incurring relocation
7293 costs at library load time.
7295 However, if a library or main executable is supposed to throw or catch
7296 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7297 for the languages used in the program, or using the option
7298 @option{-shared-libgcc}, such that it is linked with the shared
7303 Bind references to global symbols when building a shared object. Warn
7304 about any unresolved references (unless overridden by the link editor
7305 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7308 @item -Xlinker @var{option}
7310 Pass @var{option} as an option to the linker. You can use this to
7311 supply system-specific linker options which GCC does not know how to
7314 If you want to pass an option that takes an argument, you must use
7315 @option{-Xlinker} twice, once for the option and once for the argument.
7316 For example, to pass @option{-assert definitions}, you must write
7317 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7318 @option{-Xlinker "-assert definitions"}, because this passes the entire
7319 string as a single argument, which is not what the linker expects.
7321 @item -Wl,@var{option}
7323 Pass @var{option} as an option to the linker. If @var{option} contains
7324 commas, it is split into multiple options at the commas.
7326 @item -u @var{symbol}
7328 Pretend the symbol @var{symbol} is undefined, to force linking of
7329 library modules to define it. You can use @option{-u} multiple times with
7330 different symbols to force loading of additional library modules.
7333 @node Directory Options
7334 @section Options for Directory Search
7335 @cindex directory options
7336 @cindex options, directory search
7339 These options specify directories to search for header files, for
7340 libraries and for parts of the compiler:
7345 Add the directory @var{dir} to the head of the list of directories to be
7346 searched for header files. This can be used to override a system header
7347 file, substituting your own version, since these directories are
7348 searched before the system header file directories. However, you should
7349 not use this option to add directories that contain vendor-supplied
7350 system header files (use @option{-isystem} for that). If you use more than
7351 one @option{-I} option, the directories are scanned in left-to-right
7352 order; the standard system directories come after.
7354 If a standard system include directory, or a directory specified with
7355 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7356 option will be ignored. The directory will still be searched but as a
7357 system directory at its normal position in the system include chain.
7358 This is to ensure that GCC's procedure to fix buggy system headers and
7359 the ordering for the include_next directive are not inadvertently changed.
7360 If you really need to change the search order for system directories,
7361 use the @option{-nostdinc} and/or @option{-isystem} options.
7363 @item -iquote@var{dir}
7365 Add the directory @var{dir} to the head of the list of directories to
7366 be searched for header files only for the case of @samp{#include
7367 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7368 otherwise just like @option{-I}.
7372 Add directory @var{dir} to the list of directories to be searched
7375 @item -B@var{prefix}
7377 This option specifies where to find the executables, libraries,
7378 include files, and data files of the compiler itself.
7380 The compiler driver program runs one or more of the subprograms
7381 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7382 @var{prefix} as a prefix for each program it tries to run, both with and
7383 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7385 For each subprogram to be run, the compiler driver first tries the
7386 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7387 was not specified, the driver tries two standard prefixes, which are
7388 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7389 those results in a file name that is found, the unmodified program
7390 name is searched for using the directories specified in your
7391 @env{PATH} environment variable.
7393 The compiler will check to see if the path provided by the @option{-B}
7394 refers to a directory, and if necessary it will add a directory
7395 separator character at the end of the path.
7397 @option{-B} prefixes that effectively specify directory names also apply
7398 to libraries in the linker, because the compiler translates these
7399 options into @option{-L} options for the linker. They also apply to
7400 includes files in the preprocessor, because the compiler translates these
7401 options into @option{-isystem} options for the preprocessor. In this case,
7402 the compiler appends @samp{include} to the prefix.
7404 The run-time support file @file{libgcc.a} can also be searched for using
7405 the @option{-B} prefix, if needed. If it is not found there, the two
7406 standard prefixes above are tried, and that is all. The file is left
7407 out of the link if it is not found by those means.
7409 Another way to specify a prefix much like the @option{-B} prefix is to use
7410 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7413 As a special kludge, if the path provided by @option{-B} is
7414 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7415 9, then it will be replaced by @file{[dir/]include}. This is to help
7416 with boot-strapping the compiler.
7418 @item -specs=@var{file}
7420 Process @var{file} after the compiler reads in the standard @file{specs}
7421 file, in order to override the defaults that the @file{gcc} driver
7422 program uses when determining what switches to pass to @file{cc1},
7423 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7424 @option{-specs=@var{file}} can be specified on the command line, and they
7425 are processed in order, from left to right.
7427 @item --sysroot=@var{dir}
7429 Use @var{dir} as the logical root directory for headers and libraries.
7430 For example, if the compiler would normally search for headers in
7431 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7432 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7434 If you use both this option and the @option{-isysroot} option, then
7435 the @option{--sysroot} option will apply to libraries, but the
7436 @option{-isysroot} option will apply to header files.
7438 The GNU linker (beginning with version 2.16) has the necessary support
7439 for this option. If your linker does not support this option, the
7440 header file aspect of @option{--sysroot} will still work, but the
7441 library aspect will not.
7445 This option has been deprecated. Please use @option{-iquote} instead for
7446 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7447 Any directories you specify with @option{-I} options before the @option{-I-}
7448 option are searched only for the case of @samp{#include "@var{file}"};
7449 they are not searched for @samp{#include <@var{file}>}.
7451 If additional directories are specified with @option{-I} options after
7452 the @option{-I-}, these directories are searched for all @samp{#include}
7453 directives. (Ordinarily @emph{all} @option{-I} directories are used
7456 In addition, the @option{-I-} option inhibits the use of the current
7457 directory (where the current input file came from) as the first search
7458 directory for @samp{#include "@var{file}"}. There is no way to
7459 override this effect of @option{-I-}. With @option{-I.} you can specify
7460 searching the directory which was current when the compiler was
7461 invoked. That is not exactly the same as what the preprocessor does
7462 by default, but it is often satisfactory.
7464 @option{-I-} does not inhibit the use of the standard system directories
7465 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7472 @section Specifying subprocesses and the switches to pass to them
7475 @command{gcc} is a driver program. It performs its job by invoking a
7476 sequence of other programs to do the work of compiling, assembling and
7477 linking. GCC interprets its command-line parameters and uses these to
7478 deduce which programs it should invoke, and which command-line options
7479 it ought to place on their command lines. This behavior is controlled
7480 by @dfn{spec strings}. In most cases there is one spec string for each
7481 program that GCC can invoke, but a few programs have multiple spec
7482 strings to control their behavior. The spec strings built into GCC can
7483 be overridden by using the @option{-specs=} command-line switch to specify
7486 @dfn{Spec files} are plaintext files that are used to construct spec
7487 strings. They consist of a sequence of directives separated by blank
7488 lines. The type of directive is determined by the first non-whitespace
7489 character on the line and it can be one of the following:
7492 @item %@var{command}
7493 Issues a @var{command} to the spec file processor. The commands that can
7497 @item %include <@var{file}>
7499 Search for @var{file} and insert its text at the current point in the
7502 @item %include_noerr <@var{file}>
7503 @cindex %include_noerr
7504 Just like @samp{%include}, but do not generate an error message if the include
7505 file cannot be found.
7507 @item %rename @var{old_name} @var{new_name}
7509 Rename the spec string @var{old_name} to @var{new_name}.
7513 @item *[@var{spec_name}]:
7514 This tells the compiler to create, override or delete the named spec
7515 string. All lines after this directive up to the next directive or
7516 blank line are considered to be the text for the spec string. If this
7517 results in an empty string then the spec will be deleted. (Or, if the
7518 spec did not exist, then nothing will happened.) Otherwise, if the spec
7519 does not currently exist a new spec will be created. If the spec does
7520 exist then its contents will be overridden by the text of this
7521 directive, unless the first character of that text is the @samp{+}
7522 character, in which case the text will be appended to the spec.
7524 @item [@var{suffix}]:
7525 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7526 and up to the next directive or blank line are considered to make up the
7527 spec string for the indicated suffix. When the compiler encounters an
7528 input file with the named suffix, it will processes the spec string in
7529 order to work out how to compile that file. For example:
7536 This says that any input file whose name ends in @samp{.ZZ} should be
7537 passed to the program @samp{z-compile}, which should be invoked with the
7538 command-line switch @option{-input} and with the result of performing the
7539 @samp{%i} substitution. (See below.)
7541 As an alternative to providing a spec string, the text that follows a
7542 suffix directive can be one of the following:
7545 @item @@@var{language}
7546 This says that the suffix is an alias for a known @var{language}. This is
7547 similar to using the @option{-x} command-line switch to GCC to specify a
7548 language explicitly. For example:
7555 Says that .ZZ files are, in fact, C++ source files.
7558 This causes an error messages saying:
7561 @var{name} compiler not installed on this system.
7565 GCC already has an extensive list of suffixes built into it.
7566 This directive will add an entry to the end of the list of suffixes, but
7567 since the list is searched from the end backwards, it is effectively
7568 possible to override earlier entries using this technique.
7572 GCC has the following spec strings built into it. Spec files can
7573 override these strings or create their own. Note that individual
7574 targets can also add their own spec strings to this list.
7577 asm Options to pass to the assembler
7578 asm_final Options to pass to the assembler post-processor
7579 cpp Options to pass to the C preprocessor
7580 cc1 Options to pass to the C compiler
7581 cc1plus Options to pass to the C++ compiler
7582 endfile Object files to include at the end of the link
7583 link Options to pass to the linker
7584 lib Libraries to include on the command line to the linker
7585 libgcc Decides which GCC support library to pass to the linker
7586 linker Sets the name of the linker
7587 predefines Defines to be passed to the C preprocessor
7588 signed_char Defines to pass to CPP to say whether @code{char} is signed
7590 startfile Object files to include at the start of the link
7593 Here is a small example of a spec file:
7599 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7602 This example renames the spec called @samp{lib} to @samp{old_lib} and
7603 then overrides the previous definition of @samp{lib} with a new one.
7604 The new definition adds in some extra command-line options before
7605 including the text of the old definition.
7607 @dfn{Spec strings} are a list of command-line options to be passed to their
7608 corresponding program. In addition, the spec strings can contain
7609 @samp{%}-prefixed sequences to substitute variable text or to
7610 conditionally insert text into the command line. Using these constructs
7611 it is possible to generate quite complex command lines.
7613 Here is a table of all defined @samp{%}-sequences for spec
7614 strings. Note that spaces are not generated automatically around the
7615 results of expanding these sequences. Therefore you can concatenate them
7616 together or combine them with constant text in a single argument.
7620 Substitute one @samp{%} into the program name or argument.
7623 Substitute the name of the input file being processed.
7626 Substitute the basename of the input file being processed.
7627 This is the substring up to (and not including) the last period
7628 and not including the directory.
7631 This is the same as @samp{%b}, but include the file suffix (text after
7635 Marks the argument containing or following the @samp{%d} as a
7636 temporary file name, so that that file will be deleted if GCC exits
7637 successfully. Unlike @samp{%g}, this contributes no text to the
7640 @item %g@var{suffix}
7641 Substitute a file name that has suffix @var{suffix} and is chosen
7642 once per compilation, and mark the argument in the same way as
7643 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7644 name is now chosen in a way that is hard to predict even when previously
7645 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7646 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7647 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7648 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7649 was simply substituted with a file name chosen once per compilation,
7650 without regard to any appended suffix (which was therefore treated
7651 just like ordinary text), making such attacks more likely to succeed.
7653 @item %u@var{suffix}
7654 Like @samp{%g}, but generates a new temporary file name even if
7655 @samp{%u@var{suffix}} was already seen.
7657 @item %U@var{suffix}
7658 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7659 new one if there is no such last file name. In the absence of any
7660 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7661 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7662 would involve the generation of two distinct file names, one
7663 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7664 simply substituted with a file name chosen for the previous @samp{%u},
7665 without regard to any appended suffix.
7667 @item %j@var{suffix}
7668 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7669 writable, and if save-temps is off; otherwise, substitute the name
7670 of a temporary file, just like @samp{%u}. This temporary file is not
7671 meant for communication between processes, but rather as a junk
7674 @item %|@var{suffix}
7675 @itemx %m@var{suffix}
7676 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7677 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7678 all. These are the two most common ways to instruct a program that it
7679 should read from standard input or write to standard output. If you
7680 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7681 construct: see for example @file{f/lang-specs.h}.
7683 @item %.@var{SUFFIX}
7684 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7685 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7686 terminated by the next space or %.
7689 Marks the argument containing or following the @samp{%w} as the
7690 designated output file of this compilation. This puts the argument
7691 into the sequence of arguments that @samp{%o} will substitute later.
7694 Substitutes the names of all the output files, with spaces
7695 automatically placed around them. You should write spaces
7696 around the @samp{%o} as well or the results are undefined.
7697 @samp{%o} is for use in the specs for running the linker.
7698 Input files whose names have no recognized suffix are not compiled
7699 at all, but they are included among the output files, so they will
7703 Substitutes the suffix for object files. Note that this is
7704 handled specially when it immediately follows @samp{%g, %u, or %U},
7705 because of the need for those to form complete file names. The
7706 handling is such that @samp{%O} is treated exactly as if it had already
7707 been substituted, except that @samp{%g, %u, and %U} do not currently
7708 support additional @var{suffix} characters following @samp{%O} as they would
7709 following, for example, @samp{.o}.
7712 Substitutes the standard macro predefinitions for the
7713 current target machine. Use this when running @code{cpp}.
7716 Like @samp{%p}, but puts @samp{__} before and after the name of each
7717 predefined macro, except for macros that start with @samp{__} or with
7718 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
7722 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
7723 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
7724 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
7725 and @option{-imultilib} as necessary.
7728 Current argument is the name of a library or startup file of some sort.
7729 Search for that file in a standard list of directories and substitute
7730 the full name found.
7733 Print @var{str} as an error message. @var{str} is terminated by a newline.
7734 Use this when inconsistent options are detected.
7737 Substitute the contents of spec string @var{name} at this point.
7740 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
7742 @item %x@{@var{option}@}
7743 Accumulate an option for @samp{%X}.
7746 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
7750 Output the accumulated assembler options specified by @option{-Wa}.
7753 Output the accumulated preprocessor options specified by @option{-Wp}.
7756 Process the @code{asm} spec. This is used to compute the
7757 switches to be passed to the assembler.
7760 Process the @code{asm_final} spec. This is a spec string for
7761 passing switches to an assembler post-processor, if such a program is
7765 Process the @code{link} spec. This is the spec for computing the
7766 command line passed to the linker. Typically it will make use of the
7767 @samp{%L %G %S %D and %E} sequences.
7770 Dump out a @option{-L} option for each directory that GCC believes might
7771 contain startup files. If the target supports multilibs then the
7772 current multilib directory will be prepended to each of these paths.
7775 Process the @code{lib} spec. This is a spec string for deciding which
7776 libraries should be included on the command line to the linker.
7779 Process the @code{libgcc} spec. This is a spec string for deciding
7780 which GCC support library should be included on the command line to the linker.
7783 Process the @code{startfile} spec. This is a spec for deciding which
7784 object files should be the first ones passed to the linker. Typically
7785 this might be a file named @file{crt0.o}.
7788 Process the @code{endfile} spec. This is a spec string that specifies
7789 the last object files that will be passed to the linker.
7792 Process the @code{cpp} spec. This is used to construct the arguments
7793 to be passed to the C preprocessor.
7796 Process the @code{cc1} spec. This is used to construct the options to be
7797 passed to the actual C compiler (@samp{cc1}).
7800 Process the @code{cc1plus} spec. This is used to construct the options to be
7801 passed to the actual C++ compiler (@samp{cc1plus}).
7804 Substitute the variable part of a matched option. See below.
7805 Note that each comma in the substituted string is replaced by
7809 Remove all occurrences of @code{-S} from the command line. Note---this
7810 command is position dependent. @samp{%} commands in the spec string
7811 before this one will see @code{-S}, @samp{%} commands in the spec string
7812 after this one will not.
7814 @item %:@var{function}(@var{args})
7815 Call the named function @var{function}, passing it @var{args}.
7816 @var{args} is first processed as a nested spec string, then split
7817 into an argument vector in the usual fashion. The function returns
7818 a string which is processed as if it had appeared literally as part
7819 of the current spec.
7821 The following built-in spec functions are provided:
7825 The @code{getenv} spec function takes two arguments: an environment
7826 variable name and a string. If the environment variable is not
7827 defined, a fatal error is issued. Otherwise, the return value is the
7828 value of the environment variable concatenated with the string. For
7829 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
7832 %:getenv(TOPDIR /include)
7835 expands to @file{/path/to/top/include}.
7837 @item @code{if-exists}
7838 The @code{if-exists} spec function takes one argument, an absolute
7839 pathname to a file. If the file exists, @code{if-exists} returns the
7840 pathname. Here is a small example of its usage:
7844 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7847 @item @code{if-exists-else}
7848 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7849 spec function, except that it takes two arguments. The first argument is
7850 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7851 returns the pathname. If it does not exist, it returns the second argument.
7852 This way, @code{if-exists-else} can be used to select one file or another,
7853 based on the existence of the first. Here is a small example of its usage:
7857 crt0%O%s %:if-exists(crti%O%s) \
7858 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7861 @item @code{replace-outfile}
7862 The @code{replace-outfile} spec function takes two arguments. It looks for the
7863 first argument in the outfiles array and replaces it with the second argument. Here
7864 is a small example of its usage:
7867 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7870 @item @code{print-asm-header}
7871 The @code{print-asm-header} function takes no arguments and simply
7872 prints a banner like:
7878 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
7881 It is used to separate compiler options from assembler options
7882 in the @option{--target-help} output.
7886 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7887 If that switch was not specified, this substitutes nothing. Note that
7888 the leading dash is omitted when specifying this option, and it is
7889 automatically inserted if the substitution is performed. Thus the spec
7890 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7891 and would output the command line option @option{-foo}.
7893 @item %W@{@code{S}@}
7894 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7897 @item %@{@code{S}*@}
7898 Substitutes all the switches specified to GCC whose names start
7899 with @code{-S}, but which also take an argument. This is used for
7900 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7901 GCC considers @option{-o foo} as being
7902 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7903 text, including the space. Thus two arguments would be generated.
7905 @item %@{@code{S}*&@code{T}*@}
7906 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7907 (the order of @code{S} and @code{T} in the spec is not significant).
7908 There can be any number of ampersand-separated variables; for each the
7909 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7911 @item %@{@code{S}:@code{X}@}
7912 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7914 @item %@{!@code{S}:@code{X}@}
7915 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7917 @item %@{@code{S}*:@code{X}@}
7918 Substitutes @code{X} if one or more switches whose names start with
7919 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7920 once, no matter how many such switches appeared. However, if @code{%*}
7921 appears somewhere in @code{X}, then @code{X} will be substituted once
7922 for each matching switch, with the @code{%*} replaced by the part of
7923 that switch that matched the @code{*}.
7925 @item %@{.@code{S}:@code{X}@}
7926 Substitutes @code{X}, if processing a file with suffix @code{S}.
7928 @item %@{!.@code{S}:@code{X}@}
7929 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7931 @item %@{,@code{S}:@code{X}@}
7932 Substitutes @code{X}, if processing a file for language @code{S}.
7934 @item %@{!,@code{S}:@code{X}@}
7935 Substitutes @code{X}, if not processing a file for language @code{S}.
7937 @item %@{@code{S}|@code{P}:@code{X}@}
7938 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
7939 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
7940 @code{*} sequences as well, although they have a stronger binding than
7941 the @samp{|}. If @code{%*} appears in @code{X}, all of the
7942 alternatives must be starred, and only the first matching alternative
7945 For example, a spec string like this:
7948 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7951 will output the following command-line options from the following input
7952 command-line options:
7957 -d fred.c -foo -baz -boggle
7958 -d jim.d -bar -baz -boggle
7961 @item %@{S:X; T:Y; :D@}
7963 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7964 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7965 be as many clauses as you need. This may be combined with @code{.},
7966 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
7971 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7972 construct may contain other nested @samp{%} constructs or spaces, or
7973 even newlines. They are processed as usual, as described above.
7974 Trailing white space in @code{X} is ignored. White space may also
7975 appear anywhere on the left side of the colon in these constructs,
7976 except between @code{.} or @code{*} and the corresponding word.
7978 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7979 handled specifically in these constructs. If another value of
7980 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7981 @option{-W} switch is found later in the command line, the earlier
7982 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7983 just one letter, which passes all matching options.
7985 The character @samp{|} at the beginning of the predicate text is used to
7986 indicate that a command should be piped to the following command, but
7987 only if @option{-pipe} is specified.
7989 It is built into GCC which switches take arguments and which do not.
7990 (You might think it would be useful to generalize this to allow each
7991 compiler's spec to say which switches take arguments. But this cannot
7992 be done in a consistent fashion. GCC cannot even decide which input
7993 files have been specified without knowing which switches take arguments,
7994 and it must know which input files to compile in order to tell which
7997 GCC also knows implicitly that arguments starting in @option{-l} are to be
7998 treated as compiler output files, and passed to the linker in their
7999 proper position among the other output files.
8001 @c man begin OPTIONS
8003 @node Target Options
8004 @section Specifying Target Machine and Compiler Version
8005 @cindex target options
8006 @cindex cross compiling
8007 @cindex specifying machine version
8008 @cindex specifying compiler version and target machine
8009 @cindex compiler version, specifying
8010 @cindex target machine, specifying
8012 The usual way to run GCC is to run the executable called @file{gcc}, or
8013 @file{<machine>-gcc} when cross-compiling, or
8014 @file{<machine>-gcc-<version>} to run a version other than the one that
8015 was installed last. Sometimes this is inconvenient, so GCC provides
8016 options that will switch to another cross-compiler or version.
8019 @item -b @var{machine}
8021 The argument @var{machine} specifies the target machine for compilation.
8023 The value to use for @var{machine} is the same as was specified as the
8024 machine type when configuring GCC as a cross-compiler. For
8025 example, if a cross-compiler was configured with @samp{configure
8026 arm-elf}, meaning to compile for an arm processor with elf binaries,
8027 then you would specify @option{-b arm-elf} to run that cross compiler.
8028 Because there are other options beginning with @option{-b}, the
8029 configuration must contain a hyphen.
8031 @item -V @var{version}
8033 The argument @var{version} specifies which version of GCC to run.
8034 This is useful when multiple versions are installed. For example,
8035 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8038 The @option{-V} and @option{-b} options work by running the
8039 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8040 use them if you can just run that directly.
8042 @node Submodel Options
8043 @section Hardware Models and Configurations
8044 @cindex submodel options
8045 @cindex specifying hardware config
8046 @cindex hardware models and configurations, specifying
8047 @cindex machine dependent options
8049 Earlier we discussed the standard option @option{-b} which chooses among
8050 different installed compilers for completely different target
8051 machines, such as VAX vs.@: 68000 vs.@: 80386.
8053 In addition, each of these target machine types can have its own
8054 special options, starting with @samp{-m}, to choose among various
8055 hardware models or configurations---for example, 68010 vs 68020,
8056 floating coprocessor or none. A single installed version of the
8057 compiler can compile for any model or configuration, according to the
8060 Some configurations of the compiler also support additional special
8061 options, usually for compatibility with other compilers on the same
8064 @c This list is ordered alphanumerically by subsection name.
8065 @c It should be the same order and spelling as these options are listed
8066 @c in Machine Dependent Options
8072 * Blackfin Options::
8076 * DEC Alpha Options::
8077 * DEC Alpha/VMS Options::
8079 * GNU/Linux Options::
8082 * i386 and x86-64 Options::
8095 * RS/6000 and PowerPC Options::
8096 * S/390 and zSeries Options::
8101 * System V Options::
8102 * TMS320C3x/C4x Options::
8107 * Xstormy16 Options::
8113 @subsection ARC Options
8116 These options are defined for ARC implementations:
8121 Compile code for little endian mode. This is the default.
8125 Compile code for big endian mode.
8128 @opindex mmangle-cpu
8129 Prepend the name of the cpu to all public symbol names.
8130 In multiple-processor systems, there are many ARC variants with different
8131 instruction and register set characteristics. This flag prevents code
8132 compiled for one cpu to be linked with code compiled for another.
8133 No facility exists for handling variants that are ``almost identical''.
8134 This is an all or nothing option.
8136 @item -mcpu=@var{cpu}
8138 Compile code for ARC variant @var{cpu}.
8139 Which variants are supported depend on the configuration.
8140 All variants support @option{-mcpu=base}, this is the default.
8142 @item -mtext=@var{text-section}
8143 @itemx -mdata=@var{data-section}
8144 @itemx -mrodata=@var{readonly-data-section}
8148 Put functions, data, and readonly data in @var{text-section},
8149 @var{data-section}, and @var{readonly-data-section} respectively
8150 by default. This can be overridden with the @code{section} attribute.
8151 @xref{Variable Attributes}.
8156 @subsection ARM Options
8159 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8163 @item -mabi=@var{name}
8165 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8166 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8169 @opindex mapcs-frame
8170 Generate a stack frame that is compliant with the ARM Procedure Call
8171 Standard for all functions, even if this is not strictly necessary for
8172 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8173 with this option will cause the stack frames not to be generated for
8174 leaf functions. The default is @option{-mno-apcs-frame}.
8178 This is a synonym for @option{-mapcs-frame}.
8181 @c not currently implemented
8182 @item -mapcs-stack-check
8183 @opindex mapcs-stack-check
8184 Generate code to check the amount of stack space available upon entry to
8185 every function (that actually uses some stack space). If there is
8186 insufficient space available then either the function
8187 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8188 called, depending upon the amount of stack space required. The run time
8189 system is required to provide these functions. The default is
8190 @option{-mno-apcs-stack-check}, since this produces smaller code.
8192 @c not currently implemented
8194 @opindex mapcs-float
8195 Pass floating point arguments using the float point registers. This is
8196 one of the variants of the APCS@. This option is recommended if the
8197 target hardware has a floating point unit or if a lot of floating point
8198 arithmetic is going to be performed by the code. The default is
8199 @option{-mno-apcs-float}, since integer only code is slightly increased in
8200 size if @option{-mapcs-float} is used.
8202 @c not currently implemented
8203 @item -mapcs-reentrant
8204 @opindex mapcs-reentrant
8205 Generate reentrant, position independent code. The default is
8206 @option{-mno-apcs-reentrant}.
8209 @item -mthumb-interwork
8210 @opindex mthumb-interwork
8211 Generate code which supports calling between the ARM and Thumb
8212 instruction sets. Without this option the two instruction sets cannot
8213 be reliably used inside one program. The default is
8214 @option{-mno-thumb-interwork}, since slightly larger code is generated
8215 when @option{-mthumb-interwork} is specified.
8217 @item -mno-sched-prolog
8218 @opindex mno-sched-prolog
8219 Prevent the reordering of instructions in the function prolog, or the
8220 merging of those instruction with the instructions in the function's
8221 body. This means that all functions will start with a recognizable set
8222 of instructions (or in fact one of a choice from a small set of
8223 different function prologues), and this information can be used to
8224 locate the start if functions inside an executable piece of code. The
8225 default is @option{-msched-prolog}.
8228 @opindex mhard-float
8229 Generate output containing floating point instructions. This is the
8233 @opindex msoft-float
8234 Generate output containing library calls for floating point.
8235 @strong{Warning:} the requisite libraries are not available for all ARM
8236 targets. Normally the facilities of the machine's usual C compiler are
8237 used, but this cannot be done directly in cross-compilation. You must make
8238 your own arrangements to provide suitable library functions for
8241 @option{-msoft-float} changes the calling convention in the output file;
8242 therefore, it is only useful if you compile @emph{all} of a program with
8243 this option. In particular, you need to compile @file{libgcc.a}, the
8244 library that comes with GCC, with @option{-msoft-float} in order for
8247 @item -mfloat-abi=@var{name}
8249 Specifies which ABI to use for floating point values. Permissible values
8250 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8252 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8253 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8254 of floating point instructions, but still uses the soft-float calling
8257 @item -mlittle-endian
8258 @opindex mlittle-endian
8259 Generate code for a processor running in little-endian mode. This is
8260 the default for all standard configurations.
8263 @opindex mbig-endian
8264 Generate code for a processor running in big-endian mode; the default is
8265 to compile code for a little-endian processor.
8267 @item -mwords-little-endian
8268 @opindex mwords-little-endian
8269 This option only applies when generating code for big-endian processors.
8270 Generate code for a little-endian word order but a big-endian byte
8271 order. That is, a byte order of the form @samp{32107654}. Note: this
8272 option should only be used if you require compatibility with code for
8273 big-endian ARM processors generated by versions of the compiler prior to
8276 @item -mcpu=@var{name}
8278 This specifies the name of the target ARM processor. GCC uses this name
8279 to determine what kind of instructions it can emit when generating
8280 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8281 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8282 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8283 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8284 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8285 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8286 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8287 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8288 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8289 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8290 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8291 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8292 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8293 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8294 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3},
8295 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8297 @itemx -mtune=@var{name}
8299 This option is very similar to the @option{-mcpu=} option, except that
8300 instead of specifying the actual target processor type, and hence
8301 restricting which instructions can be used, it specifies that GCC should
8302 tune the performance of the code as if the target were of the type
8303 specified in this option, but still choosing the instructions that it
8304 will generate based on the cpu specified by a @option{-mcpu=} option.
8305 For some ARM implementations better performance can be obtained by using
8308 @item -march=@var{name}
8310 This specifies the name of the target ARM architecture. GCC uses this
8311 name to determine what kind of instructions it can emit when generating
8312 assembly code. This option can be used in conjunction with or instead
8313 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8314 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8315 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8316 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv7}, @samp{armv7-a},
8317 @samp{armv7-r}, @samp{armv7-m}, @samp{iwmmxt}, @samp{ep9312}.
8319 @item -mfpu=@var{name}
8320 @itemx -mfpe=@var{number}
8321 @itemx -mfp=@var{number}
8325 This specifies what floating point hardware (or hardware emulation) is
8326 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8327 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
8328 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8329 with older versions of GCC@.
8331 If @option{-msoft-float} is specified this specifies the format of
8332 floating point values.
8334 @item -mstructure-size-boundary=@var{n}
8335 @opindex mstructure-size-boundary
8336 The size of all structures and unions will be rounded up to a multiple
8337 of the number of bits set by this option. Permissible values are 8, 32
8338 and 64. The default value varies for different toolchains. For the COFF
8339 targeted toolchain the default value is 8. A value of 64 is only allowed
8340 if the underlying ABI supports it.
