1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
15 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
151 * Running Protoize:: Automatically adding or removing function prototypes.
157 @section Option Summary
159 Here is a summary of all the options, grouped by type. Explanations are
160 in the following sections.
163 @item Overall Options
164 @xref{Overall Options,,Options Controlling the Kind of Output}.
165 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{]} --target-help @gol
167 --version -wrapper@@@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++ -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 @gol
234 -Wno-deprecated-declarations -Wdisabled-optimization @gol
235 -Wdisallowed-function-list=@var{sym},@var{sym},@dots{} @gol
236 -Wno-div-by-zero -Wempty-body -Wenum-compare -Wno-endif-labels @gol
237 -Werror -Werror=* @gol
238 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
239 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
240 -Wformat-security -Wformat-y2k @gol
241 -Wframe-larger-than=@var{len} -Wignored-qualifiers @gol
242 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
243 -Winit-self -Winline @gol
244 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
245 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
246 -Wlogical-op -Wlong-long @gol
247 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
248 -Wmissing-format-attribute -Wmissing-include-dirs @gol
249 -Wmissing-noreturn -Wno-mudflap @gol
250 -Wno-multichar -Wnonnull -Wno-overflow @gol
251 -Woverlength-strings -Wpacked -Wpadded @gol
252 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
253 -Wredundant-decls @gol
254 -Wreturn-type -Wsequence-point -Wshadow @gol
255 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
256 -Wstrict-aliasing -Wstrict-aliasing=n @gol
257 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
258 -Wswitch -Wswitch-default -Wswitch-enum @gol
259 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
260 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
261 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
262 -Wunused-value -Wunused-variable @gol
263 -Wvariadic-macros -Wvla @gol
264 -Wvolatile-register-var -Wwrite-strings}
266 @item C and Objective-C-only Warning Options
267 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
268 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
269 -Wold-style-declaration -Wold-style-definition @gol
270 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
271 -Wdeclaration-after-statement -Wpointer-sign}
273 @item Debugging Options
274 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
275 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
276 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
277 -fdump-noaddr -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
278 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
279 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
280 -fdump-statistics @gol
282 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
283 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
284 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
286 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
289 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-nrv -fdump-tree-vect @gol
295 -fdump-tree-sink @gol
296 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
298 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
299 -ftree-vectorizer-verbose=@var{n} @gol
300 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
301 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
302 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
303 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
304 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
305 -ftest-coverage -ftime-report -fvar-tracking @gol
306 -g -g@var{level} -gcoff -gdwarf-2 @gol
307 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
308 -fno-merge-debug-strings -fdebug-prefix-map=@var{old}=@var{new} @gol
309 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
310 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
311 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
312 -print-multi-directory -print-multi-lib @gol
313 -print-prog-name=@var{program} -print-search-dirs -Q @gol
314 -print-sysroot -print-sysroot-headers-suffix @gol
317 @item Optimization Options
318 @xref{Optimize Options,,Options that Control Optimization}.
320 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
321 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
322 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
323 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
324 -fcheck-data-deps -fcprop-registers -fcrossjumping -fcse-follow-jumps @gol
325 -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
326 -fdata-sections -fdce -fdce @gol
327 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
328 -fearly-inlining -fexpensive-optimizations -ffast-math @gol
329 -ffinite-math-only -ffloat-store -fforward-propagate @gol
330 -ffunction-sections -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
331 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
332 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
333 -finline-small-functions -fipa-cp -fipa-marix-reorg -fipa-pta @gol
334 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
335 -fipa-type-escape -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
336 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
337 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
338 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
339 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
340 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
341 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
342 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
343 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
344 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
345 -fprofile-dir=@var{path} -fprofile-generate -fprofile-generate=@var{path} @gol
346 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
347 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
348 -freorder-blocks-and-partition -freorder-functions @gol
349 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
350 -frounding-math -frtl-abstract-sequences -fsched2-use-superblocks @gol
351 -fsched2-use-traces -fsched-spec-load -fsched-spec-load-dangerous @gol
352 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
353 -fschedule-insns -fschedule-insns2 -fsection-anchors -fsee @gol
354 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
355 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
356 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
357 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
358 -ftree-copyrename -ftree-dce @gol
359 -ftree-dominator-opts -ftree-dse -ftree-fre -ftree-loop-im @gol
360 -ftree-loop-distribution @gol
361 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
362 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-reassoc @gol
363 -ftree-sink -ftree-sra -ftree-store-ccp -ftree-switch-conversion @gol
364 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
365 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
366 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
367 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
369 --param @var{name}=@var{value}
370 -O -O0 -O1 -O2 -O3 -Os}
372 @item Preprocessor Options
373 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
374 @gccoptlist{-A@var{question}=@var{answer} @gol
375 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
376 -C -dD -dI -dM -dN @gol
377 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
378 -idirafter @var{dir} @gol
379 -include @var{file} -imacros @var{file} @gol
380 -iprefix @var{file} -iwithprefix @var{dir} @gol
381 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
382 -imultilib @var{dir} -isysroot @var{dir} @gol
383 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
384 -P -fworking-directory -remap @gol
385 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
386 -Xpreprocessor @var{option}}
388 @item Assembler Option
389 @xref{Assembler Options,,Passing Options to the Assembler}.
390 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
393 @xref{Link Options,,Options for Linking}.
394 @gccoptlist{@var{object-file-name} -l@var{library} @gol
395 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
396 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
397 -Wl,@var{option} -Xlinker @var{option} @gol
400 @item Directory Options
401 @xref{Directory Options,,Options for Directory Search}.
402 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
403 -specs=@var{file} -I- --sysroot=@var{dir}}
406 @c I wrote this xref this way to avoid overfull hbox. -- rms
407 @xref{Target Options}.
408 @gccoptlist{-V @var{version} -b @var{machine}}
410 @item Machine Dependent Options
411 @xref{Submodel Options,,Hardware Models and Configurations}.
412 @c This list is ordered alphanumerically by subsection name.
413 @c Try and put the significant identifier (CPU or system) first,
414 @c so users have a clue at guessing where the ones they want will be.
417 @gccoptlist{-EB -EL @gol
418 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
419 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
422 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
423 -mabi=@var{name} @gol
424 -mapcs-stack-check -mno-apcs-stack-check @gol
425 -mapcs-float -mno-apcs-float @gol
426 -mapcs-reentrant -mno-apcs-reentrant @gol
427 -msched-prolog -mno-sched-prolog @gol
428 -mlittle-endian -mbig-endian -mwords-little-endian @gol
429 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
430 -mthumb-interwork -mno-thumb-interwork @gol
431 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
432 -mstructure-size-boundary=@var{n} @gol
433 -mabort-on-noreturn @gol
434 -mlong-calls -mno-long-calls @gol
435 -msingle-pic-base -mno-single-pic-base @gol
436 -mpic-register=@var{reg} @gol
437 -mnop-fun-dllimport @gol
438 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
439 -mpoke-function-name @gol
441 -mtpcs-frame -mtpcs-leaf-frame @gol
442 -mcaller-super-interworking -mcallee-super-interworking @gol
446 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
447 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
449 @emph{Blackfin Options}
450 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
451 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
452 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
453 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
454 -mno-id-shared-library -mshared-library-id=@var{n} @gol
455 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
456 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
457 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram}
460 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
461 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
462 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
463 -mstack-align -mdata-align -mconst-align @gol
464 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
465 -melf -maout -melinux -mlinux -sim -sim2 @gol
466 -mmul-bug-workaround -mno-mul-bug-workaround}
469 @gccoptlist{-mmac -mpush-args}
471 @emph{Darwin Options}
472 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
473 -arch_only -bind_at_load -bundle -bundle_loader @gol
474 -client_name -compatibility_version -current_version @gol
476 -dependency-file -dylib_file -dylinker_install_name @gol
477 -dynamic -dynamiclib -exported_symbols_list @gol
478 -filelist -flat_namespace -force_cpusubtype_ALL @gol
479 -force_flat_namespace -headerpad_max_install_names @gol
481 -image_base -init -install_name -keep_private_externs @gol
482 -multi_module -multiply_defined -multiply_defined_unused @gol
483 -noall_load -no_dead_strip_inits_and_terms @gol
484 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
485 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
486 -private_bundle -read_only_relocs -sectalign @gol
487 -sectobjectsymbols -whyload -seg1addr @gol
488 -sectcreate -sectobjectsymbols -sectorder @gol
489 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
490 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
491 -segprot -segs_read_only_addr -segs_read_write_addr @gol
492 -single_module -static -sub_library -sub_umbrella @gol
493 -twolevel_namespace -umbrella -undefined @gol
494 -unexported_symbols_list -weak_reference_mismatches @gol
495 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
496 -mkernel -mone-byte-bool}
498 @emph{DEC Alpha Options}
499 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
500 -mieee -mieee-with-inexact -mieee-conformant @gol
501 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
502 -mtrap-precision=@var{mode} -mbuild-constants @gol
503 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
504 -mbwx -mmax -mfix -mcix @gol
505 -mfloat-vax -mfloat-ieee @gol
506 -mexplicit-relocs -msmall-data -mlarge-data @gol
507 -msmall-text -mlarge-text @gol
508 -mmemory-latency=@var{time}}
510 @emph{DEC Alpha/VMS Options}
511 @gccoptlist{-mvms-return-codes}
514 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
515 -mhard-float -msoft-float @gol
516 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
517 -mdouble -mno-double @gol
518 -mmedia -mno-media -mmuladd -mno-muladd @gol
519 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
520 -mlinked-fp -mlong-calls -malign-labels @gol
521 -mlibrary-pic -macc-4 -macc-8 @gol
522 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
523 -moptimize-membar -mno-optimize-membar @gol
524 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
525 -mvliw-branch -mno-vliw-branch @gol
526 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
527 -mno-nested-cond-exec -mtomcat-stats @gol
531 @emph{GNU/Linux Options}
532 @gccoptlist{-muclibc}
534 @emph{H8/300 Options}
535 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
538 @gccoptlist{-march=@var{architecture-type} @gol
539 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
540 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
541 -mfixed-range=@var{register-range} @gol
542 -mjump-in-delay -mlinker-opt -mlong-calls @gol
543 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
544 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
545 -mno-jump-in-delay -mno-long-load-store @gol
546 -mno-portable-runtime -mno-soft-float @gol
547 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
548 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
549 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
550 -munix=@var{unix-std} -nolibdld -static -threads}
552 @emph{i386 and x86-64 Options}
553 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
554 -mfpmath=@var{unit} @gol
555 -masm=@var{dialect} -mno-fancy-math-387 @gol
556 -mno-fp-ret-in-387 -msoft-float @gol
557 -mno-wide-multiply -mrtd -malign-double @gol
558 -mpreferred-stack-boundary=@var{num}
559 -mincoming-stack-boundary=@var{num}
560 -mcld -mcx16 -msahf -mrecip @gol
561 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 @gol
563 -msse4a -m3dnow -mpopcnt -mabm -msse5 @gol
564 -mthreads -mno-align-stringops -minline-all-stringops @gol
565 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
566 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
567 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
568 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
569 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
570 -mcmodel=@var{code-model} @gol
571 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
572 -mfused-madd -mno-fused-madd}
575 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
576 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
577 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
578 -minline-float-divide-max-throughput @gol
579 -minline-int-divide-min-latency @gol
580 -minline-int-divide-max-throughput @gol
581 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
582 -mno-dwarf2-asm -mearly-stop-bits @gol
583 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
584 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
585 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
586 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
587 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
588 -mno-sched-prefer-non-data-spec-insns @gol
589 -mno-sched-prefer-non-control-spec-insns @gol
590 -mno-sched-count-spec-in-critical-path}
592 @emph{M32R/D Options}
593 @gccoptlist{-m32r2 -m32rx -m32r @gol
595 -malign-loops -mno-align-loops @gol
596 -missue-rate=@var{number} @gol
597 -mbranch-cost=@var{number} @gol
598 -mmodel=@var{code-size-model-type} @gol
599 -msdata=@var{sdata-type} @gol
600 -mno-flush-func -mflush-func=@var{name} @gol
601 -mno-flush-trap -mflush-trap=@var{number} @gol
605 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
607 @emph{M680x0 Options}
608 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
609 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
610 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
611 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
612 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
613 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
614 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
615 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
618 @emph{M68hc1x Options}
619 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
620 -mauto-incdec -minmax -mlong-calls -mshort @gol
621 -msoft-reg-count=@var{count}}
624 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
625 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
626 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
627 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
628 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
631 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
632 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
633 -mips64 -mips64r2 @gol
634 -mips16 -mno-mips16 -mflip-mips16 @gol
635 -minterlink-mips16 -mno-interlink-mips16 @gol
636 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
637 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
638 -mfp32 -mfp64 -mhard-float -msoft-float @gol
639 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
640 -msmartmips -mno-smartmips @gol
641 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
642 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
643 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
644 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
645 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
646 -membedded-data -mno-embedded-data @gol
647 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
648 -mcode-readable=@var{setting} @gol
649 -msplit-addresses -mno-split-addresses @gol
650 -mexplicit-relocs -mno-explicit-relocs @gol
651 -mcheck-zero-division -mno-check-zero-division @gol
652 -mdivide-traps -mdivide-breaks @gol
653 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
654 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
655 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
656 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
657 -mfix-sb1 -mno-fix-sb1 @gol
658 -mflush-func=@var{func} -mno-flush-func @gol
659 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
660 -mfp-exceptions -mno-fp-exceptions @gol
661 -mvr4130-align -mno-vr4130-align}
664 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
665 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
666 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
667 -mno-base-addresses -msingle-exit -mno-single-exit}
669 @emph{MN10300 Options}
670 @gccoptlist{-mmult-bug -mno-mult-bug @gol
671 -mam33 -mno-am33 @gol
672 -mam33-2 -mno-am33-2 @gol
673 -mreturn-pointer-on-d0 @gol
676 @emph{PDP-11 Options}
677 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
678 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
679 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
680 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
681 -mbranch-expensive -mbranch-cheap @gol
682 -msplit -mno-split -munix-asm -mdec-asm}
684 @emph{PowerPC Options}
685 See RS/6000 and PowerPC Options.
687 @emph{RS/6000 and PowerPC Options}
688 @gccoptlist{-mcpu=@var{cpu-type} @gol
689 -mtune=@var{cpu-type} @gol
690 -mpower -mno-power -mpower2 -mno-power2 @gol
691 -mpowerpc -mpowerpc64 -mno-powerpc @gol
692 -maltivec -mno-altivec @gol
693 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
694 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
695 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
696 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
697 -mnew-mnemonics -mold-mnemonics @gol
698 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
699 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
700 -malign-power -malign-natural @gol
701 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
702 -mstring -mno-string -mupdate -mno-update @gol
703 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
704 -mstrict-align -mno-strict-align -mrelocatable @gol
705 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
706 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
707 -mdynamic-no-pic -maltivec -mswdiv @gol
708 -mprioritize-restricted-insns=@var{priority} @gol
709 -msched-costly-dep=@var{dependence_type} @gol
710 -minsert-sched-nops=@var{scheme} @gol
711 -mcall-sysv -mcall-netbsd @gol
712 -maix-struct-return -msvr4-struct-return @gol
713 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
714 -misel -mno-isel @gol
715 -misel=yes -misel=no @gol
717 -mspe=yes -mspe=no @gol
719 -mvrsave -mno-vrsave @gol
720 -mmulhw -mno-mulhw @gol
721 -mdlmzb -mno-dlmzb @gol
722 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
723 -mprototype -mno-prototype @gol
724 -msim -mmvme -mads -myellowknife -memb -msdata @gol
725 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
727 @emph{S/390 and zSeries Options}
728 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
729 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
730 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
731 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
732 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
733 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
734 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
737 @gccoptlist{-meb -mel @gol
741 -mscore5 -mscore5u -mscore7 -mscore7d}
744 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
745 -m4-nofpu -m4-single-only -m4-single -m4 @gol
746 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
747 -m5-64media -m5-64media-nofpu @gol
748 -m5-32media -m5-32media-nofpu @gol
749 -m5-compact -m5-compact-nofpu @gol
750 -mb -ml -mdalign -mrelax @gol
751 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
752 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
753 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
754 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
755 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
759 @gccoptlist{-mcpu=@var{cpu-type} @gol
760 -mtune=@var{cpu-type} @gol
761 -mcmodel=@var{code-model} @gol
762 -m32 -m64 -mapp-regs -mno-app-regs @gol
763 -mfaster-structs -mno-faster-structs @gol
764 -mfpu -mno-fpu -mhard-float -msoft-float @gol
765 -mhard-quad-float -msoft-quad-float @gol
766 -mimpure-text -mno-impure-text -mlittle-endian @gol
767 -mstack-bias -mno-stack-bias @gol
768 -munaligned-doubles -mno-unaligned-doubles @gol
769 -mv8plus -mno-v8plus -mvis -mno-vis
770 -threads -pthreads -pthread}
773 @gccoptlist{-mwarn-reloc -merror-reloc @gol
774 -msafe-dma -munsafe-dma @gol
776 -msmall-mem -mlarge-mem -mstdmain @gol
777 -mfixed-range=@var{register-range}}
779 @emph{System V Options}
780 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
783 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
784 -mprolog-function -mno-prolog-function -mspace @gol
785 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
786 -mapp-regs -mno-app-regs @gol
787 -mdisable-callt -mno-disable-callt @gol
793 @gccoptlist{-mg -mgnu -munix}
795 @emph{VxWorks Options}
796 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
797 -Xbind-lazy -Xbind-now}
799 @emph{x86-64 Options}
800 See i386 and x86-64 Options.
802 @emph{Xstormy16 Options}
805 @emph{Xtensa Options}
806 @gccoptlist{-mconst16 -mno-const16 @gol
807 -mfused-madd -mno-fused-madd @gol
808 -mserialize-volatile -mno-serialize-volatile @gol
809 -mtext-section-literals -mno-text-section-literals @gol
810 -mtarget-align -mno-target-align @gol
811 -mlongcalls -mno-longcalls}
813 @emph{zSeries Options}
814 See S/390 and zSeries Options.
816 @item Code Generation Options
817 @xref{Code Gen Options,,Options for Code Generation Conventions}.
818 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
819 -ffixed-@var{reg} -fexceptions @gol
820 -fnon-call-exceptions -funwind-tables @gol
821 -fasynchronous-unwind-tables @gol
822 -finhibit-size-directive -finstrument-functions @gol
823 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
824 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
825 -fno-common -fno-ident @gol
826 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
827 -fno-jump-tables @gol
828 -frecord-gcc-switches @gol
829 -freg-struct-return -fshort-enums @gol
830 -fshort-double -fshort-wchar @gol
831 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
832 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
833 -fno-stack-limit -fargument-alias -fargument-noalias @gol
834 -fargument-noalias-global -fargument-noalias-anything @gol
835 -fleading-underscore -ftls-model=@var{model} @gol
836 -ftrapv -fwrapv -fbounds-check @gol
841 * Overall Options:: Controlling the kind of output:
842 an executable, object files, assembler files,
843 or preprocessed source.
844 * C Dialect Options:: Controlling the variant of C language compiled.
845 * C++ Dialect Options:: Variations on C++.
846 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
848 * Language Independent Options:: Controlling how diagnostics should be
850 * Warning Options:: How picky should the compiler be?
851 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
852 * Optimize Options:: How much optimization?
853 * Preprocessor Options:: Controlling header files and macro definitions.
854 Also, getting dependency information for Make.
855 * Assembler Options:: Passing options to the assembler.
856 * Link Options:: Specifying libraries and so on.
857 * Directory Options:: Where to find header files and libraries.
858 Where to find the compiler executable files.
859 * Spec Files:: How to pass switches to sub-processes.
860 * Target Options:: Running a cross-compiler, or an old version of GCC.
863 @node Overall Options
864 @section Options Controlling the Kind of Output
866 Compilation can involve up to four stages: preprocessing, compilation
867 proper, assembly and linking, always in that order. GCC is capable of
868 preprocessing and compiling several files either into several
869 assembler input files, or into one assembler input file; then each
870 assembler input file produces an object file, and linking combines all
871 the object files (those newly compiled, and those specified as input)
872 into an executable file.
874 @cindex file name suffix
875 For any given input file, the file name suffix determines what kind of
880 C source code which must be preprocessed.
883 C source code which should not be preprocessed.
886 C++ source code which should not be preprocessed.
889 Objective-C source code. Note that you must link with the @file{libobjc}
890 library to make an Objective-C program work.
893 Objective-C source code which should not be preprocessed.
897 Objective-C++ source code. Note that you must link with the @file{libobjc}
898 library to make an Objective-C++ program work. Note that @samp{.M} refers
899 to a literal capital M@.
902 Objective-C++ source code which should not be preprocessed.
905 C, C++, Objective-C or Objective-C++ header file to be turned into a
910 @itemx @var{file}.cxx
911 @itemx @var{file}.cpp
912 @itemx @var{file}.CPP
913 @itemx @var{file}.c++
915 C++ source code which must be preprocessed. Note that in @samp{.cxx},
916 the last two letters must both be literally @samp{x}. Likewise,
917 @samp{.C} refers to a literal capital C@.
921 Objective-C++ source code which must be preprocessed.
924 Objective-C++ source code which should not be preprocessed.
929 @itemx @var{file}.hxx
930 @itemx @var{file}.hpp
931 @itemx @var{file}.HPP
932 @itemx @var{file}.h++
933 @itemx @var{file}.tcc
934 C++ header file to be turned into a precompiled header.
937 @itemx @var{file}.for
938 @itemx @var{file}.ftn
939 Fixed form Fortran source code which should not be preprocessed.
942 @itemx @var{file}.FOR
943 @itemx @var{file}.fpp
944 @itemx @var{file}.FPP
945 @itemx @var{file}.FTN
946 Fixed form Fortran source code which must be preprocessed (with the traditional
950 @itemx @var{file}.f95
951 @itemx @var{file}.f03
952 @itemx @var{file}.f08
953 Free form Fortran source code which should not be preprocessed.
956 @itemx @var{file}.F95
957 @itemx @var{file}.F03
958 @itemx @var{file}.F08
959 Free form Fortran source code which must be preprocessed (with the
960 traditional preprocessor).
962 @c FIXME: Descriptions of Java file types.
969 Ada source code file which contains a library unit declaration (a
970 declaration of a package, subprogram, or generic, or a generic
971 instantiation), or a library unit renaming declaration (a package,
972 generic, or subprogram renaming declaration). Such files are also
976 Ada source code file containing a library unit body (a subprogram or
977 package body). Such files are also called @dfn{bodies}.
979 @c GCC also knows about some suffixes for languages not yet included:
991 Assembler code which must be preprocessed.
994 An object file to be fed straight into linking.
995 Any file name with no recognized suffix is treated this way.
999 You can specify the input language explicitly with the @option{-x} option:
1002 @item -x @var{language}
1003 Specify explicitly the @var{language} for the following input files
1004 (rather than letting the compiler choose a default based on the file
1005 name suffix). This option applies to all following input files until
1006 the next @option{-x} option. Possible values for @var{language} are:
1008 c c-header c-cpp-output
1009 c++ c++-header c++-cpp-output
1010 objective-c objective-c-header objective-c-cpp-output
1011 objective-c++ objective-c++-header objective-c++-cpp-output
1012 assembler assembler-with-cpp
1014 f77 f77-cpp-input f95 f95-cpp-input
1019 Turn off any specification of a language, so that subsequent files are
1020 handled according to their file name suffixes (as they are if @option{-x}
1021 has not been used at all).
1023 @item -pass-exit-codes
1024 @opindex pass-exit-codes
1025 Normally the @command{gcc} program will exit with the code of 1 if any
1026 phase of the compiler returns a non-success return code. If you specify
1027 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1028 numerically highest error produced by any phase that returned an error
1029 indication. The C, C++, and Fortran frontends return 4, if an internal
1030 compiler error is encountered.
1033 If you only want some of the stages of compilation, you can use
1034 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1035 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1036 @command{gcc} is to stop. Note that some combinations (for example,
1037 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1042 Compile or assemble the source files, but do not link. The linking
1043 stage simply is not done. The ultimate output is in the form of an
1044 object file for each source file.
1046 By default, the object file name for a source file is made by replacing
1047 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1049 Unrecognized input files, not requiring compilation or assembly, are
1054 Stop after the stage of compilation proper; do not assemble. The output
1055 is in the form of an assembler code file for each non-assembler input
1058 By default, the assembler file name for a source file is made by
1059 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1061 Input files that don't require compilation are ignored.
1065 Stop after the preprocessing stage; do not run the compiler proper. The
1066 output is in the form of preprocessed source code, which is sent to the
1069 Input files which don't require preprocessing are ignored.
1071 @cindex output file option
1074 Place output in file @var{file}. This applies regardless to whatever
1075 sort of output is being produced, whether it be an executable file,
1076 an object file, an assembler file or preprocessed C code.
1078 If @option{-o} is not specified, the default is to put an executable
1079 file in @file{a.out}, the object file for
1080 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1081 assembler file in @file{@var{source}.s}, a precompiled header file in
1082 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1087 Print (on standard error output) the commands executed to run the stages
1088 of compilation. Also print the version number of the compiler driver
1089 program and of the preprocessor and the compiler proper.
1093 Like @option{-v} except the commands are not executed and all command
1094 arguments are quoted. This is useful for shell scripts to capture the
1095 driver-generated command lines.
1099 Use pipes rather than temporary files for communication between the
1100 various stages of compilation. This fails to work on some systems where
1101 the assembler is unable to read from a pipe; but the GNU assembler has
1106 If you are compiling multiple source files, this option tells the driver
1107 to pass all the source files to the compiler at once (for those
1108 languages for which the compiler can handle this). This will allow
1109 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1110 language for which this is supported is C@. If you pass source files for
1111 multiple languages to the driver, using this option, the driver will invoke
1112 the compiler(s) that support IMA once each, passing each compiler all the
1113 source files appropriate for it. For those languages that do not support
1114 IMA this option will be ignored, and the compiler will be invoked once for
1115 each source file in that language. If you use this option in conjunction
1116 with @option{-save-temps}, the compiler will generate multiple
1118 (one for each source file), but only one (combined) @file{.o} or
1123 Print (on the standard output) a description of the command line options
1124 understood by @command{gcc}. If the @option{-v} option is also specified
1125 then @option{--help} will also be passed on to the various processes
1126 invoked by @command{gcc}, so that they can display the command line options
1127 they accept. If the @option{-Wextra} option has also been specified
1128 (prior to the @option{--help} option), then command line options which
1129 have no documentation associated with them will also be displayed.
1132 @opindex target-help
1133 Print (on the standard output) a description of target-specific command
1134 line options for each tool. For some targets extra target-specific
1135 information may also be printed.
1137 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1138 Print (on the standard output) a description of the command line
1139 options understood by the compiler that fit into a specific class.
1140 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1141 @samp{params}, or @var{language}:
1144 @item @samp{optimizers}
1145 This will display all of the optimization options supported by the
1148 @item @samp{warnings}
1149 This will display all of the options controlling warning messages
1150 produced by the compiler.
1153 This will display target-specific options. Unlike the
1154 @option{--target-help} option however, target-specific options of the
1155 linker and assembler will not be displayed. This is because those
1156 tools do not currently support the extended @option{--help=} syntax.
1159 This will display the values recognized by the @option{--param}
1162 @item @var{language}
1163 This will display the options supported for @var{language}, where
1164 @var{language} is the name of one of the languages supported in this
1168 This will display the options that are common to all languages.
1171 It is possible to further refine the output of the @option{--help=}
1172 option by adding a comma separated list of qualifiers after the
1173 class. These can be any from the following list:
1176 @item @samp{undocumented}
1177 Display only those options which are undocumented.
1180 Display options which take an argument that appears after an equal
1181 sign in the same continuous piece of text, such as:
1182 @samp{--help=target}.
1184 @item @samp{separate}
1185 Display options which take an argument that appears as a separate word
1186 following the original option, such as: @samp{-o output-file}.
1189 Thus for example to display all the undocumented target-specific
1190 switches supported by the compiler the following can be used:
1193 --help=target,undocumented
1196 The sense of a qualifier can be inverted by prefixing it with the
1197 @var{^} character, so for example to display all binary warning
1198 options (i.e., ones that are either on or off and that do not take an
1199 argument), which have a description the following can be used:
1202 --help=warnings,^joined,^undocumented
1205 A class can also be used as a qualifier, although this usually
1206 restricts the output by so much that there is nothing to display. One
1207 case where it does work however is when one of the classes is
1208 @var{target}. So for example to display all the target-specific
1209 optimization options the following can be used:
1212 --help=target,optimizers
1215 The @option{--help=} option can be repeated on the command line. Each
1216 successive use will display its requested class of options, skipping
1217 those that have already been displayed.
1219 If the @option{-Q} option appears on the command line before the
1220 @option{--help=} option, then the descriptive text displayed by
1221 @option{--help=} is changed. Instead of describing the displayed
1222 options, an indication is given as to whether the option is enabled,
1223 disabled or set to a specific value (assuming that the compiler
1224 knows this at the point where the @option{--help=} option is used).
1226 Here is a truncated example from the ARM port of @command{gcc}:
1229 % gcc -Q -mabi=2 --help=target -c
1230 The following options are target specific:
1232 -mabort-on-noreturn [disabled]
1236 The output is sensitive to the effects of previous command line
1237 options, so for example it is possible to find out which optimizations
1238 are enabled at @option{-O2} by using:
1241 -O2 --help=optimizers
1244 Alternatively you can discover which binary optimizations are enabled
1245 by @option{-O3} by using:
1248 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1249 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1250 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1255 Display the version number and copyrights of the invoked GCC@.
1259 Invoke all subcommands under a wrapper program. It takes a single
1260 comma separated list as an argument, which will be used to invoke
1264 gcc -c t.c -wrapper gdb,--args
1267 This will invoke all subprograms of gcc under "gdb --args",
1268 thus cc1 invocation will be "gdb --args cc1 ...".
1270 @include @value{srcdir}/../libiberty/at-file.texi
1274 @section Compiling C++ Programs
1276 @cindex suffixes for C++ source
1277 @cindex C++ source file suffixes
1278 C++ source files conventionally use one of the suffixes @samp{.C},
1279 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1280 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1281 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1282 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1283 files with these names and compiles them as C++ programs even if you
1284 call the compiler the same way as for compiling C programs (usually
1285 with the name @command{gcc}).
1289 However, the use of @command{gcc} does not add the C++ library.
1290 @command{g++} is a program that calls GCC and treats @samp{.c},
1291 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1292 files unless @option{-x} is used, and automatically specifies linking
1293 against the C++ library. This program is also useful when
1294 precompiling a C header file with a @samp{.h} extension for use in C++
1295 compilations. On many systems, @command{g++} is also installed with
1296 the name @command{c++}.
1298 @cindex invoking @command{g++}
1299 When you compile C++ programs, you may specify many of the same
1300 command-line options that you use for compiling programs in any
1301 language; or command-line options meaningful for C and related
1302 languages; or options that are meaningful only for C++ programs.
1303 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1304 explanations of options for languages related to C@.
1305 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1306 explanations of options that are meaningful only for C++ programs.
1308 @node C Dialect Options
1309 @section Options Controlling C Dialect
1310 @cindex dialect options
1311 @cindex language dialect options
1312 @cindex options, dialect
1314 The following options control the dialect of C (or languages derived
1315 from C, such as C++, Objective-C and Objective-C++) that the compiler
1319 @cindex ANSI support
1323 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1324 equivalent to @samp{-std=c++98}.
1326 This turns off certain features of GCC that are incompatible with ISO
1327 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1328 such as the @code{asm} and @code{typeof} keywords, and
1329 predefined macros such as @code{unix} and @code{vax} that identify the
1330 type of system you are using. It also enables the undesirable and
1331 rarely used ISO trigraph feature. For the C compiler,
1332 it disables recognition of C++ style @samp{//} comments as well as
1333 the @code{inline} keyword.
1335 The alternate keywords @code{__asm__}, @code{__extension__},
1336 @code{__inline__} and @code{__typeof__} continue to work despite
1337 @option{-ansi}. You would not want to use them in an ISO C program, of
1338 course, but it is useful to put them in header files that might be included
1339 in compilations done with @option{-ansi}. Alternate predefined macros
1340 such as @code{__unix__} and @code{__vax__} are also available, with or
1341 without @option{-ansi}.
1343 The @option{-ansi} option does not cause non-ISO programs to be
1344 rejected gratuitously. For that, @option{-pedantic} is required in
1345 addition to @option{-ansi}. @xref{Warning Options}.
1347 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1348 option is used. Some header files may notice this macro and refrain
1349 from declaring certain functions or defining certain macros that the
1350 ISO standard doesn't call for; this is to avoid interfering with any
1351 programs that might use these names for other things.
1353 Functions that would normally be built in but do not have semantics
1354 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1355 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1356 built-in functions provided by GCC}, for details of the functions
1361 Determine the language standard. @xref{Standards,,Language Standards
1362 Supported by GCC}, for details of these standard versions. This option
1363 is currently only supported when compiling C or C++.
1365 The compiler can accept several base standards, such as @samp{c89} or
1366 @samp{c++98}, and GNU dialects of those standards, such as
1367 @samp{gnu89} or @samp{gnu++98}. By specifing a base standard, the
1368 compiler will accept all programs following that standard and those
1369 using GNU extensions that do not contradict it. For example,
1370 @samp{-std=c89} turns off certain features of GCC that are
1371 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1372 keywords, but not other GNU extensions that do not have a meaning in
1373 ISO C90, such as omitting the middle term of a @code{?:}
1374 expression. On the other hand, by specifing a GNU dialect of a
1375 standard, all features the compiler support are enabled, even when
1376 those features change the meaning of the base standard and some
1377 strict-conforming programs may be rejected. The particular standard
1378 is used by @option{-pedantic} to identify which features are GNU
1379 extensions given that version of the standard. For example
1380 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1381 comments, while @samp{-std=gnu99 -pedantic} would not.
1383 A value for this option must be provided; possible values are
1388 Support all ISO C90 programs (certain GNU extensions that conflict
1389 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1391 @item iso9899:199409
1392 ISO C90 as modified in amendment 1.
1398 ISO C99. Note that this standard is not yet fully supported; see
1399 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1400 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1403 GNU dialect of ISO C90 (including some C99 features). This
1404 is the default for C code.
1408 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1409 this will become the default. The name @samp{gnu9x} is deprecated.
1412 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1416 GNU dialect of @option{-std=c++98}. This is the default for
1420 The working draft of the upcoming ISO C++0x standard. This option
1421 enables experimental features that are likely to be included in
1422 C++0x. The working draft is constantly changing, and any feature that is
1423 enabled by this flag may be removed from future versions of GCC if it is
1424 not part of the C++0x standard.
1427 GNU dialect of @option{-std=c++0x}. This option enables
1428 experimental features that may be removed in future versions of GCC.
1431 @item -fgnu89-inline
1432 @opindex fgnu89-inline
1433 The option @option{-fgnu89-inline} tells GCC to use the traditional
1434 GNU semantics for @code{inline} functions when in C99 mode.
1435 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1436 is accepted and ignored by GCC versions 4.1.3 up to but not including
1437 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1438 C99 mode. Using this option is roughly equivalent to adding the
1439 @code{gnu_inline} function attribute to all inline functions
1440 (@pxref{Function Attributes}).
1442 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1443 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1444 specifies the default behavior). This option was first supported in
1445 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1447 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1448 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1449 in effect for @code{inline} functions. @xref{Common Predefined
1450 Macros,,,cpp,The C Preprocessor}.
1452 @item -aux-info @var{filename}
1454 Output to the given filename prototyped declarations for all functions
1455 declared and/or defined in a translation unit, including those in header
1456 files. This option is silently ignored in any language other than C@.
1458 Besides declarations, the file indicates, in comments, the origin of
1459 each declaration (source file and line), whether the declaration was
1460 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1461 @samp{O} for old, respectively, in the first character after the line
1462 number and the colon), and whether it came from a declaration or a
1463 definition (@samp{C} or @samp{F}, respectively, in the following
1464 character). In the case of function definitions, a K&R-style list of
1465 arguments followed by their declarations is also provided, inside
1466 comments, after the declaration.
1470 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1471 keyword, so that code can use these words as identifiers. You can use
1472 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1473 instead. @option{-ansi} implies @option{-fno-asm}.
1475 In C++, this switch only affects the @code{typeof} keyword, since
1476 @code{asm} and @code{inline} are standard keywords. You may want to
1477 use the @option{-fno-gnu-keywords} flag instead, which has the same
1478 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1479 switch only affects the @code{asm} and @code{typeof} keywords, since
1480 @code{inline} is a standard keyword in ISO C99.
1483 @itemx -fno-builtin-@var{function}
1484 @opindex fno-builtin
1485 @cindex built-in functions
1486 Don't recognize built-in functions that do not begin with
1487 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1488 functions provided by GCC}, for details of the functions affected,
1489 including those which are not built-in functions when @option{-ansi} or
1490 @option{-std} options for strict ISO C conformance are used because they
1491 do not have an ISO standard meaning.
1493 GCC normally generates special code to handle certain built-in functions
1494 more efficiently; for instance, calls to @code{alloca} may become single
1495 instructions that adjust the stack directly, and calls to @code{memcpy}
1496 may become inline copy loops. The resulting code is often both smaller
1497 and faster, but since the function calls no longer appear as such, you
1498 cannot set a breakpoint on those calls, nor can you change the behavior
1499 of the functions by linking with a different library. In addition,
1500 when a function is recognized as a built-in function, GCC may use
1501 information about that function to warn about problems with calls to
1502 that function, or to generate more efficient code, even if the
1503 resulting code still contains calls to that function. For example,
1504 warnings are given with @option{-Wformat} for bad calls to
1505 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1506 known not to modify global memory.
1508 With the @option{-fno-builtin-@var{function}} option
1509 only the built-in function @var{function} is
1510 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1511 function is named this is not built-in in this version of GCC, this
1512 option is ignored. There is no corresponding
1513 @option{-fbuiltin-@var{function}} option; if you wish to enable
1514 built-in functions selectively when using @option{-fno-builtin} or
1515 @option{-ffreestanding}, you may define macros such as:
1518 #define abs(n) __builtin_abs ((n))
1519 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1524 @cindex hosted environment
1526 Assert that compilation takes place in a hosted environment. This implies
1527 @option{-fbuiltin}. A hosted environment is one in which the
1528 entire standard library is available, and in which @code{main} has a return
1529 type of @code{int}. Examples are nearly everything except a kernel.
1530 This is equivalent to @option{-fno-freestanding}.
1532 @item -ffreestanding
1533 @opindex ffreestanding
1534 @cindex hosted environment
1536 Assert that compilation takes place in a freestanding environment. This
1537 implies @option{-fno-builtin}. A freestanding environment
1538 is one in which the standard library may not exist, and program startup may
1539 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1540 This is equivalent to @option{-fno-hosted}.
1542 @xref{Standards,,Language Standards Supported by GCC}, for details of
1543 freestanding and hosted environments.
1547 @cindex openmp parallel
1548 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1549 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1550 compiler generates parallel code according to the OpenMP Application
1551 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}. This option
1552 implies @option{-pthread}, and thus is only supported on targets that
1553 have support for @option{-pthread}.
1555 @item -fms-extensions
1556 @opindex fms-extensions
1557 Accept some non-standard constructs used in Microsoft header files.
1559 Some cases of unnamed fields in structures and unions are only
1560 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1561 fields within structs/unions}, for details.
1565 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1566 options for strict ISO C conformance) implies @option{-trigraphs}.
1568 @item -no-integrated-cpp
1569 @opindex no-integrated-cpp
1570 Performs a compilation in two passes: preprocessing and compiling. This
1571 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1572 @option{-B} option. The user supplied compilation step can then add in
1573 an additional preprocessing step after normal preprocessing but before
1574 compiling. The default is to use the integrated cpp (internal cpp)
1576 The semantics of this option will change if "cc1", "cc1plus", and
1577 "cc1obj" are merged.
1579 @cindex traditional C language
1580 @cindex C language, traditional
1582 @itemx -traditional-cpp
1583 @opindex traditional-cpp
1584 @opindex traditional
1585 Formerly, these options caused GCC to attempt to emulate a pre-standard
1586 C compiler. They are now only supported with the @option{-E} switch.
1587 The preprocessor continues to support a pre-standard mode. See the GNU
1588 CPP manual for details.
1590 @item -fcond-mismatch
1591 @opindex fcond-mismatch
1592 Allow conditional expressions with mismatched types in the second and
1593 third arguments. The value of such an expression is void. This option
1594 is not supported for C++.
1596 @item -flax-vector-conversions
1597 @opindex flax-vector-conversions
1598 Allow implicit conversions between vectors with differing numbers of
1599 elements and/or incompatible element types. This option should not be
1602 @item -funsigned-char
1603 @opindex funsigned-char
1604 Let the type @code{char} be unsigned, like @code{unsigned char}.
1606 Each kind of machine has a default for what @code{char} should
1607 be. It is either like @code{unsigned char} by default or like
1608 @code{signed char} by default.
1610 Ideally, a portable program should always use @code{signed char} or
1611 @code{unsigned char} when it depends on the signedness of an object.
1612 But many programs have been written to use plain @code{char} and
1613 expect it to be signed, or expect it to be unsigned, depending on the
1614 machines they were written for. This option, and its inverse, let you
1615 make such a program work with the opposite default.
1617 The type @code{char} is always a distinct type from each of
1618 @code{signed char} or @code{unsigned char}, even though its behavior
1619 is always just like one of those two.
1622 @opindex fsigned-char
1623 Let the type @code{char} be signed, like @code{signed char}.
1625 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1626 the negative form of @option{-funsigned-char}. Likewise, the option
1627 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1629 @item -fsigned-bitfields
1630 @itemx -funsigned-bitfields
1631 @itemx -fno-signed-bitfields
1632 @itemx -fno-unsigned-bitfields
1633 @opindex fsigned-bitfields
1634 @opindex funsigned-bitfields
1635 @opindex fno-signed-bitfields
1636 @opindex fno-unsigned-bitfields
1637 These options control whether a bit-field is signed or unsigned, when the
1638 declaration does not use either @code{signed} or @code{unsigned}. By
1639 default, such a bit-field is signed, because this is consistent: the
1640 basic integer types such as @code{int} are signed types.
1643 @node C++ Dialect Options
1644 @section Options Controlling C++ Dialect
1646 @cindex compiler options, C++
1647 @cindex C++ options, command line
1648 @cindex options, C++
1649 This section describes the command-line options that are only meaningful
1650 for C++ programs; but you can also use most of the GNU compiler options
1651 regardless of what language your program is in. For example, you
1652 might compile a file @code{firstClass.C} like this:
1655 g++ -g -frepo -O -c firstClass.C
1659 In this example, only @option{-frepo} is an option meant
1660 only for C++ programs; you can use the other options with any
1661 language supported by GCC@.
1663 Here is a list of options that are @emph{only} for compiling C++ programs:
1667 @item -fabi-version=@var{n}
1668 @opindex fabi-version
1669 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1670 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1671 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1672 the version that conforms most closely to the C++ ABI specification.
1673 Therefore, the ABI obtained using version 0 will change as ABI bugs
1676 The default is version 2.
1678 @item -fno-access-control
1679 @opindex fno-access-control
1680 Turn off all access checking. This switch is mainly useful for working
1681 around bugs in the access control code.
1685 Check that the pointer returned by @code{operator new} is non-null
1686 before attempting to modify the storage allocated. This check is
1687 normally unnecessary because the C++ standard specifies that
1688 @code{operator new} will only return @code{0} if it is declared
1689 @samp{throw()}, in which case the compiler will always check the
1690 return value even without this option. In all other cases, when
1691 @code{operator new} has a non-empty exception specification, memory
1692 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1693 @samp{new (nothrow)}.
1695 @item -fconserve-space
1696 @opindex fconserve-space
1697 Put uninitialized or runtime-initialized global variables into the
1698 common segment, as C does. This saves space in the executable at the
1699 cost of not diagnosing duplicate definitions. If you compile with this
1700 flag and your program mysteriously crashes after @code{main()} has
1701 completed, you may have an object that is being destroyed twice because
1702 two definitions were merged.
1704 This option is no longer useful on most targets, now that support has
1705 been added for putting variables into BSS without making them common.
1707 @item -ffriend-injection
1708 @opindex ffriend-injection
1709 Inject friend functions into the enclosing namespace, so that they are
1710 visible outside the scope of the class in which they are declared.
1711 Friend functions were documented to work this way in the old Annotated
1712 C++ Reference Manual, and versions of G++ before 4.1 always worked
1713 that way. However, in ISO C++ a friend function which is not declared
1714 in an enclosing scope can only be found using argument dependent
1715 lookup. This option causes friends to be injected as they were in
1718 This option is for compatibility, and may be removed in a future
1721 @item -fno-elide-constructors
1722 @opindex fno-elide-constructors
1723 The C++ standard allows an implementation to omit creating a temporary
1724 which is only used to initialize another object of the same type.
1725 Specifying this option disables that optimization, and forces G++ to
1726 call the copy constructor in all cases.
1728 @item -fno-enforce-eh-specs
1729 @opindex fno-enforce-eh-specs
1730 Don't generate code to check for violation of exception specifications
1731 at runtime. This option violates the C++ standard, but may be useful
1732 for reducing code size in production builds, much like defining
1733 @samp{NDEBUG}. This does not give user code permission to throw
1734 exceptions in violation of the exception specifications; the compiler
1735 will still optimize based on the specifications, so throwing an
1736 unexpected exception will result in undefined behavior.
1739 @itemx -fno-for-scope
1741 @opindex fno-for-scope
1742 If @option{-ffor-scope} is specified, the scope of variables declared in
1743 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1744 as specified by the C++ standard.
1745 If @option{-fno-for-scope} is specified, the scope of variables declared in
1746 a @i{for-init-statement} extends to the end of the enclosing scope,
1747 as was the case in old versions of G++, and other (traditional)
1748 implementations of C++.
1750 The default if neither flag is given to follow the standard,
1751 but to allow and give a warning for old-style code that would
1752 otherwise be invalid, or have different behavior.
1754 @item -fno-gnu-keywords
1755 @opindex fno-gnu-keywords
1756 Do not recognize @code{typeof} as a keyword, so that code can use this
1757 word as an identifier. You can use the keyword @code{__typeof__} instead.
1758 @option{-ansi} implies @option{-fno-gnu-keywords}.
1760 @item -fno-implicit-templates
1761 @opindex fno-implicit-templates
1762 Never emit code for non-inline templates which are instantiated
1763 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1764 @xref{Template Instantiation}, for more information.
1766 @item -fno-implicit-inline-templates
1767 @opindex fno-implicit-inline-templates
1768 Don't emit code for implicit instantiations of inline templates, either.
1769 The default is to handle inlines differently so that compiles with and
1770 without optimization will need the same set of explicit instantiations.
1772 @item -fno-implement-inlines
1773 @opindex fno-implement-inlines
1774 To save space, do not emit out-of-line copies of inline functions
1775 controlled by @samp{#pragma implementation}. This will cause linker
1776 errors if these functions are not inlined everywhere they are called.
1778 @item -fms-extensions
1779 @opindex fms-extensions
1780 Disable pedantic warnings about constructs used in MFC, such as implicit
1781 int and getting a pointer to member function via non-standard syntax.
1783 @item -fno-nonansi-builtins
1784 @opindex fno-nonansi-builtins
1785 Disable built-in declarations of functions that are not mandated by
1786 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1787 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1789 @item -fno-operator-names
1790 @opindex fno-operator-names
1791 Do not treat the operator name keywords @code{and}, @code{bitand},
1792 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1793 synonyms as keywords.
1795 @item -fno-optional-diags
1796 @opindex fno-optional-diags
1797 Disable diagnostics that the standard says a compiler does not need to
1798 issue. Currently, the only such diagnostic issued by G++ is the one for
1799 a name having multiple meanings within a class.
1802 @opindex fpermissive
1803 Downgrade some diagnostics about nonconformant code from errors to
1804 warnings. Thus, using @option{-fpermissive} will allow some
1805 nonconforming code to compile.
1809 Enable automatic template instantiation at link time. This option also
1810 implies @option{-fno-implicit-templates}. @xref{Template
1811 Instantiation}, for more information.
1815 Disable generation of information about every class with virtual
1816 functions for use by the C++ runtime type identification features
1817 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1818 of the language, you can save some space by using this flag. Note that
1819 exception handling uses the same information, but it will generate it as
1820 needed. The @samp{dynamic_cast} operator can still be used for casts that
1821 do not require runtime type information, i.e.@: casts to @code{void *} or to
1822 unambiguous base classes.
1826 Emit statistics about front-end processing at the end of the compilation.
1827 This information is generally only useful to the G++ development team.
1829 @item -ftemplate-depth-@var{n}
1830 @opindex ftemplate-depth
1831 Set the maximum instantiation depth for template classes to @var{n}.
1832 A limit on the template instantiation depth is needed to detect
1833 endless recursions during template class instantiation. ANSI/ISO C++
1834 conforming programs must not rely on a maximum depth greater than 17.
1836 @item -fno-threadsafe-statics
1837 @opindex fno-threadsafe-statics
1838 Do not emit the extra code to use the routines specified in the C++
1839 ABI for thread-safe initialization of local statics. You can use this
1840 option to reduce code size slightly in code that doesn't need to be
1843 @item -fuse-cxa-atexit
1844 @opindex fuse-cxa-atexit
1845 Register destructors for objects with static storage duration with the
1846 @code{__cxa_atexit} function rather than the @code{atexit} function.
1847 This option is required for fully standards-compliant handling of static
1848 destructors, but will only work if your C library supports
1849 @code{__cxa_atexit}.
1851 @item -fno-use-cxa-get-exception-ptr
1852 @opindex fno-use-cxa-get-exception-ptr
1853 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1854 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1855 if the runtime routine is not available.
1857 @item -fvisibility-inlines-hidden
1858 @opindex fvisibility-inlines-hidden
1859 This switch declares that the user does not attempt to compare
1860 pointers to inline methods where the addresses of the two functions
1861 were taken in different shared objects.
1863 The effect of this is that GCC may, effectively, mark inline methods with
1864 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1865 appear in the export table of a DSO and do not require a PLT indirection
1866 when used within the DSO@. Enabling this option can have a dramatic effect
1867 on load and link times of a DSO as it massively reduces the size of the
1868 dynamic export table when the library makes heavy use of templates.
1870 The behavior of this switch is not quite the same as marking the
1871 methods as hidden directly, because it does not affect static variables
1872 local to the function or cause the compiler to deduce that
1873 the function is defined in only one shared object.
1875 You may mark a method as having a visibility explicitly to negate the
1876 effect of the switch for that method. For example, if you do want to
1877 compare pointers to a particular inline method, you might mark it as
1878 having default visibility. Marking the enclosing class with explicit
1879 visibility will have no effect.
1881 Explicitly instantiated inline methods are unaffected by this option
1882 as their linkage might otherwise cross a shared library boundary.
1883 @xref{Template Instantiation}.
1885 @item -fvisibility-ms-compat
1886 @opindex fvisibility-ms-compat
1887 This flag attempts to use visibility settings to make GCC's C++
1888 linkage model compatible with that of Microsoft Visual Studio.
1890 The flag makes these changes to GCC's linkage model:
1894 It sets the default visibility to @code{hidden}, like
1895 @option{-fvisibility=hidden}.
1898 Types, but not their members, are not hidden by default.
1901 The One Definition Rule is relaxed for types without explicit
1902 visibility specifications which are defined in more than one different
1903 shared object: those declarations are permitted if they would have
1904 been permitted when this option was not used.
1907 In new code it is better to use @option{-fvisibility=hidden} and
1908 export those classes which are intended to be externally visible.
1909 Unfortunately it is possible for code to rely, perhaps accidentally,
1910 on the Visual Studio behavior.
1912 Among the consequences of these changes are that static data members
1913 of the same type with the same name but defined in different shared
1914 objects will be different, so changing one will not change the other;
1915 and that pointers to function members defined in different shared
1916 objects may not compare equal. When this flag is given, it is a
1917 violation of the ODR to define types with the same name differently.
1921 Do not use weak symbol support, even if it is provided by the linker.
1922 By default, G++ will use weak symbols if they are available. This
1923 option exists only for testing, and should not be used by end-users;
1924 it will result in inferior code and has no benefits. This option may
1925 be removed in a future release of G++.
1929 Do not search for header files in the standard directories specific to
1930 C++, but do still search the other standard directories. (This option
1931 is used when building the C++ library.)
1934 In addition, these optimization, warning, and code generation options
1935 have meanings only for C++ programs:
1938 @item -fno-default-inline
1939 @opindex fno-default-inline
1940 Do not assume @samp{inline} for functions defined inside a class scope.
1941 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1942 functions will have linkage like inline functions; they just won't be
1945 @item -Wabi @r{(C++ and Objective-C++ only)}
1948 Warn when G++ generates code that is probably not compatible with the
1949 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1950 all such cases, there are probably some cases that are not warned about,
1951 even though G++ is generating incompatible code. There may also be
1952 cases where warnings are emitted even though the code that is generated
1955 You should rewrite your code to avoid these warnings if you are
1956 concerned about the fact that code generated by G++ may not be binary
1957 compatible with code generated by other compilers.
1959 The known incompatibilities at this point include:
1964 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1965 pack data into the same byte as a base class. For example:
1968 struct A @{ virtual void f(); int f1 : 1; @};
1969 struct B : public A @{ int f2 : 1; @};
1973 In this case, G++ will place @code{B::f2} into the same byte
1974 as@code{A::f1}; other compilers will not. You can avoid this problem
1975 by explicitly padding @code{A} so that its size is a multiple of the
1976 byte size on your platform; that will cause G++ and other compilers to
1977 layout @code{B} identically.
1980 Incorrect handling of tail-padding for virtual bases. G++ does not use
1981 tail padding when laying out virtual bases. For example:
1984 struct A @{ virtual void f(); char c1; @};
1985 struct B @{ B(); char c2; @};
1986 struct C : public A, public virtual B @{@};
1990 In this case, G++ will not place @code{B} into the tail-padding for
1991 @code{A}; other compilers will. You can avoid this problem by
1992 explicitly padding @code{A} so that its size is a multiple of its
1993 alignment (ignoring virtual base classes); that will cause G++ and other
1994 compilers to layout @code{C} identically.
1997 Incorrect handling of bit-fields with declared widths greater than that
1998 of their underlying types, when the bit-fields appear in a union. For
2002 union U @{ int i : 4096; @};
2006 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2007 union too small by the number of bits in an @code{int}.
2010 Empty classes can be placed at incorrect offsets. For example:
2020 struct C : public B, public A @{@};
2024 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2025 it should be placed at offset zero. G++ mistakenly believes that the
2026 @code{A} data member of @code{B} is already at offset zero.
2029 Names of template functions whose types involve @code{typename} or
2030 template template parameters can be mangled incorrectly.
2033 template <typename Q>
2034 void f(typename Q::X) @{@}
2036 template <template <typename> class Q>
2037 void f(typename Q<int>::X) @{@}
2041 Instantiations of these templates may be mangled incorrectly.
2045 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2046 @opindex Wctor-dtor-privacy
2047 @opindex Wno-ctor-dtor-privacy
2048 Warn when a class seems unusable because all the constructors or
2049 destructors in that class are private, and it has neither friends nor
2050 public static member functions.
2052 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2053 @opindex Wnon-virtual-dtor
2054 @opindex Wno-non-virtual-dtor
2055 Warn when a class has virtual functions and accessible non-virtual
2056 destructor, in which case it would be possible but unsafe to delete
2057 an instance of a derived class through a pointer to the base class.