8342 Specifying the larger number can produce faster, more efficient code, but
8343 can also increase the size of the program. Different values are potentially
8344 incompatible. Code compiled with one value cannot necessarily expect to
8345 work with code or libraries compiled with another value, if they exchange
8346 information using structures or unions.
8348 @item -mabort-on-noreturn
8349 @opindex mabort-on-noreturn
8350 Generate a call to the function @code{abort} at the end of a
8351 @code{noreturn} function. It will be executed if the function tries to
8355 @itemx -mno-long-calls
8356 @opindex mlong-calls
8357 @opindex mno-long-calls
8358 Tells the compiler to perform function calls by first loading the
8359 address of the function into a register and then performing a subroutine
8360 call on this register. This switch is needed if the target function
8361 will lie outside of the 64 megabyte addressing range of the offset based
8362 version of subroutine call instruction.
8364 Even if this switch is enabled, not all function calls will be turned
8365 into long calls. The heuristic is that static functions, functions
8366 which have the @samp{short-call} attribute, functions that are inside
8367 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8368 definitions have already been compiled within the current compilation
8369 unit, will not be turned into long calls. The exception to this rule is
8370 that weak function definitions, functions with the @samp{long-call}
8371 attribute or the @samp{section} attribute, and functions that are within
8372 the scope of a @samp{#pragma long_calls} directive, will always be
8373 turned into long calls.
8375 This feature is not enabled by default. Specifying
8376 @option{-mno-long-calls} will restore the default behavior, as will
8377 placing the function calls within the scope of a @samp{#pragma
8378 long_calls_off} directive. Note these switches have no effect on how
8379 the compiler generates code to handle function calls via function
8382 @item -mnop-fun-dllimport
8383 @opindex mnop-fun-dllimport
8384 Disable support for the @code{dllimport} attribute.
8386 @item -msingle-pic-base
8387 @opindex msingle-pic-base
8388 Treat the register used for PIC addressing as read-only, rather than
8389 loading it in the prologue for each function. The run-time system is
8390 responsible for initializing this register with an appropriate value
8391 before execution begins.
8393 @item -mpic-register=@var{reg}
8394 @opindex mpic-register
8395 Specify the register to be used for PIC addressing. The default is R10
8396 unless stack-checking is enabled, when R9 is used.
8398 @item -mcirrus-fix-invalid-insns
8399 @opindex mcirrus-fix-invalid-insns
8400 @opindex mno-cirrus-fix-invalid-insns
8401 Insert NOPs into the instruction stream to in order to work around
8402 problems with invalid Maverick instruction combinations. This option
8403 is only valid if the @option{-mcpu=ep9312} option has been used to
8404 enable generation of instructions for the Cirrus Maverick floating
8405 point co-processor. This option is not enabled by default, since the
8406 problem is only present in older Maverick implementations. The default
8407 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8410 @item -mpoke-function-name
8411 @opindex mpoke-function-name
8412 Write the name of each function into the text section, directly
8413 preceding the function prologue. The generated code is similar to this:
8417 .ascii "arm_poke_function_name", 0
8420 .word 0xff000000 + (t1 - t0)
8421 arm_poke_function_name
8423 stmfd sp!, @{fp, ip, lr, pc@}
8427 When performing a stack backtrace, code can inspect the value of
8428 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8429 location @code{pc - 12} and the top 8 bits are set, then we know that
8430 there is a function name embedded immediately preceding this location
8431 and has length @code{((pc[-3]) & 0xff000000)}.
8435 Generate code for the Thumb instruction set. The default is to
8436 use the 32-bit ARM instruction set.
8437 This option automatically enables either 16-bit Thumb-1 or
8438 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8439 and @option{-march=@var{name}} options.
8442 @opindex mtpcs-frame
8443 Generate a stack frame that is compliant with the Thumb Procedure Call
8444 Standard for all non-leaf functions. (A leaf function is one that does
8445 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8447 @item -mtpcs-leaf-frame
8448 @opindex mtpcs-leaf-frame
8449 Generate a stack frame that is compliant with the Thumb Procedure Call
8450 Standard for all leaf functions. (A leaf function is one that does
8451 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8453 @item -mcallee-super-interworking
8454 @opindex mcallee-super-interworking
8455 Gives all externally visible functions in the file being compiled an ARM
8456 instruction set header which switches to Thumb mode before executing the
8457 rest of the function. This allows these functions to be called from
8458 non-interworking code.
8460 @item -mcaller-super-interworking
8461 @opindex mcaller-super-interworking
8462 Allows calls via function pointers (including virtual functions) to
8463 execute correctly regardless of whether the target code has been
8464 compiled for interworking or not. There is a small overhead in the cost
8465 of executing a function pointer if this option is enabled.
8467 @item -mtp=@var{name}
8469 Specify the access model for the thread local storage pointer. The valid
8470 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8471 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8472 (supported in the arm6k architecture), and @option{auto}, which uses the
8473 best available method for the selected processor. The default setting is
8479 @subsection AVR Options
8482 These options are defined for AVR implementations:
8485 @item -mmcu=@var{mcu}
8487 Specify ATMEL AVR instruction set or MCU type.
8489 Instruction set avr1 is for the minimal AVR core, not supported by the C
8490 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8491 attiny11, attiny12, attiny15, attiny28).
8493 Instruction set avr2 (default) is for the classic AVR core with up to
8494 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8495 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8496 at90c8534, at90s8535).
8498 Instruction set avr3 is for the classic AVR core with up to 128K program
8499 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8501 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8502 memory space (MCU types: atmega8, atmega83, atmega85).
8504 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8505 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8506 atmega64, atmega128, at43usb355, at94k).
8510 Output instruction sizes to the asm file.
8512 @item -minit-stack=@var{N}
8513 @opindex minit-stack
8514 Specify the initial stack address, which may be a symbol or numeric value,
8515 @samp{__stack} is the default.
8517 @item -mno-interrupts
8518 @opindex mno-interrupts
8519 Generated code is not compatible with hardware interrupts.
8520 Code size will be smaller.
8522 @item -mcall-prologues
8523 @opindex mcall-prologues
8524 Functions prologues/epilogues expanded as call to appropriate
8525 subroutines. Code size will be smaller.
8527 @item -mno-tablejump
8528 @opindex mno-tablejump
8529 Do not generate tablejump insns which sometimes increase code size.
8532 @opindex mtiny-stack
8533 Change only the low 8 bits of the stack pointer.
8537 Assume int to be 8 bit integer. This affects the sizes of all types: A
8538 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8539 and long long will be 4 bytes. Please note that this option does not
8540 comply to the C standards, but it will provide you with smaller code
8544 @node Blackfin Options
8545 @subsection Blackfin Options
8546 @cindex Blackfin Options
8549 @item -mcpu=@var{cpu}
8551 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8552 can be one of @samp{bf531}, @samp{bf532}, @samp{bf533},
8553 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf561}.
8554 Without this option, @samp{bf532} is used as the processor by default.
8555 The corresponding predefined processor macros for @var{cpu} is to
8556 be defined. For the @samp{bfin-elf} toolchain, this causes the hardware
8557 BSP provided by libgloss to be linked in if @samp{-msim} is not given.
8558 Support for @samp{bf561} is incomplete; only the processor macro is defined.
8562 Specifies that the program will be run on the simulator. This causes
8563 the simulator BSP provided by libgloss to be linked in. This option
8564 has effect only for @samp{bfin-elf} toolchain.
8566 @item -momit-leaf-frame-pointer
8567 @opindex momit-leaf-frame-pointer
8568 Don't keep the frame pointer in a register for leaf functions. This
8569 avoids the instructions to save, set up and restore frame pointers and
8570 makes an extra register available in leaf functions. The option
8571 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8572 which might make debugging harder.
8574 @item -mspecld-anomaly
8575 @opindex mspecld-anomaly
8576 When enabled, the compiler will ensure that the generated code does not
8577 contain speculative loads after jump instructions. This option is enabled
8580 @item -mno-specld-anomaly
8581 @opindex mno-specld-anomaly
8582 Don't generate extra code to prevent speculative loads from occurring.
8584 @item -mcsync-anomaly
8585 @opindex mcsync-anomaly
8586 When enabled, the compiler will ensure that the generated code does not
8587 contain CSYNC or SSYNC instructions too soon after conditional branches.
8588 This option is enabled by default.
8590 @item -mno-csync-anomaly
8591 @opindex mno-csync-anomaly
8592 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8593 occurring too soon after a conditional branch.
8597 When enabled, the compiler is free to take advantage of the knowledge that
8598 the entire program fits into the low 64k of memory.
8601 @opindex mno-low-64k
8602 Assume that the program is arbitrarily large. This is the default.
8604 @item -mstack-check-l1
8605 @opindex mstack-check-l1
8606 Do stack checking using information placed into L1 scratchpad memory by the
8609 @item -mid-shared-library
8610 @opindex mid-shared-library
8611 Generate code that supports shared libraries via the library ID method.
8612 This allows for execute in place and shared libraries in an environment
8613 without virtual memory management. This option implies @option{-fPIC}.
8615 @item -mno-id-shared-library
8616 @opindex mno-id-shared-library
8617 Generate code that doesn't assume ID based shared libraries are being used.
8618 This is the default.
8620 @item -mleaf-id-shared-library
8621 @opindex mleaf-id-shared-library
8622 Generate code that supports shared libraries via the library ID method,
8623 but assumes that this library or executable won't link against any other
8624 ID shared libraries. That allows the compiler to use faster code for jumps
8627 @item -mno-leaf-id-shared-library
8628 @opindex mno-leaf-id-shared-library
8629 Do not assume that the code being compiled won't link against any ID shared
8630 libraries. Slower code will be generated for jump and call insns.
8632 @item -mshared-library-id=n
8633 @opindex mshared-library-id
8634 Specified the identification number of the ID based shared library being
8635 compiled. Specifying a value of 0 will generate more compact code, specifying
8636 other values will force the allocation of that number to the current
8637 library but is no more space or time efficient than omitting this option.
8641 Generate code that allows the data segment to be located in a different
8642 area of memory from the text segment. This allows for execute in place in
8643 an environment without virtual memory management by eliminating relocations
8644 against the text section.
8647 @opindex mno-sep-data
8648 Generate code that assumes that the data segment follows the text segment.
8649 This is the default.
8652 @itemx -mno-long-calls
8653 @opindex mlong-calls
8654 @opindex mno-long-calls
8655 Tells the compiler to perform function calls by first loading the
8656 address of the function into a register and then performing a subroutine
8657 call on this register. This switch is needed if the target function
8658 will lie outside of the 24 bit addressing range of the offset based
8659 version of subroutine call instruction.
8661 This feature is not enabled by default. Specifying
8662 @option{-mno-long-calls} will restore the default behavior. Note these
8663 switches have no effect on how the compiler generates code to handle
8664 function calls via function pointers.
8668 Link with the fast floating-point library. This library relaxes some of
8669 the IEEE floating-point standard's rules for checking inputs against
8670 Not-a-Number (NAN), in the interest of performance.
8673 @opindex minline-plt
8674 Enable inlining of PLT entries in function calls to functions that are
8675 not known to bind locally. It has no effect without @option{-mfdpic}.
8679 @subsection CRIS Options
8680 @cindex CRIS Options
8682 These options are defined specifically for the CRIS ports.
8685 @item -march=@var{architecture-type}
8686 @itemx -mcpu=@var{architecture-type}
8689 Generate code for the specified architecture. The choices for
8690 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
8691 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
8692 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
8695 @item -mtune=@var{architecture-type}
8697 Tune to @var{architecture-type} everything applicable about the generated
8698 code, except for the ABI and the set of available instructions. The
8699 choices for @var{architecture-type} are the same as for
8700 @option{-march=@var{architecture-type}}.
8702 @item -mmax-stack-frame=@var{n}
8703 @opindex mmax-stack-frame
8704 Warn when the stack frame of a function exceeds @var{n} bytes.
8706 @item -melinux-stacksize=@var{n}
8707 @opindex melinux-stacksize
8708 Only available with the @samp{cris-axis-aout} target. Arranges for
8709 indications in the program to the kernel loader that the stack of the
8710 program should be set to @var{n} bytes.
8716 The options @option{-metrax4} and @option{-metrax100} are synonyms for
8717 @option{-march=v3} and @option{-march=v8} respectively.
8719 @item -mmul-bug-workaround
8720 @itemx -mno-mul-bug-workaround
8721 @opindex mmul-bug-workaround
8722 @opindex mno-mul-bug-workaround
8723 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
8724 models where it applies. This option is active by default.
8728 Enable CRIS-specific verbose debug-related information in the assembly
8729 code. This option also has the effect to turn off the @samp{#NO_APP}
8730 formatted-code indicator to the assembler at the beginning of the
8735 Do not use condition-code results from previous instruction; always emit
8736 compare and test instructions before use of condition codes.
8738 @item -mno-side-effects
8739 @opindex mno-side-effects
8740 Do not emit instructions with side-effects in addressing modes other than
8744 @itemx -mno-stack-align
8746 @itemx -mno-data-align
8747 @itemx -mconst-align
8748 @itemx -mno-const-align
8749 @opindex mstack-align
8750 @opindex mno-stack-align
8751 @opindex mdata-align
8752 @opindex mno-data-align
8753 @opindex mconst-align
8754 @opindex mno-const-align
8755 These options (no-options) arranges (eliminate arrangements) for the
8756 stack-frame, individual data and constants to be aligned for the maximum
8757 single data access size for the chosen CPU model. The default is to
8758 arrange for 32-bit alignment. ABI details such as structure layout are
8759 not affected by these options.
8767 Similar to the stack- data- and const-align options above, these options
8768 arrange for stack-frame, writable data and constants to all be 32-bit,
8769 16-bit or 8-bit aligned. The default is 32-bit alignment.
8771 @item -mno-prologue-epilogue
8772 @itemx -mprologue-epilogue
8773 @opindex mno-prologue-epilogue
8774 @opindex mprologue-epilogue
8775 With @option{-mno-prologue-epilogue}, the normal function prologue and
8776 epilogue that sets up the stack-frame are omitted and no return
8777 instructions or return sequences are generated in the code. Use this
8778 option only together with visual inspection of the compiled code: no
8779 warnings or errors are generated when call-saved registers must be saved,
8780 or storage for local variable needs to be allocated.
8786 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
8787 instruction sequences that load addresses for functions from the PLT part
8788 of the GOT rather than (traditional on other architectures) calls to the
8789 PLT@. The default is @option{-mgotplt}.
8793 Legacy no-op option only recognized with the cris-axis-aout target.
8797 Legacy no-op option only recognized with the cris-axis-elf and
8798 cris-axis-linux-gnu targets.
8802 Only recognized with the cris-axis-aout target, where it selects a
8803 GNU/linux-like multilib, include files and instruction set for
8808 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
8812 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
8813 to link with input-output functions from a simulator library. Code,
8814 initialized data and zero-initialized data are allocated consecutively.
8818 Like @option{-sim}, but pass linker options to locate initialized data at
8819 0x40000000 and zero-initialized data at 0x80000000.
8823 @subsection CRX Options
8826 These options are defined specifically for the CRX ports.
8832 Enable the use of multiply-accumulate instructions. Disabled by default.
8836 Push instructions will be used to pass outgoing arguments when functions
8837 are called. Enabled by default.
8840 @node Darwin Options
8841 @subsection Darwin Options
8842 @cindex Darwin options
8844 These options are defined for all architectures running the Darwin operating
8847 FSF GCC on Darwin does not create ``fat'' object files; it will create
8848 an object file for the single architecture that it was built to
8849 target. Apple's GCC on Darwin does create ``fat'' files if multiple
8850 @option{-arch} options are used; it does so by running the compiler or
8851 linker multiple times and joining the results together with
8854 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
8855 @samp{i686}) is determined by the flags that specify the ISA
8856 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
8857 @option{-force_cpusubtype_ALL} option can be used to override this.
8859 The Darwin tools vary in their behavior when presented with an ISA
8860 mismatch. The assembler, @file{as}, will only permit instructions to
8861 be used that are valid for the subtype of the file it is generating,
8862 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8863 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8864 and print an error if asked to create a shared library with a less
8865 restrictive subtype than its input files (for instance, trying to put
8866 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8867 for executables, @file{ld}, will quietly give the executable the most
8868 restrictive subtype of any of its input files.
8873 Add the framework directory @var{dir} to the head of the list of
8874 directories to be searched for header files. These directories are
8875 interleaved with those specified by @option{-I} options and are
8876 scanned in a left-to-right order.
8878 A framework directory is a directory with frameworks in it. A
8879 framework is a directory with a @samp{"Headers"} and/or
8880 @samp{"PrivateHeaders"} directory contained directly in it that ends
8881 in @samp{".framework"}. The name of a framework is the name of this
8882 directory excluding the @samp{".framework"}. Headers associated with
8883 the framework are found in one of those two directories, with
8884 @samp{"Headers"} being searched first. A subframework is a framework
8885 directory that is in a framework's @samp{"Frameworks"} directory.
8886 Includes of subframework headers can only appear in a header of a
8887 framework that contains the subframework, or in a sibling subframework
8888 header. Two subframeworks are siblings if they occur in the same
8889 framework. A subframework should not have the same name as a
8890 framework, a warning will be issued if this is violated. Currently a
8891 subframework cannot have subframeworks, in the future, the mechanism
8892 may be extended to support this. The standard frameworks can be found
8893 in @samp{"/System/Library/Frameworks"} and
8894 @samp{"/Library/Frameworks"}. An example include looks like
8895 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8896 the name of the framework and header.h is found in the
8897 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8899 @item -iframework@var{dir}
8901 Like @option{-F} except the directory is a treated as a system
8902 directory. The main difference between this @option{-iframework} and
8903 @option{-F} is that with @option{-iframework} the compiler does not
8904 warn about constructs contained within header files found via
8905 @var{dir}. This option is valid only for the C family of languages.
8909 Emit debugging information for symbols that are used. For STABS
8910 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8911 This is by default ON@.
8915 Emit debugging information for all symbols and types.
8917 @item -mmacosx-version-min=@var{version}
8918 The earliest version of MacOS X that this executable will run on
8919 is @var{version}. Typical values of @var{version} include @code{10.1},
8920 @code{10.2}, and @code{10.3.9}.
8922 If the compiler was built to use the system's headers by default,
8923 then the default for this option is the system version on which the
8924 compiler is running, otherwise the default is to make choices which
8925 are compatible with as many systems and code bases as possible.
8929 Enable kernel development mode. The @option{-mkernel} option sets
8930 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
8931 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
8932 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
8933 applicable. This mode also sets @option{-mno-altivec},
8934 @option{-msoft-float}, @option{-fno-builtin} and
8935 @option{-mlong-branch} for PowerPC targets.
8937 @item -mone-byte-bool
8938 @opindex mone-byte-bool
8939 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8940 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8941 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8942 option has no effect on x86.
8944 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8945 to generate code that is not binary compatible with code generated
8946 without that switch. Using this switch may require recompiling all
8947 other modules in a program, including system libraries. Use this
8948 switch to conform to a non-default data model.
8950 @item -mfix-and-continue
8951 @itemx -ffix-and-continue
8952 @itemx -findirect-data
8953 @opindex mfix-and-continue
8954 @opindex ffix-and-continue
8955 @opindex findirect-data
8956 Generate code suitable for fast turn around development. Needed to
8957 enable gdb to dynamically load @code{.o} files into already running
8958 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8959 are provided for backwards compatibility.
8963 Loads all members of static archive libraries.
8964 See man ld(1) for more information.
8966 @item -arch_errors_fatal
8967 @opindex arch_errors_fatal
8968 Cause the errors having to do with files that have the wrong architecture
8972 @opindex bind_at_load
8973 Causes the output file to be marked such that the dynamic linker will
8974 bind all undefined references when the file is loaded or launched.
8978 Produce a Mach-o bundle format file.
8979 See man ld(1) for more information.
8981 @item -bundle_loader @var{executable}
8982 @opindex bundle_loader
8983 This option specifies the @var{executable} that will be loading the build
8984 output file being linked. See man ld(1) for more information.
8988 When passed this option, GCC will produce a dynamic library instead of
8989 an executable when linking, using the Darwin @file{libtool} command.
8991 @item -force_cpusubtype_ALL
8992 @opindex force_cpusubtype_ALL
8993 This causes GCC's output file to have the @var{ALL} subtype, instead of
8994 one controlled by the @option{-mcpu} or @option{-march} option.
8996 @item -allowable_client @var{client_name}
8998 @itemx -compatibility_version
8999 @itemx -current_version
9001 @itemx -dependency-file
9003 @itemx -dylinker_install_name
9005 @itemx -exported_symbols_list
9007 @itemx -flat_namespace
9008 @itemx -force_flat_namespace
9009 @itemx -headerpad_max_install_names
9012 @itemx -install_name
9013 @itemx -keep_private_externs
9014 @itemx -multi_module
9015 @itemx -multiply_defined
9016 @itemx -multiply_defined_unused
9018 @itemx -no_dead_strip_inits_and_terms
9019 @itemx -nofixprebinding
9022 @itemx -noseglinkedit
9023 @itemx -pagezero_size
9025 @itemx -prebind_all_twolevel_modules
9026 @itemx -private_bundle
9027 @itemx -read_only_relocs
9029 @itemx -sectobjectsymbols
9033 @itemx -sectobjectsymbols
9036 @itemx -segs_read_only_addr
9037 @itemx -segs_read_write_addr
9038 @itemx -seg_addr_table
9039 @itemx -seg_addr_table_filename
9042 @itemx -segs_read_only_addr
9043 @itemx -segs_read_write_addr
9044 @itemx -single_module
9047 @itemx -sub_umbrella
9048 @itemx -twolevel_namespace
9051 @itemx -unexported_symbols_list
9052 @itemx -weak_reference_mismatches
9055 @opindex allowable_client
9056 @opindex client_name
9057 @opindex compatibility_version
9058 @opindex current_version
9060 @opindex dependency-file
9062 @opindex dylinker_install_name
9064 @opindex exported_symbols_list
9066 @opindex flat_namespace
9067 @opindex force_flat_namespace
9068 @opindex headerpad_max_install_names
9071 @opindex install_name
9072 @opindex keep_private_externs
9073 @opindex multi_module
9074 @opindex multiply_defined
9075 @opindex multiply_defined_unused
9077 @opindex no_dead_strip_inits_and_terms
9078 @opindex nofixprebinding
9079 @opindex nomultidefs
9081 @opindex noseglinkedit
9082 @opindex pagezero_size
9084 @opindex prebind_all_twolevel_modules
9085 @opindex private_bundle
9086 @opindex read_only_relocs
9088 @opindex sectobjectsymbols
9092 @opindex sectobjectsymbols
9095 @opindex segs_read_only_addr
9096 @opindex segs_read_write_addr
9097 @opindex seg_addr_table
9098 @opindex seg_addr_table_filename
9099 @opindex seglinkedit
9101 @opindex segs_read_only_addr
9102 @opindex segs_read_write_addr
9103 @opindex single_module
9105 @opindex sub_library
9106 @opindex sub_umbrella
9107 @opindex twolevel_namespace
9110 @opindex unexported_symbols_list
9111 @opindex weak_reference_mismatches
9112 @opindex whatsloaded
9114 These options are passed to the Darwin linker. The Darwin linker man page
9115 describes them in detail.
9118 @node DEC Alpha Options
9119 @subsection DEC Alpha Options
9121 These @samp{-m} options are defined for the DEC Alpha implementations:
9124 @item -mno-soft-float
9126 @opindex mno-soft-float
9127 @opindex msoft-float
9128 Use (do not use) the hardware floating-point instructions for
9129 floating-point operations. When @option{-msoft-float} is specified,
9130 functions in @file{libgcc.a} will be used to perform floating-point
9131 operations. Unless they are replaced by routines that emulate the
9132 floating-point operations, or compiled in such a way as to call such
9133 emulations routines, these routines will issue floating-point
9134 operations. If you are compiling for an Alpha without floating-point
9135 operations, you must ensure that the library is built so as not to call
9138 Note that Alpha implementations without floating-point operations are
9139 required to have floating-point registers.
9144 @opindex mno-fp-regs
9145 Generate code that uses (does not use) the floating-point register set.
9146 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9147 register set is not used, floating point operands are passed in integer
9148 registers as if they were integers and floating-point results are passed
9149 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9150 so any function with a floating-point argument or return value called by code
9151 compiled with @option{-mno-fp-regs} must also be compiled with that
9154 A typical use of this option is building a kernel that does not use,
9155 and hence need not save and restore, any floating-point registers.
9159 The Alpha architecture implements floating-point hardware optimized for
9160 maximum performance. It is mostly compliant with the IEEE floating
9161 point standard. However, for full compliance, software assistance is
9162 required. This option generates code fully IEEE compliant code
9163 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9164 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9165 defined during compilation. The resulting code is less efficient but is
9166 able to correctly support denormalized numbers and exceptional IEEE
9167 values such as not-a-number and plus/minus infinity. Other Alpha
9168 compilers call this option @option{-ieee_with_no_inexact}.
9170 @item -mieee-with-inexact
9171 @opindex mieee-with-inexact
9172 This is like @option{-mieee} except the generated code also maintains
9173 the IEEE @var{inexact-flag}. Turning on this option causes the
9174 generated code to implement fully-compliant IEEE math. In addition to
9175 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9176 macro. On some Alpha implementations the resulting code may execute
9177 significantly slower than the code generated by default. Since there is
9178 very little code that depends on the @var{inexact-flag}, you should
9179 normally not specify this option. Other Alpha compilers call this
9180 option @option{-ieee_with_inexact}.
9182 @item -mfp-trap-mode=@var{trap-mode}
9183 @opindex mfp-trap-mode
9184 This option controls what floating-point related traps are enabled.
9185 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9186 The trap mode can be set to one of four values:
9190 This is the default (normal) setting. The only traps that are enabled
9191 are the ones that cannot be disabled in software (e.g., division by zero
9195 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9199 Like @samp{u}, but the instructions are marked to be safe for software
9200 completion (see Alpha architecture manual for details).
9203 Like @samp{su}, but inexact traps are enabled as well.
9206 @item -mfp-rounding-mode=@var{rounding-mode}
9207 @opindex mfp-rounding-mode
9208 Selects the IEEE rounding mode. Other Alpha compilers call this option
9209 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9214 Normal IEEE rounding mode. Floating point numbers are rounded towards
9215 the nearest machine number or towards the even machine number in case
9219 Round towards minus infinity.
9222 Chopped rounding mode. Floating point numbers are rounded towards zero.
9225 Dynamic rounding mode. A field in the floating point control register
9226 (@var{fpcr}, see Alpha architecture reference manual) controls the
9227 rounding mode in effect. The C library initializes this register for
9228 rounding towards plus infinity. Thus, unless your program modifies the
9229 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9232 @item -mtrap-precision=@var{trap-precision}
9233 @opindex mtrap-precision
9234 In the Alpha architecture, floating point traps are imprecise. This
9235 means without software assistance it is impossible to recover from a
9236 floating trap and program execution normally needs to be terminated.
9237 GCC can generate code that can assist operating system trap handlers
9238 in determining the exact location that caused a floating point trap.
9239 Depending on the requirements of an application, different levels of
9240 precisions can be selected:
9244 Program precision. This option is the default and means a trap handler
9245 can only identify which program caused a floating point exception.
9248 Function precision. The trap handler can determine the function that
9249 caused a floating point exception.
9252 Instruction precision. The trap handler can determine the exact
9253 instruction that caused a floating point exception.
9256 Other Alpha compilers provide the equivalent options called
9257 @option{-scope_safe} and @option{-resumption_safe}.
9259 @item -mieee-conformant
9260 @opindex mieee-conformant
9261 This option marks the generated code as IEEE conformant. You must not
9262 use this option unless you also specify @option{-mtrap-precision=i} and either
9263 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9264 is to emit the line @samp{.eflag 48} in the function prologue of the
9265 generated assembly file. Under DEC Unix, this has the effect that
9266 IEEE-conformant math library routines will be linked in.
9268 @item -mbuild-constants
9269 @opindex mbuild-constants
9270 Normally GCC examines a 32- or 64-bit integer constant to
9271 see if it can construct it from smaller constants in two or three
9272 instructions. If it cannot, it will output the constant as a literal and
9273 generate code to load it from the data segment at runtime.
9275 Use this option to require GCC to construct @emph{all} integer constants
9276 using code, even if it takes more instructions (the maximum is six).
9278 You would typically use this option to build a shared library dynamic
9279 loader. Itself a shared library, it must relocate itself in memory
9280 before it can find the variables and constants in its own data segment.
9286 Select whether to generate code to be assembled by the vendor-supplied
9287 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9305 Indicate whether GCC should generate code to use the optional BWX,
9306 CIX, FIX and MAX instruction sets. The default is to use the instruction
9307 sets supported by the CPU type specified via @option{-mcpu=} option or that
9308 of the CPU on which GCC was built if none was specified.
9313 @opindex mfloat-ieee
9314 Generate code that uses (does not use) VAX F and G floating point
9315 arithmetic instead of IEEE single and double precision.
9317 @item -mexplicit-relocs
9318 @itemx -mno-explicit-relocs
9319 @opindex mexplicit-relocs
9320 @opindex mno-explicit-relocs
9321 Older Alpha assemblers provided no way to generate symbol relocations
9322 except via assembler macros. Use of these macros does not allow
9323 optimal instruction scheduling. GNU binutils as of version 2.12
9324 supports a new syntax that allows the compiler to explicitly mark
9325 which relocations should apply to which instructions. This option
9326 is mostly useful for debugging, as GCC detects the capabilities of
9327 the assembler when it is built and sets the default accordingly.
9331 @opindex msmall-data
9332 @opindex mlarge-data
9333 When @option{-mexplicit-relocs} is in effect, static data is
9334 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9335 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9336 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9337 16-bit relocations off of the @code{$gp} register. This limits the
9338 size of the small data area to 64KB, but allows the variables to be
9339 directly accessed via a single instruction.
9341 The default is @option{-mlarge-data}. With this option the data area
9342 is limited to just below 2GB@. Programs that require more than 2GB of
9343 data must use @code{malloc} or @code{mmap} to allocate the data in the
9344 heap instead of in the program's data segment.
9346 When generating code for shared libraries, @option{-fpic} implies
9347 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9351 @opindex msmall-text
9352 @opindex mlarge-text
9353 When @option{-msmall-text} is used, the compiler assumes that the
9354 code of the entire program (or shared library) fits in 4MB, and is
9355 thus reachable with a branch instruction. When @option{-msmall-data}
9356 is used, the compiler can assume that all local symbols share the
9357 same @code{$gp} value, and thus reduce the number of instructions
9358 required for a function call from 4 to 1.