2058 This warning is also enabled if -Weffc++ is specified.
2060 @item -Wreorder @r{(C++ and Objective-C++ only)}
2062 @opindex Wno-reorder
2063 @cindex reordering, warning
2064 @cindex warning for reordering of member initializers
2065 Warn when the order of member initializers given in the code does not
2066 match the order in which they must be executed. For instance:
2072 A(): j (0), i (1) @{ @}
2076 The compiler will rearrange the member initializers for @samp{i}
2077 and @samp{j} to match the declaration order of the members, emitting
2078 a warning to that effect. This warning is enabled by @option{-Wall}.
2081 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2084 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2087 Warn about violations of the following style guidelines from Scott Meyers'
2088 @cite{Effective C++} book:
2092 Item 11: Define a copy constructor and an assignment operator for classes
2093 with dynamically allocated memory.
2096 Item 12: Prefer initialization to assignment in constructors.
2099 Item 14: Make destructors virtual in base classes.
2102 Item 15: Have @code{operator=} return a reference to @code{*this}.
2105 Item 23: Don't try to return a reference when you must return an object.
2109 Also warn about violations of the following style guidelines from
2110 Scott Meyers' @cite{More Effective C++} book:
2114 Item 6: Distinguish between prefix and postfix forms of increment and
2115 decrement operators.
2118 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2122 When selecting this option, be aware that the standard library
2123 headers do not obey all of these guidelines; use @samp{grep -v}
2124 to filter out those warnings.
2126 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2127 @opindex Wstrict-null-sentinel
2128 @opindex Wno-strict-null-sentinel
2129 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2130 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2131 to @code{__null}. Although it is a null pointer constant not a null pointer,
2132 it is guaranteed to of the same size as a pointer. But this use is
2133 not portable across different compilers.
2135 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2136 @opindex Wno-non-template-friend
2137 @opindex Wnon-template-friend
2138 Disable warnings when non-templatized friend functions are declared
2139 within a template. Since the advent of explicit template specification
2140 support in G++, if the name of the friend is an unqualified-id (i.e.,
2141 @samp{friend foo(int)}), the C++ language specification demands that the
2142 friend declare or define an ordinary, nontemplate function. (Section
2143 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2144 could be interpreted as a particular specialization of a templatized
2145 function. Because this non-conforming behavior is no longer the default
2146 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2147 check existing code for potential trouble spots and is on by default.
2148 This new compiler behavior can be turned off with
2149 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2150 but disables the helpful warning.
2152 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2153 @opindex Wold-style-cast
2154 @opindex Wno-old-style-cast
2155 Warn if an old-style (C-style) cast to a non-void type is used within
2156 a C++ program. The new-style casts (@samp{dynamic_cast},
2157 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2158 less vulnerable to unintended effects and much easier to search for.
2160 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2161 @opindex Woverloaded-virtual
2162 @opindex Wno-overloaded-virtual
2163 @cindex overloaded virtual fn, warning
2164 @cindex warning for overloaded virtual fn
2165 Warn when a function declaration hides virtual functions from a
2166 base class. For example, in:
2173 struct B: public A @{
2178 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2186 will fail to compile.
2188 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2189 @opindex Wno-pmf-conversions
2190 @opindex Wpmf-conversions
2191 Disable the diagnostic for converting a bound pointer to member function
2194 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2195 @opindex Wsign-promo
2196 @opindex Wno-sign-promo
2197 Warn when overload resolution chooses a promotion from unsigned or
2198 enumerated type to a signed type, over a conversion to an unsigned type of
2199 the same size. Previous versions of G++ would try to preserve
2200 unsignedness, but the standard mandates the current behavior.
2205 A& operator = (int);
2215 In this example, G++ will synthesize a default @samp{A& operator =
2216 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2219 @node Objective-C and Objective-C++ Dialect Options
2220 @section Options Controlling Objective-C and Objective-C++ Dialects
2222 @cindex compiler options, Objective-C and Objective-C++
2223 @cindex Objective-C and Objective-C++ options, command line
2224 @cindex options, Objective-C and Objective-C++
2225 (NOTE: This manual does not describe the Objective-C and Objective-C++
2226 languages themselves. See @xref{Standards,,Language Standards
2227 Supported by GCC}, for references.)
2229 This section describes the command-line options that are only meaningful
2230 for Objective-C and Objective-C++ programs, but you can also use most of
2231 the language-independent GNU compiler options.
2232 For example, you might compile a file @code{some_class.m} like this:
2235 gcc -g -fgnu-runtime -O -c some_class.m
2239 In this example, @option{-fgnu-runtime} is an option meant only for
2240 Objective-C and Objective-C++ programs; you can use the other options with
2241 any language supported by GCC@.
2243 Note that since Objective-C is an extension of the C language, Objective-C
2244 compilations may also use options specific to the C front-end (e.g.,
2245 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2246 C++-specific options (e.g., @option{-Wabi}).
2248 Here is a list of options that are @emph{only} for compiling Objective-C
2249 and Objective-C++ programs:
2252 @item -fconstant-string-class=@var{class-name}
2253 @opindex fconstant-string-class
2254 Use @var{class-name} as the name of the class to instantiate for each
2255 literal string specified with the syntax @code{@@"@dots{}"}. The default
2256 class name is @code{NXConstantString} if the GNU runtime is being used, and
2257 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2258 @option{-fconstant-cfstrings} option, if also present, will override the
2259 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2260 to be laid out as constant CoreFoundation strings.
2263 @opindex fgnu-runtime
2264 Generate object code compatible with the standard GNU Objective-C
2265 runtime. This is the default for most types of systems.
2267 @item -fnext-runtime
2268 @opindex fnext-runtime
2269 Generate output compatible with the NeXT runtime. This is the default
2270 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2271 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2274 @item -fno-nil-receivers
2275 @opindex fno-nil-receivers
2276 Assume that all Objective-C message dispatches (e.g.,
2277 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2278 is not @code{nil}. This allows for more efficient entry points in the runtime
2279 to be used. Currently, this option is only available in conjunction with
2280 the NeXT runtime on Mac OS X 10.3 and later.
2282 @item -fobjc-call-cxx-cdtors
2283 @opindex fobjc-call-cxx-cdtors
2284 For each Objective-C class, check if any of its instance variables is a
2285 C++ object with a non-trivial default constructor. If so, synthesize a
2286 special @code{- (id) .cxx_construct} instance method that will run
2287 non-trivial default constructors on any such instance variables, in order,
2288 and then return @code{self}. Similarly, check if any instance variable
2289 is a C++ object with a non-trivial destructor, and if so, synthesize a
2290 special @code{- (void) .cxx_destruct} method that will run
2291 all such default destructors, in reverse order.
2293 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2294 thusly generated will only operate on instance variables declared in the
2295 current Objective-C class, and not those inherited from superclasses. It
2296 is the responsibility of the Objective-C runtime to invoke all such methods
2297 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2298 will be invoked by the runtime immediately after a new object
2299 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2300 be invoked immediately before the runtime deallocates an object instance.
2302 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2303 support for invoking the @code{- (id) .cxx_construct} and
2304 @code{- (void) .cxx_destruct} methods.
2306 @item -fobjc-direct-dispatch
2307 @opindex fobjc-direct-dispatch
2308 Allow fast jumps to the message dispatcher. On Darwin this is
2309 accomplished via the comm page.
2311 @item -fobjc-exceptions
2312 @opindex fobjc-exceptions
2313 Enable syntactic support for structured exception handling in Objective-C,
2314 similar to what is offered by C++ and Java. This option is
2315 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2324 @@catch (AnObjCClass *exc) @{
2331 @@catch (AnotherClass *exc) @{
2334 @@catch (id allOthers) @{
2344 The @code{@@throw} statement may appear anywhere in an Objective-C or
2345 Objective-C++ program; when used inside of a @code{@@catch} block, the
2346 @code{@@throw} may appear without an argument (as shown above), in which case
2347 the object caught by the @code{@@catch} will be rethrown.
2349 Note that only (pointers to) Objective-C objects may be thrown and
2350 caught using this scheme. When an object is thrown, it will be caught
2351 by the nearest @code{@@catch} clause capable of handling objects of that type,
2352 analogously to how @code{catch} blocks work in C++ and Java. A
2353 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2354 any and all Objective-C exceptions not caught by previous @code{@@catch}
2357 The @code{@@finally} clause, if present, will be executed upon exit from the
2358 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2359 regardless of whether any exceptions are thrown, caught or rethrown
2360 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2361 of the @code{finally} clause in Java.
2363 There are several caveats to using the new exception mechanism:
2367 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2368 idioms provided by the @code{NSException} class, the new
2369 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2370 systems, due to additional functionality needed in the (NeXT) Objective-C
2374 As mentioned above, the new exceptions do not support handling
2375 types other than Objective-C objects. Furthermore, when used from
2376 Objective-C++, the Objective-C exception model does not interoperate with C++
2377 exceptions at this time. This means you cannot @code{@@throw} an exception
2378 from Objective-C and @code{catch} it in C++, or vice versa
2379 (i.e., @code{throw @dots{} @@catch}).
2382 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2383 blocks for thread-safe execution:
2386 @@synchronized (ObjCClass *guard) @{
2391 Upon entering the @code{@@synchronized} block, a thread of execution shall
2392 first check whether a lock has been placed on the corresponding @code{guard}
2393 object by another thread. If it has, the current thread shall wait until
2394 the other thread relinquishes its lock. Once @code{guard} becomes available,
2395 the current thread will place its own lock on it, execute the code contained in
2396 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2397 making @code{guard} available to other threads).
2399 Unlike Java, Objective-C does not allow for entire methods to be marked
2400 @code{@@synchronized}. Note that throwing exceptions out of
2401 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2402 to be unlocked properly.
2406 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2408 @item -freplace-objc-classes
2409 @opindex freplace-objc-classes
2410 Emit a special marker instructing @command{ld(1)} not to statically link in
2411 the resulting object file, and allow @command{dyld(1)} to load it in at
2412 run time instead. This is used in conjunction with the Fix-and-Continue
2413 debugging mode, where the object file in question may be recompiled and
2414 dynamically reloaded in the course of program execution, without the need
2415 to restart the program itself. Currently, Fix-and-Continue functionality
2416 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2421 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2422 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2423 compile time) with static class references that get initialized at load time,
2424 which improves run-time performance. Specifying the @option{-fzero-link} flag
2425 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2426 to be retained. This is useful in Zero-Link debugging mode, since it allows
2427 for individual class implementations to be modified during program execution.
2431 Dump interface declarations for all classes seen in the source file to a
2432 file named @file{@var{sourcename}.decl}.
2434 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2435 @opindex Wassign-intercept
2436 @opindex Wno-assign-intercept
2437 Warn whenever an Objective-C assignment is being intercepted by the
2440 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2441 @opindex Wno-protocol
2443 If a class is declared to implement a protocol, a warning is issued for
2444 every method in the protocol that is not implemented by the class. The
2445 default behavior is to issue a warning for every method not explicitly
2446 implemented in the class, even if a method implementation is inherited
2447 from the superclass. If you use the @option{-Wno-protocol} option, then
2448 methods inherited from the superclass are considered to be implemented,
2449 and no warning is issued for them.
2451 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2453 @opindex Wno-selector
2454 Warn if multiple methods of different types for the same selector are
2455 found during compilation. The check is performed on the list of methods
2456 in the final stage of compilation. Additionally, a check is performed
2457 for each selector appearing in a @code{@@selector(@dots{})}
2458 expression, and a corresponding method for that selector has been found
2459 during compilation. Because these checks scan the method table only at
2460 the end of compilation, these warnings are not produced if the final
2461 stage of compilation is not reached, for example because an error is
2462 found during compilation, or because the @option{-fsyntax-only} option is
2465 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2466 @opindex Wstrict-selector-match
2467 @opindex Wno-strict-selector-match
2468 Warn if multiple methods with differing argument and/or return types are
2469 found for a given selector when attempting to send a message using this
2470 selector to a receiver of type @code{id} or @code{Class}. When this flag
2471 is off (which is the default behavior), the compiler will omit such warnings
2472 if any differences found are confined to types which share the same size
2475 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2476 @opindex Wundeclared-selector
2477 @opindex Wno-undeclared-selector
2478 Warn if a @code{@@selector(@dots{})} expression referring to an
2479 undeclared selector is found. A selector is considered undeclared if no
2480 method with that name has been declared before the
2481 @code{@@selector(@dots{})} expression, either explicitly in an
2482 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2483 an @code{@@implementation} section. This option always performs its
2484 checks as soon as a @code{@@selector(@dots{})} expression is found,
2485 while @option{-Wselector} only performs its checks in the final stage of
2486 compilation. This also enforces the coding style convention
2487 that methods and selectors must be declared before being used.
2489 @item -print-objc-runtime-info
2490 @opindex print-objc-runtime-info
2491 Generate C header describing the largest structure that is passed by
2496 @node Language Independent Options
2497 @section Options to Control Diagnostic Messages Formatting
2498 @cindex options to control diagnostics formatting
2499 @cindex diagnostic messages
2500 @cindex message formatting
2502 Traditionally, diagnostic messages have been formatted irrespective of
2503 the output device's aspect (e.g.@: its width, @dots{}). The options described
2504 below can be used to control the diagnostic messages formatting
2505 algorithm, e.g.@: how many characters per line, how often source location
2506 information should be reported. Right now, only the C++ front end can
2507 honor these options. However it is expected, in the near future, that
2508 the remaining front ends would be able to digest them correctly.
2511 @item -fmessage-length=@var{n}
2512 @opindex fmessage-length
2513 Try to format error messages so that they fit on lines of about @var{n}
2514 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2515 the front ends supported by GCC@. If @var{n} is zero, then no
2516 line-wrapping will be done; each error message will appear on a single
2519 @opindex fdiagnostics-show-location
2520 @item -fdiagnostics-show-location=once
2521 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2522 reporter to emit @emph{once} source location information; that is, in
2523 case the message is too long to fit on a single physical line and has to
2524 be wrapped, the source location won't be emitted (as prefix) again,
2525 over and over, in subsequent continuation lines. This is the default
2528 @item -fdiagnostics-show-location=every-line
2529 Only meaningful in line-wrapping mode. Instructs the diagnostic
2530 messages reporter to emit the same source location information (as
2531 prefix) for physical lines that result from the process of breaking
2532 a message which is too long to fit on a single line.
2534 @item -fdiagnostics-show-option
2535 @opindex fdiagnostics-show-option
2536 This option instructs the diagnostic machinery to add text to each
2537 diagnostic emitted, which indicates which command line option directly
2538 controls that diagnostic, when such an option is known to the
2539 diagnostic machinery.
2541 @item -Wcoverage-mismatch
2542 @opindex Wcoverage-mismatch
2543 Warn if feedback profiles do not match when using the
2544 @option{-fprofile-use} option.
2545 If a source file was changed between @option{-fprofile-gen} and
2546 @option{-fprofile-use}, the files with the profile feedback can fail
2547 to match the source file and GCC can not use the profile feedback
2548 information. By default, GCC emits an error message in this case.
2549 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2550 error. GCC does not use appropriate feedback profiles, so using this
2551 option can result in poorly optimized code. This option is useful
2552 only in the case of very minor changes such as bug fixes to an
2557 @node Warning Options
2558 @section Options to Request or Suppress Warnings
2559 @cindex options to control warnings
2560 @cindex warning messages
2561 @cindex messages, warning
2562 @cindex suppressing warnings
2564 Warnings are diagnostic messages that report constructions which
2565 are not inherently erroneous but which are risky or suggest there
2566 may have been an error.
2568 The following language-independent options do not enable specific
2569 warnings but control the kinds of diagnostics produced by GCC.
2572 @cindex syntax checking
2574 @opindex fsyntax-only
2575 Check the code for syntax errors, but don't do anything beyond that.
2579 Inhibit all warning messages.
2584 Make all warnings into errors.
2589 Make the specified warning into an error. The specifier for a warning
2590 is appended, for example @option{-Werror=switch} turns the warnings
2591 controlled by @option{-Wswitch} into errors. This switch takes a
2592 negative form, to be used to negate @option{-Werror} for specific
2593 warnings, for example @option{-Wno-error=switch} makes
2594 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2595 is in effect. You can use the @option{-fdiagnostics-show-option}
2596 option to have each controllable warning amended with the option which
2597 controls it, to determine what to use with this option.
2599 Note that specifying @option{-Werror=}@var{foo} automatically implies
2600 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2603 @item -Wfatal-errors
2604 @opindex Wfatal-errors
2605 @opindex Wno-fatal-errors
2606 This option causes the compiler to abort compilation on the first error
2607 occurred rather than trying to keep going and printing further error
2612 You can request many specific warnings with options beginning
2613 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2614 implicit declarations. Each of these specific warning options also
2615 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2616 example, @option{-Wno-implicit}. This manual lists only one of the
2617 two forms, whichever is not the default. For further,
2618 language-specific options also refer to @ref{C++ Dialect Options} and
2619 @ref{Objective-C and Objective-C++ Dialect Options}.
2624 Issue all the warnings demanded by strict ISO C and ISO C++;
2625 reject all programs that use forbidden extensions, and some other
2626 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2627 version of the ISO C standard specified by any @option{-std} option used.
2629 Valid ISO C and ISO C++ programs should compile properly with or without
2630 this option (though a rare few will require @option{-ansi} or a
2631 @option{-std} option specifying the required version of ISO C)@. However,
2632 without this option, certain GNU extensions and traditional C and C++
2633 features are supported as well. With this option, they are rejected.
2635 @option{-pedantic} does not cause warning messages for use of the
2636 alternate keywords whose names begin and end with @samp{__}. Pedantic
2637 warnings are also disabled in the expression that follows
2638 @code{__extension__}. However, only system header files should use
2639 these escape routes; application programs should avoid them.
2640 @xref{Alternate Keywords}.
2642 Some users try to use @option{-pedantic} to check programs for strict ISO
2643 C conformance. They soon find that it does not do quite what they want:
2644 it finds some non-ISO practices, but not all---only those for which
2645 ISO C @emph{requires} a diagnostic, and some others for which
2646 diagnostics have been added.
2648 A feature to report any failure to conform to ISO C might be useful in
2649 some instances, but would require considerable additional work and would
2650 be quite different from @option{-pedantic}. We don't have plans to
2651 support such a feature in the near future.
2653 Where the standard specified with @option{-std} represents a GNU
2654 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2655 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2656 extended dialect is based. Warnings from @option{-pedantic} are given
2657 where they are required by the base standard. (It would not make sense
2658 for such warnings to be given only for features not in the specified GNU
2659 C dialect, since by definition the GNU dialects of C include all
2660 features the compiler supports with the given option, and there would be
2661 nothing to warn about.)
2663 @item -pedantic-errors
2664 @opindex pedantic-errors
2665 Like @option{-pedantic}, except that errors are produced rather than
2671 This enables all the warnings about constructions that some users
2672 consider questionable, and that are easy to avoid (or modify to
2673 prevent the warning), even in conjunction with macros. This also
2674 enables some language-specific warnings described in @ref{C++ Dialect
2675 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2677 @option{-Wall} turns on the following warning flags:
2679 @gccoptlist{-Waddress @gol
2680 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2682 -Wchar-subscripts @gol
2684 -Wimplicit-function-declaration @gol
2687 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2688 -Wmissing-braces @gol
2694 -Wsequence-point @gol
2695 -Wsign-compare @r{(only in C++)} @gol
2696 -Wstrict-aliasing @gol
2697 -Wstrict-overflow=1 @gol
2700 -Wuninitialized @gol
2701 -Wunknown-pragmas @gol
2702 -Wunused-function @gol
2705 -Wunused-variable @gol
2706 -Wvolatile-register-var @gol
2709 Note that some warning flags are not implied by @option{-Wall}. Some of
2710 them warn about constructions that users generally do not consider
2711 questionable, but which occasionally you might wish to check for;
2712 others warn about constructions that are necessary or hard to avoid in
2713 some cases, and there is no simple way to modify the code to suppress
2714 the warning. Some of them are enabled by @option{-Wextra} but many of
2715 them must be enabled individually.
2721 This enables some extra warning flags that are not enabled by
2722 @option{-Wall}. (This option used to be called @option{-W}. The older
2723 name is still supported, but the newer name is more descriptive.)
2725 @gccoptlist{-Wclobbered @gol
2727 -Wignored-qualifiers @gol
2728 -Wmissing-field-initializers @gol
2729 -Wmissing-parameter-type @r{(C only)} @gol
2730 -Wold-style-declaration @r{(C only)} @gol
2731 -Woverride-init @gol
2734 -Wuninitialized @gol
2735 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2738 The option @option{-Wextra} also prints warning messages for the
2744 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2745 @samp{>}, or @samp{>=}.
2748 (C++ only) An enumerator and a non-enumerator both appear in a
2749 conditional expression.
2752 (C++ only) A non-static reference or non-static @samp{const} member
2753 appears in a class without constructors.
2756 (C++ only) Ambiguous virtual bases.
2759 (C++ only) Subscripting an array which has been declared @samp{register}.
2762 (C++ only) Taking the address of a variable which has been declared
2766 (C++ only) A base class is not initialized in a derived class' copy
2771 @item -Wchar-subscripts
2772 @opindex Wchar-subscripts
2773 @opindex Wno-char-subscripts
2774 Warn if an array subscript has type @code{char}. This is a common cause
2775 of error, as programmers often forget that this type is signed on some
2777 This warning is enabled by @option{-Wall}.
2781 @opindex Wno-comment
2782 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2783 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2784 This warning is enabled by @option{-Wall}.
2789 @opindex ffreestanding
2790 @opindex fno-builtin
2791 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2792 the arguments supplied have types appropriate to the format string
2793 specified, and that the conversions specified in the format string make
2794 sense. This includes standard functions, and others specified by format
2795 attributes (@pxref{Function Attributes}), in the @code{printf},
2796 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2797 not in the C standard) families (or other target-specific families).
2798 Which functions are checked without format attributes having been
2799 specified depends on the standard version selected, and such checks of
2800 functions without the attribute specified are disabled by
2801 @option{-ffreestanding} or @option{-fno-builtin}.
2803 The formats are checked against the format features supported by GNU
2804 libc version 2.2. These include all ISO C90 and C99 features, as well
2805 as features from the Single Unix Specification and some BSD and GNU
2806 extensions. Other library implementations may not support all these
2807 features; GCC does not support warning about features that go beyond a
2808 particular library's limitations. However, if @option{-pedantic} is used
2809 with @option{-Wformat}, warnings will be given about format features not
2810 in the selected standard version (but not for @code{strfmon} formats,
2811 since those are not in any version of the C standard). @xref{C Dialect
2812 Options,,Options Controlling C Dialect}.
2814 Since @option{-Wformat} also checks for null format arguments for
2815 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2817 @option{-Wformat} is included in @option{-Wall}. For more control over some
2818 aspects of format checking, the options @option{-Wformat-y2k},
2819 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2820 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2821 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2824 @opindex Wformat-y2k
2825 @opindex Wno-format-y2k
2826 If @option{-Wformat} is specified, also warn about @code{strftime}
2827 formats which may yield only a two-digit year.
2829 @item -Wno-format-contains-nul
2830 @opindex Wno-format-contains-nul
2831 @opindex Wformat-contains-nul
2832 If @option{-Wformat} is specified, do not warn about format strings that
2835 @item -Wno-format-extra-args
2836 @opindex Wno-format-extra-args
2837 @opindex Wformat-extra-args
2838 If @option{-Wformat} is specified, do not warn about excess arguments to a
2839 @code{printf} or @code{scanf} format function. The C standard specifies
2840 that such arguments are ignored.
2842 Where the unused arguments lie between used arguments that are
2843 specified with @samp{$} operand number specifications, normally
2844 warnings are still given, since the implementation could not know what
2845 type to pass to @code{va_arg} to skip the unused arguments. However,
2846 in the case of @code{scanf} formats, this option will suppress the
2847 warning if the unused arguments are all pointers, since the Single
2848 Unix Specification says that such unused arguments are allowed.
2850 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2851 @opindex Wno-format-zero-length
2852 @opindex Wformat-zero-length
2853 If @option{-Wformat} is specified, do not warn about zero-length formats.
2854 The C standard specifies that zero-length formats are allowed.
2856 @item -Wformat-nonliteral
2857 @opindex Wformat-nonliteral
2858 @opindex Wno-format-nonliteral
2859 If @option{-Wformat} is specified, also warn if the format string is not a
2860 string literal and so cannot be checked, unless the format function
2861 takes its format arguments as a @code{va_list}.
2863 @item -Wformat-security
2864 @opindex Wformat-security
2865 @opindex Wno-format-security
2866 If @option{-Wformat} is specified, also warn about uses of format
2867 functions that represent possible security problems. At present, this
2868 warns about calls to @code{printf} and @code{scanf} functions where the
2869 format string is not a string literal and there are no format arguments,
2870 as in @code{printf (foo);}. This may be a security hole if the format
2871 string came from untrusted input and contains @samp{%n}. (This is
2872 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2873 in future warnings may be added to @option{-Wformat-security} that are not
2874 included in @option{-Wformat-nonliteral}.)
2878 @opindex Wno-format=2
2879 Enable @option{-Wformat} plus format checks not included in
2880 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2881 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2883 @item -Wnonnull @r{(C and Objective-C only)}
2885 @opindex Wno-nonnull
2886 Warn about passing a null pointer for arguments marked as
2887 requiring a non-null value by the @code{nonnull} function attribute.
2889 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2890 can be disabled with the @option{-Wno-nonnull} option.
2892 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2894 @opindex Wno-init-self
2895 Warn about uninitialized variables which are initialized with themselves.
2896 Note this option can only be used with the @option{-Wuninitialized} option.
2898 For example, GCC will warn about @code{i} being uninitialized in the
2899 following snippet only when @option{-Winit-self} has been specified:
2910 @item -Wimplicit-int @r{(C and Objective-C only)}
2911 @opindex Wimplicit-int
2912 @opindex Wno-implicit-int
2913 Warn when a declaration does not specify a type.
2914 This warning is enabled by @option{-Wall}.
2916 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
2917 @opindex Wimplicit-function-declaration
2918 @opindex Wno-implicit-function-declaration
2919 Give a warning whenever a function is used before being declared. In
2920 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2921 enabled by default and it is made into an error by
2922 @option{-pedantic-errors}. This warning is also enabled by
2927 @opindex Wno-implicit
2928 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2929 This warning is enabled by @option{-Wall}.
2931 @item -Wignored-qualifiers @r{(C and C++ only)}
2932 @opindex Wignored-qualifiers
2933 @opindex Wno-ignored-qualifiers
2934 Warn if the return type of a function has a type qualifier
2935 such as @code{const}. For ISO C such a type qualifier has no effect,
2936 since the value returned by a function is not an lvalue.
2937 For C++, the warning is only emitted for scalar types or @code{void}.
2938 ISO C prohibits qualified @code{void} return types on function
2939 definitions, so such return types always receive a warning
2940 even without this option.
2942 This warning is also enabled by @option{-Wextra}.
2947 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2948 function with external linkage, returning int, taking either zero
2949 arguments, two, or three arguments of appropriate types.
2950 This warning is enabled by @option{-Wall}.
2952 @item -Wmissing-braces
2953 @opindex Wmissing-braces
2954 @opindex Wno-missing-braces
2955 Warn if an aggregate or union initializer is not fully bracketed. In
2956 the following example, the initializer for @samp{a} is not fully
2957 bracketed, but that for @samp{b} is fully bracketed.
2960 int a[2][2] = @{ 0, 1, 2, 3 @};
2961 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2964 This warning is enabled by @option{-Wall}.
2966 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2967 @opindex Wmissing-include-dirs
2968 @opindex Wno-missing-include-dirs
2969 Warn if a user-supplied include directory does not exist.
2972 @opindex Wparentheses
2973 @opindex Wno-parentheses
2974 Warn if parentheses are omitted in certain contexts, such
2975 as when there is an assignment in a context where a truth value
2976 is expected, or when operators are nested whose precedence people
2977 often get confused about.
2979 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2980 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2981 interpretation from that of ordinary mathematical notation.
2983 Also warn about constructions where there may be confusion to which
2984 @code{if} statement an @code{else} branch belongs. Here is an example of
2999 In C/C++, every @code{else} branch belongs to the innermost possible
3000 @code{if} statement, which in this example is @code{if (b)}. This is
3001 often not what the programmer expected, as illustrated in the above
3002 example by indentation the programmer chose. When there is the
3003 potential for this confusion, GCC will issue a warning when this flag
3004 is specified. To eliminate the warning, add explicit braces around
3005 the innermost @code{if} statement so there is no way the @code{else}
3006 could belong to the enclosing @code{if}. The resulting code would
3023 This warning is enabled by @option{-Wall}.
3025 @item -Wsequence-point
3026 @opindex Wsequence-point
3027 @opindex Wno-sequence-point
3028 Warn about code that may have undefined semantics because of violations
3029 of sequence point rules in the C and C++ standards.
3031 The C and C++ standards defines the order in which expressions in a C/C++
3032 program are evaluated in terms of @dfn{sequence points}, which represent
3033 a partial ordering between the execution of parts of the program: those
3034 executed before the sequence point, and those executed after it. These
3035 occur after the evaluation of a full expression (one which is not part
3036 of a larger expression), after the evaluation of the first operand of a
3037 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3038 function is called (but after the evaluation of its arguments and the
3039 expression denoting the called function), and in certain other places.
3040 Other than as expressed by the sequence point rules, the order of
3041 evaluation of subexpressions of an expression is not specified. All
3042 these rules describe only a partial order rather than a total order,
3043 since, for example, if two functions are called within one expression
3044 with no sequence point between them, the order in which the functions
3045 are called is not specified. However, the standards committee have
3046 ruled that function calls do not overlap.
3048 It is not specified when between sequence points modifications to the
3049 values of objects take effect. Programs whose behavior depends on this
3050 have undefined behavior; the C and C++ standards specify that ``Between
3051 the previous and next sequence point an object shall have its stored
3052 value modified at most once by the evaluation of an expression.
3053 Furthermore, the prior value shall be read only to determine the value
3054 to be stored.''. If a program breaks these rules, the results on any
3055 particular implementation are entirely unpredictable.
3057 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3058 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3059 diagnosed by this option, and it may give an occasional false positive
3060 result, but in general it has been found fairly effective at detecting
3061 this sort of problem in programs.
3063 The standard is worded confusingly, therefore there is some debate
3064 over the precise meaning of the sequence point rules in subtle cases.
3065 Links to discussions of the problem, including proposed formal
3066 definitions, may be found on the GCC readings page, at
3067 @w{@uref{http://gcc.gnu.org/readings.html}}.
3069 This warning is enabled by @option{-Wall} for C and C++.
3072 @opindex Wreturn-type
3073 @opindex Wno-return-type
3074 Warn whenever a function is defined with a return-type that defaults
3075 to @code{int}. Also warn about any @code{return} statement with no
3076 return-value in a function whose return-type is not @code{void}
3077 (falling off the end of the function body is considered returning
3078 without a value), and about a @code{return} statement with a
3079 expression in a function whose return-type is @code{void}.
3081 For C++, a function without return type always produces a diagnostic
3082 message, even when @option{-Wno-return-type} is specified. The only
3083 exceptions are @samp{main} and functions defined in system headers.
3085 This warning is enabled by @option{-Wall}.
3090 Warn whenever a @code{switch} statement has an index of enumerated type
3091 and lacks a @code{case} for one or more of the named codes of that
3092 enumeration. (The presence of a @code{default} label prevents this
3093 warning.) @code{case} labels outside the enumeration range also
3094 provoke warnings when this option is used.
3095 This warning is enabled by @option{-Wall}.
3097 @item -Wswitch-default
3098 @opindex Wswitch-default
3099 @opindex Wno-switch-default
3100 Warn whenever a @code{switch} statement does not have a @code{default}
3104 @opindex Wswitch-enum
3105 @opindex Wno-switch-enum
3106 Warn whenever a @code{switch} statement has an index of enumerated type
3107 and lacks a @code{case} for one or more of the named codes of that
3108 enumeration. @code{case} labels outside the enumeration range also
3109 provoke warnings when this option is used.
3113 @opindex Wno-trigraphs
3114 Warn if any trigraphs are encountered that might change the meaning of
3115 the program (trigraphs within comments are not warned about).
3116 This warning is enabled by @option{-Wall}.
3118 @item -Wunused-function
3119 @opindex Wunused-function
3120 @opindex Wno-unused-function
3121 Warn whenever a static function is declared but not defined or a
3122 non-inline static function is unused.
3123 This warning is enabled by @option{-Wall}.
3125 @item -Wunused-label
3126 @opindex Wunused-label
3127 @opindex Wno-unused-label
3128 Warn whenever a label is declared but not used.
3129 This warning is enabled by @option{-Wall}.
3131 To suppress this warning use the @samp{unused} attribute
3132 (@pxref{Variable Attributes}).
3134 @item -Wunused-parameter
3135 @opindex Wunused-parameter
3136 @opindex Wno-unused-parameter
3137 Warn whenever a function parameter is unused aside from its declaration.
3139 To suppress this warning use the @samp{unused} attribute
3140 (@pxref{Variable Attributes}).
3142 @item -Wunused-variable
3143 @opindex Wunused-variable
3144 @opindex Wno-unused-variable
3145 Warn whenever a local variable or non-constant static variable is unused
3146 aside from its declaration.
3147 This warning is enabled by @option{-Wall}.
3149 To suppress this warning use the @samp{unused} attribute
3150 (@pxref{Variable Attributes}).
3152 @item -Wunused-value
3153 @opindex Wunused-value
3154 @opindex Wno-unused-value
3155 Warn whenever a statement computes a result that is explicitly not
3156 used. To suppress this warning cast the unused expression to
3157 @samp{void}. This includes an expression-statement or the left-hand
3158 side of a comma expression that contains no side effects. For example,
3159 an expression such as @samp{x[i,j]} will cause a warning, while
3160 @samp{x[(void)i,j]} will not.
3162 This warning is enabled by @option{-Wall}.
3167 All the above @option{-Wunused} options combined.
3169 In order to get a warning about an unused function parameter, you must
3170 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3171 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3173 @item -Wuninitialized
3174 @opindex Wuninitialized
3175 @opindex Wno-uninitialized
3176 Warn if an automatic variable is used without first being initialized or
3177 if a variable may be clobbered by a @code{setjmp} call.
3179 If you want to warn about code which uses the uninitialized value of the
3180 variable in its own initializer, use the @option{-Winit-self} option.
3182 These warnings occur for individual uninitialized or clobbered
3183 elements of structure, union or array variables as well as for
3184 variables which are uninitialized or clobbered as a whole. They do
3185 not occur for variables or elements declared @code{volatile}. Because
3186 these warnings depend on optimization, the exact variables or elements
3187 for which there are warnings will depend on the precise optimization
3188 options and version of GCC used.
3190 Note that there may be no warning about a variable that is used only
3191 to compute a value that itself is never used, because such
3192 computations may be deleted by data flow analysis before the warnings
3195 These warnings are made optional because GCC is not smart
3196 enough to see all the reasons why the code might be correct
3197 despite appearing to have an error. Here is one example of how
3218 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3219 always initialized, but GCC doesn't know this. Here is
3220 another common case:
3225 if (change_y) save_y = y, y = new_y;
3227 if (change_y) y = save_y;
3232 This has no bug because @code{save_y} is used only if it is set.
3234 @cindex @code{longjmp} warnings
3235 This option also warns when a non-volatile automatic variable might be
3236 changed by a call to @code{longjmp}. These warnings as well are possible
3237 only in optimizing compilation.
3239 The compiler sees only the calls to @code{setjmp}. It cannot know
3240 where @code{longjmp} will be called; in fact, a signal handler could
3241 call it at any point in the code. As a result, you may get a warning
3242 even when there is in fact no problem because @code{longjmp} cannot
3243 in fact be called at the place which would cause a problem.
3245 Some spurious warnings can be avoided if you declare all the functions
3246 you use that never return as @code{noreturn}. @xref{Function
3249 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3251 @item -Wunknown-pragmas
3252 @opindex Wunknown-pragmas
3253 @opindex Wno-unknown-pragmas
3254 @cindex warning for unknown pragmas
3255 @cindex unknown pragmas, warning
3256 @cindex pragmas, warning of unknown
3257 Warn when a #pragma directive is encountered which is not understood by
3258 GCC@. If this command line option is used, warnings will even be issued
3259 for unknown pragmas in system header files. This is not the case if
3260 the warnings were only enabled by the @option{-Wall} command line option.
3263 @opindex Wno-pragmas
3265 Do not warn about misuses of pragmas, such as incorrect parameters,
3266 invalid syntax, or conflicts between pragmas. See also
3267 @samp{-Wunknown-pragmas}.
3269 @item -Wstrict-aliasing
3270 @opindex Wstrict-aliasing
3271 @opindex Wno-strict-aliasing
3272 This option is only active when @option{-fstrict-aliasing} is active.
3273 It warns about code which might break the strict aliasing rules that the
3274 compiler is using for optimization. The warning does not catch all
3275 cases, but does attempt to catch the more common pitfalls. It is
3276 included in @option{-Wall}.
3277 It is equivalent to @option{-Wstrict-aliasing=3}
3279 @item -Wstrict-aliasing=n
3280 @opindex Wstrict-aliasing=n
3281 @opindex Wno-strict-aliasing=n
3282 This option is only active when @option{-fstrict-aliasing} is active.
3283 It warns about code which might break the strict aliasing rules that the
3284 compiler is using for optimization.
3285 Higher levels correspond to higher accuracy (fewer false positives).
3286 Higher levels also correspond to more effort, similar to the way -O works.
3287 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3290 Level 1: Most aggressive, quick, least accurate.
3291 Possibly useful when higher levels
3292 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3293 false negatives. However, it has many false positives.
3294 Warns for all pointer conversions between possibly incompatible types,
3295 even if never dereferenced. Runs in the frontend only.
3297 Level 2: Aggressive, quick, not too precise.
3298 May still have many false positives (not as many as level 1 though),
3299 and few false negatives (but possibly more than level 1).
3300 Unlike level 1, it only warns when an address is taken. Warns about
3301 incomplete types. Runs in the frontend only.
3303 Level 3 (default for @option{-Wstrict-aliasing}):
3304 Should have very few false positives and few false
3305 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3306 Takes care of the common punn+dereference pattern in the frontend:
3307 @code{*(int*)&some_float}.
3308 If optimization is enabled, it also runs in the backend, where it deals
3309 with multiple statement cases using flow-sensitive points-to information.
3310 Only warns when the converted pointer is dereferenced.
3311 Does not warn about incomplete types.
3313 @item -Wstrict-overflow
3314 @itemx -Wstrict-overflow=@var{n}
3315 @opindex Wstrict-overflow
3316 @opindex Wno-strict-overflow
3317 This option is only active when @option{-fstrict-overflow} is active.
3318 It warns about cases where the compiler optimizes based on the
3319 assumption that signed overflow does not occur. Note that it does not
3320 warn about all cases where the code might overflow: it only warns
3321 about cases where the compiler implements some optimization. Thus
3322 this warning depends on the optimization level.
3324 An optimization which assumes that signed overflow does not occur is
3325 perfectly safe if the values of the variables involved are such that
3326 overflow never does, in fact, occur. Therefore this warning can
3327 easily give a false positive: a warning about code which is not
3328 actually a problem. To help focus on important issues, several
3329 warning levels are defined. No warnings are issued for the use of
3330 undefined signed overflow when estimating how many iterations a loop
3331 will require, in particular when determining whether a loop will be
3335 @item -Wstrict-overflow=1
3336 Warn about cases which are both questionable and easy to avoid. For
3337 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3338 compiler will simplify this to @code{1}. This level of
3339 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3340 are not, and must be explicitly requested.
3342 @item -Wstrict-overflow=2
3343 Also warn about other cases where a comparison is simplified to a
3344 constant. For example: @code{abs (x) >= 0}. This can only be
3345 simplified when @option{-fstrict-overflow} is in effect, because
3346 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3347 zero. @option{-Wstrict-overflow} (with no level) is the same as
3348 @option{-Wstrict-overflow=2}.
3350 @item -Wstrict-overflow=3
3351 Also warn about other cases where a comparison is simplified. For
3352 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3354 @item -Wstrict-overflow=4
3355 Also warn about other simplifications not covered by the above cases.
3356 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3358 @item -Wstrict-overflow=5
3359 Also warn about cases where the compiler reduces the magnitude of a
3360 constant involved in a comparison. For example: @code{x + 2 > y} will
3361 be simplified to @code{x + 1 >= y}. This is reported only at the
3362 highest warning level because this simplification applies to many
3363 comparisons, so this warning level will give a very large number of
3367 @item -Warray-bounds
3368 @opindex Wno-array-bounds
3369 @opindex Warray-bounds
3370 This option is only active when @option{-ftree-vrp} is active
3371 (default for -O2 and above). It warns about subscripts to arrays
3372 that are always out of bounds. This warning is enabled by @option{-Wall}.
3374 @item -Wno-div-by-zero
3375 @opindex Wno-div-by-zero
3376 @opindex Wdiv-by-zero
3377 Do not warn about compile-time integer division by zero. Floating point
3378 division by zero is not warned about, as it can be a legitimate way of
3379 obtaining infinities and NaNs.
3381 @item -Wsystem-headers
3382 @opindex Wsystem-headers
3383 @opindex Wno-system-headers
3384 @cindex warnings from system headers
3385 @cindex system headers, warnings from
3386 Print warning messages for constructs found in system header files.
3387 Warnings from system headers are normally suppressed, on the assumption
3388 that they usually do not indicate real problems and would only make the
3389 compiler output harder to read. Using this command line option tells
3390 GCC to emit warnings from system headers as if they occurred in user
3391 code. However, note that using @option{-Wall} in conjunction with this
3392 option will @emph{not} warn about unknown pragmas in system
3393 headers---for that, @option{-Wunknown-pragmas} must also be used.
3396 @opindex Wfloat-equal
3397 @opindex Wno-float-equal
3398 Warn if floating point values are used in equality comparisons.
3400 The idea behind this is that sometimes it is convenient (for the
3401 programmer) to consider floating-point values as approximations to
3402 infinitely precise real numbers. If you are doing this, then you need
3403 to compute (by analyzing the code, or in some other way) the maximum or
3404 likely maximum error that the computation introduces, and allow for it
3405 when performing comparisons (and when producing output, but that's a
3406 different problem). In particular, instead of testing for equality, you
3407 would check to see whether the two values have ranges that overlap; and
3408 this is done with the relational operators, so equality comparisons are
3411 @item -Wtraditional @r{(C and Objective-C only)}
3412 @opindex Wtraditional
3413 @opindex Wno-traditional
3414 Warn about certain constructs that behave differently in traditional and
3415 ISO C@. Also warn about ISO C constructs that have no traditional C
3416 equivalent, and/or problematic constructs which should be avoided.
3420 Macro parameters that appear within string literals in the macro body.
3421 In traditional C macro replacement takes place within string literals,
3422 but does not in ISO C@.
3425 In traditional C, some preprocessor directives did not exist.
3426 Traditional preprocessors would only consider a line to be a directive
3427 if the @samp{#} appeared in column 1 on the line. Therefore
3428 @option{-Wtraditional} warns about directives that traditional C
3429 understands but would ignore because the @samp{#} does not appear as the
3430 first character on the line. It also suggests you hide directives like
3431 @samp{#pragma} not understood by traditional C by indenting them. Some
3432 traditional implementations would not recognize @samp{#elif}, so it
3433 suggests avoiding it altogether.
3436 A function-like macro that appears without arguments.
3439 The unary plus operator.
3442 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3443 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3444 constants.) Note, these suffixes appear in macros defined in the system
3445 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3446 Use of these macros in user code might normally lead to spurious
3447 warnings, however GCC's integrated preprocessor has enough context to
3448 avoid warning in these cases.
3451 A function declared external in one block and then used after the end of
3455 A @code{switch} statement has an operand of type @code{long}.
3458 A non-@code{static} function declaration follows a @code{static} one.
3459 This construct is not accepted by some traditional C compilers.
3462 The ISO type of an integer constant has a different width or
3463 signedness from its traditional type. This warning is only issued if
3464 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3465 typically represent bit patterns, are not warned about.
3468 Usage of ISO string concatenation is detected.
3471 Initialization of automatic aggregates.
3474 Identifier conflicts with labels. Traditional C lacks a separate
3475 namespace for labels.
3478 Initialization of unions. If the initializer is zero, the warning is
3479 omitted. This is done under the assumption that the zero initializer in
3480 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3481 initializer warnings and relies on default initialization to zero in the
3485 Conversions by prototypes between fixed/floating point values and vice
3486 versa. The absence of these prototypes when compiling with traditional
3487 C would cause serious problems. This is a subset of the possible
3488 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3491 Use of ISO C style function definitions. This warning intentionally is
3492 @emph{not} issued for prototype declarations or variadic functions
3493 because these ISO C features will appear in your code when using
3494 libiberty's traditional C compatibility macros, @code{PARAMS} and
3495 @code{VPARAMS}. This warning is also bypassed for nested functions
3496 because that feature is already a GCC extension and thus not relevant to
3497 traditional C compatibility.
3500 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3501 @opindex Wtraditional-conversion
3502 @opindex Wno-traditional-conversion
3503 Warn if a prototype causes a type conversion that is different from what
3504 would happen to the same argument in the absence of a prototype. This
3505 includes conversions of fixed point to floating and vice versa, and
3506 conversions changing the width or signedness of a fixed point argument
3507 except when the same as the default promotion.
3509 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3510 @opindex Wdeclaration-after-statement
3511 @opindex Wno-declaration-after-statement
3512 Warn when a declaration is found after a statement in a block. This
3513 construct, known from C++, was introduced with ISO C99 and is by default
3514 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3515 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3520 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3522 @item -Wno-endif-labels
3523 @opindex Wno-endif-labels
3524 @opindex Wendif-labels
3525 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3530 Warn whenever a local variable shadows another local variable, parameter or
3531 global variable or whenever a built-in function is shadowed.
3533 @item -Wlarger-than=@var{len}
3534 @opindex Wlarger-than=@var{len}
3535 @opindex Wlarger-than-@var{len}
3536 Warn whenever an object of larger than @var{len} bytes is defined.
3538 @item -Wframe-larger-than=@var{len}
3539 @opindex Wframe-larger-than
3540 Warn if the size of a function frame is larger than @var{len} bytes.
3541 The computation done to determine the stack frame size is approximate
3542 and not conservative.
3543 The actual requirements may be somewhat greater than @var{len}
3544 even if you do not get a warning. In addition, any space allocated
3545 via @code{alloca}, variable-length arrays, or related constructs
3546 is not included by the compiler when determining
3547 whether or not to issue a warning.
3549 @item -Wunsafe-loop-optimizations
3550 @opindex Wunsafe-loop-optimizations
3551 @opindex Wno-unsafe-loop-optimizations
3552 Warn if the loop cannot be optimized because the compiler could not
3553 assume anything on the bounds of the loop indices. With
3554 @option{-funsafe-loop-optimizations} warn if the compiler made
3557 @item -Wpointer-arith
3558 @opindex Wpointer-arith
3559 @opindex Wno-pointer-arith
3560 Warn about anything that depends on the ``size of'' a function type or
3561 of @code{void}. GNU C assigns these types a size of 1, for
3562 convenience in calculations with @code{void *} pointers and pointers
3563 to functions. In C++, warn also when an arithmetic operation involves
3564 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3567 @opindex Wtype-limits
3568 @opindex Wno-type-limits
3569 Warn if a comparison is always true or always false due to the limited
3570 range of the data type, but do not warn for constant expressions. For
3571 example, warn if an unsigned variable is compared against zero with
3572 @samp{<} or @samp{>=}. This warning is also enabled by
3575 @item -Wbad-function-cast @r{(C and Objective-C only)}
3576 @opindex Wbad-function-cast
3577 @opindex Wno-bad-function-cast
3578 Warn whenever a function call is cast to a non-matching type.
3579 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3581 @item -Wc++-compat @r{(C and Objective-C only)}
3582 Warn about ISO C constructs that are outside of the common subset of
3583 ISO C and ISO C++, e.g.@: request for implicit conversion from
3584 @code{void *} to a pointer to non-@code{void} type.
3586 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3587 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3588 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3589 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3593 @opindex Wno-cast-qual
3594 Warn whenever a pointer is cast so as to remove a type qualifier from
3595 the target type. For example, warn if a @code{const char *} is cast
3596 to an ordinary @code{char *}.
3599 @opindex Wcast-align
3600 @opindex Wno-cast-align
3601 Warn whenever a pointer is cast such that the required alignment of the
3602 target is increased. For example, warn if a @code{char *} is cast to
3603 an @code{int *} on machines where integers can only be accessed at
3604 two- or four-byte boundaries.
3606 @item -Wwrite-strings
3607 @opindex Wwrite-strings
3608 @opindex Wno-write-strings
3609 When compiling C, give string constants the type @code{const
3610 char[@var{length}]} so that
3611 copying the address of one into a non-@code{const} @code{char *}
3612 pointer will get a warning; when compiling C++, warn about the
3613 deprecated conversion from string literals to @code{char *}. This
3614 warning, by default, is enabled for C++ programs.
3615 These warnings will help you find at
3616 compile time code that can try to write into a string constant, but
3617 only if you have been very careful about using @code{const} in
3618 declarations and prototypes. Otherwise, it will just be a nuisance;
3619 this is why we did not make @option{-Wall} request these warnings.
3623 @opindex Wno-clobbered
3624 Warn for variables that might be changed by @samp{longjmp} or
3625 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3628 @opindex Wconversion
3629 @opindex Wno-conversion
3630 Warn for implicit conversions that may alter a value. This includes
3631 conversions between real and integer, like @code{abs (x)} when
3632 @code{x} is @code{double}; conversions between signed and unsigned,
3633 like @code{unsigned ui = -1}; and conversions to smaller types, like
3634 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3635 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3636 changed by the conversion like in @code{abs (2.0)}. Warnings about
3637 conversions between signed and unsigned integers can be disabled by
3638 using @option{-Wno-sign-conversion}.
3640 For C++, also warn for conversions between @code{NULL} and non-pointer
3641 types; confusing overload resolution for user-defined conversions; and
3642 conversions that will never use a type conversion operator:
3643 conversions to @code{void}, the same type, a base class or a reference
3644 to them. Warnings about conversions between signed and unsigned
3645 integers are disabled by default in C++ unless
3646 @option{-Wsign-conversion} is explicitly enabled.
3649 @opindex Wempty-body
3650 @opindex Wno-empty-body
3651 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3652 while} statement. Additionally, in C++, warn when an empty body occurs
3653 in a @samp{while} or @samp{for} statement with no whitespacing before
3654 the semicolon. This warning is also enabled by @option{-Wextra}.
3656 @item -Wenum-compare @r{(C++ and Objective-C++ only)}
3657 @opindex Wenum-compare
3658 @opindex Wno-enum-compare
3659 Warn about a comparison between values of different enum types. This
3660 warning is enabled by default.
3662 @item -Wsign-compare
3663 @opindex Wsign-compare
3664 @opindex Wno-sign-compare
3665 @cindex warning for comparison of signed and unsigned values
3666 @cindex comparison of signed and unsigned values, warning
3667 @cindex signed and unsigned values, comparison warning
3668 Warn when a comparison between signed and unsigned values could produce
3669 an incorrect result when the signed value is converted to unsigned.
3670 This warning is also enabled by @option{-Wextra}; to get the other warnings
3671 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3673 @item -Wsign-conversion
3674 @opindex Wsign-conversion
3675 @opindex Wno-sign-conversion
3676 Warn for implicit conversions that may change the sign of an integer
3677 value, like assigning a signed integer expression to an unsigned
3678 integer variable. An explicit cast silences the warning. In C, this
3679 option is enabled also by @option{-Wconversion}.
3683 @opindex Wno-address
3684 Warn about suspicious uses of memory addresses. These include using
3685 the address of a function in a conditional expression, such as
3686 @code{void func(void); if (func)}, and comparisons against the memory
3687 address of a string literal, such as @code{if (x == "abc")}. Such
3688 uses typically indicate a programmer error: the address of a function
3689 always evaluates to true, so their use in a conditional usually
3690 indicate that the programmer forgot the parentheses in a function
3691 call; and comparisons against string literals result in unspecified
3692 behavior and are not portable in C, so they usually indicate that the
3693 programmer intended to use @code{strcmp}. This warning is enabled by
3697 @opindex Wlogical-op
3698 @opindex Wno-logical-op
3699 Warn about suspicious uses of logical operators in expressions.
3700 This includes using logical operators in contexts where a
3701 bit-wise operator is likely to be expected.
3703 @item -Waggregate-return
3704 @opindex Waggregate-return
3705 @opindex Wno-aggregate-return
3706 Warn if any functions that return structures or unions are defined or
3707 called. (In languages where you can return an array, this also elicits
3710 @item -Wno-attributes
3711 @opindex Wno-attributes
3712 @opindex Wattributes
3713 Do not warn if an unexpected @code{__attribute__} is used, such as
3714 unrecognized attributes, function attributes applied to variables,
3715 etc. This will not stop errors for incorrect use of supported
3718 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3719 @opindex Wstrict-prototypes
3720 @opindex Wno-strict-prototypes
3721 Warn if a function is declared or defined without specifying the
3722 argument types. (An old-style function definition is permitted without
3723 a warning if preceded by a declaration which specifies the argument
3726 @item -Wold-style-declaration @r{(C and Objective-C only)}
3727 @opindex Wold-style-declaration
3728 @opindex Wno-old-style-declaration
3729 Warn for obsolescent usages, according to the C Standard, in a
3730 declaration. For example, warn if storage-class specifiers like
3731 @code{static} are not the first things in a declaration. This warning
3732 is also enabled by @option{-Wextra}.
3734 @item -Wold-style-definition @r{(C and Objective-C only)}
3735 @opindex Wold-style-definition
3736 @opindex Wno-old-style-definition
3737 Warn if an old-style function definition is used. A warning is given
3738 even if there is a previous prototype.
3740 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3741 @opindex Wmissing-parameter-type
3742 @opindex Wno-missing-parameter-type
3743 A function parameter is declared without a type specifier in K&R-style
3750 This warning is also enabled by @option{-Wextra}.
3752 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3753 @opindex Wmissing-prototypes
3754 @opindex Wno-missing-prototypes
3755 Warn if a global function is defined without a previous prototype
3756 declaration. This warning is issued even if the definition itself
3757 provides a prototype. The aim is to detect global functions that fail
3758 to be declared in header files.
3760 @item -Wmissing-declarations
3761 @opindex Wmissing-declarations
3762 @opindex Wno-missing-declarations
3763 Warn if a global function is defined without a previous declaration.
3764 Do so even if the definition itself provides a prototype.
3765 Use this option to detect global functions that are not declared in
3766 header files. In C++, no warnings are issued for function templates,
3767 or for inline functions, or for functions in anonymous namespaces.
3769 @item -Wmissing-field-initializers
3770 @opindex Wmissing-field-initializers
3771 @opindex Wno-missing-field-initializers
3775 Warn if a structure's initializer has some fields missing. For
3776 example, the following code would cause such a warning, because
3777 @code{x.h} is implicitly zero:
3780 struct s @{ int f, g, h; @};
3781 struct s x = @{ 3, 4 @};
3784 This option does not warn about designated initializers, so the following
3785 modification would not trigger a warning:
3788 struct s @{ int f, g, h; @};
3789 struct s x = @{ .f = 3, .g = 4 @};
3792 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3793 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3795 @item -Wmissing-noreturn
3796 @opindex Wmissing-noreturn
3797 @opindex Wno-missing-noreturn
3798 Warn about functions which might be candidates for attribute @code{noreturn}.
3799 Note these are only possible candidates, not absolute ones. Care should
3800 be taken to manually verify functions actually do not ever return before
3801 adding the @code{noreturn} attribute, otherwise subtle code generation
3802 bugs could be introduced. You will not get a warning for @code{main} in
3803 hosted C environments.