9360 The default is @option{-mlarge-text}.
9362 @item -mcpu=@var{cpu_type}
9364 Set the instruction set and instruction scheduling parameters for
9365 machine type @var{cpu_type}. You can specify either the @samp{EV}
9366 style name or the corresponding chip number. GCC supports scheduling
9367 parameters for the EV4, EV5 and EV6 family of processors and will
9368 choose the default values for the instruction set from the processor
9369 you specify. If you do not specify a processor type, GCC will default
9370 to the processor on which the compiler was built.
9372 Supported values for @var{cpu_type} are
9378 Schedules as an EV4 and has no instruction set extensions.
9382 Schedules as an EV5 and has no instruction set extensions.
9386 Schedules as an EV5 and supports the BWX extension.
9391 Schedules as an EV5 and supports the BWX and MAX extensions.
9395 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9399 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9402 @item -mtune=@var{cpu_type}
9404 Set only the instruction scheduling parameters for machine type
9405 @var{cpu_type}. The instruction set is not changed.
9407 @item -mmemory-latency=@var{time}
9408 @opindex mmemory-latency
9409 Sets the latency the scheduler should assume for typical memory
9410 references as seen by the application. This number is highly
9411 dependent on the memory access patterns used by the application
9412 and the size of the external cache on the machine.
9414 Valid options for @var{time} are
9418 A decimal number representing clock cycles.
9424 The compiler contains estimates of the number of clock cycles for
9425 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9426 (also called Dcache, Scache, and Bcache), as well as to main memory.
9427 Note that L3 is only valid for EV5.
9432 @node DEC Alpha/VMS Options
9433 @subsection DEC Alpha/VMS Options
9435 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9438 @item -mvms-return-codes
9439 @opindex mvms-return-codes
9440 Return VMS condition codes from main. The default is to return POSIX
9441 style condition (e.g.@: error) codes.
9445 @subsection FRV Options
9452 Only use the first 32 general purpose registers.
9457 Use all 64 general purpose registers.
9462 Use only the first 32 floating point registers.
9467 Use all 64 floating point registers
9470 @opindex mhard-float
9472 Use hardware instructions for floating point operations.
9475 @opindex msoft-float
9477 Use library routines for floating point operations.
9482 Dynamically allocate condition code registers.
9487 Do not try to dynamically allocate condition code registers, only
9488 use @code{icc0} and @code{fcc0}.
9493 Change ABI to use double word insns.
9498 Do not use double word instructions.
9503 Use floating point double instructions.
9508 Do not use floating point double instructions.
9513 Use media instructions.
9518 Do not use media instructions.
9523 Use multiply and add/subtract instructions.
9528 Do not use multiply and add/subtract instructions.
9533 Select the FDPIC ABI, that uses function descriptors to represent
9534 pointers to functions. Without any PIC/PIE-related options, it
9535 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9536 assumes GOT entries and small data are within a 12-bit range from the
9537 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9538 are computed with 32 bits.
9541 @opindex minline-plt
9543 Enable inlining of PLT entries in function calls to functions that are
9544 not known to bind locally. It has no effect without @option{-mfdpic}.
9545 It's enabled by default if optimizing for speed and compiling for
9546 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9547 optimization option such as @option{-O3} or above is present in the
9553 Assume a large TLS segment when generating thread-local code.
9558 Do not assume a large TLS segment when generating thread-local code.
9563 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9564 that is known to be in read-only sections. It's enabled by default,
9565 except for @option{-fpic} or @option{-fpie}: even though it may help
9566 make the global offset table smaller, it trades 1 instruction for 4.
9567 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9568 one of which may be shared by multiple symbols, and it avoids the need
9569 for a GOT entry for the referenced symbol, so it's more likely to be a
9570 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9572 @item -multilib-library-pic
9573 @opindex multilib-library-pic
9575 Link with the (library, not FD) pic libraries. It's implied by
9576 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9577 @option{-fpic} without @option{-mfdpic}. You should never have to use
9583 Follow the EABI requirement of always creating a frame pointer whenever
9584 a stack frame is allocated. This option is enabled by default and can
9585 be disabled with @option{-mno-linked-fp}.
9588 @opindex mlong-calls
9590 Use indirect addressing to call functions outside the current
9591 compilation unit. This allows the functions to be placed anywhere
9592 within the 32-bit address space.
9594 @item -malign-labels
9595 @opindex malign-labels
9597 Try to align labels to an 8-byte boundary by inserting nops into the
9598 previous packet. This option only has an effect when VLIW packing
9599 is enabled. It doesn't create new packets; it merely adds nops to
9603 @opindex mlibrary-pic
9605 Generate position-independent EABI code.
9610 Use only the first four media accumulator registers.
9615 Use all eight media accumulator registers.
9620 Pack VLIW instructions.
9625 Do not pack VLIW instructions.
9630 Do not mark ABI switches in e_flags.
9635 Enable the use of conditional-move instructions (default).
9637 This switch is mainly for debugging the compiler and will likely be removed
9638 in a future version.
9640 @item -mno-cond-move
9641 @opindex mno-cond-move
9643 Disable the use of conditional-move instructions.
9645 This switch is mainly for debugging the compiler and will likely be removed
9646 in a future version.
9651 Enable the use of conditional set instructions (default).
9653 This switch is mainly for debugging the compiler and will likely be removed
9654 in a future version.
9659 Disable the use of conditional set instructions.
9661 This switch is mainly for debugging the compiler and will likely be removed
9662 in a future version.
9667 Enable the use of conditional execution (default).
9669 This switch is mainly for debugging the compiler and will likely be removed
9670 in a future version.
9672 @item -mno-cond-exec
9673 @opindex mno-cond-exec
9675 Disable the use of conditional execution.
9677 This switch is mainly for debugging the compiler and will likely be removed
9678 in a future version.
9681 @opindex mvliw-branch
9683 Run a pass to pack branches into VLIW instructions (default).
9685 This switch is mainly for debugging the compiler and will likely be removed
9686 in a future version.
9688 @item -mno-vliw-branch
9689 @opindex mno-vliw-branch
9691 Do not run a pass to pack branches into VLIW instructions.
9693 This switch is mainly for debugging the compiler and will likely be removed
9694 in a future version.
9696 @item -mmulti-cond-exec
9697 @opindex mmulti-cond-exec
9699 Enable optimization of @code{&&} and @code{||} in conditional execution
9702 This switch is mainly for debugging the compiler and will likely be removed
9703 in a future version.
9705 @item -mno-multi-cond-exec
9706 @opindex mno-multi-cond-exec
9708 Disable optimization of @code{&&} and @code{||} in conditional execution.
9710 This switch is mainly for debugging the compiler and will likely be removed
9711 in a future version.
9713 @item -mnested-cond-exec
9714 @opindex mnested-cond-exec
9716 Enable nested conditional execution optimizations (default).
9718 This switch is mainly for debugging the compiler and will likely be removed
9719 in a future version.
9721 @item -mno-nested-cond-exec
9722 @opindex mno-nested-cond-exec
9724 Disable nested conditional execution optimizations.
9726 This switch is mainly for debugging the compiler and will likely be removed
9727 in a future version.
9729 @item -moptimize-membar
9730 @opindex moptimize-membar
9732 This switch removes redundant @code{membar} instructions from the
9733 compiler generated code. It is enabled by default.
9735 @item -mno-optimize-membar
9736 @opindex mno-optimize-membar
9738 This switch disables the automatic removal of redundant @code{membar}
9739 instructions from the generated code.
9741 @item -mtomcat-stats
9742 @opindex mtomcat-stats
9744 Cause gas to print out tomcat statistics.
9746 @item -mcpu=@var{cpu}
9749 Select the processor type for which to generate code. Possible values are
9750 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
9751 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
9755 @node GNU/Linux Options
9756 @subsection GNU/Linux Options
9758 These @samp{-m} options are defined for GNU/Linux targets:
9763 Use the GNU C library instead of uClibc. This is the default except
9764 on @samp{*-*-linux-*uclibc*} targets.
9768 Use uClibc instead of the GNU C library. This is the default on
9769 @samp{*-*-linux-*uclibc*} targets.
9772 @node H8/300 Options
9773 @subsection H8/300 Options
9775 These @samp{-m} options are defined for the H8/300 implementations:
9780 Shorten some address references at link time, when possible; uses the
9781 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9782 ld, Using ld}, for a fuller description.
9786 Generate code for the H8/300H@.
9790 Generate code for the H8S@.
9794 Generate code for the H8S and H8/300H in the normal mode. This switch
9795 must be used either with @option{-mh} or @option{-ms}.
9799 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9803 Make @code{int} data 32 bits by default.
9807 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9808 The default for the H8/300H and H8S is to align longs and floats on 4
9810 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9811 This option has no effect on the H8/300.
9815 @subsection HPPA Options
9816 @cindex HPPA Options
9818 These @samp{-m} options are defined for the HPPA family of computers:
9821 @item -march=@var{architecture-type}
9823 Generate code for the specified architecture. The choices for
9824 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9825 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9826 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9827 architecture option for your machine. Code compiled for lower numbered
9828 architectures will run on higher numbered architectures, but not the
9832 @itemx -mpa-risc-1-1
9833 @itemx -mpa-risc-2-0
9834 @opindex mpa-risc-1-0
9835 @opindex mpa-risc-1-1
9836 @opindex mpa-risc-2-0
9837 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9840 @opindex mbig-switch
9841 Generate code suitable for big switch tables. Use this option only if
9842 the assembler/linker complain about out of range branches within a switch
9845 @item -mjump-in-delay
9846 @opindex mjump-in-delay
9847 Fill delay slots of function calls with unconditional jump instructions
9848 by modifying the return pointer for the function call to be the target
9849 of the conditional jump.
9851 @item -mdisable-fpregs
9852 @opindex mdisable-fpregs
9853 Prevent floating point registers from being used in any manner. This is
9854 necessary for compiling kernels which perform lazy context switching of
9855 floating point registers. If you use this option and attempt to perform
9856 floating point operations, the compiler will abort.
9858 @item -mdisable-indexing
9859 @opindex mdisable-indexing
9860 Prevent the compiler from using indexing address modes. This avoids some
9861 rather obscure problems when compiling MIG generated code under MACH@.
9863 @item -mno-space-regs
9864 @opindex mno-space-regs
9865 Generate code that assumes the target has no space registers. This allows
9866 GCC to generate faster indirect calls and use unscaled index address modes.
9868 Such code is suitable for level 0 PA systems and kernels.
9870 @item -mfast-indirect-calls
9871 @opindex mfast-indirect-calls
9872 Generate code that assumes calls never cross space boundaries. This
9873 allows GCC to emit code which performs faster indirect calls.
9875 This option will not work in the presence of shared libraries or nested
9878 @item -mfixed-range=@var{register-range}
9879 @opindex mfixed-range
9880 Generate code treating the given register range as fixed registers.
9881 A fixed register is one that the register allocator can not use. This is
9882 useful when compiling kernel code. A register range is specified as
9883 two registers separated by a dash. Multiple register ranges can be
9884 specified separated by a comma.
9886 @item -mlong-load-store
9887 @opindex mlong-load-store
9888 Generate 3-instruction load and store sequences as sometimes required by
9889 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9892 @item -mportable-runtime
9893 @opindex mportable-runtime
9894 Use the portable calling conventions proposed by HP for ELF systems.
9898 Enable the use of assembler directives only GAS understands.
9900 @item -mschedule=@var{cpu-type}
9902 Schedule code according to the constraints for the machine type
9903 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9904 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9905 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9906 proper scheduling option for your machine. The default scheduling is
9910 @opindex mlinker-opt
9911 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9912 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9913 linkers in which they give bogus error messages when linking some programs.
9916 @opindex msoft-float
9917 Generate output containing library calls for floating point.
9918 @strong{Warning:} the requisite libraries are not available for all HPPA
9919 targets. Normally the facilities of the machine's usual C compiler are
9920 used, but this cannot be done directly in cross-compilation. You must make
9921 your own arrangements to provide suitable library functions for
9922 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9923 does provide software floating point support.
9925 @option{-msoft-float} changes the calling convention in the output file;
9926 therefore, it is only useful if you compile @emph{all} of a program with
9927 this option. In particular, you need to compile @file{libgcc.a}, the
9928 library that comes with GCC, with @option{-msoft-float} in order for
9933 Generate the predefine, @code{_SIO}, for server IO@. The default is
9934 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9935 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9936 options are available under HP-UX and HI-UX@.
9940 Use GNU ld specific options. This passes @option{-shared} to ld when
9941 building a shared library. It is the default when GCC is configured,
9942 explicitly or implicitly, with the GNU linker. This option does not
9943 have any affect on which ld is called, it only changes what parameters
9944 are passed to that ld. The ld that is called is determined by the
9945 @option{--with-ld} configure option, GCC's program search path, and
9946 finally by the user's @env{PATH}. The linker used by GCC can be printed
9947 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9948 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9952 Use HP ld specific options. This passes @option{-b} to ld when building
9953 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9954 links. It is the default when GCC is configured, explicitly or
9955 implicitly, with the HP linker. This option does not have any affect on
9956 which ld is called, it only changes what parameters are passed to that
9957 ld. The ld that is called is determined by the @option{--with-ld}
9958 configure option, GCC's program search path, and finally by the user's
9959 @env{PATH}. The linker used by GCC can be printed using @samp{which
9960 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9961 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9964 @opindex mno-long-calls
9965 Generate code that uses long call sequences. This ensures that a call
9966 is always able to reach linker generated stubs. The default is to generate
9967 long calls only when the distance from the call site to the beginning
9968 of the function or translation unit, as the case may be, exceeds a
9969 predefined limit set by the branch type being used. The limits for
9970 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9971 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9974 Distances are measured from the beginning of functions when using the
9975 @option{-ffunction-sections} option, or when using the @option{-mgas}
9976 and @option{-mno-portable-runtime} options together under HP-UX with
9979 It is normally not desirable to use this option as it will degrade
9980 performance. However, it may be useful in large applications,
9981 particularly when partial linking is used to build the application.
9983 The types of long calls used depends on the capabilities of the
9984 assembler and linker, and the type of code being generated. The
9985 impact on systems that support long absolute calls, and long pic
9986 symbol-difference or pc-relative calls should be relatively small.
9987 However, an indirect call is used on 32-bit ELF systems in pic code
9988 and it is quite long.
9990 @item -munix=@var{unix-std}
9992 Generate compiler predefines and select a startfile for the specified
9993 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9994 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9995 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9996 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9997 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10000 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10001 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10002 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10003 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10004 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10005 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10007 It is @emph{important} to note that this option changes the interfaces
10008 for various library routines. It also affects the operational behavior
10009 of the C library. Thus, @emph{extreme} care is needed in using this
10012 Library code that is intended to operate with more than one UNIX
10013 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10014 as appropriate. Most GNU software doesn't provide this capability.
10018 Suppress the generation of link options to search libdld.sl when the
10019 @option{-static} option is specified on HP-UX 10 and later.
10023 The HP-UX implementation of setlocale in libc has a dependency on
10024 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10025 when the @option{-static} option is specified, special link options
10026 are needed to resolve this dependency.
10028 On HP-UX 10 and later, the GCC driver adds the necessary options to
10029 link with libdld.sl when the @option{-static} option is specified.
10030 This causes the resulting binary to be dynamic. On the 64-bit port,
10031 the linkers generate dynamic binaries by default in any case. The
10032 @option{-nolibdld} option can be used to prevent the GCC driver from
10033 adding these link options.
10037 Add support for multithreading with the @dfn{dce thread} library
10038 under HP-UX@. This option sets flags for both the preprocessor and
10042 @node i386 and x86-64 Options
10043 @subsection Intel 386 and AMD x86-64 Options
10044 @cindex i386 Options
10045 @cindex x86-64 Options
10046 @cindex Intel 386 Options
10047 @cindex AMD x86-64 Options
10049 These @samp{-m} options are defined for the i386 and x86-64 family of
10053 @item -mtune=@var{cpu-type}
10055 Tune to @var{cpu-type} everything applicable about the generated code, except
10056 for the ABI and the set of available instructions. The choices for
10057 @var{cpu-type} are:
10060 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10061 If you know the CPU on which your code will run, then you should use
10062 the corresponding @option{-mtune} option instead of
10063 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10064 of your application will have, then you should use this option.
10066 As new processors are deployed in the marketplace, the behavior of this
10067 option will change. Therefore, if you upgrade to a newer version of
10068 GCC, the code generated option will change to reflect the processors
10069 that were most common when that version of GCC was released.
10071 There is no @option{-march=generic} option because @option{-march}
10072 indicates the instruction set the compiler can use, and there is no
10073 generic instruction set applicable to all processors. In contrast,
10074 @option{-mtune} indicates the processor (or, in this case, collection of
10075 processors) for which the code is optimized.
10077 This selects the CPU to tune for at compilation time by determining
10078 the processor type of the compiling machine. Using @option{-mtune=native}
10079 will produce code optimized for the local machine under the constraints
10080 of the selected instruction set. Using @option{-march=native} will
10081 enable all instruction subsets supported by the local machine (hence
10082 the result might not run on different machines).
10084 Original Intel's i386 CPU@.
10086 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10087 @item i586, pentium
10088 Intel Pentium CPU with no MMX support.
10090 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10092 Intel PentiumPro CPU@.
10094 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10095 instruction set will be used, so the code will run on all i686 family chips.
10097 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10098 @item pentium3, pentium3m
10099 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10102 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10103 support. Used by Centrino notebooks.
10104 @item pentium4, pentium4m
10105 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10107 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10110 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10111 SSE2 and SSE3 instruction set support.
10113 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10114 instruction set support.
10116 AMD K6 CPU with MMX instruction set support.
10118 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
10119 @item athlon, athlon-tbird
10120 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
10122 @item athlon-4, athlon-xp, athlon-mp
10123 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
10124 instruction set support.
10125 @item k8, opteron, athlon64, athlon-fx
10126 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10127 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
10128 @item k8-sse3, opteron-sse3, athlon64-sse3
10129 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10130 @item amdfam10, barcelona
10131 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10132 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10133 instruction set extensions.)
10135 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10138 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
10139 instruction set support.
10141 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
10142 implemented for this chip.)
10144 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10145 implemented for this chip.)
10147 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10150 While picking a specific @var{cpu-type} will schedule things appropriately
10151 for that particular chip, the compiler will not generate any code that
10152 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10155 @item -march=@var{cpu-type}
10157 Generate instructions for the machine type @var{cpu-type}. The choices
10158 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10159 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10161 @item -mcpu=@var{cpu-type}
10163 A deprecated synonym for @option{-mtune}.
10165 @item -mfpmath=@var{unit}
10167 Generate floating point arithmetics for selected unit @var{unit}. The choices
10168 for @var{unit} are:
10172 Use the standard 387 floating point coprocessor present majority of chips and
10173 emulated otherwise. Code compiled with this option will run almost everywhere.
10174 The temporary results are computed in 80bit precision instead of precision
10175 specified by the type resulting in slightly different results compared to most
10176 of other chips. See @option{-ffloat-store} for more detailed description.
10178 This is the default choice for i386 compiler.
10181 Use scalar floating point instructions present in the SSE instruction set.
10182 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10183 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10184 instruction set supports only single precision arithmetics, thus the double and
10185 extended precision arithmetics is still done using 387. Later version, present
10186 only in Pentium4 and the future AMD x86-64 chips supports double precision
10189 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10190 or @option{-msse2} switches to enable SSE extensions and make this option
10191 effective. For the x86-64 compiler, these extensions are enabled by default.
10193 The resulting code should be considerably faster in the majority of cases and avoid
10194 the numerical instability problems of 387 code, but may break some existing
10195 code that expects temporaries to be 80bit.
10197 This is the default choice for the x86-64 compiler.
10200 Attempt to utilize both instruction sets at once. This effectively double the
10201 amount of available registers and on chips with separate execution units for
10202 387 and SSE the execution resources too. Use this option with care, as it is
10203 still experimental, because the GCC register allocator does not model separate
10204 functional units well resulting in instable performance.
10207 @item -masm=@var{dialect}
10208 @opindex masm=@var{dialect}
10209 Output asm instructions using selected @var{dialect}. Supported
10210 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10211 not support @samp{intel}.
10214 @itemx -mno-ieee-fp
10216 @opindex mno-ieee-fp
10217 Control whether or not the compiler uses IEEE floating point
10218 comparisons. These handle correctly the case where the result of a
10219 comparison is unordered.
10222 @opindex msoft-float
10223 Generate output containing library calls for floating point.
10224 @strong{Warning:} the requisite libraries are not part of GCC@.
10225 Normally the facilities of the machine's usual C compiler are used, but
10226 this can't be done directly in cross-compilation. You must make your
10227 own arrangements to provide suitable library functions for
10230 On machines where a function returns floating point results in the 80387
10231 register stack, some floating point opcodes may be emitted even if
10232 @option{-msoft-float} is used.
10234 @item -mno-fp-ret-in-387
10235 @opindex mno-fp-ret-in-387
10236 Do not use the FPU registers for return values of functions.
10238 The usual calling convention has functions return values of types
10239 @code{float} and @code{double} in an FPU register, even if there
10240 is no FPU@. The idea is that the operating system should emulate
10243 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10244 in ordinary CPU registers instead.
10246 @item -mno-fancy-math-387
10247 @opindex mno-fancy-math-387
10248 Some 387 emulators do not support the @code{sin}, @code{cos} and
10249 @code{sqrt} instructions for the 387. Specify this option to avoid
10250 generating those instructions. This option is the default on FreeBSD,
10251 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10252 indicates that the target cpu will always have an FPU and so the
10253 instruction will not need emulation. As of revision 2.6.1, these
10254 instructions are not generated unless you also use the
10255 @option{-funsafe-math-optimizations} switch.
10257 @item -malign-double
10258 @itemx -mno-align-double
10259 @opindex malign-double
10260 @opindex mno-align-double
10261 Control whether GCC aligns @code{double}, @code{long double}, and
10262 @code{long long} variables on a two word boundary or a one word
10263 boundary. Aligning @code{double} variables on a two word boundary will
10264 produce code that runs somewhat faster on a @samp{Pentium} at the
10265 expense of more memory.
10267 On x86-64, @option{-malign-double} is enabled by default.
10269 @strong{Warning:} if you use the @option{-malign-double} switch,
10270 structures containing the above types will be aligned differently than
10271 the published application binary interface specifications for the 386
10272 and will not be binary compatible with structures in code compiled
10273 without that switch.
10275 @item -m96bit-long-double
10276 @itemx -m128bit-long-double
10277 @opindex m96bit-long-double
10278 @opindex m128bit-long-double
10279 These switches control the size of @code{long double} type. The i386
10280 application binary interface specifies the size to be 96 bits,
10281 so @option{-m96bit-long-double} is the default in 32 bit mode.
10283 Modern architectures (Pentium and newer) would prefer @code{long double}
10284 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10285 conforming to the ABI, this would not be possible. So specifying a
10286 @option{-m128bit-long-double} will align @code{long double}
10287 to a 16 byte boundary by padding the @code{long double} with an additional
10290 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10291 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10293 Notice that neither of these options enable any extra precision over the x87
10294 standard of 80 bits for a @code{long double}.
10296 @strong{Warning:} if you override the default value for your target ABI, the
10297 structures and arrays containing @code{long double} variables will change
10298 their size as well as function calling convention for function taking
10299 @code{long double} will be modified. Hence they will not be binary
10300 compatible with arrays or structures in code compiled without that switch.
10302 @item -mmlarge-data-threshold=@var{number}
10303 @opindex mlarge-data-threshold=@var{number}
10304 When @option{-mcmodel=medium} is specified, the data greater than
10305 @var{threshold} are placed in large data section. This value must be the
10306 same across all object linked into the binary and defaults to 65535.
10310 Use a different function-calling convention, in which functions that
10311 take a fixed number of arguments return with the @code{ret} @var{num}
10312 instruction, which pops their arguments while returning. This saves one
10313 instruction in the caller since there is no need to pop the arguments
10316 You can specify that an individual function is called with this calling
10317 sequence with the function attribute @samp{stdcall}. You can also
10318 override the @option{-mrtd} option by using the function attribute
10319 @samp{cdecl}. @xref{Function Attributes}.
10321 @strong{Warning:} this calling convention is incompatible with the one
10322 normally used on Unix, so you cannot use it if you need to call
10323 libraries compiled with the Unix compiler.
10325 Also, you must provide function prototypes for all functions that
10326 take variable numbers of arguments (including @code{printf});
10327 otherwise incorrect code will be generated for calls to those
10330 In addition, seriously incorrect code will result if you call a
10331 function with too many arguments. (Normally, extra arguments are
10332 harmlessly ignored.)
10334 @item -mregparm=@var{num}
10336 Control how many registers are used to pass integer arguments. By
10337 default, no registers are used to pass arguments, and at most 3
10338 registers can be used. You can control this behavior for a specific
10339 function by using the function attribute @samp{regparm}.
10340 @xref{Function Attributes}.
10342 @strong{Warning:} if you use this switch, and
10343 @var{num} is nonzero, then you must build all modules with the same
10344 value, including any libraries. This includes the system libraries and
10348 @opindex msseregparm
10349 Use SSE register passing conventions for float and double arguments
10350 and return values. You can control this behavior for a specific
10351 function by using the function attribute @samp{sseregparm}.
10352 @xref{Function Attributes}.
10354 @strong{Warning:} if you use this switch then you must build all
10355 modules with the same value, including any libraries. This includes
10356 the system libraries and startup modules.
10365 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10366 is specified, the significands of results of floating-point operations are
10367 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10368 significands of results of floating-point operations to 53 bits (double
10369 precision) and @option{-mpc80} rounds the significands of results of
10370 floating-point operations to 64 bits (extended double precision), which is
10371 the default. When this option is used, floating-point operations in higher
10372 precisions are not available to the programmer without setting the FPU
10373 control word explicitly.
10375 Setting the rounding of floating-point operations to less than the default
10376 80 bits can speed some programs by 2% or more. Note that some mathematical
10377 libraries assume that extended precision (80 bit) floating-point operations
10378 are enabled by default; routines in such libraries could suffer significant
10379 loss of accuracy, typically through so-called "catastrophic cancellation",
10380 when this option is used to set the precision to less than extended precision.
10382 @item -mstackrealign
10383 @opindex mstackrealign
10384 Realign the stack at entry. On the Intel x86, the
10385 @option{-mstackrealign} option will generate an alternate prologue and
10386 epilogue that realigns the runtime stack. This supports mixing legacy
10387 codes that keep a 4-byte aligned stack with modern codes that keep a
10388 16-byte stack for SSE compatibility. The alternate prologue and
10389 epilogue are slower and bigger than the regular ones, and the
10390 alternate prologue requires an extra scratch register; this lowers the
10391 number of registers available if used in conjunction with the
10392 @code{regparm} attribute. The @option{-mstackrealign} option is
10393 incompatible with the nested function prologue; this is considered a
10394 hard error. See also the attribute @code{force_align_arg_pointer},
10395 applicable to individual functions.
10397 @item -mpreferred-stack-boundary=@var{num}
10398 @opindex mpreferred-stack-boundary
10399 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10400 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10401 the default is 4 (16 bytes or 128 bits).
10403 On Pentium and PentiumPro, @code{double} and @code{long double} values
10404 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10405 suffer significant run time performance penalties. On Pentium III, the
10406 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10407 properly if it is not 16 byte aligned.
10409 To ensure proper alignment of this values on the stack, the stack boundary
10410 must be as aligned as that required by any value stored on the stack.
10411 Further, every function must be generated such that it keeps the stack
10412 aligned. Thus calling a function compiled with a higher preferred
10413 stack boundary from a function compiled with a lower preferred stack
10414 boundary will most likely misalign the stack. It is recommended that
10415 libraries that use callbacks always use the default setting.
10417 This extra alignment does consume extra stack space, and generally
10418 increases code size. Code that is sensitive to stack space usage, such
10419 as embedded systems and operating system kernels, may want to reduce the
10420 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10452 These switches enable or disable the use of instructions in the MMX,
10453 SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4A, ABM or 3DNow! extended
10455 These extensions are also available as built-in functions: see
10456 @ref{X86 Built-in Functions}, for details of the functions enabled and
10457 disabled by these switches.
10459 To have SSE/SSE2 instructions generated automatically from floating-point
10460 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10462 These options will enable GCC to use these extended instructions in
10463 generated code, even without @option{-mfpmath=sse}. Applications which
10464 perform runtime CPU detection must compile separate files for each
10465 supported architecture, using the appropriate flags. In particular,
10466 the file containing the CPU detection code should be compiled without
10471 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10472 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10473 data types. This is useful for high resolution counters that could be updated
10474 by multiple processors (or cores). This instruction is generated as part of
10475 atomic built-in functions: see @ref{Atomic Builtins} for details.
10479 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10480 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10481 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10482 SAHF are load and store instructions, respectively, for certain status flags.
10483 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10484 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10488 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10489 vectorized variants RCPPS and RSQRTPS) instead of DIVSS and SQRTSS (and their
10490 vectorized variants). These instructions will be generated only when
10491 @option{-funsafe-math-optimizations} is enabled.
10493 @item -mveclibabi=@var{type}
10494 @opindex mveclibabi
10495 Specifies the ABI type to use for vectorizing intrinsics using an
10496 external library. Supported types are @code{acml} for the AMD
10497 math core library style of interfacing. GCC will currently emit
10498 calls to @code{__vrd2_sin}, @code{__vrd2_cos}, @code{__vrd2_exp},
10499 @code{__vrd2_log}, @code{__vrd2_log2}, @code{__vrd2_log10},
10500 @code{__vrs4_sinf}, @code{__vrs4_cosf}, @code{__vrs4_expf},
10501 @code{__vrs4_logf}, @code{__vrs4_log2f}, @code{__vrs4_log10f}
10502 and @code{__vrs4_powf} when using this type and @option{-ftree-vectorize}
10503 is enabled. A ACML ABI compatible library will have to be specified
10507 @itemx -mno-push-args
10508 @opindex mpush-args
10509 @opindex mno-push-args
10510 Use PUSH operations to store outgoing parameters. This method is shorter
10511 and usually equally fast as method using SUB/MOV operations and is enabled
10512 by default. In some cases disabling it may improve performance because of
10513 improved scheduling and reduced dependencies.
10515 @item -maccumulate-outgoing-args
10516 @opindex maccumulate-outgoing-args
10517 If enabled, the maximum amount of space required for outgoing arguments will be
10518 computed in the function prologue. This is faster on most modern CPUs
10519 because of reduced dependencies, improved scheduling and reduced stack usage
10520 when preferred stack boundary is not equal to 2. The drawback is a notable
10521 increase in code size. This switch implies @option{-mno-push-args}.
10525 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10526 on thread-safe exception handling must compile and link all code with the
10527 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10528 @option{-D_MT}; when linking, it links in a special thread helper library
10529 @option{-lmingwthrd} which cleans up per thread exception handling data.
10531 @item -mno-align-stringops
10532 @opindex mno-align-stringops
10533 Do not align destination of inlined string operations. This switch reduces
10534 code size and improves performance in case the destination is already aligned,
10535 but GCC doesn't know about it.
10537 @item -minline-all-stringops
10538 @opindex minline-all-stringops
10539 By default GCC inlines string operations only when destination is known to be
10540 aligned at least to 4 byte boundary. This enables more inlining, increase code
10541 size, but may improve performance of code that depends on fast memcpy, strlen
10542 and memset for short lengths.
10544 @item -minline-stringops-dynamically
10545 @opindex minline-stringops-dynamically
10546 For string operation of unknown size, inline runtime checks so for small
10547 blocks inline code is used, while for large blocks library call is used.
10549 @item -mstringop-strategy=@var{alg}
10550 @opindex mstringop-strategy=@var{alg}
10551 Overwrite internal decision heuristic about particular algorithm to inline
10552 string operation with. The allowed values are @code{rep_byte},
10553 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10554 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10555 expanding inline loop, @code{libcall} for always expanding library call.
10557 @item -momit-leaf-frame-pointer
10558 @opindex momit-leaf-frame-pointer
10559 Don't keep the frame pointer in a register for leaf functions. This
10560 avoids the instructions to save, set up and restore frame pointers and
10561 makes an extra register available in leaf functions. The option
10562 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10563 which might make debugging harder.
10565 @item -mtls-direct-seg-refs
10566 @itemx -mno-tls-direct-seg-refs
10567 @opindex mtls-direct-seg-refs
10568 Controls whether TLS variables may be accessed with offsets from the
10569 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10570 or whether the thread base pointer must be added. Whether or not this
10571 is legal depends on the operating system, and whether it maps the
10572 segment to cover the entire TLS area.
10574 For systems that use GNU libc, the default is on.
10577 These @samp{-m} switches are supported in addition to the above
10578 on AMD x86-64 processors in 64-bit environments.
10585 Generate code for a 32-bit or 64-bit environment.
10586 The 32-bit environment sets int, long and pointer to 32 bits and
10587 generates code that runs on any i386 system.
10588 The 64-bit environment sets int to 32 bits and long and pointer
10589 to 64 bits and generates code for AMD's x86-64 architecture. For
10590 darwin only the -m64 option turns off the @option{-fno-pic} and
10591 @option{-mdynamic-no-pic} options.
10593 @item -mno-red-zone
10594 @opindex no-red-zone
10595 Do not use a so called red zone for x86-64 code. The red zone is mandated
10596 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10597 stack pointer that will not be modified by signal or interrupt handlers
10598 and therefore can be used for temporary data without adjusting the stack
10599 pointer. The flag @option{-mno-red-zone} disables this red zone.
10601 @item -mcmodel=small
10602 @opindex mcmodel=small
10603 Generate code for the small code model: the program and its symbols must
10604 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
10605 Programs can be statically or dynamically linked. This is the default
10608 @item -mcmodel=kernel
10609 @opindex mcmodel=kernel
10610 Generate code for the kernel code model. The kernel runs in the
10611 negative 2 GB of the address space.
10612 This model has to be used for Linux kernel code.
10614 @item -mcmodel=medium
10615 @opindex mcmodel=medium
10616 Generate code for the medium model: The program is linked in the lower 2
10617 GB of the address space but symbols can be located anywhere in the
10618 address space. Programs can be statically or dynamically linked, but
10619 building of shared libraries are not supported with the medium model.
10621 @item -mcmodel=large
10622 @opindex mcmodel=large
10623 Generate code for the large model: This model makes no assumptions
10624 about addresses and sizes of sections.
10627 @node IA-64 Options
10628 @subsection IA-64 Options
10629 @cindex IA-64 Options
10631 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10635 @opindex mbig-endian
10636 Generate code for a big endian target. This is the default for HP-UX@.
10638 @item -mlittle-endian
10639 @opindex mlittle-endian
10640 Generate code for a little endian target. This is the default for AIX5
10646 @opindex mno-gnu-as
10647 Generate (or don't) code for the GNU assembler. This is the default.
10648 @c Also, this is the default if the configure option @option{--with-gnu-as}
10654 @opindex mno-gnu-ld
10655 Generate (or don't) code for the GNU linker. This is the default.
10656 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10661 Generate code that does not use a global pointer register. The result
10662 is not position independent code, and violates the IA-64 ABI@.
10664 @item -mvolatile-asm-stop
10665 @itemx -mno-volatile-asm-stop
10666 @opindex mvolatile-asm-stop
10667 @opindex mno-volatile-asm-stop
10668 Generate (or don't) a stop bit immediately before and after volatile asm
10671 @item -mregister-names
10672 @itemx -mno-register-names
10673 @opindex mregister-names
10674 @opindex mno-register-names
10675 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10676 the stacked registers. This may make assembler output more readable.
10682 Disable (or enable) optimizations that use the small data section. This may
10683 be useful for working around optimizer bugs.
10685 @item -mconstant-gp
10686 @opindex mconstant-gp
10687 Generate code that uses a single constant global pointer value. This is
10688 useful when compiling kernel code.
10692 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10693 This is useful when compiling firmware code.
10695 @item -minline-float-divide-min-latency
10696 @opindex minline-float-divide-min-latency
10697 Generate code for inline divides of floating point values
10698 using the minimum latency algorithm.
10700 @item -minline-float-divide-max-throughput
10701 @opindex minline-float-divide-max-throughput
10702 Generate code for inline divides of floating point values
10703 using the maximum throughput algorithm.
10705 @item -minline-int-divide-min-latency
10706 @opindex minline-int-divide-min-latency
10707 Generate code for inline divides of integer values
10708 using the minimum latency algorithm.
10710 @item -minline-int-divide-max-throughput
10711 @opindex minline-int-divide-max-throughput
10712 Generate code for inline divides of integer values
10713 using the maximum throughput algorithm.
10715 @item -minline-sqrt-min-latency
10716 @opindex minline-sqrt-min-latency
10717 Generate code for inline square roots
10718 using the minimum latency algorithm.
10720 @item -minline-sqrt-max-throughput
10721 @opindex minline-sqrt-max-throughput
10722 Generate code for inline square roots
10723 using the maximum throughput algorithm.
10725 @item -mno-dwarf2-asm
10726 @itemx -mdwarf2-asm
10727 @opindex mno-dwarf2-asm
10728 @opindex mdwarf2-asm
10729 Don't (or do) generate assembler code for the DWARF2 line number debugging
10730 info. This may be useful when not using the GNU assembler.
10732 @item -mearly-stop-bits
10733 @itemx -mno-early-stop-bits
10734 @opindex mearly-stop-bits
10735 @opindex mno-early-stop-bits
10736 Allow stop bits to be placed earlier than immediately preceding the
10737 instruction that triggered the stop bit. This can improve instruction
10738 scheduling, but does not always do so.
10740 @item -mfixed-range=@var{register-range}
10741 @opindex mfixed-range
10742 Generate code treating the given register range as fixed registers.
10743 A fixed register is one that the register allocator can not use. This is
10744 useful when compiling kernel code. A register range is specified as
10745 two registers separated by a dash. Multiple register ranges can be
10746 specified separated by a comma.
10748 @item -mtls-size=@var{tls-size}
10750 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
10753 @item -mtune=@var{cpu-type}
10755 Tune the instruction scheduling for a particular CPU, Valid values are
10756 itanium, itanium1, merced, itanium2, and mckinley.
10762 Add support for multithreading using the POSIX threads library. This
10763 option sets flags for both the preprocessor and linker. It does
10764 not affect the thread safety of object code produced by the compiler or
10765 that of libraries supplied with it. These are HP-UX specific flags.
10771 Generate code for a 32-bit or 64-bit environment.
10772 The 32-bit environment sets int, long and pointer to 32 bits.
10773 The 64-bit environment sets int to 32 bits and long and pointer
10774 to 64 bits. These are HP-UX specific flags.
10776 @item -mno-sched-br-data-spec
10777 @itemx -msched-br-data-spec
10778 @opindex mno-sched-br-data-spec
10779 @opindex msched-br-data-spec
10780 (Dis/En)able data speculative scheduling before reload.
10781 This will result in generation of the ld.a instructions and
10782 the corresponding check instructions (ld.c / chk.a).
10783 The default is 'disable'.
10785 @item -msched-ar-data-spec
10786 @itemx -mno-sched-ar-data-spec
10787 @opindex msched-ar-data-spec
10788 @opindex mno-sched-ar-data-spec
10789 (En/Dis)able data speculative scheduling after reload.
10790 This will result in generation of the ld.a instructions and
10791 the corresponding check instructions (ld.c / chk.a).
10792 The default is 'enable'.
10794 @item -mno-sched-control-spec
10795 @itemx -msched-control-spec
10796 @opindex mno-sched-control-spec
10797 @opindex msched-control-spec
10798 (Dis/En)able control speculative scheduling. This feature is
10799 available only during region scheduling (i.e. before reload).
10800 This will result in generation of the ld.s instructions and
10801 the corresponding check instructions chk.s .
10802 The default is 'disable'.
10804 @item -msched-br-in-data-spec
10805 @itemx -mno-sched-br-in-data-spec
10806 @opindex msched-br-in-data-spec
10807 @opindex mno-sched-br-in-data-spec
10808 (En/Dis)able speculative scheduling of the instructions that
10809 are dependent on the data speculative loads before reload.
10810 This is effective only with @option{-msched-br-data-spec} enabled.
10811 The default is 'enable'.
10813 @item -msched-ar-in-data-spec
10814 @itemx -mno-sched-ar-in-data-spec
10815 @opindex msched-ar-in-data-spec
10816 @opindex mno-sched-ar-in-data-spec
10817 (En/Dis)able speculative scheduling of the instructions that
10818 are dependent on the data speculative loads after reload.
10819 This is effective only with @option{-msched-ar-data-spec} enabled.
10820 The default is 'enable'.
10822 @item -msched-in-control-spec
10823 @itemx -mno-sched-in-control-spec
10824 @opindex msched-in-control-spec
10825 @opindex mno-sched-in-control-spec
10826 (En/Dis)able speculative scheduling of the instructions that
10827 are dependent on the control speculative loads.
10828 This is effective only with @option{-msched-control-spec} enabled.
10829 The default is 'enable'.
10832 @itemx -mno-sched-ldc
10833 @opindex msched-ldc
10834 @opindex mno-sched-ldc
10835 (En/Dis)able use of simple data speculation checks ld.c .
10836 If disabled, only chk.a instructions will be emitted to check
10837 data speculative loads.
10838 The default is 'enable'.
10840 @item -mno-sched-control-ldc
10841 @itemx -msched-control-ldc
10842 @opindex mno-sched-control-ldc
10843 @opindex msched-control-ldc
10844 (Dis/En)able use of ld.c instructions to check control speculative loads.
10845 If enabled, in case of control speculative load with no speculatively
10846 scheduled dependent instructions this load will be emitted as ld.sa and
10847 ld.c will be used to check it.
10848 The default is 'disable'.
10850 @item -mno-sched-spec-verbose
10851 @itemx -msched-spec-verbose
10852 @opindex mno-sched-spec-verbose
10853 @opindex msched-spec-verbose
10854 (Dis/En)able printing of the information about speculative motions.
10856 @item -mno-sched-prefer-non-data-spec-insns
10857 @itemx -msched-prefer-non-data-spec-insns
10858 @opindex mno-sched-prefer-non-data-spec-insns
10859 @opindex msched-prefer-non-data-spec-insns
10860 If enabled, data speculative instructions will be chosen for schedule
10861 only if there are no other choices at the moment. This will make
10862 the use of the data speculation much more conservative.
10863 The default is 'disable'.
10865 @item -mno-sched-prefer-non-control-spec-insns
10866 @itemx -msched-prefer-non-control-spec-insns
10867 @opindex mno-sched-prefer-non-control-spec-insns
10868 @opindex msched-prefer-non-control-spec-insns
10869 If enabled, control speculative instructions will be chosen for schedule
10870 only if there are no other choices at the moment. This will make
10871 the use of the control speculation much more conservative.
10872 The default is 'disable'.
10874 @item -mno-sched-count-spec-in-critical-path
10875 @itemx -msched-count-spec-in-critical-path
10876 @opindex mno-sched-count-spec-in-critical-path
10877 @opindex msched-count-spec-in-critical-path
10878 If enabled, speculative dependencies will be considered during
10879 computation of the instructions priorities. This will make the use of the
10880 speculation a bit more conservative.
10881 The default is 'disable'.
10886 @subsection M32C Options
10887 @cindex M32C options
10890 @item -mcpu=@var{name}
10892 Select the CPU for which code is generated. @var{name} may be one of
10893 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
10894 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
10895 the M32C/80 series.
10899 Specifies that the program will be run on the simulator. This causes
10900 an alternate runtime library to be linked in which supports, for
10901 example, file I/O. You must not use this option when generating
10902 programs that will run on real hardware; you must provide your own
10903 runtime library for whatever I/O functions are needed.
10905 @item -memregs=@var{number}
10907 Specifies the number of memory-based pseudo-registers GCC will use
10908 during code generation. These pseudo-registers will be used like real
10909 registers, so there is a tradeoff between GCC's ability to fit the
10910 code into available registers, and the performance penalty of using
10911 memory instead of registers. Note that all modules in a program must
10912 be compiled with the same value for this option. Because of that, you
10913 must not use this option with the default runtime libraries gcc
10918 @node M32R/D Options
10919 @subsection M32R/D Options
10920 @cindex M32R/D options
10922 These @option{-m} options are defined for Renesas M32R/D architectures:
10927 Generate code for the M32R/2@.
10931 Generate code for the M32R/X@.
10935 Generate code for the M32R@. This is the default.
10937 @item -mmodel=small
10938 @opindex mmodel=small
10939 Assume all objects live in the lower 16MB of memory (so that their addresses
10940 can be loaded with the @code{ld24} instruction), and assume all subroutines
10941 are reachable with the @code{bl} instruction.
10942 This is the default.
10944 The addressability of a particular object can be set with the
10945 @code{model} attribute.
10947 @item -mmodel=medium
10948 @opindex mmodel=medium
10949 Assume objects may be anywhere in the 32-bit address space (the compiler
10950 will generate @code{seth/add3} instructions to load their addresses), and
10951 assume all subroutines are reachable with the @code{bl} instruction.
10953 @item -mmodel=large
10954 @opindex mmodel=large
10955 Assume objects may be anywhere in the 32-bit address space (the compiler
10956 will generate @code{seth/add3} instructions to load their addresses), and
10957 assume subroutines may not be reachable with the @code{bl} instruction
10958 (the compiler will generate the much slower @code{seth/add3/jl}
10959 instruction sequence).
10962 @opindex msdata=none
10963 Disable use of the small data area. Variables will be put into
10964 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10965 @code{section} attribute has been specified).
10966 This is the default.
10968 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10969 Objects may be explicitly put in the small data area with the
10970 @code{section} attribute using one of these sections.
10972 @item -msdata=sdata
10973 @opindex msdata=sdata
10974 Put small global and static data in the small data area, but do not
10975 generate special code to reference them.
10978 @opindex msdata=use
10979 Put small global and static data in the small data area, and generate
10980 special instructions to reference them.
10984 @cindex smaller data references
10985 Put global and static objects less than or equal to @var{num} bytes
10986 into the small data or bss sections instead of the normal data or bss
10987 sections. The default value of @var{num} is 8.
10988 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10989 for this option to have any effect.
10991 All modules should be compiled with the same @option{-G @var{num}} value.
10992 Compiling with different values of @var{num} may or may not work; if it
10993 doesn't the linker will give an error message---incorrect code will not be
10998 Makes the M32R specific code in the compiler display some statistics
10999 that might help in debugging programs.
11001 @item -malign-loops
11002 @opindex malign-loops
11003 Align all loops to a 32-byte boundary.
11005 @item -mno-align-loops
11006 @opindex mno-align-loops
11007 Do not enforce a 32-byte alignment for loops. This is the default.
11009 @item -missue-rate=@var{number}
11010 @opindex missue-rate=@var{number}
11011 Issue @var{number} instructions per cycle. @var{number} can only be 1
11014 @item -mbranch-cost=@var{number}
11015 @opindex mbranch-cost=@var{number}
11016 @var{number} can only be 1 or 2. If it is 1 then branches will be
11017 preferred over conditional code, if it is 2, then the opposite will
11020 @item -mflush-trap=@var{number}
11021 @opindex mflush-trap=@var{number}
11022 Specifies the trap number to use to flush the cache. The default is
11023 12. Valid numbers are between 0 and 15 inclusive.
11025 @item -mno-flush-trap
11026 @opindex mno-flush-trap
11027 Specifies that the cache cannot be flushed by using a trap.
11029 @item -mflush-func=@var{name}
11030 @opindex mflush-func=@var{name}
11031 Specifies the name of the operating system function to call to flush
11032 the cache. The default is @emph{_flush_cache}, but a function call
11033 will only be used if a trap is not available.
11035 @item -mno-flush-func
11036 @opindex mno-flush-func
11037 Indicates that there is no OS function for flushing the cache.
11041 @node M680x0 Options
11042 @subsection M680x0 Options
11043 @cindex M680x0 options
11045 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11046 The default settings depend on which architecture was selected when
11047 the compiler was configured; the defaults for the most common choices
11051 @item -march=@var{arch}
11053 Generate code for a specific M680x0 or ColdFire instruction set
11054 architecture. Permissible values of @var{arch} for M680x0
11055 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11056 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11057 architectures are selected according to Freescale's ISA classification
11058 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11059 @samp{isab} and @samp{isac}.
11061 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11062 code for a ColdFire target. The @var{arch} in this macro is one of the
11063 @option{-march} arguments given above.
11065 When used together, @option{-march} and @option{-mtune} select code
11066 that runs on a family of similar processors but that is optimized
11067 for a particular microarchitecture.
11069 @item -mcpu=@var{cpu}
11071 Generate code for a specific M680x0 or ColdFire processor.
11072 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11073 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11074 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11075 below, which also classifies the CPUs into families:
11077 @multitable @columnfractions 0.20 0.80
11078 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11079 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11080 @item @samp{5206e} @tab @samp{5206e}
11081 @item @samp{5208} @tab @samp{5207} @samp{5208}
11082 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11083 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11084 @item @samp{5216} @tab @samp{5214} @samp{5216}
11085 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11086 @item @samp{5225} @tab @samp{5224} @samp{5225}
11087 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11088 @item @samp{5249} @tab @samp{5249}
11089 @item @samp{5250} @tab @samp{5250}
11090 @item @samp{5271} @tab @samp{5270} @samp{5271}
11091 @item @samp{5272} @tab @samp{5272}
11092 @item @samp{5275} @tab @samp{5274} @samp{5275}
11093 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11094 @item @samp{5307} @tab @samp{5307}
11095 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11096 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11097 @item @samp{5407} @tab @samp{5407}
11098 @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}
11101 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11102 @var{arch} is compatible with @var{cpu}. Other combinations of
11103 @option{-mcpu} and @option{-march} are rejected.
11105 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11106 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11107 where the value of @var{family} is given by the table above.
11109 @item -mtune=@var{tune}
11111 Tune the code for a particular microarchitecture, within the
11112 constraints set by @option{-march} and @option{-mcpu}.
11113 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11114 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11115 and @samp{cpu32}. The ColdFire microarchitectures
11116 are: @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11118 You can also use @option{-mtune=68020-40} for code that needs
11119 to run relatively well on 68020, 68030 and 68040 targets.
11120 @option{-mtune=68020-60} is similar but includes 68060 targets
11121 as well. These two options select the same tuning decisions as
11122 @option{-m68020-40} and @option{-m68020-60} respectively.
11124 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11125 when tuning for 680x0 architecture @var{arch}. It also defines
11126 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11127 option is used. If gcc is tuning for a range of architectures,
11128 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11129 it defines the macros for every architecture in the range.
11131 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11132 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11133 of the arguments given above.
11139 Generate output for a 68000. This is the default
11140 when the compiler is configured for 68000-based systems.
11141 It is equivalent to @option{-march=68000}.
11143 Use this option for microcontrollers with a 68000 or EC000 core,
11144 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11148 Generate output for a 68010. This is the default
11149 when the compiler is configured for 68010-based systems.
11150 It is equivalent to @option{-march=68010}.
11156 Generate output for a 68020. This is the default
11157 when the compiler is configured for 68020-based systems.
11158 It is equivalent to @option{-march=68020}.
11162 Generate output for a 68030. This is the default when the compiler is
11163 configured for 68030-based systems. It is equivalent to
11164 @option{-march=68030}.
11168 Generate output for a 68040. This is the default when the compiler is
11169 configured for 68040-based systems. It is equivalent to
11170 @option{-march=68040}.
11172 This option inhibits the use of 68881/68882 instructions that have to be
11173 emulated by software on the 68040. Use this option if your 68040 does not
11174 have code to emulate those instructions.
11178 Generate output for a 68060. This is the default when the compiler is
11179 configured for 68060-based systems. It is equivalent to
11180 @option{-march=68060}.
11182 This option inhibits the use of 68020 and 68881/68882 instructions that
11183 have to be emulated by software on the 68060. Use this option if your 68060
11184 does not have code to emulate those instructions.
11188 Generate output for a CPU32. This is the default
11189 when the compiler is configured for CPU32-based systems.
11190 It is equivalent to @option{-march=cpu32}.
11192 Use this option for microcontrollers with a
11193 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11194 68336, 68340, 68341, 68349 and 68360.
11198 Generate output for a 520X ColdFire CPU. This is the default
11199 when the compiler is configured for 520X-based systems.
11200 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11201 in favor of that option.
11203 Use this option for microcontroller with a 5200 core, including
11204 the MCF5202, MCF5203, MCF5204 and MCF5206.
11208 Generate output for a 5206e ColdFire CPU. The option is now
11209 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11213 Generate output for a member of the ColdFire 528X family.
11214 The option is now deprecated in favor of the equivalent
11215 @option{-mcpu=528x}.
11219 Generate output for a ColdFire 5307 CPU. The option is now deprecated
11220 in favor of the equivalent @option{-mcpu=5307}.
11224 Generate output for a ColdFire 5407 CPU. The option is now deprecated
11225 in favor of the equivalent @option{-mcpu=5407}.
11229 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11230 This includes use of hardware floating point instructions.
11231 The option is equivalent to @option{-mcpu=547x}, and is now
11232 deprecated in favor of that option.
11236 Generate output for a 68040, without using any of the new instructions.
11237 This results in code which can run relatively efficiently on either a
11238 68020/68881 or a 68030 or a 68040. The generated code does use the
11239 68881 instructions that are emulated on the 68040.
11241 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11245 Generate output for a 68060, without using any of the new instructions.
11246 This results in code which can run relatively efficiently on either a
11247 68020/68881 or a 68030 or a 68040. The generated code does use the
11248 68881 instructions that are emulated on the 68060.
11250 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11254 @opindex mhard-float
11256 Generate floating-point instructions. This is the default for 68020
11257 and above, and for ColdFire devices that have an FPU. It defines the
11258 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11259 on ColdFire targets.
11262 @opindex msoft-float
11263 Do not generate floating-point instructions; use library calls instead.
11264 This is the default for 68000, 68010, and 68832 targets. It is also
11265 the default for ColdFire devices that have no FPU.
11271 Generate (do not generate) ColdFire hardware divide and remainder
11272 instructions. If @option{-march} is used without @option{-mcpu},
11273 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11274 architectures. Otherwise, the default is taken from the target CPU
11275 (either the default CPU, or the one specified by @option{-mcpu}). For
11276 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11277 @option{-mcpu=5206e}.
11279 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11283 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11284 Additionally, parameters passed on the stack are also aligned to a
11285 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11289 Do not consider type @code{int} to be 16 bits wide. This is the default.
11292 @itemx -mno-bitfield
11293 @opindex mnobitfield
11294 @opindex mno-bitfield
11295 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11296 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11300 Do use the bit-field instructions. The @option{-m68020} option implies
11301 @option{-mbitfield}. This is the default if you use a configuration
11302 designed for a 68020.
11306 Use a different function-calling convention, in which functions
11307 that take a fixed number of arguments return with the @code{rtd}
11308 instruction, which pops their arguments while returning. This
11309 saves one instruction in the caller since there is no need to pop
11310 the arguments there.
11312 This calling convention is incompatible with the one normally
11313 used on Unix, so you cannot use it if you need to call libraries
11314 compiled with the Unix compiler.
11316 Also, you must provide function prototypes for all functions that
11317 take variable numbers of arguments (including @code{printf});
11318 otherwise incorrect code will be generated for calls to those
11321 In addition, seriously incorrect code will result if you call a
11322 function with too many arguments. (Normally, extra arguments are
11323 harmlessly ignored.)
11325 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11326 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11330 Do not use the calling conventions selected by @option{-mrtd}.
11331 This is the default.
11334 @itemx -mno-align-int
11335 @opindex malign-int
11336 @opindex mno-align-int
11337 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11338 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11339 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11340 Aligning variables on 32-bit boundaries produces code that runs somewhat
11341 faster on processors with 32-bit busses at the expense of more memory.
11343 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11344 align structures containing the above types differently than
11345 most published application binary interface specifications for the m68k.
11349 Use the pc-relative addressing mode of the 68000 directly, instead of
11350 using a global offset table. At present, this option implies @option{-fpic},
11351 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11352 not presently supported with @option{-mpcrel}, though this could be supported for
11353 68020 and higher processors.
11355 @item -mno-strict-align
11356 @itemx -mstrict-align
11357 @opindex mno-strict-align
11358 @opindex mstrict-align
11359 Do not (do) assume that unaligned memory references will be handled by
11363 Generate code that allows the data segment to be located in a different
11364 area of memory from the text segment. This allows for execute in place in
11365 an environment without virtual memory management. This option implies
11368 @item -mno-sep-data
11369 Generate code that assumes that the data segment follows the text segment.
11370 This is the default.
11372 @item -mid-shared-library
11373 Generate code that supports shared libraries via the library ID method.
11374 This allows for execute in place and shared libraries in an environment
11375 without virtual memory management. This option implies @option{-fPIC}.
11377 @item -mno-id-shared-library
11378 Generate code that doesn't assume ID based shared libraries are being used.
11379 This is the default.
11381 @item -mshared-library-id=n
11382 Specified the identification number of the ID based shared library being
11383 compiled. Specifying a value of 0 will generate more compact code, specifying
11384 other values will force the allocation of that number to the current
11385 library but is no more space or time efficient than omitting this option.
11389 @node M68hc1x Options
11390 @subsection M68hc1x Options
11391 @cindex M68hc1x options
11393 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11394 microcontrollers. The default values for these options depends on
11395 which style of microcontroller was selected when the compiler was configured;
11396 the defaults for the most common choices are given below.
11403 Generate output for a 68HC11. This is the default
11404 when the compiler is configured for 68HC11-based systems.
11410 Generate output for a 68HC12. This is the default
11411 when the compiler is configured for 68HC12-based systems.
11417 Generate output for a 68HCS12.
11419 @item -mauto-incdec
11420 @opindex mauto-incdec
11421 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11428 Enable the use of 68HC12 min and max instructions.
11431 @itemx -mno-long-calls
11432 @opindex mlong-calls
11433 @opindex mno-long-calls
11434 Treat all calls as being far away (near). If calls are assumed to be
11435 far away, the compiler will use the @code{call} instruction to
11436 call a function and the @code{rtc} instruction for returning.
11440 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11442 @item -msoft-reg-count=@var{count}
11443 @opindex msoft-reg-count
11444 Specify the number of pseudo-soft registers which are used for the
11445 code generation. The maximum number is 32. Using more pseudo-soft
11446 register may or may not result in better code depending on the program.
11447 The default is 4 for 68HC11 and 2 for 68HC12.
11451 @node MCore Options
11452 @subsection MCore Options
11453 @cindex MCore options
11455 These are the @samp{-m} options defined for the Motorola M*Core
11461 @itemx -mno-hardlit
11463 @opindex mno-hardlit
11464 Inline constants into the code stream if it can be done in two
11465 instructions or less.
11471 Use the divide instruction. (Enabled by default).
11473 @item -mrelax-immediate
11474 @itemx -mno-relax-immediate
11475 @opindex mrelax-immediate
11476 @opindex mno-relax-immediate
11477 Allow arbitrary sized immediates in bit operations.
11479 @item -mwide-bitfields
11480 @itemx -mno-wide-bitfields
11481 @opindex mwide-bitfields
11482 @opindex mno-wide-bitfields
11483 Always treat bit-fields as int-sized.
11485 @item -m4byte-functions
11486 @itemx -mno-4byte-functions
11487 @opindex m4byte-functions
11488 @opindex mno-4byte-functions
11489 Force all functions to be aligned to a four byte boundary.
11491 @item -mcallgraph-data
11492 @itemx -mno-callgraph-data
11493 @opindex mcallgraph-data
11494 @opindex mno-callgraph-data
11495 Emit callgraph information.
11498 @itemx -mno-slow-bytes
11499 @opindex mslow-bytes
11500 @opindex mno-slow-bytes
11501 Prefer word access when reading byte quantities.
11503 @item -mlittle-endian
11504 @itemx -mbig-endian
11505 @opindex mlittle-endian
11506 @opindex mbig-endian
11507 Generate code for a little endian target.
11513 Generate code for the 210 processor.
11517 @subsection MIPS Options
11518 @cindex MIPS options
11524 Generate big-endian code.
11528 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11531 @item -march=@var{arch}
11533 Generate code that will run on @var{arch}, which can be the name of a
11534 generic MIPS ISA, or the name of a particular processor.
11536 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11537 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11538 The processor names are:
11539 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11540 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11541 @samp{5kc}, @samp{5kf},
11543 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11544 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11545 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11546 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11549 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11550 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11551 @samp{rm7000}, @samp{rm9000},
11554 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11555 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
11556 The special value @samp{from-abi} selects the
11557 most compatible architecture for the selected ABI (that is,
11558 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11560 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11561 (for example, @samp{-march=r2k}). Prefixes are optional, and
11562 @samp{vr} may be written @samp{r}.