3805 @item -Wmissing-format-attribute
3806 @opindex Wmissing-format-attribute
3807 @opindex Wno-missing-format-attribute
3810 Warn about function pointers which might be candidates for @code{format}
3811 attributes. Note these are only possible candidates, not absolute ones.
3812 GCC will guess that function pointers with @code{format} attributes that
3813 are used in assignment, initialization, parameter passing or return
3814 statements should have a corresponding @code{format} attribute in the
3815 resulting type. I.e.@: the left-hand side of the assignment or
3816 initialization, the type of the parameter variable, or the return type
3817 of the containing function respectively should also have a @code{format}
3818 attribute to avoid the warning.
3820 GCC will also warn about function definitions which might be
3821 candidates for @code{format} attributes. Again, these are only
3822 possible candidates. GCC will guess that @code{format} attributes
3823 might be appropriate for any function that calls a function like
3824 @code{vprintf} or @code{vscanf}, but this might not always be the
3825 case, and some functions for which @code{format} attributes are
3826 appropriate may not be detected.
3828 @item -Wno-multichar
3829 @opindex Wno-multichar
3831 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3832 Usually they indicate a typo in the user's code, as they have
3833 implementation-defined values, and should not be used in portable code.
3835 @item -Wnormalized=<none|id|nfc|nfkc>
3836 @opindex Wnormalized=
3839 @cindex character set, input normalization
3840 In ISO C and ISO C++, two identifiers are different if they are
3841 different sequences of characters. However, sometimes when characters
3842 outside the basic ASCII character set are used, you can have two
3843 different character sequences that look the same. To avoid confusion,
3844 the ISO 10646 standard sets out some @dfn{normalization rules} which
3845 when applied ensure that two sequences that look the same are turned into
3846 the same sequence. GCC can warn you if you are using identifiers which
3847 have not been normalized; this option controls that warning.
3849 There are four levels of warning that GCC supports. The default is
3850 @option{-Wnormalized=nfc}, which warns about any identifier which is
3851 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3852 recommended form for most uses.
3854 Unfortunately, there are some characters which ISO C and ISO C++ allow
3855 in identifiers that when turned into NFC aren't allowable as
3856 identifiers. That is, there's no way to use these symbols in portable
3857 ISO C or C++ and have all your identifiers in NFC@.
3858 @option{-Wnormalized=id} suppresses the warning for these characters.
3859 It is hoped that future versions of the standards involved will correct
3860 this, which is why this option is not the default.
3862 You can switch the warning off for all characters by writing
3863 @option{-Wnormalized=none}. You would only want to do this if you
3864 were using some other normalization scheme (like ``D''), because
3865 otherwise you can easily create bugs that are literally impossible to see.
3867 Some characters in ISO 10646 have distinct meanings but look identical
3868 in some fonts or display methodologies, especially once formatting has
3869 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3870 LETTER N'', will display just like a regular @code{n} which has been
3871 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3872 normalization scheme to convert all these into a standard form as
3873 well, and GCC will warn if your code is not in NFKC if you use
3874 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3875 about every identifier that contains the letter O because it might be
3876 confused with the digit 0, and so is not the default, but may be
3877 useful as a local coding convention if the programming environment is
3878 unable to be fixed to display these characters distinctly.
3880 @item -Wno-deprecated
3881 @opindex Wno-deprecated
3882 @opindex Wdeprecated
3883 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
3885 @item -Wno-deprecated-declarations
3886 @opindex Wno-deprecated-declarations
3887 @opindex Wdeprecated-declarations
3888 Do not warn about uses of functions (@pxref{Function Attributes}),
3889 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3890 Attributes}) marked as deprecated by using the @code{deprecated}
3894 @opindex Wno-overflow
3896 Do not warn about compile-time overflow in constant expressions.
3898 @item -Woverride-init @r{(C and Objective-C only)}
3899 @opindex Woverride-init
3900 @opindex Wno-override-init
3904 Warn if an initialized field without side effects is overridden when
3905 using designated initializers (@pxref{Designated Inits, , Designated
3908 This warning is included in @option{-Wextra}. To get other
3909 @option{-Wextra} warnings without this one, use @samp{-Wextra
3910 -Wno-override-init}.
3915 Warn if a structure is given the packed attribute, but the packed
3916 attribute has no effect on the layout or size of the structure.
3917 Such structures may be mis-aligned for little benefit. For
3918 instance, in this code, the variable @code{f.x} in @code{struct bar}
3919 will be misaligned even though @code{struct bar} does not itself
3920 have the packed attribute:
3927 @} __attribute__((packed));
3938 Warn if padding is included in a structure, either to align an element
3939 of the structure or to align the whole structure. Sometimes when this
3940 happens it is possible to rearrange the fields of the structure to
3941 reduce the padding and so make the structure smaller.
3943 @item -Wredundant-decls
3944 @opindex Wredundant-decls
3945 @opindex Wno-redundant-decls
3946 Warn if anything is declared more than once in the same scope, even in
3947 cases where multiple declaration is valid and changes nothing.
3949 @item -Wnested-externs @r{(C and Objective-C only)}
3950 @opindex Wnested-externs
3951 @opindex Wno-nested-externs
3952 Warn if an @code{extern} declaration is encountered within a function.
3954 @item -Wunreachable-code
3955 @opindex Wunreachable-code
3956 @opindex Wno-unreachable-code
3957 Warn if the compiler detects that code will never be executed.
3959 This option is intended to warn when the compiler detects that at
3960 least a whole line of source code will never be executed, because
3961 some condition is never satisfied or because it is after a
3962 procedure that never returns.
3964 It is possible for this option to produce a warning even though there
3965 are circumstances under which part of the affected line can be executed,
3966 so care should be taken when removing apparently-unreachable code.
3968 For instance, when a function is inlined, a warning may mean that the
3969 line is unreachable in only one inlined copy of the function.
3971 This option is not made part of @option{-Wall} because in a debugging
3972 version of a program there is often substantial code which checks
3973 correct functioning of the program and is, hopefully, unreachable
3974 because the program does work. Another common use of unreachable
3975 code is to provide behavior which is selectable at compile-time.
3980 Warn if a function can not be inlined and it was declared as inline.
3981 Even with this option, the compiler will not warn about failures to
3982 inline functions declared in system headers.
3984 The compiler uses a variety of heuristics to determine whether or not
3985 to inline a function. For example, the compiler takes into account
3986 the size of the function being inlined and the amount of inlining
3987 that has already been done in the current function. Therefore,
3988 seemingly insignificant changes in the source program can cause the
3989 warnings produced by @option{-Winline} to appear or disappear.
3991 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
3992 @opindex Wno-invalid-offsetof
3993 @opindex Winvalid-offsetof
3994 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3995 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3996 to a non-POD type is undefined. In existing C++ implementations,
3997 however, @samp{offsetof} typically gives meaningful results even when
3998 applied to certain kinds of non-POD types. (Such as a simple
3999 @samp{struct} that fails to be a POD type only by virtue of having a
4000 constructor.) This flag is for users who are aware that they are
4001 writing nonportable code and who have deliberately chosen to ignore the
4004 The restrictions on @samp{offsetof} may be relaxed in a future version
4005 of the C++ standard.
4007 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4008 @opindex Wno-int-to-pointer-cast
4009 @opindex Wint-to-pointer-cast
4010 Suppress warnings from casts to pointer type of an integer of a
4013 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4014 @opindex Wno-pointer-to-int-cast
4015 @opindex Wpointer-to-int-cast
4016 Suppress warnings from casts from a pointer to an integer type of a
4020 @opindex Winvalid-pch
4021 @opindex Wno-invalid-pch
4022 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4023 the search path but can't be used.
4027 @opindex Wno-long-long
4028 Warn if @samp{long long} type is used. This is default. To inhibit
4029 the warning messages, use @option{-Wno-long-long}. Flags
4030 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
4031 only when @option{-pedantic} flag is used.
4033 @item -Wvariadic-macros
4034 @opindex Wvariadic-macros
4035 @opindex Wno-variadic-macros
4036 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4037 alternate syntax when in pedantic ISO C99 mode. This is default.
4038 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4043 Warn if variable length array is used in the code.
4044 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4045 the variable length array.
4047 @item -Wvolatile-register-var
4048 @opindex Wvolatile-register-var
4049 @opindex Wno-volatile-register-var
4050 Warn if a register variable is declared volatile. The volatile
4051 modifier does not inhibit all optimizations that may eliminate reads
4052 and/or writes to register variables. This warning is enabled by
4055 @item -Wdisabled-optimization
4056 @opindex Wdisabled-optimization
4057 @opindex Wno-disabled-optimization
4058 Warn if a requested optimization pass is disabled. This warning does
4059 not generally indicate that there is anything wrong with your code; it
4060 merely indicates that GCC's optimizers were unable to handle the code
4061 effectively. Often, the problem is that your code is too big or too
4062 complex; GCC will refuse to optimize programs when the optimization
4063 itself is likely to take inordinate amounts of time.
4065 @item -Wpointer-sign @r{(C and Objective-C only)}
4066 @opindex Wpointer-sign
4067 @opindex Wno-pointer-sign
4068 Warn for pointer argument passing or assignment with different signedness.
4069 This option is only supported for C and Objective-C@. It is implied by
4070 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4071 @option{-Wno-pointer-sign}.
4073 @item -Wstack-protector
4074 @opindex Wstack-protector
4075 @opindex Wno-stack-protector
4076 This option is only active when @option{-fstack-protector} is active. It
4077 warns about functions that will not be protected against stack smashing.
4080 @opindex Wno-mudflap
4081 Suppress warnings about constructs that cannot be instrumented by
4084 @item -Woverlength-strings
4085 @opindex Woverlength-strings
4086 @opindex Wno-overlength-strings
4087 Warn about string constants which are longer than the ``minimum
4088 maximum'' length specified in the C standard. Modern compilers
4089 generally allow string constants which are much longer than the
4090 standard's minimum limit, but very portable programs should avoid
4091 using longer strings.
4093 The limit applies @emph{after} string constant concatenation, and does
4094 not count the trailing NUL@. In C89, the limit was 509 characters; in
4095 C99, it was raised to 4095. C++98 does not specify a normative
4096 minimum maximum, so we do not diagnose overlength strings in C++@.
4098 This option is implied by @option{-pedantic}, and can be disabled with
4099 @option{-Wno-overlength-strings}.
4101 @item -Wdisallowed-function-list=@var{sym},@var{sym},@dots{}
4102 @opindex Wdisallowed-function-list
4104 If any of @var{sym} is called, GCC will issue a warning. This can be useful
4105 in enforcing coding conventions that ban calls to certain functions, for
4106 example, @code{alloca}, @code{malloc}, etc.
4109 @node Debugging Options
4110 @section Options for Debugging Your Program or GCC
4111 @cindex options, debugging
4112 @cindex debugging information options
4114 GCC has various special options that are used for debugging
4115 either your program or GCC:
4120 Produce debugging information in the operating system's native format
4121 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4124 On most systems that use stabs format, @option{-g} enables use of extra
4125 debugging information that only GDB can use; this extra information
4126 makes debugging work better in GDB but will probably make other debuggers
4128 refuse to read the program. If you want to control for certain whether
4129 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4130 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4132 GCC allows you to use @option{-g} with
4133 @option{-O}. The shortcuts taken by optimized code may occasionally
4134 produce surprising results: some variables you declared may not exist
4135 at all; flow of control may briefly move where you did not expect it;
4136 some statements may not be executed because they compute constant
4137 results or their values were already at hand; some statements may
4138 execute in different places because they were moved out of loops.
4140 Nevertheless it proves possible to debug optimized output. This makes
4141 it reasonable to use the optimizer for programs that might have bugs.
4143 The following options are useful when GCC is generated with the
4144 capability for more than one debugging format.
4148 Produce debugging information for use by GDB@. This means to use the
4149 most expressive format available (DWARF 2, stabs, or the native format
4150 if neither of those are supported), including GDB extensions if at all
4155 Produce debugging information in stabs format (if that is supported),
4156 without GDB extensions. This is the format used by DBX on most BSD
4157 systems. On MIPS, Alpha and System V Release 4 systems this option
4158 produces stabs debugging output which is not understood by DBX or SDB@.
4159 On System V Release 4 systems this option requires the GNU assembler.
4161 @item -feliminate-unused-debug-symbols
4162 @opindex feliminate-unused-debug-symbols
4163 Produce debugging information in stabs format (if that is supported),
4164 for only symbols that are actually used.
4166 @item -femit-class-debug-always
4167 Instead of emitting debugging information for a C++ class in only one
4168 object file, emit it in all object files using the class. This option
4169 should be used only with debuggers that are unable to handle the way GCC
4170 normally emits debugging information for classes because using this
4171 option will increase the size of debugging information by as much as a
4176 Produce debugging information in stabs format (if that is supported),
4177 using GNU extensions understood only by the GNU debugger (GDB)@. The
4178 use of these extensions is likely to make other debuggers crash or
4179 refuse to read the program.
4183 Produce debugging information in COFF format (if that is supported).
4184 This is the format used by SDB on most System V systems prior to
4189 Produce debugging information in XCOFF format (if that is supported).
4190 This is the format used by the DBX debugger on IBM RS/6000 systems.
4194 Produce debugging information in XCOFF format (if that is supported),
4195 using GNU extensions understood only by the GNU debugger (GDB)@. The
4196 use of these extensions is likely to make other debuggers crash or
4197 refuse to read the program, and may cause assemblers other than the GNU
4198 assembler (GAS) to fail with an error.
4202 Produce debugging information in DWARF version 2 format (if that is
4203 supported). This is the format used by DBX on IRIX 6. With this
4204 option, GCC uses features of DWARF version 3 when they are useful;
4205 version 3 is upward compatible with version 2, but may still cause
4206 problems for older debuggers.
4210 Produce debugging information in VMS debug format (if that is
4211 supported). This is the format used by DEBUG on VMS systems.
4214 @itemx -ggdb@var{level}
4215 @itemx -gstabs@var{level}
4216 @itemx -gcoff@var{level}
4217 @itemx -gxcoff@var{level}
4218 @itemx -gvms@var{level}
4219 Request debugging information and also use @var{level} to specify how
4220 much information. The default level is 2.
4222 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4225 Level 1 produces minimal information, enough for making backtraces in
4226 parts of the program that you don't plan to debug. This includes
4227 descriptions of functions and external variables, but no information
4228 about local variables and no line numbers.
4230 Level 3 includes extra information, such as all the macro definitions
4231 present in the program. Some debuggers support macro expansion when
4232 you use @option{-g3}.
4234 @option{-gdwarf-2} does not accept a concatenated debug level, because
4235 GCC used to support an option @option{-gdwarf} that meant to generate
4236 debug information in version 1 of the DWARF format (which is very
4237 different from version 2), and it would have been too confusing. That
4238 debug format is long obsolete, but the option cannot be changed now.
4239 Instead use an additional @option{-g@var{level}} option to change the
4240 debug level for DWARF2.
4242 @item -feliminate-dwarf2-dups
4243 @opindex feliminate-dwarf2-dups
4244 Compress DWARF2 debugging information by eliminating duplicated
4245 information about each symbol. This option only makes sense when
4246 generating DWARF2 debugging information with @option{-gdwarf-2}.
4248 @item -femit-struct-debug-baseonly
4249 Emit debug information for struct-like types
4250 only when the base name of the compilation source file
4251 matches the base name of file in which the struct was defined.
4253 This option substantially reduces the size of debugging information,
4254 but at significant potential loss in type information to the debugger.
4255 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4256 See @option{-femit-struct-debug-detailed} for more detailed control.
4258 This option works only with DWARF 2.
4260 @item -femit-struct-debug-reduced
4261 Emit debug information for struct-like types
4262 only when the base name of the compilation source file
4263 matches the base name of file in which the type was defined,
4264 unless the struct is a template or defined in a system header.
4266 This option significantly reduces the size of debugging information,
4267 with some potential loss in type information to the debugger.
4268 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4269 See @option{-femit-struct-debug-detailed} for more detailed control.
4271 This option works only with DWARF 2.
4273 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4274 Specify the struct-like types
4275 for which the compiler will generate debug information.
4276 The intent is to reduce duplicate struct debug information
4277 between different object files within the same program.
4279 This option is a detailed version of
4280 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4281 which will serve for most needs.
4283 A specification has the syntax
4284 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4286 The optional first word limits the specification to
4287 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4288 A struct type is used directly when it is the type of a variable, member.
4289 Indirect uses arise through pointers to structs.
4290 That is, when use of an incomplete struct would be legal, the use is indirect.
4292 @samp{struct one direct; struct two * indirect;}.
4294 The optional second word limits the specification to
4295 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4296 Generic structs are a bit complicated to explain.
4297 For C++, these are non-explicit specializations of template classes,
4298 or non-template classes within the above.
4299 Other programming languages have generics,
4300 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4302 The third word specifies the source files for those
4303 structs for which the compiler will emit debug information.
4304 The values @samp{none} and @samp{any} have the normal meaning.
4305 The value @samp{base} means that
4306 the base of name of the file in which the type declaration appears
4307 must match the base of the name of the main compilation file.
4308 In practice, this means that
4309 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4310 but types declared in other header will not.
4311 The value @samp{sys} means those types satisfying @samp{base}
4312 or declared in system or compiler headers.
4314 You may need to experiment to determine the best settings for your application.
4316 The default is @samp{-femit-struct-debug-detailed=all}.
4318 This option works only with DWARF 2.
4320 @item -fno-merge-debug-strings
4321 @opindex fmerge-debug-strings
4322 @opindex fno-merge-debug-strings
4323 Direct the linker to merge together strings which are identical in
4324 different object files. This is not supported by all assemblers or
4325 linker. This decreases the size of the debug information in the
4326 output file at the cost of increasing link processing time. This is
4329 @item -fdebug-prefix-map=@var{old}=@var{new}
4330 @opindex fdebug-prefix-map
4331 When compiling files in directory @file{@var{old}}, record debugging
4332 information describing them as in @file{@var{new}} instead.
4334 @cindex @command{prof}
4337 Generate extra code to write profile information suitable for the
4338 analysis program @command{prof}. You must use this option when compiling
4339 the source files you want data about, and you must also use it when
4342 @cindex @command{gprof}
4345 Generate extra code to write profile information suitable for the
4346 analysis program @command{gprof}. You must use this option when compiling
4347 the source files you want data about, and you must also use it when
4352 Makes the compiler print out each function name as it is compiled, and
4353 print some statistics about each pass when it finishes.
4356 @opindex ftime-report
4357 Makes the compiler print some statistics about the time consumed by each
4358 pass when it finishes.
4361 @opindex fmem-report
4362 Makes the compiler print some statistics about permanent memory
4363 allocation when it finishes.
4365 @item -fpre-ipa-mem-report
4366 @opindex fpre-ipa-mem-report
4367 @item -fpost-ipa-mem-report
4368 @opindex fpost-ipa-mem-report
4369 Makes the compiler print some statistics about permanent memory
4370 allocation before or after interprocedural optimization.
4372 @item -fprofile-arcs
4373 @opindex fprofile-arcs
4374 Add code so that program flow @dfn{arcs} are instrumented. During
4375 execution the program records how many times each branch and call is
4376 executed and how many times it is taken or returns. When the compiled
4377 program exits it saves this data to a file called
4378 @file{@var{auxname}.gcda} for each source file. The data may be used for
4379 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4380 test coverage analysis (@option{-ftest-coverage}). Each object file's
4381 @var{auxname} is generated from the name of the output file, if
4382 explicitly specified and it is not the final executable, otherwise it is
4383 the basename of the source file. In both cases any suffix is removed
4384 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4385 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4386 @xref{Cross-profiling}.
4388 @cindex @command{gcov}
4392 This option is used to compile and link code instrumented for coverage
4393 analysis. The option is a synonym for @option{-fprofile-arcs}
4394 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4395 linking). See the documentation for those options for more details.
4400 Compile the source files with @option{-fprofile-arcs} plus optimization
4401 and code generation options. For test coverage analysis, use the
4402 additional @option{-ftest-coverage} option. You do not need to profile
4403 every source file in a program.
4406 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4407 (the latter implies the former).
4410 Run the program on a representative workload to generate the arc profile
4411 information. This may be repeated any number of times. You can run
4412 concurrent instances of your program, and provided that the file system
4413 supports locking, the data files will be correctly updated. Also
4414 @code{fork} calls are detected and correctly handled (double counting
4418 For profile-directed optimizations, compile the source files again with
4419 the same optimization and code generation options plus
4420 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4421 Control Optimization}).
4424 For test coverage analysis, use @command{gcov} to produce human readable
4425 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4426 @command{gcov} documentation for further information.
4430 With @option{-fprofile-arcs}, for each function of your program GCC
4431 creates a program flow graph, then finds a spanning tree for the graph.
4432 Only arcs that are not on the spanning tree have to be instrumented: the
4433 compiler adds code to count the number of times that these arcs are
4434 executed. When an arc is the only exit or only entrance to a block, the
4435 instrumentation code can be added to the block; otherwise, a new basic
4436 block must be created to hold the instrumentation code.
4439 @item -ftest-coverage
4440 @opindex ftest-coverage
4441 Produce a notes file that the @command{gcov} code-coverage utility
4442 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4443 show program coverage. Each source file's note file is called
4444 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4445 above for a description of @var{auxname} and instructions on how to
4446 generate test coverage data. Coverage data will match the source files
4447 more closely, if you do not optimize.
4449 @item -fdbg-cnt-list
4450 @opindex fdbg-cnt-list
4451 Print the name and the counter upperbound for all debug counters.
4453 @item -fdbg-cnt=@var{counter-value-list}
4455 Set the internal debug counter upperbound. @var{counter-value-list}
4456 is a comma-separated list of @var{name}:@var{value} pairs
4457 which sets the upperbound of each debug counter @var{name} to @var{value}.
4458 All debug counters have the initial upperbound of @var{UINT_MAX},
4459 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4460 e.g. With -fdbg-cnt=dce:10,tail_call:0
4461 dbg_cnt(dce) will return true only for first 10 invocations
4462 and dbg_cnt(tail_call) will return false always.
4464 @item -d@var{letters}
4465 @itemx -fdump-rtl-@var{pass}
4467 Says to make debugging dumps during compilation at times specified by
4468 @var{letters}. This is used for debugging the RTL-based passes of the
4469 compiler. The file names for most of the dumps are made by appending a
4470 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4471 from the name of the output file, if explicitly specified and it is not
4472 an executable, otherwise it is the basename of the source file. These
4473 switches may have different effects when @option{-E} is used for
4476 Most debug dumps can be enabled either passing a letter to the @option{-d}
4477 option, or with a long @option{-fdump-rtl} switch; here are the possible
4478 letters for use in @var{letters} and @var{pass}, and their meanings:
4483 Annotate the assembler output with miscellaneous debugging information.
4485 @item -fdump-rtl-bbro
4486 @opindex fdump-rtl-bbro
4487 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4489 @item -fdump-rtl-combine
4490 @opindex fdump-rtl-combine
4491 Dump after the RTL instruction combination pass, to the file
4492 @file{@var{file}.129r.combine}.
4494 @item -fdump-rtl-ce1
4495 @itemx -fdump-rtl-ce2
4496 @opindex fdump-rtl-ce1
4497 @opindex fdump-rtl-ce2
4498 @option{-fdump-rtl-ce1} enable dumping after the
4499 first if conversion, to the file @file{@var{file}.117r.ce1}.
4500 @option{-fdump-rtl-ce2} enable dumping after the second if
4501 conversion, to the file @file{@var{file}.130r.ce2}.
4503 @item -fdump-rtl-btl
4504 @itemx -fdump-rtl-dbr
4505 @opindex fdump-rtl-btl
4506 @opindex fdump-rtl-dbr
4507 @option{-fdump-rtl-btl} enable dumping after branch
4508 target load optimization, to @file{@var{file}.31.btl}.
4509 @option{-fdump-rtl-dbr} enable dumping after delayed branch
4510 scheduling, to @file{@var{file}.36.dbr}.
4514 Dump all macro definitions, at the end of preprocessing, in addition to
4517 @item -fdump-rtl-ce3
4518 @opindex fdump-rtl-ce3
4519 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4521 @item -fdump-rtl-cfg
4522 @itemx -fdump-rtl-life
4523 @opindex fdump-rtl-cfg
4524 @opindex fdump-rtl-life
4525 @option{-fdump-rtl-cfg} enable dumping after control
4526 and data flow analysis, to @file{@var{file}.116r.cfg}.
4527 @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4528 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4530 @item -fdump-rtl-greg
4531 @opindex fdump-rtl-greg
4532 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4534 @item -fdump-rtl-gcse
4535 @itemx -fdump-rtl-bypass
4536 @opindex fdump-rtl-gcse
4537 @opindex fdump-rtl-bypass
4538 @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4539 @file{@var{file}.114r.gcse}. @option{-fdump-rtl-bypass}
4540 enable dumping after jump bypassing and control flow optimizations, to
4541 @file{@var{file}.115r.bypass}.
4544 @opindex fdump-rtl-eh
4545 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4547 @item -fdump-rtl-sibling
4548 @opindex fdump-rtl-sibling
4549 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4551 @item -fdump-rtl-jump
4552 @opindex fdump-rtl-jump
4553 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4555 @item -fdump-rtl-stack
4556 @opindex fdump-rtl-stack
4557 Dump after conversion from GCC's "flat register file" registers to the
4558 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4560 @item -fdump-rtl-lreg
4561 @opindex fdump-rtl-lreg
4562 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4564 @item -fdump-rtl-loop2
4565 @opindex fdump-rtl-loop2
4566 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4567 loop optimization pass, to @file{@var{file}.119r.loop2},
4568 @file{@var{file}.120r.loop2_init},
4569 @file{@var{file}.121r.loop2_invariant}, and
4570 @file{@var{file}.125r.loop2_done}.
4572 @item -fdump-rtl-sms
4573 @opindex fdump-rtl-sms
4574 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4576 @item -fdump-rtl-mach
4577 @opindex fdump-rtl-mach
4578 Dump after performing the machine dependent reorganization pass, to
4579 @file{@var{file}.155r.mach} if that pass exists.
4581 @item -fdump-rtl-rnreg
4582 @opindex fdump-rtl-rnreg
4583 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4585 @item -fdump-rtl-regmove
4586 @opindex fdump-rtl-regmove
4587 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4589 @item -fdump-rtl-postreload
4590 @opindex fdump-rtl-postreload
4591 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4593 @item -fdump-rtl-expand
4594 @opindex fdump-rtl-expand
4595 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4597 @item -fdump-rtl-sched2
4598 @opindex fdump-rtl-sched2
4599 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4601 @item -fdump-rtl-cse
4602 @opindex fdump-rtl-cse
4603 Dump after CSE (including the jump optimization that sometimes follows
4604 CSE), to @file{@var{file}.113r.cse}.
4606 @item -fdump-rtl-sched1
4607 @opindex fdump-rtl-sched1
4608 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4610 @item -fdump-rtl-cse2
4611 @opindex fdump-rtl-cse2
4612 Dump after the second CSE pass (including the jump optimization that
4613 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4615 @item -fdump-rtl-tracer
4616 @opindex fdump-rtl-tracer
4617 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4619 @item -fdump-rtl-vpt
4620 @itemx -fdump-rtl-vartrack
4621 @opindex fdump-rtl-vpt
4622 @opindex fdump-rtl-vartrack
4623 @option{-fdump-rtl-vpt} enable dumping after the value
4624 profile transformations, to @file{@var{file}.10.vpt}.
4625 @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4626 to @file{@var{file}.154r.vartrack}.
4628 @item -fdump-rtl-flow2
4629 @opindex fdump-rtl-flow2
4630 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4632 @item -fdump-rtl-peephole2
4633 @opindex fdump-rtl-peephole2
4634 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4636 @item -fdump-rtl-web
4637 @opindex fdump-rtl-web
4638 Dump after live range splitting, to @file{@var{file}.126r.web}.
4640 @item -fdump-rtl-all
4641 @opindex fdump-rtl-all
4642 Produce all the dumps listed above.
4646 Produce a core dump whenever an error occurs.
4650 Print statistics on memory usage, at the end of the run, to
4655 Annotate the assembler output with a comment indicating which
4656 pattern and alternative was used. The length of each instruction is
4661 Dump the RTL in the assembler output as a comment before each instruction.
4662 Also turns on @option{-dp} annotation.
4666 For each of the other indicated dump files (either with @option{-d} or
4667 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4668 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4672 Just generate RTL for a function instead of compiling it. Usually used
4673 with @samp{r} (@option{-fdump-rtl-expand}).
4677 Dump debugging information during parsing, to standard error.
4681 @opindex fdump-noaddr
4682 When doing debugging dumps (see @option{-d} option above), suppress
4683 address output. This makes it more feasible to use diff on debugging
4684 dumps for compiler invocations with different compiler binaries and/or
4685 different text / bss / data / heap / stack / dso start locations.
4687 @item -fdump-unnumbered
4688 @opindex fdump-unnumbered
4689 When doing debugging dumps (see @option{-d} option above), suppress instruction
4690 numbers and address output. This makes it more feasible to
4691 use diff on debugging dumps for compiler invocations with different
4692 options, in particular with and without @option{-g}.
4694 @item -fdump-translation-unit @r{(C++ only)}
4695 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4696 @opindex fdump-translation-unit
4697 Dump a representation of the tree structure for the entire translation
4698 unit to a file. The file name is made by appending @file{.tu} to the
4699 source file name. If the @samp{-@var{options}} form is used, @var{options}
4700 controls the details of the dump as described for the
4701 @option{-fdump-tree} options.
4703 @item -fdump-class-hierarchy @r{(C++ only)}
4704 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4705 @opindex fdump-class-hierarchy
4706 Dump a representation of each class's hierarchy and virtual function
4707 table layout to a file. The file name is made by appending @file{.class}
4708 to the source file name. If the @samp{-@var{options}} form is used,
4709 @var{options} controls the details of the dump as described for the
4710 @option{-fdump-tree} options.
4712 @item -fdump-ipa-@var{switch}
4714 Control the dumping at various stages of inter-procedural analysis
4715 language tree to a file. The file name is generated by appending a switch
4716 specific suffix to the source file name. The following dumps are possible:
4720 Enables all inter-procedural analysis dumps.
4723 Dumps information about call-graph optimization, unused function removal,
4724 and inlining decisions.
4727 Dump after function inlining.
4731 @item -fdump-statistics-@var{option}
4732 @opindex -fdump-statistics
4733 Enable and control dumping of pass statistics in a separate file. The
4734 file name is generated by appending a suffix ending in @samp{.statistics}
4735 to the source file name. If the @samp{-@var{option}} form is used,
4736 @samp{-stats} will cause counters to be summed over the whole compilation unit
4737 while @samp{-details} will dump every event as the passes generate them.
4738 The default with no option is to sum counters for each function compiled.
4740 @item -fdump-tree-@var{switch}
4741 @itemx -fdump-tree-@var{switch}-@var{options}
4743 Control the dumping at various stages of processing the intermediate
4744 language tree to a file. The file name is generated by appending a switch
4745 specific suffix to the source file name. If the @samp{-@var{options}}
4746 form is used, @var{options} is a list of @samp{-} separated options that
4747 control the details of the dump. Not all options are applicable to all
4748 dumps, those which are not meaningful will be ignored. The following
4749 options are available
4753 Print the address of each node. Usually this is not meaningful as it
4754 changes according to the environment and source file. Its primary use
4755 is for tying up a dump file with a debug environment.
4757 Inhibit dumping of members of a scope or body of a function merely
4758 because that scope has been reached. Only dump such items when they
4759 are directly reachable by some other path. When dumping pretty-printed
4760 trees, this option inhibits dumping the bodies of control structures.
4762 Print a raw representation of the tree. By default, trees are
4763 pretty-printed into a C-like representation.
4765 Enable more detailed dumps (not honored by every dump option).
4767 Enable dumping various statistics about the pass (not honored by every dump
4770 Enable showing basic block boundaries (disabled in raw dumps).
4772 Enable showing virtual operands for every statement.
4774 Enable showing line numbers for statements.
4776 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4778 Enable showing the tree dump for each statement.
4780 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
4781 and @option{lineno}.
4784 The following tree dumps are possible:
4788 Dump before any tree based optimization, to @file{@var{file}.original}.
4791 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4794 @opindex fdump-tree-gimple
4795 Dump each function before and after the gimplification pass to a file. The
4796 file name is made by appending @file{.gimple} to the source file name.
4799 @opindex fdump-tree-cfg
4800 Dump the control flow graph of each function to a file. The file name is
4801 made by appending @file{.cfg} to the source file name.
4804 @opindex fdump-tree-vcg
4805 Dump the control flow graph of each function to a file in VCG format. The
4806 file name is made by appending @file{.vcg} to the source file name. Note
4807 that if the file contains more than one function, the generated file cannot
4808 be used directly by VCG@. You will need to cut and paste each function's
4809 graph into its own separate file first.
4812 @opindex fdump-tree-ch
4813 Dump each function after copying loop headers. The file name is made by
4814 appending @file{.ch} to the source file name.
4817 @opindex fdump-tree-ssa
4818 Dump SSA related information to a file. The file name is made by appending
4819 @file{.ssa} to the source file name.
4822 @opindex fdump-tree-alias
4823 Dump aliasing information for each function. The file name is made by
4824 appending @file{.alias} to the source file name.
4827 @opindex fdump-tree-ccp
4828 Dump each function after CCP@. The file name is made by appending
4829 @file{.ccp} to the source file name.
4832 @opindex fdump-tree-storeccp
4833 Dump each function after STORE-CCP@. The file name is made by appending
4834 @file{.storeccp} to the source file name.
4837 @opindex fdump-tree-pre
4838 Dump trees after partial redundancy elimination. The file name is made
4839 by appending @file{.pre} to the source file name.
4842 @opindex fdump-tree-fre
4843 Dump trees after full redundancy elimination. The file name is made
4844 by appending @file{.fre} to the source file name.
4847 @opindex fdump-tree-copyprop
4848 Dump trees after copy propagation. The file name is made
4849 by appending @file{.copyprop} to the source file name.
4851 @item store_copyprop
4852 @opindex fdump-tree-store_copyprop
4853 Dump trees after store copy-propagation. The file name is made
4854 by appending @file{.store_copyprop} to the source file name.
4857 @opindex fdump-tree-dce
4858 Dump each function after dead code elimination. The file name is made by
4859 appending @file{.dce} to the source file name.
4862 @opindex fdump-tree-mudflap
4863 Dump each function after adding mudflap instrumentation. The file name is
4864 made by appending @file{.mudflap} to the source file name.
4867 @opindex fdump-tree-sra
4868 Dump each function after performing scalar replacement of aggregates. The
4869 file name is made by appending @file{.sra} to the source file name.
4872 @opindex fdump-tree-sink
4873 Dump each function after performing code sinking. The file name is made
4874 by appending @file{.sink} to the source file name.
4877 @opindex fdump-tree-dom
4878 Dump each function after applying dominator tree optimizations. The file
4879 name is made by appending @file{.dom} to the source file name.
4882 @opindex fdump-tree-dse
4883 Dump each function after applying dead store elimination. The file
4884 name is made by appending @file{.dse} to the source file name.
4887 @opindex fdump-tree-phiopt
4888 Dump each function after optimizing PHI nodes into straightline code. The file
4889 name is made by appending @file{.phiopt} to the source file name.
4892 @opindex fdump-tree-forwprop
4893 Dump each function after forward propagating single use variables. The file
4894 name is made by appending @file{.forwprop} to the source file name.
4897 @opindex fdump-tree-copyrename
4898 Dump each function after applying the copy rename optimization. The file
4899 name is made by appending @file{.copyrename} to the source file name.
4902 @opindex fdump-tree-nrv
4903 Dump each function after applying the named return value optimization on
4904 generic trees. The file name is made by appending @file{.nrv} to the source
4908 @opindex fdump-tree-vect
4909 Dump each function after applying vectorization of loops. The file name is
4910 made by appending @file{.vect} to the source file name.
4913 @opindex fdump-tree-vrp
4914 Dump each function after Value Range Propagation (VRP). The file name
4915 is made by appending @file{.vrp} to the source file name.
4918 @opindex fdump-tree-all
4919 Enable all the available tree dumps with the flags provided in this option.
4922 @item -ftree-vectorizer-verbose=@var{n}
4923 @opindex ftree-vectorizer-verbose
4924 This option controls the amount of debugging output the vectorizer prints.
4925 This information is written to standard error, unless
4926 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4927 in which case it is output to the usual dump listing file, @file{.vect}.
4928 For @var{n}=0 no diagnostic information is reported.
4929 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4930 and the total number of loops that got vectorized.
4931 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4932 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
4933 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4934 level that @option{-fdump-tree-vect-stats} uses.
4935 Higher verbosity levels mean either more information dumped for each
4936 reported loop, or same amount of information reported for more loops:
4937 If @var{n}=3, alignment related information is added to the reports.
4938 If @var{n}=4, data-references related information (e.g.@: memory dependences,
4939 memory access-patterns) is added to the reports.
4940 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4941 that did not pass the first analysis phase (i.e., may not be countable, or
4942 may have complicated control-flow).
4943 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4944 For @var{n}=7, all the information the vectorizer generates during its
4945 analysis and transformation is reported. This is the same verbosity level
4946 that @option{-fdump-tree-vect-details} uses.
4948 @item -frandom-seed=@var{string}
4949 @opindex frandom-string
4950 This option provides a seed that GCC uses when it would otherwise use
4951 random numbers. It is used to generate certain symbol names
4952 that have to be different in every compiled file. It is also used to
4953 place unique stamps in coverage data files and the object files that
4954 produce them. You can use the @option{-frandom-seed} option to produce
4955 reproducibly identical object files.
4957 The @var{string} should be different for every file you compile.
4959 @item -fsched-verbose=@var{n}
4960 @opindex fsched-verbose
4961 On targets that use instruction scheduling, this option controls the
4962 amount of debugging output the scheduler prints. This information is
4963 written to standard error, unless @option{-dS} or @option{-dR} is
4964 specified, in which case it is output to the usual dump
4965 listing file, @file{.sched} or @file{.sched2} respectively. However
4966 for @var{n} greater than nine, the output is always printed to standard
4969 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4970 same information as @option{-dRS}. For @var{n} greater than one, it
4971 also output basic block probabilities, detailed ready list information
4972 and unit/insn info. For @var{n} greater than two, it includes RTL
4973 at abort point, control-flow and regions info. And for @var{n} over
4974 four, @option{-fsched-verbose} also includes dependence info.
4978 Store the usual ``temporary'' intermediate files permanently; place them
4979 in the current directory and name them based on the source file. Thus,
4980 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4981 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4982 preprocessed @file{foo.i} output file even though the compiler now
4983 normally uses an integrated preprocessor.
4985 When used in combination with the @option{-x} command line option,
4986 @option{-save-temps} is sensible enough to avoid over writing an
4987 input source file with the same extension as an intermediate file.
4988 The corresponding intermediate file may be obtained by renaming the
4989 source file before using @option{-save-temps}.
4993 Report the CPU time taken by each subprocess in the compilation
4994 sequence. For C source files, this is the compiler proper and assembler
4995 (plus the linker if linking is done). The output looks like this:
5002 The first number on each line is the ``user time'', that is time spent
5003 executing the program itself. The second number is ``system time'',
5004 time spent executing operating system routines on behalf of the program.
5005 Both numbers are in seconds.
5007 @item -fvar-tracking
5008 @opindex fvar-tracking
5009 Run variable tracking pass. It computes where variables are stored at each
5010 position in code. Better debugging information is then generated
5011 (if the debugging information format supports this information).
5013 It is enabled by default when compiling with optimization (@option{-Os},
5014 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5015 the debug info format supports it.
5017 @item -print-file-name=@var{library}
5018 @opindex print-file-name
5019 Print the full absolute name of the library file @var{library} that
5020 would be used when linking---and don't do anything else. With this
5021 option, GCC does not compile or link anything; it just prints the
5024 @item -print-multi-directory
5025 @opindex print-multi-directory
5026 Print the directory name corresponding to the multilib selected by any
5027 other switches present in the command line. This directory is supposed
5028 to exist in @env{GCC_EXEC_PREFIX}.
5030 @item -print-multi-lib
5031 @opindex print-multi-lib
5032 Print the mapping from multilib directory names to compiler switches
5033 that enable them. The directory name is separated from the switches by
5034 @samp{;}, and each switch starts with an @samp{@@} instead of the
5035 @samp{-}, without spaces between multiple switches. This is supposed to
5036 ease shell-processing.
5038 @item -print-prog-name=@var{program}
5039 @opindex print-prog-name
5040 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5042 @item -print-libgcc-file-name
5043 @opindex print-libgcc-file-name
5044 Same as @option{-print-file-name=libgcc.a}.
5046 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5047 but you do want to link with @file{libgcc.a}. You can do
5050 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5053 @item -print-search-dirs
5054 @opindex print-search-dirs
5055 Print the name of the configured installation directory and a list of
5056 program and library directories @command{gcc} will search---and don't do anything else.
5058 This is useful when @command{gcc} prints the error message
5059 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5060 To resolve this you either need to put @file{cpp0} and the other compiler
5061 components where @command{gcc} expects to find them, or you can set the environment
5062 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5063 Don't forget the trailing @samp{/}.
5064 @xref{Environment Variables}.
5066 @item -print-sysroot
5067 @opindex print-sysroot
5068 Print the target sysroot directory that will be used during
5069 compilation. This is the target sysroot specified either at configure
5070 time or or using the @option{--sysroot} option, possibly with an extra
5071 suffix that depends on compilation options. If no target sysroot is
5072 specified, the option prints nothing.
5074 @item -print-sysroot-headers-suffix
5075 @opindex print-sysroot-headers-suffix
5076 Print the suffix added to the target sysroot when searching for
5077 headers, or give an error if the compiler is not configured with such
5078 a suffix---and don't do anything else.
5081 @opindex dumpmachine
5082 Print the compiler's target machine (for example,
5083 @samp{i686-pc-linux-gnu})---and don't do anything else.
5086 @opindex dumpversion
5087 Print the compiler version (for example, @samp{3.0})---and don't do
5092 Print the compiler's built-in specs---and don't do anything else. (This
5093 is used when GCC itself is being built.) @xref{Spec Files}.
5095 @item -feliminate-unused-debug-types
5096 @opindex feliminate-unused-debug-types
5097 Normally, when producing DWARF2 output, GCC will emit debugging
5098 information for all types declared in a compilation
5099 unit, regardless of whether or not they are actually used
5100 in that compilation unit. Sometimes this is useful, such as
5101 if, in the debugger, you want to cast a value to a type that is
5102 not actually used in your program (but is declared). More often,
5103 however, this results in a significant amount of wasted space.
5104 With this option, GCC will avoid producing debug symbol output
5105 for types that are nowhere used in the source file being compiled.
5108 @node Optimize Options
5109 @section Options That Control Optimization
5110 @cindex optimize options
5111 @cindex options, optimization
5113 These options control various sorts of optimizations.
5115 Without any optimization option, the compiler's goal is to reduce the
5116 cost of compilation and to make debugging produce the expected
5117 results. Statements are independent: if you stop the program with a
5118 breakpoint between statements, you can then assign a new value to any
5119 variable or change the program counter to any other statement in the
5120 function and get exactly the results you would expect from the source
5123 Turning on optimization flags makes the compiler attempt to improve
5124 the performance and/or code size at the expense of compilation time
5125 and possibly the ability to debug the program.
5127 The compiler performs optimization based on the knowledge it has of the
5128 program. Compiling multiple files at once to a single output file mode allows
5129 the compiler to use information gained from all of the files when compiling
5132 Not all optimizations are controlled directly by a flag. Only
5133 optimizations that have a flag are listed.
5140 Optimize. Optimizing compilation takes somewhat more time, and a lot
5141 more memory for a large function.
5143 With @option{-O}, the compiler tries to reduce code size and execution
5144 time, without performing any optimizations that take a great deal of
5147 @option{-O} turns on the following optimization flags:
5150 -fcprop-registers @gol
5153 -fdelayed-branch @gol
5155 -fguess-branch-probability @gol
5156 -fif-conversion2 @gol
5157 -fif-conversion @gol
5158 -finline-small-functions @gol
5159 -fipa-pure-const @gol
5160 -fipa-reference @gol
5162 -fsplit-wide-types @gol
5163 -ftree-builtin-call-dce @gol
5166 -ftree-copyrename @gol
5168 -ftree-dominator-opts @gol
5175 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5176 where doing so does not interfere with debugging.
5180 Optimize even more. GCC performs nearly all supported optimizations
5181 that do not involve a space-speed tradeoff. The compiler does not
5182 perform loop unrolling or function inlining when you specify @option{-O2}.
5183 As compared to @option{-O}, this option increases both compilation time
5184 and the performance of the generated code.
5186 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5187 also turns on the following optimization flags:
5188 @gccoptlist{-fthread-jumps @gol
5189 -falign-functions -falign-jumps @gol
5190 -falign-loops -falign-labels @gol
5193 -fcse-follow-jumps -fcse-skip-blocks @gol
5194 -fdelete-null-pointer-checks @gol
5195 -fexpensive-optimizations @gol
5196 -fgcse -fgcse-lm @gol
5197 -findirect-inlining @gol
5198 -foptimize-sibling-calls @gol
5201 -freorder-blocks -freorder-functions @gol
5202 -frerun-cse-after-loop @gol
5203 -fsched-interblock -fsched-spec @gol
5204 -fschedule-insns -fschedule-insns2 @gol
5205 -fstrict-aliasing -fstrict-overflow @gol
5206 -ftree-switch-conversion @gol
5210 Please note the warning under @option{-fgcse} about
5211 invoking @option{-O2} on programs that use computed gotos.
5215 Optimize yet more. @option{-O3} turns on all optimizations specified
5216 by @option{-O2} and also turns on the @option{-finline-functions},
5217 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5218 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5222 Reduce compilation time and make debugging produce the expected
5223 results. This is the default.
5227 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5228 do not typically increase code size. It also performs further
5229 optimizations designed to reduce code size.
5231 @option{-Os} disables the following optimization flags:
5232 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5233 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5234 -fprefetch-loop-arrays -ftree-vect-loop-version}
5236 If you use multiple @option{-O} options, with or without level numbers,
5237 the last such option is the one that is effective.
5240 Options of the form @option{-f@var{flag}} specify machine-independent
5241 flags. Most flags have both positive and negative forms; the negative
5242 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5243 below, only one of the forms is listed---the one you typically will
5244 use. You can figure out the other form by either removing @samp{no-}
5247 The following options control specific optimizations. They are either
5248 activated by @option{-O} options or are related to ones that are. You
5249 can use the following flags in the rare cases when ``fine-tuning'' of
5250 optimizations to be performed is desired.
5253 @item -fno-default-inline
5254 @opindex fno-default-inline
5255 Do not make member functions inline by default merely because they are
5256 defined inside the class scope (C++ only). Otherwise, when you specify
5257 @w{@option{-O}}, member functions defined inside class scope are compiled
5258 inline by default; i.e., you don't need to add @samp{inline} in front of
5259 the member function name.
5261 @item -fno-defer-pop
5262 @opindex fno-defer-pop
5263 Always pop the arguments to each function call as soon as that function
5264 returns. For machines which must pop arguments after a function call,
5265 the compiler normally lets arguments accumulate on the stack for several
5266 function calls and pops them all at once.
5268 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5270 @item -fforward-propagate
5271 @opindex fforward-propagate
5272 Perform a forward propagation pass on RTL@. The pass tries to combine two
5273 instructions and checks if the result can be simplified. If loop unrolling
5274 is active, two passes are performed and the second is scheduled after
5277 This option is enabled by default at optimization levels @option{-O2},
5278 @option{-O3}, @option{-Os}.
5280 @item -fomit-frame-pointer
5281 @opindex fomit-frame-pointer
5282 Don't keep the frame pointer in a register for functions that
5283 don't need one. This avoids the instructions to save, set up and
5284 restore frame pointers; it also makes an extra register available
5285 in many functions. @strong{It also makes debugging impossible on
5288 On some machines, such as the VAX, this flag has no effect, because
5289 the standard calling sequence automatically handles the frame pointer
5290 and nothing is saved by pretending it doesn't exist. The
5291 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5292 whether a target machine supports this flag. @xref{Registers,,Register
5293 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5295 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5297 @item -foptimize-sibling-calls
5298 @opindex foptimize-sibling-calls
5299 Optimize sibling and tail recursive calls.
5301 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5305 Don't pay attention to the @code{inline} keyword. Normally this option
5306 is used to keep the compiler from expanding any functions inline.
5307 Note that if you are not optimizing, no functions can be expanded inline.
5309 @item -finline-small-functions
5310 @opindex finline-small-functions
5311 Integrate functions into their callers when their body is smaller than expected
5312 function call code (so overall size of program gets smaller). The compiler
5313 heuristically decides which functions are simple enough to be worth integrating
5316 Enabled at level @option{-O2}.
5318 @item -findirect-inlining
5319 @opindex findirect-inlining
5320 Inline also indirect calls that are discovered to be known at compile
5321 time thanks to previous inlining. This option has any effect only
5322 when inlining itself is turned on by the @option{-finline-functions}
5323 or @option{-finline-small-functions} options.
5325 Enabled at level @option{-O2}.
5327 @item -finline-functions
5328 @opindex finline-functions
5329 Integrate all simple functions into their callers. The compiler
5330 heuristically decides which functions are simple enough to be worth
5331 integrating in this way.
5333 If all calls to a given function are integrated, and the function is
5334 declared @code{static}, then the function is normally not output as
5335 assembler code in its own right.
5337 Enabled at level @option{-O3}.
5339 @item -finline-functions-called-once
5340 @opindex finline-functions-called-once
5341 Consider all @code{static} functions called once for inlining into their
5342 caller even if they are not marked @code{inline}. If a call to a given
5343 function is integrated, then the function is not output as assembler code
5346 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5348 @item -fearly-inlining
5349 @opindex fearly-inlining
5350 Inline functions marked by @code{always_inline} and functions whose body seems
5351 smaller than the function call overhead early before doing
5352 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5353 makes profiling significantly cheaper and usually inlining faster on programs
5354 having large chains of nested wrapper functions.
5358 @item -finline-limit=@var{n}
5359 @opindex finline-limit
5360 By default, GCC limits the size of functions that can be inlined. This flag
5361 allows coarse control of this limit. @var{n} is the size of functions that
5362 can be inlined in number of pseudo instructions.
5364 Inlining is actually controlled by a number of parameters, which may be
5365 specified individually by using @option{--param @var{name}=@var{value}}.
5366 The @option{-finline-limit=@var{n}} option sets some of these parameters
5370 @item max-inline-insns-single
5371 is set to @var{n}/2.
5372 @item max-inline-insns-auto
5373 is set to @var{n}/2.
5376 See below for a documentation of the individual
5377 parameters controlling inlining and for the defaults of these parameters.
5379 @emph{Note:} there may be no value to @option{-finline-limit} that results
5380 in default behavior.
5382 @emph{Note:} pseudo instruction represents, in this particular context, an
5383 abstract measurement of function's size. In no way does it represent a count
5384 of assembly instructions and as such its exact meaning might change from one
5385 release to an another.
5387 @item -fkeep-inline-functions
5388 @opindex fkeep-inline-functions
5389 In C, emit @code{static} functions that are declared @code{inline}
5390 into the object file, even if the function has been inlined into all
5391 of its callers. This switch does not affect functions using the
5392 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5393 inline functions into the object file.
5395 @item -fkeep-static-consts
5396 @opindex fkeep-static-consts
5397 Emit variables declared @code{static const} when optimization isn't turned
5398 on, even if the variables aren't referenced.
5400 GCC enables this option by default. If you want to force the compiler to
5401 check if the variable was referenced, regardless of whether or not
5402 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5404 @item -fmerge-constants
5405 @opindex fmerge-constants
5406 Attempt to merge identical constants (string constants and floating point
5407 constants) across compilation units.
5409 This option is the default for optimized compilation if the assembler and
5410 linker support it. Use @option{-fno-merge-constants} to inhibit this
5413 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5415 @item -fmerge-all-constants
5416 @opindex fmerge-all-constants
5417 Attempt to merge identical constants and identical variables.
5419 This option implies @option{-fmerge-constants}. In addition to
5420 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5421 arrays or initialized constant variables with integral or floating point
5422 types. Languages like C or C++ require each non-automatic variable to
5423 have distinct location, so using this option will result in non-conforming
5426 @item -fmodulo-sched
5427 @opindex fmodulo-sched
5428 Perform swing modulo scheduling immediately before the first scheduling
5429 pass. This pass looks at innermost loops and reorders their
5430 instructions by overlapping different iterations.
5432 @item -fmodulo-sched-allow-regmoves
5433 @opindex fmodulo-sched-allow-regmoves
5434 Perform more aggressive SMS based modulo scheduling with register moves
5435 allowed. By setting this flag certain anti-dependences edges will be
5436 deleted which will trigger the generation of reg-moves based on the
5437 life-range analysis. This option is effective only with
5438 @option{-fmodulo-sched} enabled.
5440 @item -fno-branch-count-reg
5441 @opindex fno-branch-count-reg
5442 Do not use ``decrement and branch'' instructions on a count register,
5443 but instead generate a sequence of instructions that decrement a
5444 register, compare it against zero, then branch based upon the result.
5445 This option is only meaningful on architectures that support such
5446 instructions, which include x86, PowerPC, IA-64 and S/390.
5448 The default is @option{-fbranch-count-reg}.
5450 @item -fno-function-cse
5451 @opindex fno-function-cse
5452 Do not put function addresses in registers; make each instruction that
5453 calls a constant function contain the function's address explicitly.
5455 This option results in less efficient code, but some strange hacks
5456 that alter the assembler output may be confused by the optimizations
5457 performed when this option is not used.
5459 The default is @option{-ffunction-cse}
5461 @item -fno-zero-initialized-in-bss
5462 @opindex fno-zero-initialized-in-bss
5463 If the target supports a BSS section, GCC by default puts variables that
5464 are initialized to zero into BSS@. This can save space in the resulting
5467 This option turns off this behavior because some programs explicitly
5468 rely on variables going to the data section. E.g., so that the
5469 resulting executable can find the beginning of that section and/or make
5470 assumptions based on that.
5472 The default is @option{-fzero-initialized-in-bss}.
5474 @item -fmudflap -fmudflapth -fmudflapir
5478 @cindex bounds checking
5480 For front-ends that support it (C and C++), instrument all risky
5481 pointer/array dereferencing operations, some standard library
5482 string/heap functions, and some other associated constructs with
5483 range/validity tests. Modules so instrumented should be immune to
5484 buffer overflows, invalid heap use, and some other classes of C/C++
5485 programming errors. The instrumentation relies on a separate runtime
5486 library (@file{libmudflap}), which will be linked into a program if
5487 @option{-fmudflap} is given at link time. Run-time behavior of the
5488 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5489 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5492 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5493 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5494 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5495 instrumentation should ignore pointer reads. This produces less
5496 instrumentation (and therefore faster execution) and still provides
5497 some protection against outright memory corrupting writes, but allows
5498 erroneously read data to propagate within a program.