11564 Names of the form @samp{@var{n}f2_1} refer to processors with
11565 FPUs clocked at half the rate of the core, names of the form
11566 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
11567 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
11568 processors with FPUs clocked a ratio of 3:2 with respect to the core.
11569 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
11570 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
11571 accepted as synonyms for @samp{@var{n}f1_1}.
11573 GCC defines two macros based on the value of this option. The first
11574 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
11575 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
11576 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
11577 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
11578 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
11580 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
11581 above. In other words, it will have the full prefix and will not
11582 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
11583 the macro names the resolved architecture (either @samp{"mips1"} or
11584 @samp{"mips3"}). It names the default architecture when no
11585 @option{-march} option is given.
11587 @item -mtune=@var{arch}
11589 Optimize for @var{arch}. Among other things, this option controls
11590 the way instructions are scheduled, and the perceived cost of arithmetic
11591 operations. The list of @var{arch} values is the same as for
11594 When this option is not used, GCC will optimize for the processor
11595 specified by @option{-march}. By using @option{-march} and
11596 @option{-mtune} together, it is possible to generate code that will
11597 run on a family of processors, but optimize the code for one
11598 particular member of that family.
11600 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
11601 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
11602 @samp{-march} ones described above.
11606 Equivalent to @samp{-march=mips1}.
11610 Equivalent to @samp{-march=mips2}.
11614 Equivalent to @samp{-march=mips3}.
11618 Equivalent to @samp{-march=mips4}.
11622 Equivalent to @samp{-march=mips32}.
11626 Equivalent to @samp{-march=mips32r2}.
11630 Equivalent to @samp{-march=mips64}.
11635 @opindex mno-mips16
11636 Generate (do not generate) MIPS16 code. If GCC is targetting a
11637 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
11639 MIPS16 code generation can also be controlled on a per-function basis
11640 by means of @code{mips16} and @code{nomips16} attributes.
11641 @xref{Function Attributes}, for more information.
11643 @item -mflip-mips16
11644 @opindex mflip-mips16
11645 Generate MIPS16 code on alternating functions. This option is provided
11646 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
11647 not intended for ordinary use in compiling user code.
11659 Generate code for the given ABI@.
11661 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
11662 generates 64-bit code when you select a 64-bit architecture, but you
11663 can use @option{-mgp32} to get 32-bit code instead.
11665 For information about the O64 ABI, see
11666 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
11668 GCC supports a variant of the o32 ABI in which floating-point registers
11669 are 64 rather than 32 bits wide. You can select this combination with
11670 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
11671 and @samp{mfhc1} instructions and is therefore only supported for
11672 MIPS32R2 processors.
11674 The register assignments for arguments and return values remain the
11675 same, but each scalar value is passed in a single 64-bit register
11676 rather than a pair of 32-bit registers. For example, scalar
11677 floating-point values are returned in @samp{$f0} only, not a
11678 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
11679 remains the same, but all 64 bits are saved.
11682 @itemx -mno-abicalls
11684 @opindex mno-abicalls
11685 Generate (do not generate) code that is suitable for SVR4-style
11686 dynamic objects. @option{-mabicalls} is the default for SVR4-based
11691 Generate (do not generate) code that is fully position-independent,
11692 and that can therefore be linked into shared libraries. This option
11693 only affects @option{-mabicalls}.
11695 All @option{-mabicalls} code has traditionally been position-independent,
11696 regardless of options like @option{-fPIC} and @option{-fpic}. However,
11697 as an extension, the GNU toolchain allows executables to use absolute
11698 accesses for locally-binding symbols. It can also use shorter GP
11699 initialization sequences and generate direct calls to locally-defined
11700 functions. This mode is selected by @option{-mno-shared}.
11702 @option{-mno-shared} depends on binutils 2.16 or higher and generates
11703 objects that can only be linked by the GNU linker. However, the option
11704 does not affect the ABI of the final executable; it only affects the ABI
11705 of relocatable objects. Using @option{-mno-shared} will generally make
11706 executables both smaller and quicker.
11708 @option{-mshared} is the default.
11714 Lift (do not lift) the usual restrictions on the size of the global
11717 GCC normally uses a single instruction to load values from the GOT@.
11718 While this is relatively efficient, it will only work if the GOT
11719 is smaller than about 64k. Anything larger will cause the linker
11720 to report an error such as:
11722 @cindex relocation truncated to fit (MIPS)
11724 relocation truncated to fit: R_MIPS_GOT16 foobar
11727 If this happens, you should recompile your code with @option{-mxgot}.
11728 It should then work with very large GOTs, although it will also be
11729 less efficient, since it will take three instructions to fetch the
11730 value of a global symbol.
11732 Note that some linkers can create multiple GOTs. If you have such a
11733 linker, you should only need to use @option{-mxgot} when a single object
11734 file accesses more than 64k's worth of GOT entries. Very few do.
11736 These options have no effect unless GCC is generating position
11741 Assume that general-purpose registers are 32 bits wide.
11745 Assume that general-purpose registers are 64 bits wide.
11749 Assume that floating-point registers are 32 bits wide.
11753 Assume that floating-point registers are 64 bits wide.
11756 @opindex mhard-float
11757 Use floating-point coprocessor instructions.
11760 @opindex msoft-float
11761 Do not use floating-point coprocessor instructions. Implement
11762 floating-point calculations using library calls instead.
11764 @item -msingle-float
11765 @opindex msingle-float
11766 Assume that the floating-point coprocessor only supports single-precision
11769 @item -mdouble-float
11770 @opindex mdouble-float
11771 Assume that the floating-point coprocessor supports double-precision
11772 operations. This is the default.
11778 Use (do not use) revision 1 of the MIPS DSP ASE.
11779 @xref{MIPS DSP Built-in Functions}. This option defines the
11780 preprocessor macro @samp{__mips_dsp}. It also defines
11781 @samp{__mips_dsp_rev} to 1.
11787 Use (do not use) revision 2 of the MIPS DSP ASE.
11788 @xref{MIPS DSP Built-in Functions}. This option defines the
11789 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
11790 It also defines @samp{__mips_dsp_rev} to 2.
11793 @itemx -mno-smartmips
11794 @opindex msmartmips
11795 @opindex mno-smartmips
11796 Use (do not use) the MIPS SmartMIPS ASE.
11798 @item -mpaired-single
11799 @itemx -mno-paired-single
11800 @opindex mpaired-single
11801 @opindex mno-paired-single
11802 Use (do not use) paired-single floating-point instructions.
11803 @xref{MIPS Paired-Single Support}. This option can only be used
11804 when generating 64-bit code and requires hardware floating-point
11805 support to be enabled.
11811 Use (do not use) MIPS Digital Media Extension instructions.
11812 This option can only be used when generating 64-bit code and requires
11813 hardware floating-point support to be enabled.
11818 @opindex mno-mips3d
11819 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
11820 The option @option{-mips3d} implies @option{-mpaired-single}.
11826 Use (do not use) MT Multithreading instructions.
11830 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
11831 an explanation of the default and the way that the pointer size is
11836 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
11838 The default size of @code{int}s, @code{long}s and pointers depends on
11839 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
11840 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
11841 32-bit @code{long}s. Pointers are the same size as @code{long}s,
11842 or the same size as integer registers, whichever is smaller.
11848 Assume (do not assume) that all symbols have 32-bit values, regardless
11849 of the selected ABI@. This option is useful in combination with
11850 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
11851 to generate shorter and faster references to symbolic addresses.
11855 Put definitions of externally-visible data in a small data section
11856 if that data is no bigger than @var{num} bytes. GCC can then access
11857 the data more efficiently; see @option{-mgpopt} for details.
11859 The default @option{-G} option depends on the configuration.
11861 @item -mlocal-sdata
11862 @itemx -mno-local-sdata
11863 @opindex mlocal-sdata
11864 @opindex mno-local-sdata
11865 Extend (do not extend) the @option{-G} behavior to local data too,
11866 such as to static variables in C. @option{-mlocal-sdata} is the
11867 default for all configurations.
11869 If the linker complains that an application is using too much small data,
11870 you might want to try rebuilding the less performance-critical parts with
11871 @option{-mno-local-sdata}. You might also want to build large
11872 libraries with @option{-mno-local-sdata}, so that the libraries leave
11873 more room for the main program.
11875 @item -mextern-sdata
11876 @itemx -mno-extern-sdata
11877 @opindex mextern-sdata
11878 @opindex mno-extern-sdata
11879 Assume (do not assume) that externally-defined data will be in
11880 a small data section if that data is within the @option{-G} limit.
11881 @option{-mextern-sdata} is the default for all configurations.
11883 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
11884 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
11885 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
11886 is placed in a small data section. If @var{Var} is defined by another
11887 module, you must either compile that module with a high-enough
11888 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
11889 definition. If @var{Var} is common, you must link the application
11890 with a high-enough @option{-G} setting.
11892 The easiest way of satisfying these restrictions is to compile
11893 and link every module with the same @option{-G} option. However,
11894 you may wish to build a library that supports several different
11895 small data limits. You can do this by compiling the library with
11896 the highest supported @option{-G} setting and additionally using
11897 @option{-mno-extern-sdata} to stop the library from making assumptions
11898 about externally-defined data.
11904 Use (do not use) GP-relative accesses for symbols that are known to be
11905 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
11906 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
11909 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
11910 might not hold the value of @code{_gp}. For example, if the code is
11911 part of a library that might be used in a boot monitor, programs that
11912 call boot monitor routines will pass an unknown value in @code{$gp}.
11913 (In such situations, the boot monitor itself would usually be compiled
11914 with @option{-G0}.)
11916 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
11917 @option{-mno-extern-sdata}.
11919 @item -membedded-data
11920 @itemx -mno-embedded-data
11921 @opindex membedded-data
11922 @opindex mno-embedded-data
11923 Allocate variables to the read-only data section first if possible, then
11924 next in the small data section if possible, otherwise in data. This gives
11925 slightly slower code than the default, but reduces the amount of RAM required
11926 when executing, and thus may be preferred for some embedded systems.
11928 @item -muninit-const-in-rodata
11929 @itemx -mno-uninit-const-in-rodata
11930 @opindex muninit-const-in-rodata
11931 @opindex mno-uninit-const-in-rodata
11932 Put uninitialized @code{const} variables in the read-only data section.
11933 This option is only meaningful in conjunction with @option{-membedded-data}.
11935 @item -mcode-readable=@var{setting}
11936 @opindex mcode-readable
11937 Specify whether GCC may generate code that reads from executable sections.
11938 There are three possible settings:
11941 @item -mcode-readable=yes
11942 Instructions may freely access executable sections. This is the
11945 @item -mcode-readable=pcrel
11946 MIPS16 PC-relative load instructions can access executable sections,
11947 but other instructions must not do so. This option is useful on 4KSc
11948 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
11949 It is also useful on processors that can be configured to have a dual
11950 instruction/data SRAM interface and that, like the M4K, automatically
11951 redirect PC-relative loads to the instruction RAM.
11953 @item -mcode-readable=no
11954 Instructions must not access executable sections. This option can be
11955 useful on targets that are configured to have a dual instruction/data
11956 SRAM interface but that (unlike the M4K) do not automatically redirect
11957 PC-relative loads to the instruction RAM.
11960 @item -msplit-addresses
11961 @itemx -mno-split-addresses
11962 @opindex msplit-addresses
11963 @opindex mno-split-addresses
11964 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
11965 relocation operators. This option has been superseded by
11966 @option{-mexplicit-relocs} but is retained for backwards compatibility.
11968 @item -mexplicit-relocs
11969 @itemx -mno-explicit-relocs
11970 @opindex mexplicit-relocs
11971 @opindex mno-explicit-relocs
11972 Use (do not use) assembler relocation operators when dealing with symbolic
11973 addresses. The alternative, selected by @option{-mno-explicit-relocs},
11974 is to use assembler macros instead.
11976 @option{-mexplicit-relocs} is the default if GCC was configured
11977 to use an assembler that supports relocation operators.
11979 @item -mcheck-zero-division
11980 @itemx -mno-check-zero-division
11981 @opindex mcheck-zero-division
11982 @opindex mno-check-zero-division
11983 Trap (do not trap) on integer division by zero.
11985 The default is @option{-mcheck-zero-division}.
11987 @item -mdivide-traps
11988 @itemx -mdivide-breaks
11989 @opindex mdivide-traps
11990 @opindex mdivide-breaks
11991 MIPS systems check for division by zero by generating either a
11992 conditional trap or a break instruction. Using traps results in
11993 smaller code, but is only supported on MIPS II and later. Also, some
11994 versions of the Linux kernel have a bug that prevents trap from
11995 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
11996 allow conditional traps on architectures that support them and
11997 @option{-mdivide-breaks} to force the use of breaks.
11999 The default is usually @option{-mdivide-traps}, but this can be
12000 overridden at configure time using @option{--with-divide=breaks}.
12001 Divide-by-zero checks can be completely disabled using
12002 @option{-mno-check-zero-division}.
12007 @opindex mno-memcpy
12008 Force (do not force) the use of @code{memcpy()} for non-trivial block
12009 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
12010 most constant-sized copies.
12013 @itemx -mno-long-calls
12014 @opindex mlong-calls
12015 @opindex mno-long-calls
12016 Disable (do not disable) use of the @code{jal} instruction. Calling
12017 functions using @code{jal} is more efficient but requires the caller
12018 and callee to be in the same 256 megabyte segment.
12020 This option has no effect on abicalls code. The default is
12021 @option{-mno-long-calls}.
12027 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
12028 instructions, as provided by the R4650 ISA@.
12031 @itemx -mno-fused-madd
12032 @opindex mfused-madd
12033 @opindex mno-fused-madd
12034 Enable (disable) use of the floating point multiply-accumulate
12035 instructions, when they are available. The default is
12036 @option{-mfused-madd}.
12038 When multiply-accumulate instructions are used, the intermediate
12039 product is calculated to infinite precision and is not subject to
12040 the FCSR Flush to Zero bit. This may be undesirable in some
12045 Tell the MIPS assembler to not run its preprocessor over user
12046 assembler files (with a @samp{.s} suffix) when assembling them.
12049 @itemx -mno-fix-r4000
12050 @opindex mfix-r4000
12051 @opindex mno-fix-r4000
12052 Work around certain R4000 CPU errata:
12055 A double-word or a variable shift may give an incorrect result if executed
12056 immediately after starting an integer division.
12058 A double-word or a variable shift may give an incorrect result if executed
12059 while an integer multiplication is in progress.
12061 An integer division may give an incorrect result if started in a delay slot
12062 of a taken branch or a jump.
12066 @itemx -mno-fix-r4400
12067 @opindex mfix-r4400
12068 @opindex mno-fix-r4400
12069 Work around certain R4400 CPU errata:
12072 A double-word or a variable shift may give an incorrect result if executed
12073 immediately after starting an integer division.
12077 @itemx -mno-fix-vr4120
12078 @opindex mfix-vr4120
12079 Work around certain VR4120 errata:
12082 @code{dmultu} does not always produce the correct result.
12084 @code{div} and @code{ddiv} do not always produce the correct result if one
12085 of the operands is negative.
12087 The workarounds for the division errata rely on special functions in
12088 @file{libgcc.a}. At present, these functions are only provided by
12089 the @code{mips64vr*-elf} configurations.
12091 Other VR4120 errata require a nop to be inserted between certain pairs of
12092 instructions. These errata are handled by the assembler, not by GCC itself.
12095 @opindex mfix-vr4130
12096 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
12097 workarounds are implemented by the assembler rather than by GCC,
12098 although GCC will avoid using @code{mflo} and @code{mfhi} if the
12099 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
12100 instructions are available instead.
12103 @itemx -mno-fix-sb1
12105 Work around certain SB-1 CPU core errata.
12106 (This flag currently works around the SB-1 revision 2
12107 ``F1'' and ``F2'' floating point errata.)
12109 @item -mflush-func=@var{func}
12110 @itemx -mno-flush-func
12111 @opindex mflush-func
12112 Specifies the function to call to flush the I and D caches, or to not
12113 call any such function. If called, the function must take the same
12114 arguments as the common @code{_flush_func()}, that is, the address of the
12115 memory range for which the cache is being flushed, the size of the
12116 memory range, and the number 3 (to flush both caches). The default
12117 depends on the target GCC was configured for, but commonly is either
12118 @samp{_flush_func} or @samp{__cpu_flush}.
12120 @item mbranch-cost=@var{num}
12121 @opindex mbranch-cost
12122 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12123 This cost is only a heuristic and is not guaranteed to produce
12124 consistent results across releases. A zero cost redundantly selects
12125 the default, which is based on the @option{-mtune} setting.
12127 @item -mbranch-likely
12128 @itemx -mno-branch-likely
12129 @opindex mbranch-likely
12130 @opindex mno-branch-likely
12131 Enable or disable use of Branch Likely instructions, regardless of the
12132 default for the selected architecture. By default, Branch Likely
12133 instructions may be generated if they are supported by the selected
12134 architecture. An exception is for the MIPS32 and MIPS64 architectures
12135 and processors which implement those architectures; for those, Branch
12136 Likely instructions will not be generated by default because the MIPS32
12137 and MIPS64 architectures specifically deprecate their use.
12139 @item -mfp-exceptions
12140 @itemx -mno-fp-exceptions
12141 @opindex mfp-exceptions
12142 Specifies whether FP exceptions are enabled. This affects how we schedule
12143 FP instructions for some processors. The default is that FP exceptions are
12146 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12147 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12150 @item -mvr4130-align
12151 @itemx -mno-vr4130-align
12152 @opindex mvr4130-align
12153 The VR4130 pipeline is two-way superscalar, but can only issue two
12154 instructions together if the first one is 8-byte aligned. When this
12155 option is enabled, GCC will align pairs of instructions that it
12156 thinks should execute in parallel.
12158 This option only has an effect when optimizing for the VR4130.
12159 It normally makes code faster, but at the expense of making it bigger.
12160 It is enabled by default at optimization level @option{-O3}.
12164 @subsection MMIX Options
12165 @cindex MMIX Options
12167 These options are defined for the MMIX:
12171 @itemx -mno-libfuncs
12173 @opindex mno-libfuncs
12174 Specify that intrinsic library functions are being compiled, passing all
12175 values in registers, no matter the size.
12178 @itemx -mno-epsilon
12180 @opindex mno-epsilon
12181 Generate floating-point comparison instructions that compare with respect
12182 to the @code{rE} epsilon register.
12184 @item -mabi=mmixware
12186 @opindex mabi-mmixware
12188 Generate code that passes function parameters and return values that (in
12189 the called function) are seen as registers @code{$0} and up, as opposed to
12190 the GNU ABI which uses global registers @code{$231} and up.
12192 @item -mzero-extend
12193 @itemx -mno-zero-extend
12194 @opindex mzero-extend
12195 @opindex mno-zero-extend
12196 When reading data from memory in sizes shorter than 64 bits, use (do not
12197 use) zero-extending load instructions by default, rather than
12198 sign-extending ones.
12201 @itemx -mno-knuthdiv
12203 @opindex mno-knuthdiv
12204 Make the result of a division yielding a remainder have the same sign as
12205 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12206 remainder follows the sign of the dividend. Both methods are
12207 arithmetically valid, the latter being almost exclusively used.
12209 @item -mtoplevel-symbols
12210 @itemx -mno-toplevel-symbols
12211 @opindex mtoplevel-symbols
12212 @opindex mno-toplevel-symbols
12213 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12214 code can be used with the @code{PREFIX} assembly directive.
12218 Generate an executable in the ELF format, rather than the default
12219 @samp{mmo} format used by the @command{mmix} simulator.
12221 @item -mbranch-predict
12222 @itemx -mno-branch-predict
12223 @opindex mbranch-predict
12224 @opindex mno-branch-predict
12225 Use (do not use) the probable-branch instructions, when static branch
12226 prediction indicates a probable branch.
12228 @item -mbase-addresses
12229 @itemx -mno-base-addresses
12230 @opindex mbase-addresses
12231 @opindex mno-base-addresses
12232 Generate (do not generate) code that uses @emph{base addresses}. Using a
12233 base address automatically generates a request (handled by the assembler
12234 and the linker) for a constant to be set up in a global register. The
12235 register is used for one or more base address requests within the range 0
12236 to 255 from the value held in the register. The generally leads to short
12237 and fast code, but the number of different data items that can be
12238 addressed is limited. This means that a program that uses lots of static
12239 data may require @option{-mno-base-addresses}.
12241 @item -msingle-exit
12242 @itemx -mno-single-exit
12243 @opindex msingle-exit
12244 @opindex mno-single-exit
12245 Force (do not force) generated code to have a single exit point in each
12249 @node MN10300 Options
12250 @subsection MN10300 Options
12251 @cindex MN10300 options
12253 These @option{-m} options are defined for Matsushita MN10300 architectures:
12258 Generate code to avoid bugs in the multiply instructions for the MN10300
12259 processors. This is the default.
12261 @item -mno-mult-bug
12262 @opindex mno-mult-bug
12263 Do not generate code to avoid bugs in the multiply instructions for the
12264 MN10300 processors.
12268 Generate code which uses features specific to the AM33 processor.
12272 Do not generate code which uses features specific to the AM33 processor. This
12275 @item -mreturn-pointer-on-d0
12276 @opindex mreturn-pointer-on-d0
12277 When generating a function which returns a pointer, return the pointer
12278 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12279 only in a0, and attempts to call such functions without a prototype
12280 would result in errors. Note that this option is on by default; use
12281 @option{-mno-return-pointer-on-d0} to disable it.
12285 Do not link in the C run-time initialization object file.
12289 Indicate to the linker that it should perform a relaxation optimization pass
12290 to shorten branches, calls and absolute memory addresses. This option only
12291 has an effect when used on the command line for the final link step.
12293 This option makes symbolic debugging impossible.
12297 @subsection MT Options
12300 These @option{-m} options are defined for Morpho MT architectures:
12304 @item -march=@var{cpu-type}
12306 Generate code that will run on @var{cpu-type}, which is the name of a system
12307 representing a certain processor type. Possible values for
12308 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
12309 @samp{ms1-16-003} and @samp{ms2}.
12311 When this option is not used, the default is @option{-march=ms1-16-002}.
12315 Use byte loads and stores when generating code.
12319 Do not use byte loads and stores when generating code.
12323 Use simulator runtime
12327 Do not link in the C run-time initialization object file
12328 @file{crti.o}. Other run-time initialization and termination files
12329 such as @file{startup.o} and @file{exit.o} are still included on the
12330 linker command line.
12334 @node PDP-11 Options
12335 @subsection PDP-11 Options
12336 @cindex PDP-11 Options
12338 These options are defined for the PDP-11:
12343 Use hardware FPP floating point. This is the default. (FIS floating
12344 point on the PDP-11/40 is not supported.)
12347 @opindex msoft-float
12348 Do not use hardware floating point.
12352 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12356 Return floating-point results in memory. This is the default.
12360 Generate code for a PDP-11/40.
12364 Generate code for a PDP-11/45. This is the default.
12368 Generate code for a PDP-11/10.
12370 @item -mbcopy-builtin
12371 @opindex bcopy-builtin
12372 Use inline @code{movmemhi} patterns for copying memory. This is the
12377 Do not use inline @code{movmemhi} patterns for copying memory.
12383 Use 16-bit @code{int}. This is the default.
12389 Use 32-bit @code{int}.
12392 @itemx -mno-float32
12394 @opindex mno-float32
12395 Use 64-bit @code{float}. This is the default.
12398 @itemx -mno-float64
12400 @opindex mno-float64
12401 Use 32-bit @code{float}.
12405 Use @code{abshi2} pattern. This is the default.
12409 Do not use @code{abshi2} pattern.
12411 @item -mbranch-expensive
12412 @opindex mbranch-expensive
12413 Pretend that branches are expensive. This is for experimenting with
12414 code generation only.
12416 @item -mbranch-cheap
12417 @opindex mbranch-cheap
12418 Do not pretend that branches are expensive. This is the default.
12422 Generate code for a system with split I&D@.
12426 Generate code for a system without split I&D@. This is the default.
12430 Use Unix assembler syntax. This is the default when configured for
12431 @samp{pdp11-*-bsd}.
12435 Use DEC assembler syntax. This is the default when configured for any
12436 PDP-11 target other than @samp{pdp11-*-bsd}.
12439 @node PowerPC Options
12440 @subsection PowerPC Options
12441 @cindex PowerPC options
12443 These are listed under @xref{RS/6000 and PowerPC Options}.
12445 @node RS/6000 and PowerPC Options
12446 @subsection IBM RS/6000 and PowerPC Options
12447 @cindex RS/6000 and PowerPC Options
12448 @cindex IBM RS/6000 and PowerPC Options
12450 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12457 @itemx -mno-powerpc
12458 @itemx -mpowerpc-gpopt
12459 @itemx -mno-powerpc-gpopt
12460 @itemx -mpowerpc-gfxopt
12461 @itemx -mno-powerpc-gfxopt
12463 @itemx -mno-powerpc64
12467 @itemx -mno-popcntb
12479 @opindex mno-power2
12481 @opindex mno-powerpc
12482 @opindex mpowerpc-gpopt
12483 @opindex mno-powerpc-gpopt
12484 @opindex mpowerpc-gfxopt
12485 @opindex mno-powerpc-gfxopt
12486 @opindex mpowerpc64
12487 @opindex mno-powerpc64
12491 @opindex mno-popcntb
12497 @opindex mno-mfpgpr
12500 GCC supports two related instruction set architectures for the
12501 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12502 instructions supported by the @samp{rios} chip set used in the original
12503 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12504 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12505 the IBM 4xx, 6xx, and follow-on microprocessors.
12507 Neither architecture is a subset of the other. However there is a
12508 large common subset of instructions supported by both. An MQ
12509 register is included in processors supporting the POWER architecture.
12511 You use these options to specify which instructions are available on the
12512 processor you are using. The default value of these options is
12513 determined when configuring GCC@. Specifying the
12514 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12515 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12516 rather than the options listed above.
12518 The @option{-mpower} option allows GCC to generate instructions that
12519 are found only in the POWER architecture and to use the MQ register.
12520 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12521 to generate instructions that are present in the POWER2 architecture but
12522 not the original POWER architecture.
12524 The @option{-mpowerpc} option allows GCC to generate instructions that
12525 are found only in the 32-bit subset of the PowerPC architecture.
12526 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12527 GCC to use the optional PowerPC architecture instructions in the
12528 General Purpose group, including floating-point square root. Specifying
12529 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12530 use the optional PowerPC architecture instructions in the Graphics
12531 group, including floating-point select.
12533 The @option{-mmfcrf} option allows GCC to generate the move from
12534 condition register field instruction implemented on the POWER4
12535 processor and other processors that support the PowerPC V2.01
12537 The @option{-mpopcntb} option allows GCC to generate the popcount and
12538 double precision FP reciprocal estimate instruction implemented on the
12539 POWER5 processor and other processors that support the PowerPC V2.02
12541 The @option{-mfprnd} option allows GCC to generate the FP round to
12542 integer instructions implemented on the POWER5+ processor and other
12543 processors that support the PowerPC V2.03 architecture.
12544 The @option{-mcmpb} option allows GCC to generate the compare bytes
12545 instruction implemented on the POWER6 processor and other processors
12546 that support the PowerPC V2.05 architecture.
12547 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12548 general purpose register instructions implemented on the POWER6X
12549 processor and other processors that support the extended PowerPC V2.05
12551 The @option{-mdfp} option allows GCC to generate the decimal floating
12552 point instructions implemented on some POWER processors.
12554 The @option{-mpowerpc64} option allows GCC to generate the additional
12555 64-bit instructions that are found in the full PowerPC64 architecture
12556 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12557 @option{-mno-powerpc64}.
12559 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12560 will use only the instructions in the common subset of both
12561 architectures plus some special AIX common-mode calls, and will not use
12562 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12563 permits GCC to use any instruction from either architecture and to
12564 allow use of the MQ register; specify this for the Motorola MPC601.
12566 @item -mnew-mnemonics
12567 @itemx -mold-mnemonics
12568 @opindex mnew-mnemonics
12569 @opindex mold-mnemonics
12570 Select which mnemonics to use in the generated assembler code. With
12571 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12572 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12573 assembler mnemonics defined for the POWER architecture. Instructions
12574 defined in only one architecture have only one mnemonic; GCC uses that
12575 mnemonic irrespective of which of these options is specified.
12577 GCC defaults to the mnemonics appropriate for the architecture in
12578 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
12579 value of these option. Unless you are building a cross-compiler, you
12580 should normally not specify either @option{-mnew-mnemonics} or
12581 @option{-mold-mnemonics}, but should instead accept the default.
12583 @item -mcpu=@var{cpu_type}
12585 Set architecture type, register usage, choice of mnemonics, and
12586 instruction scheduling parameters for machine type @var{cpu_type}.
12587 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
12588 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
12589 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
12590 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
12591 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
12592 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
12593 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
12594 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
12595 @samp{power6x}, @samp{common}, @samp{powerpc}, @samp{powerpc64},
12596 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
12598 @option{-mcpu=common} selects a completely generic processor. Code
12599 generated under this option will run on any POWER or PowerPC processor.
12600 GCC will use only the instructions in the common subset of both
12601 architectures, and will not use the MQ register. GCC assumes a generic
12602 processor model for scheduling purposes.
12604 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
12605 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
12606 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
12607 types, with an appropriate, generic processor model assumed for
12608 scheduling purposes.
12610 The other options specify a specific processor. Code generated under
12611 those options will run best on that processor, and may not run at all on
12614 The @option{-mcpu} options automatically enable or disable the
12617 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
12618 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
12619 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
12621 The particular options set for any particular CPU will vary between
12622 compiler versions, depending on what setting seems to produce optimal
12623 code for that CPU; it doesn't necessarily reflect the actual hardware's
12624 capabilities. If you wish to set an individual option to a particular
12625 value, you may specify it after the @option{-mcpu} option, like
12626 @samp{-mcpu=970 -mno-altivec}.
12628 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
12629 not enabled or disabled by the @option{-mcpu} option at present because
12630 AIX does not have full support for these options. You may still
12631 enable or disable them individually if you're sure it'll work in your
12634 @item -mtune=@var{cpu_type}
12636 Set the instruction scheduling parameters for machine type
12637 @var{cpu_type}, but do not set the architecture type, register usage, or
12638 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
12639 values for @var{cpu_type} are used for @option{-mtune} as for
12640 @option{-mcpu}. If both are specified, the code generated will use the
12641 architecture, registers, and mnemonics set by @option{-mcpu}, but the
12642 scheduling parameters set by @option{-mtune}.