5500 @item -fthread-jumps
5501 @opindex fthread-jumps
5502 Perform optimizations where we check to see if a jump branches to a
5503 location where another comparison subsumed by the first is found. If
5504 so, the first branch is redirected to either the destination of the
5505 second branch or a point immediately following it, depending on whether
5506 the condition is known to be true or false.
5508 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5510 @item -fsplit-wide-types
5511 @opindex fsplit-wide-types
5512 When using a type that occupies multiple registers, such as @code{long
5513 long} on a 32-bit system, split the registers apart and allocate them
5514 independently. This normally generates better code for those types,
5515 but may make debugging more difficult.
5517 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5520 @item -fcse-follow-jumps
5521 @opindex fcse-follow-jumps
5522 In common subexpression elimination (CSE), scan through jump instructions
5523 when the target of the jump is not reached by any other path. For
5524 example, when CSE encounters an @code{if} statement with an
5525 @code{else} clause, CSE will follow the jump when the condition
5528 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5530 @item -fcse-skip-blocks
5531 @opindex fcse-skip-blocks
5532 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5533 follow jumps which conditionally skip over blocks. When CSE
5534 encounters a simple @code{if} statement with no else clause,
5535 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5536 body of the @code{if}.
5538 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5540 @item -frerun-cse-after-loop
5541 @opindex frerun-cse-after-loop
5542 Re-run common subexpression elimination after loop optimizations has been
5545 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5549 Perform a global common subexpression elimination pass.
5550 This pass also performs global constant and copy propagation.
5552 @emph{Note:} When compiling a program using computed gotos, a GCC
5553 extension, you may get better runtime performance if you disable
5554 the global common subexpression elimination pass by adding
5555 @option{-fno-gcse} to the command line.
5557 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5561 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5562 attempt to move loads which are only killed by stores into themselves. This
5563 allows a loop containing a load/store sequence to be changed to a load outside
5564 the loop, and a copy/store within the loop.
5566 Enabled by default when gcse is enabled.
5570 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5571 global common subexpression elimination. This pass will attempt to move
5572 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5573 loops containing a load/store sequence can be changed to a load before
5574 the loop and a store after the loop.
5576 Not enabled at any optimization level.
5580 When @option{-fgcse-las} is enabled, the global common subexpression
5581 elimination pass eliminates redundant loads that come after stores to the
5582 same memory location (both partial and full redundancies).
5584 Not enabled at any optimization level.
5586 @item -fgcse-after-reload
5587 @opindex fgcse-after-reload
5588 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5589 pass is performed after reload. The purpose of this pass is to cleanup
5592 @item -funsafe-loop-optimizations
5593 @opindex funsafe-loop-optimizations
5594 If given, the loop optimizer will assume that loop indices do not
5595 overflow, and that the loops with nontrivial exit condition are not
5596 infinite. This enables a wider range of loop optimizations even if
5597 the loop optimizer itself cannot prove that these assumptions are valid.
5598 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5599 if it finds this kind of loop.
5601 @item -fcrossjumping
5602 @opindex fcrossjumping
5603 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5604 resulting code may or may not perform better than without cross-jumping.
5606 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5608 @item -fauto-inc-dec
5609 @opindex fauto-inc-dec
5610 Combine increments or decrements of addresses with memory accesses.
5611 This pass is always skipped on architectures that do not have
5612 instructions to support this. Enabled by default at @option{-O} and
5613 higher on architectures that support this.
5617 Perform dead code elimination (DCE) on RTL@.
5618 Enabled by default at @option{-O} and higher.
5622 Perform dead store elimination (DSE) on RTL@.
5623 Enabled by default at @option{-O} and higher.
5625 @item -fif-conversion
5626 @opindex fif-conversion
5627 Attempt to transform conditional jumps into branch-less equivalents. This
5628 include use of conditional moves, min, max, set flags and abs instructions, and
5629 some tricks doable by standard arithmetics. The use of conditional execution
5630 on chips where it is available is controlled by @code{if-conversion2}.
5632 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5634 @item -fif-conversion2
5635 @opindex fif-conversion2
5636 Use conditional execution (where available) to transform conditional jumps into
5637 branch-less equivalents.
5639 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5641 @item -fdelete-null-pointer-checks
5642 @opindex fdelete-null-pointer-checks
5643 Use global dataflow analysis to identify and eliminate useless checks
5644 for null pointers. The compiler assumes that dereferencing a null
5645 pointer would have halted the program. If a pointer is checked after
5646 it has already been dereferenced, it cannot be null.
5648 In some environments, this assumption is not true, and programs can
5649 safely dereference null pointers. Use
5650 @option{-fno-delete-null-pointer-checks} to disable this optimization
5651 for programs which depend on that behavior.
5653 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5655 @item -fexpensive-optimizations
5656 @opindex fexpensive-optimizations
5657 Perform a number of minor optimizations that are relatively expensive.
5659 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5661 @item -foptimize-register-move
5663 @opindex foptimize-register-move
5665 Attempt to reassign register numbers in move instructions and as
5666 operands of other simple instructions in order to maximize the amount of
5667 register tying. This is especially helpful on machines with two-operand
5670 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5673 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5675 @item -fdelayed-branch
5676 @opindex fdelayed-branch
5677 If supported for the target machine, attempt to reorder instructions
5678 to exploit instruction slots available after delayed branch
5681 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5683 @item -fschedule-insns
5684 @opindex fschedule-insns
5685 If supported for the target machine, attempt to reorder instructions to
5686 eliminate execution stalls due to required data being unavailable. This
5687 helps machines that have slow floating point or memory load instructions
5688 by allowing other instructions to be issued until the result of the load
5689 or floating point instruction is required.
5691 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5693 @item -fschedule-insns2
5694 @opindex fschedule-insns2
5695 Similar to @option{-fschedule-insns}, but requests an additional pass of
5696 instruction scheduling after register allocation has been done. This is
5697 especially useful on machines with a relatively small number of
5698 registers and where memory load instructions take more than one cycle.
5700 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5702 @item -fno-sched-interblock
5703 @opindex fno-sched-interblock
5704 Don't schedule instructions across basic blocks. This is normally
5705 enabled by default when scheduling before register allocation, i.e.@:
5706 with @option{-fschedule-insns} or at @option{-O2} or higher.
5708 @item -fno-sched-spec
5709 @opindex fno-sched-spec
5710 Don't allow speculative motion of non-load instructions. This is normally
5711 enabled by default when scheduling before register allocation, i.e.@:
5712 with @option{-fschedule-insns} or at @option{-O2} or higher.
5714 @item -fsched-spec-load
5715 @opindex fsched-spec-load
5716 Allow speculative motion of some load instructions. This only makes
5717 sense when scheduling before register allocation, i.e.@: with
5718 @option{-fschedule-insns} or at @option{-O2} or higher.
5720 @item -fsched-spec-load-dangerous
5721 @opindex fsched-spec-load-dangerous
5722 Allow speculative motion of more load instructions. This only makes
5723 sense when scheduling before register allocation, i.e.@: with
5724 @option{-fschedule-insns} or at @option{-O2} or higher.
5726 @item -fsched-stalled-insns
5727 @itemx -fsched-stalled-insns=@var{n}
5728 @opindex fsched-stalled-insns
5729 Define how many insns (if any) can be moved prematurely from the queue
5730 of stalled insns into the ready list, during the second scheduling pass.
5731 @option{-fno-sched-stalled-insns} means that no insns will be moved
5732 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5733 on how many queued insns can be moved prematurely.
5734 @option{-fsched-stalled-insns} without a value is equivalent to
5735 @option{-fsched-stalled-insns=1}.
5737 @item -fsched-stalled-insns-dep
5738 @itemx -fsched-stalled-insns-dep=@var{n}
5739 @opindex fsched-stalled-insns-dep
5740 Define how many insn groups (cycles) will be examined for a dependency
5741 on a stalled insn that is candidate for premature removal from the queue
5742 of stalled insns. This has an effect only during the second scheduling pass,
5743 and only if @option{-fsched-stalled-insns} is used.
5744 @option{-fno-sched-stalled-insns-dep} is equivalent to
5745 @option{-fsched-stalled-insns-dep=0}.
5746 @option{-fsched-stalled-insns-dep} without a value is equivalent to
5747 @option{-fsched-stalled-insns-dep=1}.
5749 @item -fsched2-use-superblocks
5750 @opindex fsched2-use-superblocks
5751 When scheduling after register allocation, do use superblock scheduling
5752 algorithm. Superblock scheduling allows motion across basic block boundaries
5753 resulting on faster schedules. This option is experimental, as not all machine
5754 descriptions used by GCC model the CPU closely enough to avoid unreliable
5755 results from the algorithm.
5757 This only makes sense when scheduling after register allocation, i.e.@: with
5758 @option{-fschedule-insns2} or at @option{-O2} or higher.
5760 @item -fsched2-use-traces
5761 @opindex fsched2-use-traces
5762 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5763 allocation and additionally perform code duplication in order to increase the
5764 size of superblocks using tracer pass. See @option{-ftracer} for details on
5767 This mode should produce faster but significantly longer programs. Also
5768 without @option{-fbranch-probabilities} the traces constructed may not
5769 match the reality and hurt the performance. This only makes
5770 sense when scheduling after register allocation, i.e.@: with
5771 @option{-fschedule-insns2} or at @option{-O2} or higher.
5775 Eliminate redundant sign extension instructions and move the non-redundant
5776 ones to optimal placement using lazy code motion (LCM).
5778 @item -freschedule-modulo-scheduled-loops
5779 @opindex freschedule-modulo-scheduled-loops
5780 The modulo scheduling comes before the traditional scheduling, if a loop
5781 was modulo scheduled we may want to prevent the later scheduling passes
5782 from changing its schedule, we use this option to control that.
5784 @item -fcaller-saves
5785 @opindex fcaller-saves
5786 Enable values to be allocated in registers that will be clobbered by
5787 function calls, by emitting extra instructions to save and restore the
5788 registers around such calls. Such allocation is done only when it
5789 seems to result in better code than would otherwise be produced.
5791 This option is always enabled by default on certain machines, usually
5792 those which have no call-preserved registers to use instead.
5794 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5796 @item -ftree-reassoc
5797 @opindex ftree-reassoc
5798 Perform reassociation on trees. This flag is enabled by default
5799 at @option{-O} and higher.
5803 Perform partial redundancy elimination (PRE) on trees. This flag is
5804 enabled by default at @option{-O2} and @option{-O3}.
5808 Perform full redundancy elimination (FRE) on trees. The difference
5809 between FRE and PRE is that FRE only considers expressions
5810 that are computed on all paths leading to the redundant computation.
5811 This analysis is faster than PRE, though it exposes fewer redundancies.
5812 This flag is enabled by default at @option{-O} and higher.
5814 @item -ftree-copy-prop
5815 @opindex ftree-copy-prop
5816 Perform copy propagation on trees. This pass eliminates unnecessary
5817 copy operations. This flag is enabled by default at @option{-O} and
5820 @item -fipa-pure-const
5821 @opindex fipa-pure-const
5822 Discover which functions are pure or constant.
5823 Enabled by default at @option{-O} and higher.
5825 @item -fipa-reference
5826 @opindex fipa-reference
5827 Discover which static variables do not escape cannot escape the
5829 Enabled by default at @option{-O} and higher.
5831 @item -fipa-struct-reorg
5832 @opindex fipa-struct-reorg
5833 Perform structure reorganization optimization, that change C-like structures
5834 layout in order to better utilize spatial locality. This transformation is
5835 affective for programs containing arrays of structures. Available in two
5836 compilation modes: profile-based (enabled with @option{-fprofile-generate})
5837 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
5838 to provide the safety of this transformation. It works only in whole program
5839 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
5840 enabled. Structures considered @samp{cold} by this transformation are not
5841 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
5843 With this flag, the program debug info reflects a new structure layout.
5847 Perform interprocedural pointer analysis.
5851 Perform interprocedural constant propagation.
5852 This optimization analyzes the program to determine when values passed
5853 to functions are constants and then optimizes accordingly.
5854 This optimization can substantially increase performance
5855 if the application has constants passed to functions, but
5856 because this optimization can create multiple copies of functions,
5857 it may significantly increase code size.
5859 @item -fipa-matrix-reorg
5860 @opindex fipa-matrix-reorg
5861 Perform matrix flattening and transposing.
5862 Matrix flattening tries to replace a m-dimensional matrix
5863 with its equivalent n-dimensional matrix, where n < m.
5864 This reduces the level of indirection needed for accessing the elements
5865 of the matrix. The second optimization is matrix transposing that
5866 attemps to change the order of the matrix's dimensions in order to
5867 improve cache locality.
5868 Both optimizations need fwhole-program flag.
5869 Transposing is enabled only if profiling information is avaliable.
5874 Perform forward store motion on trees. This flag is
5875 enabled by default at @option{-O} and higher.
5879 Perform sparse conditional constant propagation (CCP) on trees. This
5880 pass only operates on local scalar variables and is enabled by default
5881 at @option{-O} and higher.
5883 @item -ftree-store-ccp
5884 @opindex ftree-store-ccp
5885 Perform sparse conditional constant propagation (CCP) on trees. This
5886 pass operates on both local scalar variables and memory stores and
5887 loads (global variables, structures, arrays, etc). This flag is
5888 enabled by default at @option{-O2} and higher.
5890 @item -ftree-switch-conversion
5891 Perform conversion of simple initializations in a switch to
5892 initializations from a scalar array. This flag is enabled by default
5893 at @option{-O2} and higher.
5897 Perform dead code elimination (DCE) on trees. This flag is enabled by
5898 default at @option{-O} and higher.
5900 @item -ftree-builtin-call-dce
5901 @opindex ftree-builtin-call-dce
5902 Perform conditional dead code elimination (DCE) for calls to builtin functions
5903 that may set @code{errno} but are otherwise side-effect free. This flag is
5904 enabled by default at @option{-O2} and higher if @option{-Os} is not also
5907 @item -ftree-dominator-opts
5908 @opindex ftree-dominator-opts
5909 Perform a variety of simple scalar cleanups (constant/copy
5910 propagation, redundancy elimination, range propagation and expression
5911 simplification) based on a dominator tree traversal. This also
5912 performs jump threading (to reduce jumps to jumps). This flag is
5913 enabled by default at @option{-O} and higher.
5917 Perform dead store elimination (DSE) on trees. A dead store is a store into
5918 a memory location which will later be overwritten by another store without
5919 any intervening loads. In this case the earlier store can be deleted. This
5920 flag is enabled by default at @option{-O} and higher.
5924 Perform loop header copying on trees. This is beneficial since it increases
5925 effectiveness of code motion optimizations. It also saves one jump. This flag
5926 is enabled by default at @option{-O} and higher. It is not enabled
5927 for @option{-Os}, since it usually increases code size.
5929 @item -ftree-loop-optimize
5930 @opindex ftree-loop-optimize
5931 Perform loop optimizations on trees. This flag is enabled by default
5932 at @option{-O} and higher.
5934 @item -ftree-loop-linear
5935 @opindex ftree-loop-linear
5936 Perform linear loop transformations on tree. This flag can improve cache
5937 performance and allow further loop optimizations to take place.
5939 @item -fcheck-data-deps
5940 @opindex fcheck-data-deps
5941 Compare the results of several data dependence analyzers. This option
5942 is used for debugging the data dependence analyzers.
5944 @item -ftree-loop-distribution
5945 Perform loop distribution. This flag can improve cache performance on
5946 big loop bodies and allow further loop optimizations, like
5947 parallelization or vectorization, to take place. For example, the loop
5964 @item -ftree-loop-im
5965 @opindex ftree-loop-im
5966 Perform loop invariant motion on trees. This pass moves only invariants that
5967 would be hard to handle at RTL level (function calls, operations that expand to
5968 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5969 operands of conditions that are invariant out of the loop, so that we can use
5970 just trivial invariantness analysis in loop unswitching. The pass also includes
5973 @item -ftree-loop-ivcanon
5974 @opindex ftree-loop-ivcanon
5975 Create a canonical counter for number of iterations in the loop for that
5976 determining number of iterations requires complicated analysis. Later
5977 optimizations then may determine the number easily. Useful especially
5978 in connection with unrolling.
5982 Perform induction variable optimizations (strength reduction, induction
5983 variable merging and induction variable elimination) on trees.
5985 @item -ftree-parallelize-loops=n
5986 @opindex ftree-parallelize-loops
5987 Parallelize loops, i.e., split their iteration space to run in n threads.
5988 This is only possible for loops whose iterations are independent
5989 and can be arbitrarily reordered. The optimization is only
5990 profitable on multiprocessor machines, for loops that are CPU-intensive,
5991 rather than constrained e.g.@: by memory bandwidth. This option
5992 implies @option{-pthread}, and thus is only supported on targets
5993 that have support for @option{-pthread}.
5997 Perform scalar replacement of aggregates. This pass replaces structure
5998 references with scalars to prevent committing structures to memory too
5999 early. This flag is enabled by default at @option{-O} and higher.
6001 @item -ftree-copyrename
6002 @opindex ftree-copyrename
6003 Perform copy renaming on trees. This pass attempts to rename compiler
6004 temporaries to other variables at copy locations, usually resulting in
6005 variable names which more closely resemble the original variables. This flag
6006 is enabled by default at @option{-O} and higher.
6010 Perform temporary expression replacement during the SSA->normal phase. Single
6011 use/single def temporaries are replaced at their use location with their
6012 defining expression. This results in non-GIMPLE code, but gives the expanders
6013 much more complex trees to work on resulting in better RTL generation. This is
6014 enabled by default at @option{-O} and higher.
6016 @item -ftree-vectorize
6017 @opindex ftree-vectorize
6018 Perform loop vectorization on trees. This flag is enabled by default at
6021 @item -ftree-vect-loop-version
6022 @opindex ftree-vect-loop-version
6023 Perform loop versioning when doing loop vectorization on trees. When a loop
6024 appears to be vectorizable except that data alignment or data dependence cannot
6025 be determined at compile time then vectorized and non-vectorized versions of
6026 the loop are generated along with runtime checks for alignment or dependence
6027 to control which version is executed. This option is enabled by default
6028 except at level @option{-Os} where it is disabled.
6030 @item -fvect-cost-model
6031 @opindex fvect-cost-model
6032 Enable cost model for vectorization.
6036 Perform Value Range Propagation on trees. This is similar to the
6037 constant propagation pass, but instead of values, ranges of values are
6038 propagated. This allows the optimizers to remove unnecessary range
6039 checks like array bound checks and null pointer checks. This is
6040 enabled by default at @option{-O2} and higher. Null pointer check
6041 elimination is only done if @option{-fdelete-null-pointer-checks} is
6046 Perform tail duplication to enlarge superblock size. This transformation
6047 simplifies the control flow of the function allowing other optimizations to do
6050 @item -funroll-loops
6051 @opindex funroll-loops
6052 Unroll loops whose number of iterations can be determined at compile
6053 time or upon entry to the loop. @option{-funroll-loops} implies
6054 @option{-frerun-cse-after-loop}. This option makes code larger,
6055 and may or may not make it run faster.
6057 @item -funroll-all-loops
6058 @opindex funroll-all-loops
6059 Unroll all loops, even if their number of iterations is uncertain when
6060 the loop is entered. This usually makes programs run more slowly.
6061 @option{-funroll-all-loops} implies the same options as
6062 @option{-funroll-loops},
6064 @item -fsplit-ivs-in-unroller
6065 @opindex fsplit-ivs-in-unroller
6066 Enables expressing of values of induction variables in later iterations
6067 of the unrolled loop using the value in the first iteration. This breaks
6068 long dependency chains, thus improving efficiency of the scheduling passes.
6070 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6071 same effect. However in cases the loop body is more complicated than
6072 a single basic block, this is not reliable. It also does not work at all
6073 on some of the architectures due to restrictions in the CSE pass.
6075 This optimization is enabled by default.
6077 @item -fvariable-expansion-in-unroller
6078 @opindex fvariable-expansion-in-unroller
6079 With this option, the compiler will create multiple copies of some
6080 local variables when unrolling a loop which can result in superior code.
6082 @item -fpredictive-commoning
6083 @opindex fpredictive-commoning
6084 Perform predictive commoning optimization, i.e., reusing computations
6085 (especially memory loads and stores) performed in previous
6086 iterations of loops.
6088 This option is enabled at level @option{-O3}.
6090 @item -fprefetch-loop-arrays
6091 @opindex fprefetch-loop-arrays
6092 If supported by the target machine, generate instructions to prefetch
6093 memory to improve the performance of loops that access large arrays.
6095 This option may generate better or worse code; results are highly
6096 dependent on the structure of loops within the source code.
6098 Disabled at level @option{-Os}.
6101 @itemx -fno-peephole2
6102 @opindex fno-peephole
6103 @opindex fno-peephole2
6104 Disable any machine-specific peephole optimizations. The difference
6105 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6106 are implemented in the compiler; some targets use one, some use the
6107 other, a few use both.
6109 @option{-fpeephole} is enabled by default.
6110 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6112 @item -fno-guess-branch-probability
6113 @opindex fno-guess-branch-probability
6114 Do not guess branch probabilities using heuristics.
6116 GCC will use heuristics to guess branch probabilities if they are
6117 not provided by profiling feedback (@option{-fprofile-arcs}). These
6118 heuristics are based on the control flow graph. If some branch probabilities
6119 are specified by @samp{__builtin_expect}, then the heuristics will be
6120 used to guess branch probabilities for the rest of the control flow graph,
6121 taking the @samp{__builtin_expect} info into account. The interactions
6122 between the heuristics and @samp{__builtin_expect} can be complex, and in
6123 some cases, it may be useful to disable the heuristics so that the effects
6124 of @samp{__builtin_expect} are easier to understand.
6126 The default is @option{-fguess-branch-probability} at levels
6127 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6129 @item -freorder-blocks
6130 @opindex freorder-blocks
6131 Reorder basic blocks in the compiled function in order to reduce number of
6132 taken branches and improve code locality.
6134 Enabled at levels @option{-O2}, @option{-O3}.
6136 @item -freorder-blocks-and-partition
6137 @opindex freorder-blocks-and-partition
6138 In addition to reordering basic blocks in the compiled function, in order
6139 to reduce number of taken branches, partitions hot and cold basic blocks
6140 into separate sections of the assembly and .o files, to improve
6141 paging and cache locality performance.
6143 This optimization is automatically turned off in the presence of
6144 exception handling, for linkonce sections, for functions with a user-defined
6145 section attribute and on any architecture that does not support named
6148 @item -freorder-functions
6149 @opindex freorder-functions
6150 Reorder functions in the object file in order to
6151 improve code locality. This is implemented by using special
6152 subsections @code{.text.hot} for most frequently executed functions and
6153 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6154 the linker so object file format must support named sections and linker must
6155 place them in a reasonable way.
6157 Also profile feedback must be available in to make this option effective. See
6158 @option{-fprofile-arcs} for details.
6160 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6162 @item -fstrict-aliasing
6163 @opindex fstrict-aliasing
6164 Allows the compiler to assume the strictest aliasing rules applicable to
6165 the language being compiled. For C (and C++), this activates
6166 optimizations based on the type of expressions. In particular, an
6167 object of one type is assumed never to reside at the same address as an
6168 object of a different type, unless the types are almost the same. For
6169 example, an @code{unsigned int} can alias an @code{int}, but not a
6170 @code{void*} or a @code{double}. A character type may alias any other
6173 @anchor{Type-punning}Pay special attention to code like this:
6186 The practice of reading from a different union member than the one most
6187 recently written to (called ``type-punning'') is common. Even with
6188 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6189 is accessed through the union type. So, the code above will work as
6190 expected. @xref{Structures unions enumerations and bit-fields
6191 implementation}. However, this code might not:
6202 Similarly, access by taking the address, casting the resulting pointer
6203 and dereferencing the result has undefined behavior, even if the cast
6204 uses a union type, e.g.:
6208 return ((union a_union *) &d)->i;
6212 The @option{-fstrict-aliasing} option is enabled at levels
6213 @option{-O2}, @option{-O3}, @option{-Os}.
6215 @item -fstrict-overflow
6216 @opindex fstrict-overflow
6217 Allow the compiler to assume strict signed overflow rules, depending
6218 on the language being compiled. For C (and C++) this means that
6219 overflow when doing arithmetic with signed numbers is undefined, which
6220 means that the compiler may assume that it will not happen. This
6221 permits various optimizations. For example, the compiler will assume
6222 that an expression like @code{i + 10 > i} will always be true for
6223 signed @code{i}. This assumption is only valid if signed overflow is
6224 undefined, as the expression is false if @code{i + 10} overflows when
6225 using twos complement arithmetic. When this option is in effect any
6226 attempt to determine whether an operation on signed numbers will
6227 overflow must be written carefully to not actually involve overflow.
6229 This option also allows the compiler to assume strict pointer
6230 semantics: given a pointer to an object, if adding an offset to that
6231 pointer does not produce a pointer to the same object, the addition is
6232 undefined. This permits the compiler to conclude that @code{p + u >
6233 p} is always true for a pointer @code{p} and unsigned integer
6234 @code{u}. This assumption is only valid because pointer wraparound is
6235 undefined, as the expression is false if @code{p + u} overflows using
6236 twos complement arithmetic.
6238 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6239 that integer signed overflow is fully defined: it wraps. When
6240 @option{-fwrapv} is used, there is no difference between
6241 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
6242 integers. With @option{-fwrapv} certain types of overflow are
6243 permitted. For example, if the compiler gets an overflow when doing
6244 arithmetic on constants, the overflowed value can still be used with
6245 @option{-fwrapv}, but not otherwise.
6247 The @option{-fstrict-overflow} option is enabled at levels
6248 @option{-O2}, @option{-O3}, @option{-Os}.
6250 @item -falign-functions
6251 @itemx -falign-functions=@var{n}
6252 @opindex falign-functions
6253 Align the start of functions to the next power-of-two greater than
6254 @var{n}, skipping up to @var{n} bytes. For instance,
6255 @option{-falign-functions=32} aligns functions to the next 32-byte
6256 boundary, but @option{-falign-functions=24} would align to the next
6257 32-byte boundary only if this can be done by skipping 23 bytes or less.
6259 @option{-fno-align-functions} and @option{-falign-functions=1} are
6260 equivalent and mean that functions will not be aligned.
6262 Some assemblers only support this flag when @var{n} is a power of two;
6263 in that case, it is rounded up.
6265 If @var{n} is not specified or is zero, use a machine-dependent default.
6267 Enabled at levels @option{-O2}, @option{-O3}.
6269 @item -falign-labels
6270 @itemx -falign-labels=@var{n}
6271 @opindex falign-labels
6272 Align all branch targets to a power-of-two boundary, skipping up to
6273 @var{n} bytes like @option{-falign-functions}. This option can easily
6274 make code slower, because it must insert dummy operations for when the
6275 branch target is reached in the usual flow of the code.
6277 @option{-fno-align-labels} and @option{-falign-labels=1} are
6278 equivalent and mean that labels will not be aligned.
6280 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6281 are greater than this value, then their values are used instead.
6283 If @var{n} is not specified or is zero, use a machine-dependent default
6284 which is very likely to be @samp{1}, meaning no alignment.
6286 Enabled at levels @option{-O2}, @option{-O3}.
6289 @itemx -falign-loops=@var{n}
6290 @opindex falign-loops
6291 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6292 like @option{-falign-functions}. The hope is that the loop will be
6293 executed many times, which will make up for any execution of the dummy
6296 @option{-fno-align-loops} and @option{-falign-loops=1} are
6297 equivalent and mean that loops will not be aligned.