12648 Generate code to compute division as reciprocal estimate and iterative
12649 refinement, creating opportunities for increased throughput. This
12650 feature requires: optional PowerPC Graphics instruction set for single
12651 precision and FRE instruction for double precision, assuming divides
12652 cannot generate user-visible traps, and the domain values not include
12653 Infinities, denormals or zero denominator.
12656 @itemx -mno-altivec
12658 @opindex mno-altivec
12659 Generate code that uses (does not use) AltiVec instructions, and also
12660 enable the use of built-in functions that allow more direct access to
12661 the AltiVec instruction set. You may also need to set
12662 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
12668 @opindex mno-vrsave
12669 Generate VRSAVE instructions when generating AltiVec code.
12672 @opindex msecure-plt
12673 Generate code that allows ld and ld.so to build executables and shared
12674 libraries with non-exec .plt and .got sections. This is a PowerPC
12675 32-bit SYSV ABI option.
12679 Generate code that uses a BSS .plt section that ld.so fills in, and
12680 requires .plt and .got sections that are both writable and executable.
12681 This is a PowerPC 32-bit SYSV ABI option.
12687 This switch enables or disables the generation of ISEL instructions.
12689 @item -misel=@var{yes/no}
12690 This switch has been deprecated. Use @option{-misel} and
12691 @option{-mno-isel} instead.
12697 This switch enables or disables the generation of SPE simd
12703 @opindex mno-paired
12704 This switch enables or disables the generation of PAIRED simd
12707 @item -mspe=@var{yes/no}
12708 This option has been deprecated. Use @option{-mspe} and
12709 @option{-mno-spe} instead.
12711 @item -mfloat-gprs=@var{yes/single/double/no}
12712 @itemx -mfloat-gprs
12713 @opindex mfloat-gprs
12714 This switch enables or disables the generation of floating point
12715 operations on the general purpose registers for architectures that
12718 The argument @var{yes} or @var{single} enables the use of
12719 single-precision floating point operations.
12721 The argument @var{double} enables the use of single and
12722 double-precision floating point operations.
12724 The argument @var{no} disables floating point operations on the
12725 general purpose registers.
12727 This option is currently only available on the MPC854x.
12733 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
12734 targets (including GNU/Linux). The 32-bit environment sets int, long
12735 and pointer to 32 bits and generates code that runs on any PowerPC
12736 variant. The 64-bit environment sets int to 32 bits and long and
12737 pointer to 64 bits, and generates code for PowerPC64, as for
12738 @option{-mpowerpc64}.
12741 @itemx -mno-fp-in-toc
12742 @itemx -mno-sum-in-toc
12743 @itemx -mminimal-toc
12745 @opindex mno-fp-in-toc
12746 @opindex mno-sum-in-toc
12747 @opindex mminimal-toc
12748 Modify generation of the TOC (Table Of Contents), which is created for
12749 every executable file. The @option{-mfull-toc} option is selected by
12750 default. In that case, GCC will allocate at least one TOC entry for
12751 each unique non-automatic variable reference in your program. GCC
12752 will also place floating-point constants in the TOC@. However, only
12753 16,384 entries are available in the TOC@.
12755 If you receive a linker error message that saying you have overflowed
12756 the available TOC space, you can reduce the amount of TOC space used
12757 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
12758 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
12759 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
12760 generate code to calculate the sum of an address and a constant at
12761 run-time instead of putting that sum into the TOC@. You may specify one
12762 or both of these options. Each causes GCC to produce very slightly
12763 slower and larger code at the expense of conserving TOC space.
12765 If you still run out of space in the TOC even when you specify both of
12766 these options, specify @option{-mminimal-toc} instead. This option causes
12767 GCC to make only one TOC entry for every file. When you specify this
12768 option, GCC will produce code that is slower and larger but which
12769 uses extremely little TOC space. You may wish to use this option
12770 only on files that contain less frequently executed code.
12776 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
12777 @code{long} type, and the infrastructure needed to support them.
12778 Specifying @option{-maix64} implies @option{-mpowerpc64} and
12779 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
12780 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
12783 @itemx -mno-xl-compat
12784 @opindex mxl-compat
12785 @opindex mno-xl-compat
12786 Produce code that conforms more closely to IBM XL compiler semantics
12787 when using AIX-compatible ABI. Pass floating-point arguments to
12788 prototyped functions beyond the register save area (RSA) on the stack
12789 in addition to argument FPRs. Do not assume that most significant
12790 double in 128-bit long double value is properly rounded when comparing
12791 values and converting to double. Use XL symbol names for long double
12794 The AIX calling convention was extended but not initially documented to
12795 handle an obscure K&R C case of calling a function that takes the
12796 address of its arguments with fewer arguments than declared. IBM XL
12797 compilers access floating point arguments which do not fit in the
12798 RSA from the stack when a subroutine is compiled without
12799 optimization. Because always storing floating-point arguments on the
12800 stack is inefficient and rarely needed, this option is not enabled by
12801 default and only is necessary when calling subroutines compiled by IBM
12802 XL compilers without optimization.
12806 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
12807 application written to use message passing with special startup code to
12808 enable the application to run. The system must have PE installed in the
12809 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
12810 must be overridden with the @option{-specs=} option to specify the
12811 appropriate directory location. The Parallel Environment does not
12812 support threads, so the @option{-mpe} option and the @option{-pthread}
12813 option are incompatible.
12815 @item -malign-natural
12816 @itemx -malign-power
12817 @opindex malign-natural
12818 @opindex malign-power
12819 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
12820 @option{-malign-natural} overrides the ABI-defined alignment of larger
12821 types, such as floating-point doubles, on their natural size-based boundary.
12822 The option @option{-malign-power} instructs GCC to follow the ABI-specified
12823 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
12825 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
12829 @itemx -mhard-float
12830 @opindex msoft-float
12831 @opindex mhard-float
12832 Generate code that does not use (uses) the floating-point register set.
12833 Software floating point emulation is provided if you use the
12834 @option{-msoft-float} option, and pass the option to GCC when linking.
12837 @itemx -mno-multiple
12839 @opindex mno-multiple
12840 Generate code that uses (does not use) the load multiple word
12841 instructions and the store multiple word instructions. These
12842 instructions are generated by default on POWER systems, and not
12843 generated on PowerPC systems. Do not use @option{-mmultiple} on little
12844 endian PowerPC systems, since those instructions do not work when the
12845 processor is in little endian mode. The exceptions are PPC740 and
12846 PPC750 which permit the instructions usage in little endian mode.
12851 @opindex mno-string
12852 Generate code that uses (does not use) the load string instructions
12853 and the store string word instructions to save multiple registers and
12854 do small block moves. These instructions are generated by default on
12855 POWER systems, and not generated on PowerPC systems. Do not use
12856 @option{-mstring} on little endian PowerPC systems, since those
12857 instructions do not work when the processor is in little endian mode.
12858 The exceptions are PPC740 and PPC750 which permit the instructions
12859 usage in little endian mode.
12864 @opindex mno-update
12865 Generate code that uses (does not use) the load or store instructions
12866 that update the base register to the address of the calculated memory
12867 location. These instructions are generated by default. If you use
12868 @option{-mno-update}, there is a small window between the time that the
12869 stack pointer is updated and the address of the previous frame is
12870 stored, which means code that walks the stack frame across interrupts or
12871 signals may get corrupted data.
12874 @itemx -mno-fused-madd
12875 @opindex mfused-madd
12876 @opindex mno-fused-madd
12877 Generate code that uses (does not use) the floating point multiply and
12878 accumulate instructions. These instructions are generated by default if
12879 hardware floating is used.
12885 Generate code that uses (does not use) the half-word multiply and
12886 multiply-accumulate instructions on the IBM 405 and 440 processors.
12887 These instructions are generated by default when targetting those
12894 Generate code that uses (does not use) the string-search @samp{dlmzb}
12895 instruction on the IBM 405 and 440 processors. This instruction is
12896 generated by default when targetting those processors.
12898 @item -mno-bit-align
12900 @opindex mno-bit-align
12901 @opindex mbit-align
12902 On System V.4 and embedded PowerPC systems do not (do) force structures
12903 and unions that contain bit-fields to be aligned to the base type of the
12906 For example, by default a structure containing nothing but 8
12907 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
12908 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
12909 the structure would be aligned to a 1 byte boundary and be one byte in
12912 @item -mno-strict-align
12913 @itemx -mstrict-align
12914 @opindex mno-strict-align
12915 @opindex mstrict-align
12916 On System V.4 and embedded PowerPC systems do not (do) assume that
12917 unaligned memory references will be handled by the system.
12919 @item -mrelocatable
12920 @itemx -mno-relocatable
12921 @opindex mrelocatable
12922 @opindex mno-relocatable
12923 On embedded PowerPC systems generate code that allows (does not allow)
12924 the program to be relocated to a different address at runtime. If you
12925 use @option{-mrelocatable} on any module, all objects linked together must
12926 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
12928 @item -mrelocatable-lib
12929 @itemx -mno-relocatable-lib
12930 @opindex mrelocatable-lib
12931 @opindex mno-relocatable-lib
12932 On embedded PowerPC systems generate code that allows (does not allow)
12933 the program to be relocated to a different address at runtime. Modules
12934 compiled with @option{-mrelocatable-lib} can be linked with either modules
12935 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
12936 with modules compiled with the @option{-mrelocatable} options.
12942 On System V.4 and embedded PowerPC systems do not (do) assume that
12943 register 2 contains a pointer to a global area pointing to the addresses
12944 used in the program.
12947 @itemx -mlittle-endian
12949 @opindex mlittle-endian
12950 On System V.4 and embedded PowerPC systems compile code for the
12951 processor in little endian mode. The @option{-mlittle-endian} option is
12952 the same as @option{-mlittle}.
12955 @itemx -mbig-endian
12957 @opindex mbig-endian
12958 On System V.4 and embedded PowerPC systems compile code for the
12959 processor in big endian mode. The @option{-mbig-endian} option is
12960 the same as @option{-mbig}.
12962 @item -mdynamic-no-pic
12963 @opindex mdynamic-no-pic
12964 On Darwin and Mac OS X systems, compile code so that it is not
12965 relocatable, but that its external references are relocatable. The
12966 resulting code is suitable for applications, but not shared
12969 @item -mprioritize-restricted-insns=@var{priority}
12970 @opindex mprioritize-restricted-insns
12971 This option controls the priority that is assigned to
12972 dispatch-slot restricted instructions during the second scheduling
12973 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
12974 @var{no/highest/second-highest} priority to dispatch slot restricted
12977 @item -msched-costly-dep=@var{dependence_type}
12978 @opindex msched-costly-dep
12979 This option controls which dependences are considered costly
12980 by the target during instruction scheduling. The argument
12981 @var{dependence_type} takes one of the following values:
12982 @var{no}: no dependence is costly,
12983 @var{all}: all dependences are costly,
12984 @var{true_store_to_load}: a true dependence from store to load is costly,
12985 @var{store_to_load}: any dependence from store to load is costly,
12986 @var{number}: any dependence which latency >= @var{number} is costly.
12988 @item -minsert-sched-nops=@var{scheme}
12989 @opindex minsert-sched-nops
12990 This option controls which nop insertion scheme will be used during
12991 the second scheduling pass. The argument @var{scheme} takes one of the
12993 @var{no}: Don't insert nops.
12994 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
12995 according to the scheduler's grouping.
12996 @var{regroup_exact}: Insert nops to force costly dependent insns into
12997 separate groups. Insert exactly as many nops as needed to force an insn
12998 to a new group, according to the estimated processor grouping.
12999 @var{number}: Insert nops to force costly dependent insns into
13000 separate groups. Insert @var{number} nops to force an insn to a new group.
13003 @opindex mcall-sysv
13004 On System V.4 and embedded PowerPC systems compile code using calling
13005 conventions that adheres to the March 1995 draft of the System V
13006 Application Binary Interface, PowerPC processor supplement. This is the
13007 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
13009 @item -mcall-sysv-eabi
13010 @opindex mcall-sysv-eabi
13011 Specify both @option{-mcall-sysv} and @option{-meabi} options.
13013 @item -mcall-sysv-noeabi
13014 @opindex mcall-sysv-noeabi
13015 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
13017 @item -mcall-solaris
13018 @opindex mcall-solaris
13019 On System V.4 and embedded PowerPC systems compile code for the Solaris
13023 @opindex mcall-linux
13024 On System V.4 and embedded PowerPC systems compile code for the
13025 Linux-based GNU system.
13029 On System V.4 and embedded PowerPC systems compile code for the
13030 Hurd-based GNU system.
13032 @item -mcall-netbsd
13033 @opindex mcall-netbsd
13034 On System V.4 and embedded PowerPC systems compile code for the
13035 NetBSD operating system.
13037 @item -maix-struct-return
13038 @opindex maix-struct-return
13039 Return all structures in memory (as specified by the AIX ABI)@.
13041 @item -msvr4-struct-return
13042 @opindex msvr4-struct-return
13043 Return structures smaller than 8 bytes in registers (as specified by the
13046 @item -mabi=@var{abi-type}
13048 Extend the current ABI with a particular extension, or remove such extension.
13049 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
13050 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
13054 Extend the current ABI with SPE ABI extensions. This does not change
13055 the default ABI, instead it adds the SPE ABI extensions to the current
13059 @opindex mabi=no-spe
13060 Disable Booke SPE ABI extensions for the current ABI@.
13062 @item -mabi=ibmlongdouble
13063 @opindex mabi=ibmlongdouble
13064 Change the current ABI to use IBM extended precision long double.
13065 This is a PowerPC 32-bit SYSV ABI option.
13067 @item -mabi=ieeelongdouble
13068 @opindex mabi=ieeelongdouble
13069 Change the current ABI to use IEEE extended precision long double.
13070 This is a PowerPC 32-bit Linux ABI option.
13073 @itemx -mno-prototype
13074 @opindex mprototype
13075 @opindex mno-prototype
13076 On System V.4 and embedded PowerPC systems assume that all calls to
13077 variable argument functions are properly prototyped. Otherwise, the
13078 compiler must insert an instruction before every non prototyped call to
13079 set or clear bit 6 of the condition code register (@var{CR}) to
13080 indicate whether floating point values were passed in the floating point
13081 registers in case the function takes a variable arguments. With
13082 @option{-mprototype}, only calls to prototyped variable argument functions
13083 will set or clear the bit.
13087 On embedded PowerPC systems, assume that the startup module is called
13088 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
13089 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
13094 On embedded PowerPC systems, assume that the startup module is called
13095 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
13100 On embedded PowerPC systems, assume that the startup module is called
13101 @file{crt0.o} and the standard C libraries are @file{libads.a} and
13104 @item -myellowknife
13105 @opindex myellowknife
13106 On embedded PowerPC systems, assume that the startup module is called
13107 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
13112 On System V.4 and embedded PowerPC systems, specify that you are
13113 compiling for a VxWorks system.
13117 Specify that you are compiling for the WindISS simulation environment.
13121 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
13122 header to indicate that @samp{eabi} extended relocations are used.
13128 On System V.4 and embedded PowerPC systems do (do not) adhere to the
13129 Embedded Applications Binary Interface (eabi) which is a set of
13130 modifications to the System V.4 specifications. Selecting @option{-meabi}
13131 means that the stack is aligned to an 8 byte boundary, a function
13132 @code{__eabi} is called to from @code{main} to set up the eabi
13133 environment, and the @option{-msdata} option can use both @code{r2} and
13134 @code{r13} to point to two separate small data areas. Selecting
13135 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
13136 do not call an initialization function from @code{main}, and the
13137 @option{-msdata} option will only use @code{r13} to point to a single
13138 small data area. The @option{-meabi} option is on by default if you
13139 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
13142 @opindex msdata=eabi
13143 On System V.4 and embedded PowerPC systems, put small initialized
13144 @code{const} global and static data in the @samp{.sdata2} section, which
13145 is pointed to by register @code{r2}. Put small initialized
13146 non-@code{const} global and static data in the @samp{.sdata} section,
13147 which is pointed to by register @code{r13}. Put small uninitialized
13148 global and static data in the @samp{.sbss} section, which is adjacent to
13149 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13150 incompatible with the @option{-mrelocatable} option. The
13151 @option{-msdata=eabi} option also sets the @option{-memb} option.
13154 @opindex msdata=sysv
13155 On System V.4 and embedded PowerPC systems, put small global and static
13156 data in the @samp{.sdata} section, which is pointed to by register
13157 @code{r13}. Put small uninitialized global and static data in the
13158 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13159 The @option{-msdata=sysv} option is incompatible with the
13160 @option{-mrelocatable} option.
13162 @item -msdata=default
13164 @opindex msdata=default
13166 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13167 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13168 same as @option{-msdata=sysv}.
13171 @opindex msdata-data
13172 On System V.4 and embedded PowerPC systems, put small global
13173 data in the @samp{.sdata} section. Put small uninitialized global
13174 data in the @samp{.sbss} section. Do not use register @code{r13}
13175 to address small data however. This is the default behavior unless
13176 other @option{-msdata} options are used.
13180 @opindex msdata=none
13182 On embedded PowerPC systems, put all initialized global and static data
13183 in the @samp{.data} section, and all uninitialized data in the
13184 @samp{.bss} section.
13188 @cindex smaller data references (PowerPC)
13189 @cindex .sdata/.sdata2 references (PowerPC)
13190 On embedded PowerPC systems, put global and static items less than or
13191 equal to @var{num} bytes into the small data or bss sections instead of
13192 the normal data or bss section. By default, @var{num} is 8. The
13193 @option{-G @var{num}} switch is also passed to the linker.
13194 All modules should be compiled with the same @option{-G @var{num}} value.
13197 @itemx -mno-regnames
13199 @opindex mno-regnames
13200 On System V.4 and embedded PowerPC systems do (do not) emit register
13201 names in the assembly language output using symbolic forms.
13204 @itemx -mno-longcall
13206 @opindex mno-longcall
13207 By default assume that all calls are far away so that a longer more
13208 expensive calling sequence is required. This is required for calls
13209 further than 32 megabytes (33,554,432 bytes) from the current location.
13210 A short call will be generated if the compiler knows
13211 the call cannot be that far away. This setting can be overridden by
13212 the @code{shortcall} function attribute, or by @code{#pragma
13215 Some linkers are capable of detecting out-of-range calls and generating
13216 glue code on the fly. On these systems, long calls are unnecessary and
13217 generate slower code. As of this writing, the AIX linker can do this,
13218 as can the GNU linker for PowerPC/64. It is planned to add this feature
13219 to the GNU linker for 32-bit PowerPC systems as well.
13221 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13222 callee, L42'', plus a ``branch island'' (glue code). The two target
13223 addresses represent the callee and the ``branch island''. The
13224 Darwin/PPC linker will prefer the first address and generate a ``bl
13225 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13226 otherwise, the linker will generate ``bl L42'' to call the ``branch
13227 island''. The ``branch island'' is appended to the body of the
13228 calling function; it computes the full 32-bit address of the callee
13231 On Mach-O (Darwin) systems, this option directs the compiler emit to
13232 the glue for every direct call, and the Darwin linker decides whether
13233 to use or discard it.
13235 In the future, we may cause GCC to ignore all longcall specifications
13236 when the linker is known to generate glue.
13240 Adds support for multithreading with the @dfn{pthreads} library.
13241 This option sets flags for both the preprocessor and linker.
13245 @node S/390 and zSeries Options
13246 @subsection S/390 and zSeries Options
13247 @cindex S/390 and zSeries Options
13249 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13253 @itemx -msoft-float
13254 @opindex mhard-float
13255 @opindex msoft-float
13256 Use (do not use) the hardware floating-point instructions and registers
13257 for floating-point operations. When @option{-msoft-float} is specified,
13258 functions in @file{libgcc.a} will be used to perform floating-point
13259 operations. When @option{-mhard-float} is specified, the compiler
13260 generates IEEE floating-point instructions. This is the default.
13262 @item -mlong-double-64
13263 @itemx -mlong-double-128
13264 @opindex mlong-double-64
13265 @opindex mlong-double-128
13266 These switches control the size of @code{long double} type. A size
13267 of 64bit makes the @code{long double} type equivalent to the @code{double}
13268 type. This is the default.
13271 @itemx -mno-backchain
13272 @opindex mbackchain
13273 @opindex mno-backchain
13274 Store (do not store) the address of the caller's frame as backchain pointer
13275 into the callee's stack frame.
13276 A backchain may be needed to allow debugging using tools that do not understand
13277 DWARF-2 call frame information.
13278 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13279 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13280 the backchain is placed into the topmost word of the 96/160 byte register
13283 In general, code compiled with @option{-mbackchain} is call-compatible with
13284 code compiled with @option{-mmo-backchain}; however, use of the backchain
13285 for debugging purposes usually requires that the whole binary is built with
13286 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13287 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13288 to build a linux kernel use @option{-msoft-float}.
13290 The default is to not maintain the backchain.
13292 @item -mpacked-stack
13293 @item -mno-packed-stack
13294 @opindex mpacked-stack
13295 @opindex mno-packed-stack
13296 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13297 specified, the compiler uses the all fields of the 96/160 byte register save
13298 area only for their default purpose; unused fields still take up stack space.
13299 When @option{-mpacked-stack} is specified, register save slots are densely
13300 packed at the top of the register save area; unused space is reused for other
13301 purposes, allowing for more efficient use of the available stack space.
13302 However, when @option{-mbackchain} is also in effect, the topmost word of
13303 the save area is always used to store the backchain, and the return address
13304 register is always saved two words below the backchain.
13306 As long as the stack frame backchain is not used, code generated with
13307 @option{-mpacked-stack} is call-compatible with code generated with
13308 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13309 S/390 or zSeries generated code that uses the stack frame backchain at run
13310 time, not just for debugging purposes. Such code is not call-compatible
13311 with code compiled with @option{-mpacked-stack}. Also, note that the
13312 combination of @option{-mbackchain},
13313 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13314 to build a linux kernel use @option{-msoft-float}.
13316 The default is to not use the packed stack layout.
13319 @itemx -mno-small-exec
13320 @opindex msmall-exec
13321 @opindex mno-small-exec
13322 Generate (or do not generate) code using the @code{bras} instruction
13323 to do subroutine calls.
13324 This only works reliably if the total executable size does not
13325 exceed 64k. The default is to use the @code{basr} instruction instead,
13326 which does not have this limitation.
13332 When @option{-m31} is specified, generate code compliant to the
13333 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13334 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13335 particular to generate 64-bit instructions. For the @samp{s390}
13336 targets, the default is @option{-m31}, while the @samp{s390x}
13337 targets default to @option{-m64}.
13343 When @option{-mzarch} is specified, generate code using the
13344 instructions available on z/Architecture.
13345 When @option{-mesa} is specified, generate code using the
13346 instructions available on ESA/390. Note that @option{-mesa} is
13347 not possible with @option{-m64}.
13348 When generating code compliant to the GNU/Linux for S/390 ABI,
13349 the default is @option{-mesa}. When generating code compliant
13350 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13356 Generate (or do not generate) code using the @code{mvcle} instruction
13357 to perform block moves. When @option{-mno-mvcle} is specified,
13358 use a @code{mvc} loop instead. This is the default unless optimizing for
13365 Print (or do not print) additional debug information when compiling.
13366 The default is to not print debug information.
13368 @item -march=@var{cpu-type}
13370 Generate code that will run on @var{cpu-type}, which is the name of a system
13371 representing a certain processor type. Possible values for
13372 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13373 When generating code using the instructions available on z/Architecture,
13374 the default is @option{-march=z900}. Otherwise, the default is
13375 @option{-march=g5}.
13377 @item -mtune=@var{cpu-type}
13379 Tune to @var{cpu-type} everything applicable about the generated code,
13380 except for the ABI and the set of available instructions.
13381 The list of @var{cpu-type} values is the same as for @option{-march}.
13382 The default is the value used for @option{-march}.
13385 @itemx -mno-tpf-trace
13386 @opindex mtpf-trace
13387 @opindex mno-tpf-trace
13388 Generate code that adds (does not add) in TPF OS specific branches to trace
13389 routines in the operating system. This option is off by default, even
13390 when compiling for the TPF OS@.
13393 @itemx -mno-fused-madd
13394 @opindex mfused-madd
13395 @opindex mno-fused-madd
13396 Generate code that uses (does not use) the floating point multiply and
13397 accumulate instructions. These instructions are generated by default if
13398 hardware floating point is used.
13400 @item -mwarn-framesize=@var{framesize}
13401 @opindex mwarn-framesize
13402 Emit a warning if the current function exceeds the given frame size. Because
13403 this is a compile time check it doesn't need to be a real problem when the program
13404 runs. It is intended to identify functions which most probably cause
13405 a stack overflow. It is useful to be used in an environment with limited stack
13406 size e.g.@: the linux kernel.
13408 @item -mwarn-dynamicstack
13409 @opindex mwarn-dynamicstack
13410 Emit a warning if the function calls alloca or uses dynamically
13411 sized arrays. This is generally a bad idea with a limited stack size.
13413 @item -mstack-guard=@var{stack-guard}
13414 @item -mstack-size=@var{stack-size}
13415 @opindex mstack-guard
13416 @opindex mstack-size
13417 If these options are provided the s390 back end emits additional instructions in
13418 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13419 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13420 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13421 the frame size of the compiled function is chosen.
13422 These options are intended to be used to help debugging stack overflow problems.
13423 The additionally emitted code causes only little overhead and hence can also be
13424 used in production like systems without greater performance degradation. The given
13425 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13426 @var{stack-guard} without exceeding 64k.
13427 In order to be efficient the extra code makes the assumption that the stack starts
13428 at an address aligned to the value given by @var{stack-size}.
13429 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13432 @node Score Options
13433 @subsection Score Options
13434 @cindex Score Options
13436 These options are defined for Score implementations:
13441 Compile code for big endian mode. This is the default.
13445 Compile code for little endian mode.
13449 Disable generate bcnz instruction.
13453 Enable generate unaligned load and store instruction.
13457 Enable the use of multiply-accumulate instructions. Disabled by default.
13461 Specify the SCORE5 as the target architecture.
13465 Specify the SCORE5U of the target architecture.
13469 Specify the SCORE7 as the target architecture. This is the default.
13473 Specify the SCORE7D as the target architecture.
13477 @subsection SH Options
13479 These @samp{-m} options are defined for the SH implementations:
13484 Generate code for the SH1.
13488 Generate code for the SH2.
13491 Generate code for the SH2e.
13495 Generate code for the SH3.
13499 Generate code for the SH3e.
13503 Generate code for the SH4 without a floating-point unit.
13505 @item -m4-single-only
13506 @opindex m4-single-only
13507 Generate code for the SH4 with a floating-point unit that only
13508 supports single-precision arithmetic.
13512 Generate code for the SH4 assuming the floating-point unit is in
13513 single-precision mode by default.
13517 Generate code for the SH4.
13521 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13522 floating-point unit is not used.
13524 @item -m4a-single-only
13525 @opindex m4a-single-only
13526 Generate code for the SH4a, in such a way that no double-precision
13527 floating point operations are used.
13530 @opindex m4a-single
13531 Generate code for the SH4a assuming the floating-point unit is in
13532 single-precision mode by default.
13536 Generate code for the SH4a.
13540 Same as @option{-m4a-nofpu}, except that it implicitly passes
13541 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13542 instructions at the moment.
13546 Compile code for the processor in big endian mode.
13550 Compile code for the processor in little endian mode.
13554 Align doubles at 64-bit boundaries. Note that this changes the calling
13555 conventions, and thus some functions from the standard C library will
13556 not work unless you recompile it first with @option{-mdalign}.
13560 Shorten some address references at link time, when possible; uses the
13561 linker option @option{-relax}.
13565 Use 32-bit offsets in @code{switch} tables. The default is to use
13570 Enable the use of the instruction @code{fmovd}.
13574 Comply with the calling conventions defined by Renesas.
13578 Comply with the calling conventions defined by Renesas.
13582 Comply with the calling conventions defined for GCC before the Renesas
13583 conventions were available. This option is the default for all
13584 targets of the SH toolchain except for @samp{sh-symbianelf}.
13587 @opindex mnomacsave
13588 Mark the @code{MAC} register as call-clobbered, even if
13589 @option{-mhitachi} is given.
13593 Increase IEEE-compliance of floating-point code.
13594 At the moment, this is equivalent to @option{-fno-finite-math-only}.
13595 When generating 16 bit SH opcodes, getting IEEE-conforming results for
13596 comparisons of NANs / infinities incurs extra overhead in every
13597 floating point comparison, therefore the default is set to
13598 @option{-ffinite-math-only}.
13600 @item -minline-ic_invalidate
13601 @opindex minline-ic_invalidate
13602 Inline code to invalidate instruction cache entries after setting up
13603 nested function trampolines.
13604 This option has no effect if -musermode is in effect and the selected
13605 code generation option (e.g. -m4) does not allow the use of the icbi
13607 If the selected code generation option does not allow the use of the icbi
13608 instruction, and -musermode is not in effect, the inlined code will
13609 manipulate the instruction cache address array directly with an associative
13610 write. This not only requires privileged mode, but it will also
13611 fail if the cache line had been mapped via the TLB and has become unmapped.
13615 Dump instruction size and location in the assembly code.
13618 @opindex mpadstruct
13619 This option is deprecated. It pads structures to multiple of 4 bytes,
13620 which is incompatible with the SH ABI@.
13624 Optimize for space instead of speed. Implied by @option{-Os}.
13627 @opindex mprefergot
13628 When generating position-independent code, emit function calls using
13629 the Global Offset Table instead of the Procedure Linkage Table.
13633 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
13634 if the inlined code would not work in user mode.
13635 This is the default when the target is @code{sh-*-linux*}.
13637 @item -multcost=@var{number}
13638 @opindex multcost=@var{number}
13639 Set the cost to assume for a multiply insn.
13641 @item -mdiv=@var{strategy}
13642 @opindex mdiv=@var{strategy}
13643 Set the division strategy to use for SHmedia code. @var{strategy} must be
13644 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
13645 inv:call2, inv:fp .
13646 "fp" performs the operation in floating point. This has a very high latency,
13647 but needs only a few instructions, so it might be a good choice if
13648 your code has enough easily exploitable ILP to allow the compiler to
13649 schedule the floating point instructions together with other instructions.
13650 Division by zero causes a floating point exception.
13651 "inv" uses integer operations to calculate the inverse of the divisor,
13652 and then multiplies the dividend with the inverse. This strategy allows
13653 cse and hoisting of the inverse calculation. Division by zero calculates
13654 an unspecified result, but does not trap.
13655 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
13656 have been found, or if the entire operation has been hoisted to the same
13657 place, the last stages of the inverse calculation are intertwined with the
13658 final multiply to reduce the overall latency, at the expense of using a few
13659 more instructions, and thus offering fewer scheduling opportunities with
13661 "call" calls a library function that usually implements the inv:minlat
13663 This gives high code density for m5-*media-nofpu compilations.
13664 "call2" uses a different entry point of the same library function, where it
13665 assumes that a pointer to a lookup table has already been set up, which
13666 exposes the pointer load to cse / code hoisting optimizations.
13667 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
13668 code generation, but if the code stays unoptimized, revert to the "call",
13669 "call2", or "fp" strategies, respectively. Note that the
13670 potentially-trapping side effect of division by zero is carried by a
13671 separate instruction, so it is possible that all the integer instructions
13672 are hoisted out, but the marker for the side effect stays where it is.
13673 A recombination to fp operations or a call is not possible in that case.
13674 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
13675 that the inverse calculation was nor separated from the multiply, they speed
13676 up division where the dividend fits into 20 bits (plus sign where applicable),
13677 by inserting a test to skip a number of operations in this case; this test
13678 slows down the case of larger dividends. inv20u assumes the case of a such
13679 a small dividend to be unlikely, and inv20l assumes it to be likely.
13681 @item -mdivsi3_libfunc=@var{name}
13682 @opindex mdivsi3_libfunc=@var{name}
13683 Set the name of the library function used for 32 bit signed division to
13684 @var{name}. This only affect the name used in the call and inv:call
13685 division strategies, and the compiler will still expect the same
13686 sets of input/output/clobbered registers as if this option was not present.
13688 @item -madjust-unroll
13689 @opindex madjust-unroll
13690 Throttle unrolling to avoid thrashing target registers.
13691 This option only has an effect if the gcc code base supports the
13692 TARGET_ADJUST_UNROLL_MAX target hook.
13694 @item -mindexed-addressing
13695 @opindex mindexed-addressing
13696 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
13697 This is only safe if the hardware and/or OS implement 32 bit wrap-around
13698 semantics for the indexed addressing mode. The architecture allows the
13699 implementation of processors with 64 bit MMU, which the OS could use to
13700 get 32 bit addressing, but since no current hardware implementation supports
13701 this or any other way to make the indexed addressing mode safe to use in
13702 the 32 bit ABI, the default is -mno-indexed-addressing.
13704 @item -mgettrcost=@var{number}
13705 @opindex mgettrcost=@var{number}
13706 Set the cost assumed for the gettr instruction to @var{number}.
13707 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
13711 Assume pt* instructions won't trap. This will generally generate better
13712 scheduled code, but is unsafe on current hardware. The current architecture
13713 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
13714 This has the unintentional effect of making it unsafe to schedule ptabs /
13715 ptrel before a branch, or hoist it out of a loop. For example,
13716 __do_global_ctors, a part of libgcc that runs constructors at program
13717 startup, calls functions in a list which is delimited by @minus{}1. With the
13718 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
13719 That means that all the constructors will be run a bit quicker, but when
13720 the loop comes to the end of the list, the program crashes because ptabs
13721 loads @minus{}1 into a target register. Since this option is unsafe for any
13722 hardware implementing the current architecture specification, the default
13723 is -mno-pt-fixed. Unless the user specifies a specific cost with
13724 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
13725 this deters register allocation using target registers for storing
13728 @item -minvalid-symbols
13729 @opindex minvalid-symbols
13730 Assume symbols might be invalid. Ordinary function symbols generated by
13731 the compiler will always be valid to load with movi/shori/ptabs or
13732 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
13733 to generate symbols that will cause ptabs / ptrel to trap.
13734 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
13735 It will then prevent cross-basic-block cse, hoisting and most scheduling
13736 of symbol loads. The default is @option{-mno-invalid-symbols}.
13739 @node SPARC Options
13740 @subsection SPARC Options
13741 @cindex SPARC options
13743 These @samp{-m} options are supported on the SPARC:
13746 @item -mno-app-regs
13748 @opindex mno-app-regs
13750 Specify @option{-mapp-regs} to generate output using the global registers
13751 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
13754 To be fully SVR4 ABI compliant at the cost of some performance loss,
13755 specify @option{-mno-app-regs}. You should compile libraries and system
13756 software with this option.
13759 @itemx -mhard-float
13761 @opindex mhard-float
13762 Generate output containing floating point instructions. This is the
13766 @itemx -msoft-float
13768 @opindex msoft-float
13769 Generate output containing library calls for floating point.
13770 @strong{Warning:} the requisite libraries are not available for all SPARC
13771 targets. Normally the facilities of the machine's usual C compiler are
13772 used, but this cannot be done directly in cross-compilation. You must make
13773 your own arrangements to provide suitable library functions for
13774 cross-compilation. The embedded targets @samp{sparc-*-aout} and
13775 @samp{sparclite-*-*} do provide software floating point support.
13777 @option{-msoft-float} changes the calling convention in the output file;
13778 therefore, it is only useful if you compile @emph{all} of a program with
13779 this option. In particular, you need to compile @file{libgcc.a}, the
13780 library that comes with GCC, with @option{-msoft-float} in order for
13783 @item -mhard-quad-float
13784 @opindex mhard-quad-float
13785 Generate output containing quad-word (long double) floating point
13788 @item -msoft-quad-float
13789 @opindex msoft-quad-float
13790 Generate output containing library calls for quad-word (long double)
13791 floating point instructions. The functions called are those specified
13792 in the SPARC ABI@. This is the default.
13794 As of this writing, there are no SPARC implementations that have hardware
13795 support for the quad-word floating point instructions. They all invoke
13796 a trap handler for one of these instructions, and then the trap handler
13797 emulates the effect of the instruction. Because of the trap handler overhead,
13798 this is much slower than calling the ABI library routines. Thus the
13799 @option{-msoft-quad-float} option is the default.
13801 @item -mno-unaligned-doubles
13802 @itemx -munaligned-doubles
13803 @opindex mno-unaligned-doubles
13804 @opindex munaligned-doubles
13805 Assume that doubles have 8 byte alignment. This is the default.
13807 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
13808 alignment only if they are contained in another type, or if they have an
13809 absolute address. Otherwise, it assumes they have 4 byte alignment.
13810 Specifying this option avoids some rare compatibility problems with code
13811 generated by other compilers. It is not the default because it results
13812 in a performance loss, especially for floating point code.
13814 @item -mno-faster-structs
13815 @itemx -mfaster-structs
13816 @opindex mno-faster-structs
13817 @opindex mfaster-structs
13818 With @option{-mfaster-structs}, the compiler assumes that structures
13819 should have 8 byte alignment. This enables the use of pairs of
13820 @code{ldd} and @code{std} instructions for copies in structure
13821 assignment, in place of twice as many @code{ld} and @code{st} pairs.
13822 However, the use of this changed alignment directly violates the SPARC
13823 ABI@. Thus, it's intended only for use on targets where the developer
13824 acknowledges that their resulting code will not be directly in line with
13825 the rules of the ABI@.
13827 @item -mimpure-text
13828 @opindex mimpure-text
13829 @option{-mimpure-text}, used in addition to @option{-shared}, tells
13830 the compiler to not pass @option{-z text} to the linker when linking a
13831 shared object. Using this option, you can link position-dependent
13832 code into a shared object.
13834 @option{-mimpure-text} suppresses the ``relocations remain against
13835 allocatable but non-writable sections'' linker error message.
13836 However, the necessary relocations will trigger copy-on-write, and the
13837 shared object is not actually shared across processes. Instead of
13838 using @option{-mimpure-text}, you should compile all source code with
13839 @option{-fpic} or @option{-fPIC}.
13841 This option is only available on SunOS and Solaris.
13843 @item -mcpu=@var{cpu_type}
13845 Set the instruction set, register set, and instruction scheduling parameters
13846 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
13847 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
13848 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
13849 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
13850 @samp{ultrasparc3}, and @samp{niagara}.
13852 Default instruction scheduling parameters are used for values that select
13853 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
13854 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
13856 Here is a list of each supported architecture and their supported
13861 v8: supersparc, hypersparc
13862 sparclite: f930, f934, sparclite86x
13864 v9: ultrasparc, ultrasparc3, niagara
13867 By default (unless configured otherwise), GCC generates code for the V7
13868 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
13869 additionally optimizes it for the Cypress CY7C602 chip, as used in the
13870 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
13871 SPARCStation 1, 2, IPX etc.
13873 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
13874 architecture. The only difference from V7 code is that the compiler emits
13875 the integer multiply and integer divide instructions which exist in SPARC-V8
13876 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
13877 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
13880 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
13881 the SPARC architecture. This adds the integer multiply, integer divide step
13882 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
13883 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
13884 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
13885 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
13886 MB86934 chip, which is the more recent SPARClite with FPU@.
13888 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
13889 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
13890 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
13891 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
13892 optimizes it for the TEMIC SPARClet chip.
13894 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
13895 architecture. This adds 64-bit integer and floating-point move instructions,
13896 3 additional floating-point condition code registers and conditional move
13897 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
13898 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
13899 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
13900 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
13901 @option{-mcpu=niagara}, the compiler additionally optimizes it for
13902 Sun UltraSPARC T1 chips.
13904 @item -mtune=@var{cpu_type}
13906 Set the instruction scheduling parameters for machine type
13907 @var{cpu_type}, but do not set the instruction set or register set that the
13908 option @option{-mcpu=@var{cpu_type}} would.
13910 The same values for @option{-mcpu=@var{cpu_type}} can be used for
13911 @option{-mtune=@var{cpu_type}}, but the only useful values are those
13912 that select a particular cpu implementation. Those are @samp{cypress},
13913 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
13914 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
13915 @samp{ultrasparc3}, and @samp{niagara}.
13920 @opindex mno-v8plus
13921 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
13922 difference from the V8 ABI is that the global and out registers are
13923 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
13924 mode for all SPARC-V9 processors.
13930 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
13931 Visual Instruction Set extensions. The default is @option{-mno-vis}.
13934 These @samp{-m} options are supported in addition to the above
13935 on SPARC-V9 processors in 64-bit environments:
13938 @item -mlittle-endian
13939 @opindex mlittle-endian
13940 Generate code for a processor running in little-endian mode. It is only
13941 available for a few configurations and most notably not on Solaris and Linux.
13947 Generate code for a 32-bit or 64-bit environment.
13948 The 32-bit environment sets int, long and pointer to 32 bits.
13949 The 64-bit environment sets int to 32 bits and long and pointer
13952 @item -mcmodel=medlow
13953 @opindex mcmodel=medlow
13954 Generate code for the Medium/Low code model: 64-bit addresses, programs
13955 must be linked in the low 32 bits of memory. Programs can be statically
13956 or dynamically linked.
13958 @item -mcmodel=medmid
13959 @opindex mcmodel=medmid
13960 Generate code for the Medium/Middle code model: 64-bit addresses, programs
13961 must be linked in the low 44 bits of memory, the text and data segments must
13962 be less than 2GB in size and the data segment must be located within 2GB of
13965 @item -mcmodel=medany
13966 @opindex mcmodel=medany
13967 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
13968 may be linked anywhere in memory, the text and data segments must be less
13969 than 2GB in size and the data segment must be located within 2GB of the
13972 @item -mcmodel=embmedany
13973 @opindex mcmodel=embmedany
13974 Generate code for the Medium/Anywhere code model for embedded systems:
13975 64-bit addresses, the text and data segments must be less than 2GB in
13976 size, both starting anywhere in memory (determined at link time). The
13977 global register %g4 points to the base of the data segment. Programs
13978 are statically linked and PIC is not supported.
13981 @itemx -mno-stack-bias
13982 @opindex mstack-bias
13983 @opindex mno-stack-bias
13984 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
13985 frame pointer if present, are offset by @minus{}2047 which must be added back
13986 when making stack frame references. This is the default in 64-bit mode.
13987 Otherwise, assume no such offset is present.
13990 These switches are supported in addition to the above on Solaris:
13995 Add support for multithreading using the Solaris threads library. This
13996 option sets flags for both the preprocessor and linker. This option does
13997 not affect the thread safety of object code produced by the compiler or
13998 that of libraries supplied with it.
14002 Add support for multithreading using the POSIX threads library. This
14003 option sets flags for both the preprocessor and linker. This option does
14004 not affect the thread safety of object code produced by the compiler or
14005 that of libraries supplied with it.
14009 This is a synonym for @option{-pthreads}.
14013 @subsection SPU Options
14014 @cindex SPU options
14016 These @samp{-m} options are supported on the SPU:
14020 @itemx -merror-reloc
14021 @opindex mwarn-reloc
14022 @opindex merror-reloc
14024 The loader for SPU does not handle dynamic relocations. By default, GCC
14025 will give an error when it generates code that requires a dynamic
14026 relocation. @option{-mno-error-reloc} disables the error,
14027 @option{-mwarn-reloc} will generate a warning instead.
14030 @itemx -munsafe-dma
14032 @opindex munsafe-dma
14034 Instructions which initiate or test completion of DMA must not be
14035 reordered with respect to loads and stores of the memory which is being
14036 accessed. Users typically address this problem using the volatile
14037 keyword, but that can lead to inefficient code in places where the
14038 memory is known to not change. Rather than mark the memory as volatile
14039 we treat the DMA instructions as potentially effecting all memory. With
14040 @option{-munsafe-dma} users must use the volatile keyword to protect
14043 @item -mbranch-hints
14044 @opindex mbranch-hints
14046 By default, GCC will generate a branch hint instruction to avoid
14047 pipeline stalls for always taken or probably taken branches. A hint
14048 will not be generated closer than 8 instructions away from its branch.
14049 There is little reason to disable them, except for debugging purposes,
14050 or to make an object a little bit smaller.
14054 @opindex msmall-mem
14055 @opindex mlarge-mem
14057 By default, GCC generates code assuming that addresses are never larger
14058 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
14059 a full 32 bit address.
14064 By default, GCC links against startup code that assumes the SPU-style
14065 main function interface (which has an unconventional parameter list).
14066 With @option{-mstdmain}, GCC will link your program against startup
14067 code that assumes a C99-style interface to @code{main}, including a
14068 local copy of @code{argv} strings.
14070 @item -mfixed-range=@var{register-range}
14071 @opindex mfixed-range
14072 Generate code treating the given register range as fixed registers.
14073 A fixed register is one that the register allocator can not use. This is
14074 useful when compiling kernel code. A register range is specified as
14075 two registers separated by a dash. Multiple register ranges can be
14076 specified separated by a comma.
14080 @node System V Options
14081 @subsection Options for System V
14083 These additional options are available on System V Release 4 for
14084 compatibility with other compilers on those systems:
14089 Create a shared object.
14090 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
14094 Identify the versions of each tool used by the compiler, in a
14095 @code{.ident} assembler directive in the output.
14099 Refrain from adding @code{.ident} directives to the output file (this is
14102 @item -YP,@var{dirs}
14104 Search the directories @var{dirs}, and no others, for libraries
14105 specified with @option{-l}.
14107 @item -Ym,@var{dir}
14109 Look in the directory @var{dir} to find the M4 preprocessor.
14110 The assembler uses this option.
14111 @c This is supposed to go with a -Yd for predefined M4 macro files, but
14112 @c the generic assembler that comes with Solaris takes just -Ym.
14115 @node TMS320C3x/C4x Options
14116 @subsection TMS320C3x/C4x Options
14117 @cindex TMS320C3x/C4x Options
14119 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
14123 @item -mcpu=@var{cpu_type}
14125 Set the instruction set, register set, and instruction scheduling
14126 parameters for machine type @var{cpu_type}. Supported values for
14127 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
14128 @samp{c44}. The default is @samp{c40} to generate code for the
14133 @itemx -msmall-memory
14135 @opindex mbig-memory
14137 @opindex msmall-memory
14139 Generates code for the big or small memory model. The small memory
14140 model assumed that all data fits into one 64K word page. At run-time
14141 the data page (DP) register must be set to point to the 64K page
14142 containing the .bss and .data program sections. The big memory model is
14143 the default and requires reloading of the DP register for every direct
14150 Allow (disallow) allocation of general integer operands into the block
14151 count register BK@.
14157 Enable (disable) generation of code using decrement and branch,
14158 DBcond(D), instructions. This is enabled by default for the C4x. To be
14159 on the safe side, this is disabled for the C3x, since the maximum
14160 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
14161 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
14162 that it can utilize the decrement and branch instruction, but will give
14163 up if there is more than one memory reference in the loop. Thus a loop
14164 where the loop counter is decremented can generate slightly more
14165 efficient code, in cases where the RPTB instruction cannot be utilized.
14167 @item -mdp-isr-reload
14169 @opindex mdp-isr-reload
14171 Force the DP register to be saved on entry to an interrupt service
14172 routine (ISR), reloaded to point to the data section, and restored on
14173 exit from the ISR@. This should not be required unless someone has
14174 violated the small memory model by modifying the DP register, say within
14181 For the C3x use the 24-bit MPYI instruction for integer multiplies
14182 instead of a library call to guarantee 32-bit results. Note that if one
14183 of the operands is a constant, then the multiplication will be performed
14184 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
14185 then squaring operations are performed inline instead of a library call.
14188 @itemx -mno-fast-fix
14190 @opindex mno-fast-fix
14191 The C3x/C4x FIX instruction to convert a floating point value to an
14192 integer value chooses the nearest integer less than or equal to the
14193 floating point value rather than to the nearest integer. Thus if the
14194 floating point number is negative, the result will be incorrectly
14195 truncated an additional code is necessary to detect and correct this
14196 case. This option can be used to disable generation of the additional
14197 code required to correct the result.
14203 Enable (disable) generation of repeat block sequences using the RPTB
14204 instruction for zero overhead looping. The RPTB construct is only used
14205 for innermost loops that do not call functions or jump across the loop
14206 boundaries. There is no advantage having nested RPTB loops due to the
14207 overhead required to save and restore the RC, RS, and RE registers.
14208 This is enabled by default with @option{-O2}.
14210 @item -mrpts=@var{count}
14214 Enable (disable) the use of the single instruction repeat instruction
14215 RPTS@. If a repeat block contains a single instruction, and the loop
14216 count can be guaranteed to be less than the value @var{count}, GCC will
14217 emit a RPTS instruction instead of a RPTB@. If no value is specified,
14218 then a RPTS will be emitted even if the loop count cannot be determined
14219 at compile time. Note that the repeated instruction following RPTS does
14220 not have to be reloaded from memory each iteration, thus freeing up the
14221 CPU buses for operands. However, since interrupts are blocked by this
14222 instruction, it is disabled by default.
14224 @item -mloop-unsigned
14225 @itemx -mno-loop-unsigned
14226 @opindex mloop-unsigned
14227 @opindex mno-loop-unsigned
14228 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
14229 is @math{2^{31} + 1} since these instructions test if the iteration count is
14230 negative to terminate the loop. If the iteration count is unsigned
14231 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
14232 exceeded. This switch allows an unsigned iteration count.
14236 Try to emit an assembler syntax that the TI assembler (asm30) is happy
14237 with. This also enforces compatibility with the API employed by the TI
14238 C3x C compiler. For example, long doubles are passed as structures
14239 rather than in floating point registers.
14245 Generate code that uses registers (stack) for passing arguments to functions.
14246 By default, arguments are passed in registers where possible rather
14247 than by pushing arguments on to the stack.
14249 @item -mparallel-insns
14250 @itemx -mno-parallel-insns
14251 @opindex mparallel-insns
14252 @opindex mno-parallel-insns
14253 Allow the generation of parallel instructions. This is enabled by
14254 default with @option{-O2}.
14256 @item -mparallel-mpy
14257 @itemx -mno-parallel-mpy
14258 @opindex mparallel-mpy
14259 @opindex mno-parallel-mpy
14260 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
14261 provided @option{-mparallel-insns} is also specified. These instructions have
14262 tight register constraints which can pessimize the code generation
14263 of large functions.
14268 @subsection V850 Options
14269 @cindex V850 Options
14271 These @samp{-m} options are defined for V850 implementations:
14275 @itemx -mno-long-calls
14276 @opindex mlong-calls
14277 @opindex mno-long-calls
14278 Treat all calls as being far away (near). If calls are assumed to be
14279 far away, the compiler will always load the functions address up into a
14280 register, and call indirect through the pointer.
14286 Do not optimize (do optimize) basic blocks that use the same index
14287 pointer 4 or more times to copy pointer into the @code{ep} register, and
14288 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14289 option is on by default if you optimize.
14291 @item -mno-prolog-function
14292 @itemx -mprolog-function
14293 @opindex mno-prolog-function
14294 @opindex mprolog-function
14295 Do not use (do use) external functions to save and restore registers
14296 at the prologue and epilogue of a function. The external functions
14297 are slower, but use less code space if more than one function saves
14298 the same number of registers. The @option{-mprolog-function} option
14299 is on by default if you optimize.
14303 Try to make the code as small as possible. At present, this just turns
14304 on the @option{-mep} and @option{-mprolog-function} options.
14306 @item -mtda=@var{n}
14308 Put static or global variables whose size is @var{n} bytes or less into
14309 the tiny data area that register @code{ep} points to. The tiny data
14310 area can hold up to 256 bytes in total (128 bytes for byte references).
14312 @item -msda=@var{n}
14314 Put static or global variables whose size is @var{n} bytes or less into
14315 the small data area that register @code{gp} points to. The small data
14316 area can hold up to 64 kilobytes.
14318 @item -mzda=@var{n}
14320 Put static or global variables whose size is @var{n} bytes or less into
14321 the first 32 kilobytes of memory.
14325 Specify that the target processor is the V850.
14328 @opindex mbig-switch
14329 Generate code suitable for big switch tables. Use this option only if
14330 the assembler/linker complain about out of range branches within a switch
14335 This option will cause r2 and r5 to be used in the code generated by
14336 the compiler. This setting is the default.
14338 @item -mno-app-regs
14339 @opindex mno-app-regs
14340 This option will cause r2 and r5 to be treated as fixed registers.
14344 Specify that the target processor is the V850E1. The preprocessor
14345 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14346 this option is used.
14350 Specify that the target processor is the V850E@. The preprocessor
14351 constant @samp{__v850e__} will be defined if this option is used.
14353 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14354 are defined then a default target processor will be chosen and the
14355 relevant @samp{__v850*__} preprocessor constant will be defined.
14357 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14358 defined, regardless of which processor variant is the target.
14360 @item -mdisable-callt
14361 @opindex mdisable-callt
14362 This option will suppress generation of the CALLT instruction for the
14363 v850e and v850e1 flavors of the v850 architecture. The default is
14364 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14369 @subsection VAX Options
14370 @cindex VAX options
14372 These @samp{-m} options are defined for the VAX:
14377 Do not output certain jump instructions (@code{aobleq} and so on)
14378 that the Unix assembler for the VAX cannot handle across long
14383 Do output those jump instructions, on the assumption that you
14384 will assemble with the GNU assembler.
14388 Output code for g-format floating point numbers instead of d-format.
14391 @node VxWorks Options
14392 @subsection VxWorks Options
14393 @cindex VxWorks Options
14395 The options in this section are defined for all VxWorks targets.
14396 Options specific to the target hardware are listed with the other
14397 options for that target.
14402 GCC can generate code for both VxWorks kernels and real time processes
14403 (RTPs). This option switches from the former to the latter. It also
14404 defines the preprocessor macro @code{__RTP__}.
14407 @opindex non-static
14408 Link an RTP executable against shared libraries rather than static
14409 libraries. The options @option{-static} and @option{-shared} can
14410 also be used for RTPs (@pxref{Link Options}); @option{-static}
14417 These options are passed down to the linker. They are defined for
14418 compatibility with Diab.
14421 @opindex Xbind-lazy
14422 Enable lazy binding of function calls. This option is equivalent to
14423 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14427 Disable lazy binding of function calls. This option is the default and
14428 is defined for compatibility with Diab.
14431 @node x86-64 Options
14432 @subsection x86-64 Options
14433 @cindex x86-64 options
14435 These are listed under @xref{i386 and x86-64 Options}.
14437 @node Xstormy16 Options
14438 @subsection Xstormy16 Options
14439 @cindex Xstormy16 Options
14441 These options are defined for Xstormy16:
14446 Choose startup files and linker script suitable for the simulator.
14449 @node Xtensa Options
14450 @subsection Xtensa Options
14451 @cindex Xtensa Options
14453 These options are supported for Xtensa targets:
14457 @itemx -mno-const16
14459 @opindex mno-const16
14460 Enable or disable use of @code{CONST16} instructions for loading
14461 constant values. The @code{CONST16} instruction is currently not a
14462 standard option from Tensilica. When enabled, @code{CONST16}
14463 instructions are always used in place of the standard @code{L32R}
14464 instructions. The use of @code{CONST16} is enabled by default only if
14465 the @code{L32R} instruction is not available.
14468 @itemx -mno-fused-madd
14469 @opindex mfused-madd
14470 @opindex mno-fused-madd
14471 Enable or disable use of fused multiply/add and multiply/subtract
14472 instructions in the floating-point option. This has no effect if the
14473 floating-point option is not also enabled. Disabling fused multiply/add
14474 and multiply/subtract instructions forces the compiler to use separate
14475 instructions for the multiply and add/subtract operations. This may be
14476 desirable in some cases where strict IEEE 754-compliant results are
14477 required: the fused multiply add/subtract instructions do not round the
14478 intermediate result, thereby producing results with @emph{more} bits of
14479 precision than specified by the IEEE standard. Disabling fused multiply
14480 add/subtract instructions also ensures that the program output is not
14481 sensitive to the compiler's ability to combine multiply and add/subtract
14484 @item -mtext-section-literals
14485 @itemx -mno-text-section-literals
14486 @opindex mtext-section-literals
14487 @opindex mno-text-section-literals
14488 Control the treatment of literal pools. The default is
14489 @option{-mno-text-section-literals}, which places literals in a separate
14490 section in the output file. This allows the literal pool to be placed
14491 in a data RAM/ROM, and it also allows the linker to combine literal
14492 pools from separate object files to remove redundant literals and
14493 improve code size. With @option{-mtext-section-literals}, the literals
14494 are interspersed in the text section in order to keep them as close as
14495 possible to their references. This may be necessary for large assembly
14498 @item -mtarget-align
14499 @itemx -mno-target-align
14500 @opindex mtarget-align
14501 @opindex mno-target-align
14502 When this option is enabled, GCC instructs the assembler to
14503 automatically align instructions to reduce branch penalties at the
14504 expense of some code density. The assembler attempts to widen density
14505 instructions to align branch targets and the instructions following call
14506 instructions. If there are not enough preceding safe density
14507 instructions to align a target, no widening will be performed. The
14508 default is @option{-mtarget-align}. These options do not affect the
14509 treatment of auto-aligned instructions like @code{LOOP}, which the
14510 assembler will always align, either by widening density instructions or
14511 by inserting no-op instructions.
14514 @itemx -mno-longcalls
14515 @opindex mlongcalls
14516 @opindex mno-longcalls
14517 When this option is enabled, GCC instructs the assembler to translate
14518 direct calls to indirect calls unless it can determine that the target
14519 of a direct call is in the range allowed by the call instruction. This
14520 translation typically occurs for calls to functions in other source
14521 files. Specifically, the assembler translates a direct @code{CALL}
14522 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14523 The default is @option{-mno-longcalls}. This option should be used in
14524 programs where the call target can potentially be out of range. This
14525 option is implemented in the assembler, not the compiler, so the
14526 assembly code generated by GCC will still show direct call
14527 instructions---look at the disassembled object code to see the actual
14528 instructions. Note that the assembler will use an indirect call for
14529 every cross-file call, not just those that really will be out of range.
14532 @node zSeries Options
14533 @subsection zSeries Options
14534 @cindex zSeries options
14536 These are listed under @xref{S/390 and zSeries Options}.
14538 @node Code Gen Options
14539 @section Options for Code Generation Conventions
14540 @cindex code generation conventions
14541 @cindex options, code generation
14542 @cindex run-time options
14544 These machine-independent options control the interface conventions
14545 used in code generation.
14547 Most of them have both positive and negative forms; the negative form
14548 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14549 one of the forms is listed---the one which is not the default. You
14550 can figure out the other form by either removing @samp{no-} or adding
14554 @item -fbounds-check
14555 @opindex fbounds-check
14556 For front-ends that support it, generate additional code to check that
14557 indices used to access arrays are within the declared range. This is
14558 currently only supported by the Java and Fortran front-ends, where
14559 this option defaults to true and false respectively.
14563 This option generates traps for signed overflow on addition, subtraction,
14564 multiplication operations.
14568 This option instructs the compiler to assume that signed arithmetic
14569 overflow of addition, subtraction and multiplication wraps around
14570 using twos-complement representation. This flag enables some optimizations
14571 and disables others. This option is enabled by default for the Java
14572 front-end, as required by the Java language specification.
14575 @opindex fexceptions
14576 Enable exception handling. Generates extra code needed to propagate
14577 exceptions. For some targets, this implies GCC will generate frame
14578 unwind information for all functions, which can produce significant data
14579 size overhead, although it does not affect execution. If you do not
14580 specify this option, GCC will enable it by default for languages like
14581 C++ which normally require exception handling, and disable it for
14582 languages like C that do not normally require it. However, you may need
14583 to enable this option when compiling C code that needs to interoperate
14584 properly with exception handlers written in C++. You may also wish to
14585 disable this option if you are compiling older C++ programs that don't
14586 use exception handling.
14588 @item -fnon-call-exceptions
14589 @opindex fnon-call-exceptions
14590 Generate code that allows trapping instructions to throw exceptions.
14591 Note that this requires platform-specific runtime support that does
14592 not exist everywhere. Moreover, it only allows @emph{trapping}
14593 instructions to throw exceptions, i.e.@: memory references or floating
14594 point instructions. It does not allow exceptions to be thrown from
14595 arbitrary signal handlers such as @code{SIGALRM}.
14597 @item -funwind-tables
14598 @opindex funwind-tables
14599 Similar to @option{-fexceptions}, except that it will just generate any needed
14600 static data, but will not affect the generated code in any other way.
14601 You will normally not enable this option; instead, a language processor
14602 that needs this handling would enable it on your behalf.
14604 @item -fasynchronous-unwind-tables
14605 @opindex fasynchronous-unwind-tables
14606 Generate unwind table in dwarf2 format, if supported by target machine. The
14607 table is exact at each instruction boundary, so it can be used for stack
14608 unwinding from asynchronous events (such as debugger or garbage collector).
14610 @item -fpcc-struct-return
14611 @opindex fpcc-struct-return
14612 Return ``short'' @code{struct} and @code{union} values in memory like
14613 longer ones, rather than in registers. This convention is less
14614 efficient, but it has the advantage of allowing intercallability between
14615 GCC-compiled files and files compiled with other compilers, particularly
14616 the Portable C Compiler (pcc).
14618 The precise convention for returning structures in memory depends
14619 on the target configuration macros.
14621 Short structures and unions are those whose size and alignment match
14622 that of some integer type.
14624 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14625 switch is not binary compatible with code compiled with the
14626 @option{-freg-struct-return} switch.
14627 Use it to conform to a non-default application binary interface.
14629 @item -freg-struct-return
14630 @opindex freg-struct-return
14631 Return @code{struct} and @code{union} values in registers when possible.
14632 This is more efficient for small structures than
14633 @option{-fpcc-struct-return}.
14635 If you specify neither @option{-fpcc-struct-return} nor
14636 @option{-freg-struct-return}, GCC defaults to whichever convention is
14637 standard for the target. If there is no standard convention, GCC
14638 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14639 the principal compiler. In those cases, we can choose the standard, and
14640 we chose the more efficient register return alternative.
14642 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14643 switch is not binary compatible with code compiled with the
14644 @option{-fpcc-struct-return} switch.
14645 Use it to conform to a non-default application binary interface.
14647 @item -fshort-enums
14648 @opindex fshort-enums
14649 Allocate to an @code{enum} type only as many bytes as it needs for the
14650 declared range of possible values. Specifically, the @code{enum} type
14651 will be equivalent to the smallest integer type which has enough room.
14653 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14654 code that is not binary compatible with code generated without that switch.
14655 Use it to conform to a non-default application binary interface.
14657 @item -fshort-double
14658 @opindex fshort-double
14659 Use the same size for @code{double} as for @code{float}.
14661 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14662 code that is not binary compatible with code generated without that switch.
14663 Use it to conform to a non-default application binary interface.
14665 @item -fshort-wchar
14666 @opindex fshort-wchar
14667 Override the underlying type for @samp{wchar_t} to be @samp{short
14668 unsigned int} instead of the default for the target. This option is
14669 useful for building programs to run under WINE@.
14671 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14672 code that is not binary compatible with code generated without that switch.
14673 Use it to conform to a non-default application binary interface.
14676 @opindex fno-common
14677 In C, allocate even uninitialized global variables in the data section of the
14678 object file, rather than generating them as common blocks. This has the
14679 effect that if the same variable is declared (without @code{extern}) in
14680 two different compilations, you will get an error when you link them.
14681 The only reason this might be useful is if you wish to verify that the
14682 program will work on other systems which always work this way.
14686 Ignore the @samp{#ident} directive.
14688 @item -finhibit-size-directive
14689 @opindex finhibit-size-directive
14690 Don't output a @code{.size} assembler directive, or anything else that
14691 would cause trouble if the function is split in the middle, and the
14692 two halves are placed at locations far apart in memory. This option is
14693 used when compiling @file{crtstuff.c}; you should not need to use it
14696 @item -fverbose-asm
14697 @opindex fverbose-asm
14698 Put extra commentary information in the generated assembly code to
14699 make it more readable. This option is generally only of use to those
14700 who actually need to read the generated assembly code (perhaps while
14701 debugging the compiler itself).
14703 @option{-fno-verbose-asm}, the default, causes the
14704 extra information to be omitted and is useful when comparing two assembler
14707 @item -frecord-gcc-switches
14708 @opindex frecord-gcc-switches
14709 This switch causes the command line that was used to invoke the
14710 compiler to be recorded into the object file that is being created.
14711 This switch is only implemented on some targets and the exact format
14712 of the recording is target and binary file format dependent, but it
14713 usually takes the form of a section containing ASCII text. This
14714 switch is related to the @option{-fverbose-asm} switch, but that
14715 switch only records information in the assembler output file as
14716 comments, so it never reaches the object file.
14720 @cindex global offset table
14722 Generate position-independent code (PIC) suitable for use in a shared
14723 library, if supported for the target machine. Such code accesses all
14724 constant addresses through a global offset table (GOT)@. The dynamic
14725 loader resolves the GOT entries when the program starts (the dynamic
14726 loader is not part of GCC; it is part of the operating system). If
14727 the GOT size for the linked executable exceeds a machine-specific
14728 maximum size, you get an error message from the linker indicating that
14729 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
14730 instead. (These maximums are 8k on the SPARC and 32k
14731 on the m68k and RS/6000. The 386 has no such limit.)
14733 Position-independent code requires special support, and therefore works
14734 only on certain machines. For the 386, GCC supports PIC for System V
14735 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
14736 position-independent.
14738 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14743 If supported for the target machine, emit position-independent code,
14744 suitable for dynamic linking and avoiding any limit on the size of the
14745 global offset table. This option makes a difference on the m68k,
14746 PowerPC and SPARC@.
14748 Position-independent code requires special support, and therefore works
14749 only on certain machines.
14751 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
14758 These options are similar to @option{-fpic} and @option{-fPIC}, but
14759 generated position independent code can be only linked into executables.
14760 Usually these options are used when @option{-pie} GCC option will be
14761 used during linking.
14763 @option{-fpie} and @option{-fPIE} both define the macros
14764 @code{__pie__} and @code{__PIE__}. The macros have the value 1
14765 for @option{-fpie} and 2 for @option{-fPIE}.
14767 @item -fno-jump-tables
14768 @opindex fno-jump-tables
14769 Do not use jump tables for switch statements even where it would be
14770 more efficient than other code generation strategies. This option is
14771 of use in conjunction with @option{-fpic} or @option{-fPIC} for
14772 building code which forms part of a dynamic linker and cannot
14773 reference the address of a jump table. On some targets, jump tables
14774 do not require a GOT and this option is not needed.
14776 @item -ffixed-@var{reg}
14778 Treat the register named @var{reg} as a fixed register; generated code
14779 should never refer to it (except perhaps as a stack pointer, frame
14780 pointer or in some other fixed role).
14782 @var{reg} must be the name of a register. The register names accepted
14783 are machine-specific and are defined in the @code{REGISTER_NAMES}
14784 macro in the machine description macro file.
14786 This flag does not have a negative form, because it specifies a
14789 @item -fcall-used-@var{reg}
14790 @opindex fcall-used
14791 Treat the register named @var{reg} as an allocable register that is
14792 clobbered by function calls. It may be allocated for temporaries or
14793 variables that do not live across a call. Functions compiled this way
14794 will not save and restore the register @var{reg}.
14796 It is an error to used this flag with the frame pointer or stack pointer.
14797 Use of this flag for other registers that have fixed pervasive roles in
14798 the machine's execution model will produce disastrous results.
14800 This flag does not have a negative form, because it specifies a
14803 @item -fcall-saved-@var{reg}
14804 @opindex fcall-saved
14805 Treat the register named @var{reg} as an allocable register saved by
14806 functions. It may be allocated even for temporaries or variables that
14807 live across a call. Functions compiled this way will save and restore
14808 the register @var{reg} if they use it.
14810 It is an error to used this flag with the frame pointer or stack pointer.
14811 Use of this flag for other registers that have fixed pervasive roles in
14812 the machine's execution model will produce disastrous results.
14814 A different sort of disaster will result from the use of this flag for
14815 a register in which function values may be returned.
14817 This flag does not have a negative form, because it specifies a
14820 @item -fpack-struct[=@var{n}]
14821 @opindex fpack-struct
14822 Without a value specified, pack all structure members together without
14823 holes. When a value is specified (which must be a small power of two), pack
14824 structure members according to this value, representing the maximum
14825 alignment (that is, objects with default alignment requirements larger than
14826 this will be output potentially unaligned at the next fitting location.
14828 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
14829 code that is not binary compatible with code generated without that switch.
14830 Additionally, it makes the code suboptimal.
14831 Use it to conform to a non-default application binary interface.
14833 @item -finstrument-functions
14834 @opindex finstrument-functions
14835 Generate instrumentation calls for entry and exit to functions. Just
14836 after function entry and just before function exit, the following
14837 profiling functions will be called with the address of the current
14838 function and its call site. (On some platforms,
14839 @code{__builtin_return_address} does not work beyond the current
14840 function, so the call site information may not be available to the
14841 profiling functions otherwise.)
14844 void __cyg_profile_func_enter (void *this_fn,
14846 void __cyg_profile_func_exit (void *this_fn,
14850 The first argument is the address of the start of the current function,
14851 which may be looked up exactly in the symbol table.
14853 This instrumentation is also done for functions expanded inline in other
14854 functions. The profiling calls will indicate where, conceptually, the
14855 inline function is entered and exited. This means that addressable
14856 versions of such functions must be available. If all your uses of a
14857 function are expanded inline, this may mean an additional expansion of
14858 code size. If you use @samp{extern inline} in your C code, an
14859 addressable version of such functions must be provided. (This is
14860 normally the case anyways, but if you get lucky and the optimizer always
14861 expands the functions inline, you might have gotten away without
14862 providing static copies.)
14864 A function may be given the attribute @code{no_instrument_function}, in
14865 which case this instrumentation will not be done. This can be used, for
14866 example, for the profiling functions listed above, high-priority
14867 interrupt routines, and any functions from which the profiling functions
14868 cannot safely be called (perhaps signal handlers, if the profiling
14869 routines generate output or allocate memory).
14871 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
14872 @opindex finstrument-functions-exclude-file-list
14874 Set the list of functions that are excluded from instrumentation (see
14875 the description of @code{-finstrument-functions}). If the file that
14876 contains a function definition matches with one of @var{file}, then
14877 that function is not instrumented. The match is done on substrings:
14878 if the @var{file} parameter is a substring of the file name, it is
14879 considered to be a match.
14882 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
14883 will exclude any inline function defined in files whose pathnames
14884 contain @code{/bits/stl} or @code{include/sys}.
14886 If, for some reason, you want to include letter @code{','} in one of
14887 @var{sym}, write @code{'\,'}. For example,
14888 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
14889 (note the single quote surrounding the option).
14891 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
14892 @opindex finstrument-functions-exclude-function-list
14894 This is similar to @code{-finstrument-functions-exclude-file-list},
14895 but this option sets the list of function names to be excluded from
14896 instrumentation. The function name to be matched is its user-visible
14897 name, such as @code{vector<int> blah(const vector<int> &)}, not the
14898 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
14899 match is done on substrings: if the @var{sym} parameter is a substring
14900 of the function name, it is considered to be a match.
14902 @item -fstack-check
14903 @opindex fstack-check
14904 Generate code to verify that you do not go beyond the boundary of the
14905 stack. You should specify this flag if you are running in an
14906 environment with multiple threads, but only rarely need to specify it in
14907 a single-threaded environment since stack overflow is automatically
14908 detected on nearly all systems if there is only one stack.
14910 Note that this switch does not actually cause checking to be done; the
14911 operating system must do that. The switch causes generation of code
14912 to ensure that the operating system sees the stack being extended.
14914 @item -fstack-limit-register=@var{reg}
14915 @itemx -fstack-limit-symbol=@var{sym}
14916 @itemx -fno-stack-limit
14917 @opindex fstack-limit-register
14918 @opindex fstack-limit-symbol
14919 @opindex fno-stack-limit
14920 Generate code to ensure that the stack does not grow beyond a certain value,
14921 either the value of a register or the address of a symbol. If the stack
14922 would grow beyond the value, a signal is raised. For most targets,
14923 the signal is raised before the stack overruns the boundary, so
14924 it is possible to catch the signal without taking special precautions.
14926 For instance, if the stack starts at absolute address @samp{0x80000000}
14927 and grows downwards, you can use the flags
14928 @option{-fstack-limit-symbol=__stack_limit} and
14929 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
14930 of 128KB@. Note that this may only work with the GNU linker.
14932 @cindex aliasing of parameters
14933 @cindex parameters, aliased
14934 @item -fargument-alias
14935 @itemx -fargument-noalias
14936 @itemx -fargument-noalias-global
14937 @itemx -fargument-noalias-anything
14938 @opindex fargument-alias
14939 @opindex fargument-noalias
14940 @opindex fargument-noalias-global
14941 @opindex fargument-noalias-anything
14942 Specify the possible relationships among parameters and between
14943 parameters and global data.
14945 @option{-fargument-alias} specifies that arguments (parameters) may
14946 alias each other and may alias global storage.@*
14947 @option{-fargument-noalias} specifies that arguments do not alias
14948 each other, but may alias global storage.@*
14949 @option{-fargument-noalias-global} specifies that arguments do not
14950 alias each other and do not alias global storage.
14951 @option{-fargument-noalias-anything} specifies that arguments do not
14952 alias any other storage.
14954 Each language will automatically use whatever option is required by
14955 the language standard. You should not need to use these options yourself.
14957 @item -fleading-underscore
14958 @opindex fleading-underscore
14959 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
14960 change the way C symbols are represented in the object file. One use
14961 is to help link with legacy assembly code.
14963 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
14964 generate code that is not binary compatible with code generated without that
14965 switch. Use it to conform to a non-default application binary interface.
14966 Not all targets provide complete support for this switch.
14968 @item -ftls-model=@var{model}
14969 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
14970 The @var{model} argument should be one of @code{global-dynamic},
14971 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
14973 The default without @option{-fpic} is @code{initial-exec}; with
14974 @option{-fpic} the default is @code{global-dynamic}.
14976 @item -fvisibility=@var{default|internal|hidden|protected}
14977 @opindex fvisibility
14978 Set the default ELF image symbol visibility to the specified option---all
14979 symbols will be marked with this unless overridden within the code.
14980 Using this feature can very substantially improve linking and
14981 load times of shared object libraries, produce more optimized
14982 code, provide near-perfect API export and prevent symbol clashes.
14983 It is @strong{strongly} recommended that you use this in any shared objects
14986 Despite the nomenclature, @code{default} always means public ie;
14987 available to be linked against from outside the shared object.
14988 @code{protected} and @code{internal} are pretty useless in real-world
14989 usage so the only other commonly used option will be @code{hidden}.
14990 The default if @option{-fvisibility} isn't specified is
14991 @code{default}, i.e., make every
14992 symbol public---this causes the same behavior as previous versions of
14995 A good explanation of the benefits offered by ensuring ELF
14996 symbols have the correct visibility is given by ``How To Write
14997 Shared Libraries'' by Ulrich Drepper (which can be found at
14998 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
14999 solution made possible by this option to marking things hidden when
15000 the default is public is to make the default hidden and mark things
15001 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
15002 and @code{__attribute__ ((visibility("default")))} instead of
15003 @code{__declspec(dllexport)} you get almost identical semantics with
15004 identical syntax. This is a great boon to those working with
15005 cross-platform projects.
15007 For those adding visibility support to existing code, you may find
15008 @samp{#pragma GCC visibility} of use. This works by you enclosing
15009 the declarations you wish to set visibility for with (for example)
15010 @samp{#pragma GCC visibility push(hidden)} and
15011 @samp{#pragma GCC visibility pop}.
15012 Bear in mind that symbol visibility should be viewed @strong{as
15013 part of the API interface contract} and thus all new code should
15014 always specify visibility when it is not the default ie; declarations
15015 only for use within the local DSO should @strong{always} be marked explicitly
15016 as hidden as so to avoid PLT indirection overheads---making this
15017 abundantly clear also aids readability and self-documentation of the code.
15018 Note that due to ISO C++ specification requirements, operator new and
15019 operator delete must always be of default visibility.
15021 Be aware that headers from outside your project, in particular system
15022 headers and headers from any other library you use, may not be
15023 expecting to be compiled with visibility other than the default. You
15024 may need to explicitly say @samp{#pragma GCC visibility push(default)}
15025 before including any such headers.
15027 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
15028 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
15029 no modifications. However, this means that calls to @samp{extern}
15030 functions with no explicit visibility will use the PLT, so it is more
15031 effective to use @samp{__attribute ((visibility))} and/or
15032 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
15033 declarations should be treated as hidden.
15035 Note that @samp{-fvisibility} does affect C++ vague linkage
15036 entities. This means that, for instance, an exception class that will
15037 be thrown between DSOs must be explicitly marked with default
15038 visibility so that the @samp{type_info} nodes will be unified between
15041 An overview of these techniques, their benefits and how to use them
15042 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
15048 @node Environment Variables
15049 @section Environment Variables Affecting GCC
15050 @cindex environment variables
15052 @c man begin ENVIRONMENT
15053 This section describes several environment variables that affect how GCC
15054 operates. Some of them work by specifying directories or prefixes to use
15055 when searching for various kinds of files. Some are used to specify other
15056 aspects of the compilation environment.
15058 Note that you can also specify places to search using options such as
15059 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
15060 take precedence over places specified using environment variables, which
15061 in turn take precedence over those specified by the configuration of GCC@.
15062 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
15063 GNU Compiler Collection (GCC) Internals}.
15068 @c @itemx LC_COLLATE
15070 @c @itemx LC_MONETARY
15071 @c @itemx LC_NUMERIC
15076 @c @findex LC_COLLATE
15077 @findex LC_MESSAGES
15078 @c @findex LC_MONETARY
15079 @c @findex LC_NUMERIC
15083 These environment variables control the way that GCC uses
15084 localization information that allow GCC to work with different
15085 national conventions. GCC inspects the locale categories
15086 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
15087 so. These locale categories can be set to any value supported by your
15088 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
15089 Kingdom encoded in UTF-8.
15091 The @env{LC_CTYPE} environment variable specifies character
15092 classification. GCC uses it to determine the character boundaries in
15093 a string; this is needed for some multibyte encodings that contain quote
15094 and escape characters that would otherwise be interpreted as a string
15097 The @env{LC_MESSAGES} environment variable specifies the language to
15098 use in diagnostic messages.
15100 If the @env{LC_ALL} environment variable is set, it overrides the value
15101 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
15102 and @env{LC_MESSAGES} default to the value of the @env{LANG}
15103 environment variable. If none of these variables are set, GCC
15104 defaults to traditional C English behavior.
15108 If @env{TMPDIR} is set, it specifies the directory to use for temporary
15109 files. GCC uses temporary files to hold the output of one stage of
15110 compilation which is to be used as input to the next stage: for example,
15111 the output of the preprocessor, which is the input to the compiler
15114 @item GCC_EXEC_PREFIX
15115 @findex GCC_EXEC_PREFIX
15116 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
15117 names of the subprograms executed by the compiler. No slash is added
15118 when this prefix is combined with the name of a subprogram, but you can
15119 specify a prefix that ends with a slash if you wish.
15121 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
15122 an appropriate prefix to use based on the pathname it was invoked with.
15124 If GCC cannot find the subprogram using the specified prefix, it
15125 tries looking in the usual places for the subprogram.
15127 The default value of @env{GCC_EXEC_PREFIX} is
15128 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
15129 the installed compiler. In many cases @var{prefix} is the value
15130 of @code{prefix} when you ran the @file{configure} script.
15132 Other prefixes specified with @option{-B} take precedence over this prefix.
15134 This prefix is also used for finding files such as @file{crt0.o} that are
15137 In addition, the prefix is used in an unusual way in finding the
15138 directories to search for header files. For each of the standard
15139 directories whose name normally begins with @samp{/usr/local/lib/gcc}
15140 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
15141 replacing that beginning with the specified prefix to produce an
15142 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
15143 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
15144 These alternate directories are searched first; the standard directories
15145 come next. If a standard directory begins with the configured
15146 @var{prefix} then the value of @var{prefix} is replaced by
15147 @env{GCC_EXEC_PREFIX} when looking for header files.
15149 @item COMPILER_PATH
15150 @findex COMPILER_PATH
15151 The value of @env{COMPILER_PATH} is a colon-separated list of
15152 directories, much like @env{PATH}. GCC tries the directories thus
15153 specified when searching for subprograms, if it can't find the
15154 subprograms using @env{GCC_EXEC_PREFIX}.
15157 @findex LIBRARY_PATH
15158 The value of @env{LIBRARY_PATH} is a colon-separated list of
15159 directories, much like @env{PATH}. When configured as a native compiler,
15160 GCC tries the directories thus specified when searching for special
15161 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15162 using GCC also uses these directories when searching for ordinary
15163 libraries for the @option{-l} option (but directories specified with
15164 @option{-L} come first).
15168 @cindex locale definition
15169 This variable is used to pass locale information to the compiler. One way in
15170 which this information is used is to determine the character set to be used
15171 when character literals, string literals and comments are parsed in C and C++.
15172 When the compiler is configured to allow multibyte characters,
15173 the following values for @env{LANG} are recognized:
15177 Recognize JIS characters.
15179 Recognize SJIS characters.
15181 Recognize EUCJP characters.
15184 If @env{LANG} is not defined, or if it has some other value, then the
15185 compiler will use mblen and mbtowc as defined by the default locale to
15186 recognize and translate multibyte characters.
15190 Some additional environments variables affect the behavior of the
15193 @include cppenv.texi
15197 @node Precompiled Headers
15198 @section Using Precompiled Headers
15199 @cindex precompiled headers
15200 @cindex speed of compilation
15202 Often large projects have many header files that are included in every
15203 source file. The time the compiler takes to process these header files
15204 over and over again can account for nearly all of the time required to
15205 build the project. To make builds faster, GCC allows users to
15206 `precompile' a header file; then, if builds can use the precompiled
15207 header file they will be much faster.
15209 To create a precompiled header file, simply compile it as you would any
15210 other file, if necessary using the @option{-x} option to make the driver
15211 treat it as a C or C++ header file. You will probably want to use a
15212 tool like @command{make} to keep the precompiled header up-to-date when
15213 the headers it contains change.
15215 A precompiled header file will be searched for when @code{#include} is
15216 seen in the compilation. As it searches for the included file
15217 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15218 compiler looks for a precompiled header in each directory just before it
15219 looks for the include file in that directory. The name searched for is
15220 the name specified in the @code{#include} with @samp{.gch} appended. If
15221 the precompiled header file can't be used, it is ignored.
15223 For instance, if you have @code{#include "all.h"}, and you have
15224 @file{all.h.gch} in the same directory as @file{all.h}, then the
15225 precompiled header file will be used if possible, and the original
15226 header will be used otherwise.
15228 Alternatively, you might decide to put the precompiled header file in a
15229 directory and use @option{-I} to ensure that directory is searched
15230 before (or instead of) the directory containing the original header.
15231 Then, if you want to check that the precompiled header file is always
15232 used, you can put a file of the same name as the original header in this
15233 directory containing an @code{#error} command.
15235 This also works with @option{-include}. So yet another way to use
15236 precompiled headers, good for projects not designed with precompiled
15237 header files in mind, is to simply take most of the header files used by
15238 a project, include them from another header file, precompile that header
15239 file, and @option{-include} the precompiled header. If the header files
15240 have guards against multiple inclusion, they will be skipped because
15241 they've already been included (in the precompiled header).
15243 If you need to precompile the same header file for different
15244 languages, targets, or compiler options, you can instead make a
15245 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15246 header in the directory, perhaps using @option{-o}. It doesn't matter
15247 what you call the files in the directory, every precompiled header in
15248 the directory will be considered. The first precompiled header
15249 encountered in the directory that is valid for this compilation will
15250 be used; they're searched in no particular order.
15252 There are many other possibilities, limited only by your imagination,
15253 good sense, and the constraints of your build system.
15255 A precompiled header file can be used only when these conditions apply:
15259 Only one precompiled header can be used in a particular compilation.
15262 A precompiled header can't be used once the first C token is seen. You
15263 can have preprocessor directives before a precompiled header; you can
15264 even include a precompiled header from inside another header, so long as
15265 there are no C tokens before the @code{#include}.
15268 The precompiled header file must be produced for the same language as
15269 the current compilation. You can't use a C precompiled header for a C++
15273 The precompiled header file must have been produced by the same compiler
15274 binary as the current compilation is using.
15277 Any macros defined before the precompiled header is included must
15278 either be defined in the same way as when the precompiled header was
15279 generated, or must not affect the precompiled header, which usually
15280 means that they don't appear in the precompiled header at all.
15282 The @option{-D} option is one way to define a macro before a
15283 precompiled header is included; using a @code{#define} can also do it.
15284 There are also some options that define macros implicitly, like
15285 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15288 @item If debugging information is output when using the precompiled
15289 header, using @option{-g} or similar, the same kind of debugging information
15290 must have been output when building the precompiled header. However,
15291 a precompiled header built using @option{-g} can be used in a compilation
15292 when no debugging information is being output.
15294 @item The same @option{-m} options must generally be used when building
15295 and using the precompiled header. @xref{Submodel Options},
15296 for any cases where this rule is relaxed.
15298 @item Each of the following options must be the same when building and using
15299 the precompiled header:
15301 @gccoptlist{-fexceptions -funit-at-a-time}
15304 Some other command-line options starting with @option{-f},
15305 @option{-p}, or @option{-O} must be defined in the same way as when
15306 the precompiled header was generated. At present, it's not clear
15307 which options are safe to change and which are not; the safest choice
15308 is to use exactly the same options when generating and using the
15309 precompiled header. The following are known to be safe:
15311 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15312 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15313 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15318 For all of these except the last, the compiler will automatically
15319 ignore the precompiled header if the conditions aren't met. If you
15320 find an option combination that doesn't work and doesn't cause the
15321 precompiled header to be ignored, please consider filing a bug report,
15324 If you do use differing options when generating and using the
15325 precompiled header, the actual behavior will be a mixture of the
15326 behavior for the options. For instance, if you use @option{-g} to
15327 generate the precompiled header but not when using it, you may or may
15328 not get debugging information for routines in the precompiled header.
15330 @node Running Protoize
15331 @section Running Protoize
15333 The program @code{protoize} is an optional part of GCC@. You can use
15334 it to add prototypes to a program, thus converting the program to ISO
15335 C in one respect. The companion program @code{unprotoize} does the
15336 reverse: it removes argument types from any prototypes that are found.
15338 When you run these programs, you must specify a set of source files as
15339 command line arguments. The conversion programs start out by compiling
15340 these files to see what functions they define. The information gathered
15341 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15343 After scanning comes actual conversion. The specified files are all
15344 eligible to be converted; any files they include (whether sources or
15345 just headers) are eligible as well.
15347 But not all the eligible files are converted. By default,
15348 @code{protoize} and @code{unprotoize} convert only source and header
15349 files in the current directory. You can specify additional directories
15350 whose files should be converted with the @option{-d @var{directory}}
15351 option. You can also specify particular files to exclude with the
15352 @option{-x @var{file}} option. A file is converted if it is eligible, its
15353 directory name matches one of the specified directory names, and its
15354 name within the directory has not been excluded.
15356 Basic conversion with @code{protoize} consists of rewriting most
15357 function definitions and function declarations to specify the types of
15358 the arguments. The only ones not rewritten are those for varargs
15361 @code{protoize} optionally inserts prototype declarations at the
15362 beginning of the source file, to make them available for any calls that
15363 precede the function's definition. Or it can insert prototype
15364 declarations with block scope in the blocks where undeclared functions
15367 Basic conversion with @code{unprotoize} consists of rewriting most
15368 function declarations to remove any argument types, and rewriting
15369 function definitions to the old-style pre-ISO form.
15371 Both conversion programs print a warning for any function declaration or
15372 definition that they can't convert. You can suppress these warnings
15375 The output from @code{protoize} or @code{unprotoize} replaces the
15376 original source file. The original file is renamed to a name ending
15377 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15378 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15379 for DOS) file already exists, then the source file is simply discarded.
15381 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15382 scan the program and collect information about the functions it uses.
15383 So neither of these programs will work until GCC is installed.
15385 Here is a table of the options you can use with @code{protoize} and
15386 @code{unprotoize}. Each option works with both programs unless
15390 @item -B @var{directory}
15391 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15392 usual directory (normally @file{/usr/local/lib}). This file contains
15393 prototype information about standard system functions. This option
15394 applies only to @code{protoize}.
15396 @item -c @var{compilation-options}
15397 Use @var{compilation-options} as the options when running @command{gcc} to
15398 produce the @samp{.X} files. The special option @option{-aux-info} is
15399 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15401 Note that the compilation options must be given as a single argument to
15402 @code{protoize} or @code{unprotoize}. If you want to specify several
15403 @command{gcc} options, you must quote the entire set of compilation options
15404 to make them a single word in the shell.
15406 There are certain @command{gcc} arguments that you cannot use, because they
15407 would produce the wrong kind of output. These include @option{-g},
15408 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15409 the @var{compilation-options}, they are ignored.
15412 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15413 systems) instead of @samp{.c}. This is convenient if you are converting
15414 a C program to C++. This option applies only to @code{protoize}.
15417 Add explicit global declarations. This means inserting explicit
15418 declarations at the beginning of each source file for each function
15419 that is called in the file and was not declared. These declarations
15420 precede the first function definition that contains a call to an
15421 undeclared function. This option applies only to @code{protoize}.
15423 @item -i @var{string}
15424 Indent old-style parameter declarations with the string @var{string}.
15425 This option applies only to @code{protoize}.
15427 @code{unprotoize} converts prototyped function definitions to old-style
15428 function definitions, where the arguments are declared between the
15429 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15430 uses five spaces as the indentation. If you want to indent with just
15431 one space instead, use @option{-i " "}.
15434 Keep the @samp{.X} files. Normally, they are deleted after conversion
15438 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15439 a prototype declaration for each function in each block which calls the
15440 function without any declaration. This option applies only to
15444 Make no real changes. This mode just prints information about the conversions
15445 that would have been done without @option{-n}.
15448 Make no @samp{.save} files. The original files are simply deleted.
15449 Use this option with caution.
15451 @item -p @var{program}
15452 Use the program @var{program} as the compiler. Normally, the name
15453 @file{gcc} is used.
15456 Work quietly. Most warnings are suppressed.
15459 Print the version number, just like @option{-v} for @command{gcc}.
15462 If you need special compiler options to compile one of your program's
15463 source files, then you should generate that file's @samp{.X} file
15464 specially, by running @command{gcc} on that source file with the
15465 appropriate options and the option @option{-aux-info}. Then run
15466 @code{protoize} on the entire set of files. @code{protoize} will use
15467 the existing @samp{.X} file because it is newer than the source file.
15471 gcc -Dfoo=bar file1.c -aux-info file1.X
15476 You need to include the special files along with the rest in the
15477 @code{protoize} command, even though their @samp{.X} files already
15478 exist, because otherwise they won't get converted.
15480 @xref{Protoize Caveats}, for more information on how to use
15481 @code{protoize} successfully.