6299 If @var{n} is not specified or is zero, use a machine-dependent default.
6301 Enabled at levels @option{-O2}, @option{-O3}.
6304 @itemx -falign-jumps=@var{n}
6305 @opindex falign-jumps
6306 Align branch targets to a power-of-two boundary, for branch targets
6307 where the targets can only be reached by jumping, skipping up to @var{n}
6308 bytes like @option{-falign-functions}. In this case, no dummy operations
6311 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6312 equivalent and mean that loops will not be aligned.
6314 If @var{n} is not specified or is zero, use a machine-dependent default.
6316 Enabled at levels @option{-O2}, @option{-O3}.
6318 @item -funit-at-a-time
6319 @opindex funit-at-a-time
6320 This option is left for compatibility reasons. @option{-funit-at-a-time}
6321 has no effect, while @option{-fno-unit-at-a-time} implies
6322 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
6326 @item -fno-toplevel-reorder
6327 @opindex fno-toplevel-reorder
6328 Do not reorder top-level functions, variables, and @code{asm}
6329 statements. Output them in the same order that they appear in the
6330 input file. When this option is used, unreferenced static variables
6331 will not be removed. This option is intended to support existing code
6332 which relies on a particular ordering. For new code, it is better to
6335 Enabled at level @option{-O0}. When disabled explicitly, it also imply
6336 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
6341 Constructs webs as commonly used for register allocation purposes and assign
6342 each web individual pseudo register. This allows the register allocation pass
6343 to operate on pseudos directly, but also strengthens several other optimization
6344 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6345 however, make debugging impossible, since variables will no longer stay in a
6348 Enabled by default with @option{-funroll-loops}.
6350 @item -fwhole-program
6351 @opindex fwhole-program
6352 Assume that the current compilation unit represents whole program being
6353 compiled. All public functions and variables with the exception of @code{main}
6354 and those merged by attribute @code{externally_visible} become static functions
6355 and in a affect gets more aggressively optimized by interprocedural optimizers.
6356 While this option is equivalent to proper use of @code{static} keyword for
6357 programs consisting of single file, in combination with option
6358 @option{--combine} this flag can be used to compile most of smaller scale C
6359 programs since the functions and variables become local for the whole combined
6360 compilation unit, not for the single source file itself.
6362 This option is not supported for Fortran programs.
6364 @item -fcprop-registers
6365 @opindex fcprop-registers
6366 After register allocation and post-register allocation instruction splitting,
6367 we perform a copy-propagation pass to try to reduce scheduling dependencies
6368 and occasionally eliminate the copy.
6370 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6372 @item -fprofile-dir=@var{path}
6373 @opindex fprofile-dir
6375 Set the directory to search the profile data files in to @var{path}.
6376 This option affects only the profile data generated by
6377 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
6378 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
6379 and its related options.
6380 By default, GCC will use the current directory as @var{path}
6381 thus the profile data file will appear in the same directory as the object file.
6383 @item -fprofile-generate
6384 @itemx -fprofile-generate=@var{path}
6385 @opindex fprofile-generate
6387 Enable options usually used for instrumenting application to produce
6388 profile useful for later recompilation with profile feedback based
6389 optimization. You must use @option{-fprofile-generate} both when
6390 compiling and when linking your program.
6392 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6394 If @var{path} is specified, GCC will look at the @var{path} to find
6395 the profile feeedback data files. See @option{-fprofile-dir}.
6398 @itemx -fprofile-use=@var{path}
6399 @opindex fprofile-use
6400 Enable profile feedback directed optimizations, and optimizations
6401 generally profitable only with profile feedback available.
6403 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6404 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6406 By default, GCC emits an error message if the feedback profiles do not
6407 match the source code. This error can be turned into a warning by using
6408 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6411 If @var{path} is specified, GCC will look at the @var{path} to find
6412 the profile feedback data files. See @option{-fprofile-dir}.
6415 The following options control compiler behavior regarding floating
6416 point arithmetic. These options trade off between speed and
6417 correctness. All must be specifically enabled.
6421 @opindex ffloat-store
6422 Do not store floating point variables in registers, and inhibit other
6423 options that might change whether a floating point value is taken from a
6426 @cindex floating point precision
6427 This option prevents undesirable excess precision on machines such as
6428 the 68000 where the floating registers (of the 68881) keep more
6429 precision than a @code{double} is supposed to have. Similarly for the
6430 x86 architecture. For most programs, the excess precision does only
6431 good, but a few programs rely on the precise definition of IEEE floating
6432 point. Use @option{-ffloat-store} for such programs, after modifying
6433 them to store all pertinent intermediate computations into variables.
6437 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6438 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6439 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6441 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6443 This option is not turned on by any @option{-O} option since
6444 it can result in incorrect output for programs which depend on
6445 an exact implementation of IEEE or ISO rules/specifications for
6446 math functions. It may, however, yield faster code for programs
6447 that do not require the guarantees of these specifications.
6449 @item -fno-math-errno
6450 @opindex fno-math-errno
6451 Do not set ERRNO after calling math functions that are executed
6452 with a single instruction, e.g., sqrt. A program that relies on
6453 IEEE exceptions for math error handling may want to use this flag
6454 for speed while maintaining IEEE arithmetic compatibility.
6456 This option is not turned on by any @option{-O} option since
6457 it can result in incorrect output for programs which depend on
6458 an exact implementation of IEEE or ISO rules/specifications for
6459 math functions. It may, however, yield faster code for programs
6460 that do not require the guarantees of these specifications.
6462 The default is @option{-fmath-errno}.
6464 On Darwin systems, the math library never sets @code{errno}. There is
6465 therefore no reason for the compiler to consider the possibility that
6466 it might, and @option{-fno-math-errno} is the default.
6468 @item -funsafe-math-optimizations
6469 @opindex funsafe-math-optimizations
6471 Allow optimizations for floating-point arithmetic that (a) assume
6472 that arguments and results are valid and (b) may violate IEEE or
6473 ANSI standards. When used at link-time, it may include libraries
6474 or startup files that change the default FPU control word or other
6475 similar optimizations.
6477 This option is not turned on by any @option{-O} option since
6478 it can result in incorrect output for programs which depend on
6479 an exact implementation of IEEE or ISO rules/specifications for
6480 math functions. It may, however, yield faster code for programs
6481 that do not require the guarantees of these specifications.
6482 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6483 @option{-fassociative-math} and @option{-freciprocal-math}.
6485 The default is @option{-fno-unsafe-math-optimizations}.
6487 @item -fassociative-math
6488 @opindex fassociative-math
6490 Allow re-association of operands in series of floating-point operations.
6491 This violates the ISO C and C++ language standard by possibly changing
6492 computation result. NOTE: re-ordering may change the sign of zero as
6493 well as ignore NaNs and inhibit or create underflow or overflow (and
6494 thus cannot be used on a code which relies on rounding behavior like
6495 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6496 and thus may not be used when ordered comparisons are required.
6497 This option requires that both @option{-fno-signed-zeros} and
6498 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6499 much sense with @option{-frounding-math}.
6501 The default is @option{-fno-associative-math}.
6503 @item -freciprocal-math
6504 @opindex freciprocal-math
6506 Allow the reciprocal of a value to be used instead of dividing by
6507 the value if this enables optimizations. For example @code{x / y}
6508 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6509 is subject to common subexpression elimination. Note that this loses
6510 precision and increases the number of flops operating on the value.
6512 The default is @option{-fno-reciprocal-math}.
6514 @item -ffinite-math-only
6515 @opindex ffinite-math-only
6516 Allow optimizations for floating-point arithmetic that assume
6517 that arguments and results are not NaNs or +-Infs.
6519 This option is not turned on by any @option{-O} option since
6520 it can result in incorrect output for programs which depend on
6521 an exact implementation of IEEE or ISO rules/specifications for
6522 math functions. It may, however, yield faster code for programs
6523 that do not require the guarantees of these specifications.
6525 The default is @option{-fno-finite-math-only}.
6527 @item -fno-signed-zeros
6528 @opindex fno-signed-zeros
6529 Allow optimizations for floating point arithmetic that ignore the
6530 signedness of zero. IEEE arithmetic specifies the behavior of
6531 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6532 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6533 This option implies that the sign of a zero result isn't significant.
6535 The default is @option{-fsigned-zeros}.
6537 @item -fno-trapping-math
6538 @opindex fno-trapping-math
6539 Compile code assuming that floating-point operations cannot generate
6540 user-visible traps. These traps include division by zero, overflow,
6541 underflow, inexact result and invalid operation. This option requires
6542 that @option{-fno-signaling-nans} be in effect. Setting this option may
6543 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6545 This option should never be turned on by any @option{-O} option since
6546 it can result in incorrect output for programs which depend on
6547 an exact implementation of IEEE or ISO rules/specifications for
6550 The default is @option{-ftrapping-math}.
6552 @item -frounding-math
6553 @opindex frounding-math
6554 Disable transformations and optimizations that assume default floating
6555 point rounding behavior. This is round-to-zero for all floating point
6556 to integer conversions, and round-to-nearest for all other arithmetic
6557 truncations. This option should be specified for programs that change
6558 the FP rounding mode dynamically, or that may be executed with a
6559 non-default rounding mode. This option disables constant folding of
6560 floating point expressions at compile-time (which may be affected by
6561 rounding mode) and arithmetic transformations that are unsafe in the
6562 presence of sign-dependent rounding modes.
6564 The default is @option{-fno-rounding-math}.
6566 This option is experimental and does not currently guarantee to
6567 disable all GCC optimizations that are affected by rounding mode.
6568 Future versions of GCC may provide finer control of this setting
6569 using C99's @code{FENV_ACCESS} pragma. This command line option
6570 will be used to specify the default state for @code{FENV_ACCESS}.
6572 @item -frtl-abstract-sequences
6573 @opindex frtl-abstract-sequences
6574 It is a size optimization method. This option is to find identical
6575 sequences of code, which can be turned into pseudo-procedures and
6576 then replace all occurrences with calls to the newly created
6577 subroutine. It is kind of an opposite of @option{-finline-functions}.
6578 This optimization runs at RTL level.
6580 @item -fsignaling-nans
6581 @opindex fsignaling-nans
6582 Compile code assuming that IEEE signaling NaNs may generate user-visible
6583 traps during floating-point operations. Setting this option disables
6584 optimizations that may change the number of exceptions visible with
6585 signaling NaNs. This option implies @option{-ftrapping-math}.
6587 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6590 The default is @option{-fno-signaling-nans}.
6592 This option is experimental and does not currently guarantee to
6593 disable all GCC optimizations that affect signaling NaN behavior.
6595 @item -fsingle-precision-constant
6596 @opindex fsingle-precision-constant
6597 Treat floating point constant as single precision constant instead of
6598 implicitly converting it to double precision constant.
6600 @item -fcx-limited-range
6601 @opindex fcx-limited-range
6602 When enabled, this option states that a range reduction step is not
6603 needed when performing complex division. Also, there is no checking
6604 whether the result of a complex multiplication or division is @code{NaN
6605 + I*NaN}, with an attempt to rescue the situation in that case. The
6606 default is @option{-fno-cx-limited-range}, but is enabled by
6607 @option{-ffast-math}.
6609 This option controls the default setting of the ISO C99
6610 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6613 @item -fcx-fortran-rules
6614 @opindex fcx-fortran-rules
6615 Complex multiplication and division follow Fortran rules. Range
6616 reduction is done as part of complex division, but there is no checking
6617 whether the result of a complex multiplication or division is @code{NaN
6618 + I*NaN}, with an attempt to rescue the situation in that case.
6620 The default is @option{-fno-cx-fortran-rules}.
6624 The following options control optimizations that may improve
6625 performance, but are not enabled by any @option{-O} options. This
6626 section includes experimental options that may produce broken code.
6629 @item -fbranch-probabilities
6630 @opindex fbranch-probabilities
6631 After running a program compiled with @option{-fprofile-arcs}
6632 (@pxref{Debugging Options,, Options for Debugging Your Program or
6633 @command{gcc}}), you can compile it a second time using
6634 @option{-fbranch-probabilities}, to improve optimizations based on
6635 the number of times each branch was taken. When the program
6636 compiled with @option{-fprofile-arcs} exits it saves arc execution
6637 counts to a file called @file{@var{sourcename}.gcda} for each source
6638 file. The information in this data file is very dependent on the
6639 structure of the generated code, so you must use the same source code
6640 and the same optimization options for both compilations.
6642 With @option{-fbranch-probabilities}, GCC puts a
6643 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6644 These can be used to improve optimization. Currently, they are only
6645 used in one place: in @file{reorg.c}, instead of guessing which path a
6646 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6647 exactly determine which path is taken more often.
6649 @item -fprofile-values
6650 @opindex fprofile-values
6651 If combined with @option{-fprofile-arcs}, it adds code so that some
6652 data about values of expressions in the program is gathered.
6654 With @option{-fbranch-probabilities}, it reads back the data gathered
6655 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6656 notes to instructions for their later usage in optimizations.
6658 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6662 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6663 a code to gather information about values of expressions.
6665 With @option{-fbranch-probabilities}, it reads back the data gathered
6666 and actually performs the optimizations based on them.
6667 Currently the optimizations include specialization of division operation
6668 using the knowledge about the value of the denominator.
6670 @item -frename-registers
6671 @opindex frename-registers
6672 Attempt to avoid false dependencies in scheduled code by making use
6673 of registers left over after register allocation. This optimization
6674 will most benefit processors with lots of registers. Depending on the
6675 debug information format adopted by the target, however, it can
6676 make debugging impossible, since variables will no longer stay in
6677 a ``home register''.
6679 Enabled by default with @option{-funroll-loops}.
6683 Perform tail duplication to enlarge superblock size. This transformation
6684 simplifies the control flow of the function allowing other optimizations to do
6687 Enabled with @option{-fprofile-use}.
6689 @item -funroll-loops
6690 @opindex funroll-loops
6691 Unroll loops whose number of iterations can be determined at compile time or
6692 upon entry to the loop. @option{-funroll-loops} implies
6693 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6694 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6695 small constant number of iterations). This option makes code larger, and may
6696 or may not make it run faster.
6698 Enabled with @option{-fprofile-use}.
6700 @item -funroll-all-loops
6701 @opindex funroll-all-loops
6702 Unroll all loops, even if their number of iterations is uncertain when
6703 the loop is entered. This usually makes programs run more slowly.
6704 @option{-funroll-all-loops} implies the same options as
6705 @option{-funroll-loops}.
6708 @opindex fpeel-loops
6709 Peels the loops for that there is enough information that they do not
6710 roll much (from profile feedback). It also turns on complete loop peeling
6711 (i.e.@: complete removal of loops with small constant number of iterations).
6713 Enabled with @option{-fprofile-use}.
6715 @item -fmove-loop-invariants
6716 @opindex fmove-loop-invariants
6717 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6718 at level @option{-O1}
6720 @item -funswitch-loops
6721 @opindex funswitch-loops
6722 Move branches with loop invariant conditions out of the loop, with duplicates
6723 of the loop on both branches (modified according to result of the condition).
6725 @item -ffunction-sections
6726 @itemx -fdata-sections
6727 @opindex ffunction-sections
6728 @opindex fdata-sections
6729 Place each function or data item into its own section in the output
6730 file if the target supports arbitrary sections. The name of the
6731 function or the name of the data item determines the section's name
6734 Use these options on systems where the linker can perform optimizations
6735 to improve locality of reference in the instruction space. Most systems
6736 using the ELF object format and SPARC processors running Solaris 2 have
6737 linkers with such optimizations. AIX may have these optimizations in
6740 Only use these options when there are significant benefits from doing
6741 so. When you specify these options, the assembler and linker will
6742 create larger object and executable files and will also be slower.
6743 You will not be able to use @code{gprof} on all systems if you
6744 specify this option and you may have problems with debugging if
6745 you specify both this option and @option{-g}.
6747 @item -fbranch-target-load-optimize
6748 @opindex fbranch-target-load-optimize
6749 Perform branch target register load optimization before prologue / epilogue
6751 The use of target registers can typically be exposed only during reload,
6752 thus hoisting loads out of loops and doing inter-block scheduling needs
6753 a separate optimization pass.
6755 @item -fbranch-target-load-optimize2
6756 @opindex fbranch-target-load-optimize2
6757 Perform branch target register load optimization after prologue / epilogue
6760 @item -fbtr-bb-exclusive
6761 @opindex fbtr-bb-exclusive
6762 When performing branch target register load optimization, don't reuse
6763 branch target registers in within any basic block.
6765 @item -fstack-protector
6766 @opindex fstack-protector
6767 Emit extra code to check for buffer overflows, such as stack smashing
6768 attacks. This is done by adding a guard variable to functions with
6769 vulnerable objects. This includes functions that call alloca, and
6770 functions with buffers larger than 8 bytes. The guards are initialized
6771 when a function is entered and then checked when the function exits.
6772 If a guard check fails, an error message is printed and the program exits.
6774 @item -fstack-protector-all
6775 @opindex fstack-protector-all
6776 Like @option{-fstack-protector} except that all functions are protected.
6778 @item -fsection-anchors
6779 @opindex fsection-anchors
6780 Try to reduce the number of symbolic address calculations by using
6781 shared ``anchor'' symbols to address nearby objects. This transformation
6782 can help to reduce the number of GOT entries and GOT accesses on some
6785 For example, the implementation of the following function @code{foo}:
6789 int foo (void) @{ return a + b + c; @}
6792 would usually calculate the addresses of all three variables, but if you
6793 compile it with @option{-fsection-anchors}, it will access the variables
6794 from a common anchor point instead. The effect is similar to the
6795 following pseudocode (which isn't valid C):
6800 register int *xr = &x;
6801 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6805 Not all targets support this option.
6807 @item --param @var{name}=@var{value}
6809 In some places, GCC uses various constants to control the amount of
6810 optimization that is done. For example, GCC will not inline functions
6811 that contain more that a certain number of instructions. You can
6812 control some of these constants on the command-line using the
6813 @option{--param} option.
6815 The names of specific parameters, and the meaning of the values, are
6816 tied to the internals of the compiler, and are subject to change
6817 without notice in future releases.
6819 In each case, the @var{value} is an integer. The allowable choices for
6820 @var{name} are given in the following table:
6823 @item sra-max-structure-size
6824 The maximum structure size, in bytes, at which the scalar replacement
6825 of aggregates (SRA) optimization will perform block copies. The
6826 default value, 0, implies that GCC will select the most appropriate
6829 @item sra-field-structure-ratio
6830 The threshold ratio (as a percentage) between instantiated fields and
6831 the complete structure size. We say that if the ratio of the number
6832 of bytes in instantiated fields to the number of bytes in the complete
6833 structure exceeds this parameter, then block copies are not used. The
6836 @item struct-reorg-cold-struct-ratio
6837 The threshold ratio (as a percentage) between a structure frequency
6838 and the frequency of the hottest structure in the program. This parameter
6839 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
6840 We say that if the ratio of a structure frequency, calculated by profiling,
6841 to the hottest structure frequency in the program is less than this
6842 parameter, then structure reorganization is not applied to this structure.
6845 @item max-crossjump-edges
6846 The maximum number of incoming edges to consider for crossjumping.
6847 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6848 the number of edges incoming to each block. Increasing values mean
6849 more aggressive optimization, making the compile time increase with
6850 probably small improvement in executable size.
6852 @item min-crossjump-insns
6853 The minimum number of instructions which must be matched at the end
6854 of two blocks before crossjumping will be performed on them. This
6855 value is ignored in the case where all instructions in the block being
6856 crossjumped from are matched. The default value is 5.
6858 @item max-grow-copy-bb-insns
6859 The maximum code size expansion factor when copying basic blocks
6860 instead of jumping. The expansion is relative to a jump instruction.
6861 The default value is 8.
6863 @item max-goto-duplication-insns
6864 The maximum number of instructions to duplicate to a block that jumps
6865 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6866 passes, GCC factors computed gotos early in the compilation process,
6867 and unfactors them as late as possible. Only computed jumps at the
6868 end of a basic blocks with no more than max-goto-duplication-insns are
6869 unfactored. The default value is 8.
6871 @item max-delay-slot-insn-search
6872 The maximum number of instructions to consider when looking for an
6873 instruction to fill a delay slot. If more than this arbitrary number of
6874 instructions is searched, the time savings from filling the delay slot
6875 will be minimal so stop searching. Increasing values mean more
6876 aggressive optimization, making the compile time increase with probably
6877 small improvement in executable run time.
6879 @item max-delay-slot-live-search
6880 When trying to fill delay slots, the maximum number of instructions to
6881 consider when searching for a block with valid live register
6882 information. Increasing this arbitrarily chosen value means more
6883 aggressive optimization, increasing the compile time. This parameter
6884 should be removed when the delay slot code is rewritten to maintain the
6887 @item max-gcse-memory
6888 The approximate maximum amount of memory that will be allocated in
6889 order to perform the global common subexpression elimination
6890 optimization. If more memory than specified is required, the
6891 optimization will not be done.
6893 @item max-gcse-passes
6894 The maximum number of passes of GCSE to run. The default is 1.
6896 @item max-pending-list-length
6897 The maximum number of pending dependencies scheduling will allow
6898 before flushing the current state and starting over. Large functions
6899 with few branches or calls can create excessively large lists which
6900 needlessly consume memory and resources.
6902 @item max-inline-insns-single
6903 Several parameters control the tree inliner used in gcc.
6904 This number sets the maximum number of instructions (counted in GCC's
6905 internal representation) in a single function that the tree inliner
6906 will consider for inlining. This only affects functions declared
6907 inline and methods implemented in a class declaration (C++).
6908 The default value is 450.
6910 @item max-inline-insns-auto
6911 When you use @option{-finline-functions} (included in @option{-O3}),
6912 a lot of functions that would otherwise not be considered for inlining
6913 by the compiler will be investigated. To those functions, a different
6914 (more restrictive) limit compared to functions declared inline can
6916 The default value is 90.
6918 @item large-function-insns
6919 The limit specifying really large functions. For functions larger than this
6920 limit after inlining inlining is constrained by
6921 @option{--param large-function-growth}. This parameter is useful primarily
6922 to avoid extreme compilation time caused by non-linear algorithms used by the
6924 The default value is 2700.
6926 @item large-function-growth
6927 Specifies maximal growth of large function caused by inlining in percents.
6928 The default value is 100 which limits large function growth to 2.0 times
6931 @item large-unit-insns
6932 The limit specifying large translation unit. Growth caused by inlining of
6933 units larger than this limit is limited by @option{--param inline-unit-growth}.
6934 For small units this might be too tight (consider unit consisting of function A
6935 that is inline and B that just calls A three time. If B is small relative to
6936 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6937 large units consisting of small inlineable functions however the overall unit
6938 growth limit is needed to avoid exponential explosion of code size. Thus for
6939 smaller units, the size is increased to @option{--param large-unit-insns}
6940 before applying @option{--param inline-unit-growth}. The default is 10000
6942 @item inline-unit-growth
6943 Specifies maximal overall growth of the compilation unit caused by inlining.
6944 The default value is 30 which limits unit growth to 1.3 times the original
6947 @item large-stack-frame
6948 The limit specifying large stack frames. While inlining the algorithm is trying
6949 to not grow past this limit too much. Default value is 256 bytes.
6951 @item large-stack-frame-growth
6952 Specifies maximal growth of large stack frames caused by inlining in percents.
6953 The default value is 1000 which limits large stack frame growth to 11 times
6956 @item max-inline-insns-recursive
6957 @itemx max-inline-insns-recursive-auto
6958 Specifies maximum number of instructions out-of-line copy of self recursive inline
6959 function can grow into by performing recursive inlining.
6961 For functions declared inline @option{--param max-inline-insns-recursive} is
6962 taken into account. For function not declared inline, recursive inlining
6963 happens only when @option{-finline-functions} (included in @option{-O3}) is
6964 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6965 default value is 450.
6967 @item max-inline-recursive-depth
6968 @itemx max-inline-recursive-depth-auto
6969 Specifies maximum recursion depth used by the recursive inlining.
6971 For functions declared inline @option{--param max-inline-recursive-depth} is
6972 taken into account. For function not declared inline, recursive inlining
6973 happens only when @option{-finline-functions} (included in @option{-O3}) is
6974 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6977 @item min-inline-recursive-probability
6978 Recursive inlining is profitable only for function having deep recursion
6979 in average and can hurt for function having little recursion depth by
6980 increasing the prologue size or complexity of function body to other
6983 When profile feedback is available (see @option{-fprofile-generate}) the actual
6984 recursion depth can be guessed from probability that function will recurse via
6985 given call expression. This parameter limits inlining only to call expression
6986 whose probability exceeds given threshold (in percents). The default value is
6989 @item inline-call-cost
6990 Specify cost of call instruction relative to simple arithmetics operations
6991 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6992 functions and at the same time increases size of leaf function that is believed to
6993 reduce function size by being inlined. In effect it increases amount of
6994 inlining for code having large abstraction penalty (many functions that just
6995 pass the arguments to other functions) and decrease inlining for code with low
6996 abstraction penalty. The default value is 12.
6998 @item min-vect-loop-bound
6999 The minimum number of iterations under which a loop will not get vectorized
7000 when @option{-ftree-vectorize} is used. The number of iterations after
7001 vectorization needs to be greater than the value specified by this option
7002 to allow vectorization. The default value is 0.
7004 @item max-unrolled-insns
7005 The maximum number of instructions that a loop should have if that loop
7006 is unrolled, and if the loop is unrolled, it determines how many times
7007 the loop code is unrolled.
7009 @item max-average-unrolled-insns
7010 The maximum number of instructions biased by probabilities of their execution
7011 that a loop should have if that loop is unrolled, and if the loop is unrolled,
7012 it determines how many times the loop code is unrolled.
7014 @item max-unroll-times
7015 The maximum number of unrollings of a single loop.
7017 @item max-peeled-insns
7018 The maximum number of instructions that a loop should have if that loop
7019 is peeled, and if the loop is peeled, it determines how many times
7020 the loop code is peeled.
7022 @item max-peel-times
7023 The maximum number of peelings of a single loop.
7025 @item max-completely-peeled-insns
7026 The maximum number of insns of a completely peeled loop.
7028 @item max-completely-peel-times
7029 The maximum number of iterations of a loop to be suitable for complete peeling.
7031 @item max-unswitch-insns
7032 The maximum number of insns of an unswitched loop.
7034 @item max-unswitch-level
7035 The maximum number of branches unswitched in a single loop.
7038 The minimum cost of an expensive expression in the loop invariant motion.
7040 @item iv-consider-all-candidates-bound
7041 Bound on number of candidates for induction variables below that
7042 all candidates are considered for each use in induction variable
7043 optimizations. Only the most relevant candidates are considered
7044 if there are more candidates, to avoid quadratic time complexity.
7046 @item iv-max-considered-uses
7047 The induction variable optimizations give up on loops that contain more
7048 induction variable uses.
7050 @item iv-always-prune-cand-set-bound
7051 If number of candidates in the set is smaller than this value,
7052 we always try to remove unnecessary ivs from the set during its
7053 optimization when a new iv is added to the set.
7055 @item scev-max-expr-size
7056 Bound on size of expressions used in the scalar evolutions analyzer.
7057 Large expressions slow the analyzer.
7059 @item omega-max-vars
7060 The maximum number of variables in an Omega constraint system.
7061 The default value is 128.
7063 @item omega-max-geqs
7064 The maximum number of inequalities in an Omega constraint system.
7065 The default value is 256.
7068 The maximum number of equalities in an Omega constraint system.
7069 The default value is 128.
7071 @item omega-max-wild-cards
7072 The maximum number of wildcard variables that the Omega solver will
7073 be able to insert. The default value is 18.
7075 @item omega-hash-table-size
7076 The size of the hash table in the Omega solver. The default value is
7079 @item omega-max-keys
7080 The maximal number of keys used by the Omega solver. The default
7083 @item omega-eliminate-redundant-constraints
7084 When set to 1, use expensive methods to eliminate all redundant
7085 constraints. The default value is 0.
7087 @item vect-max-version-for-alignment-checks
7088 The maximum number of runtime checks that can be performed when
7089 doing loop versioning for alignment in the vectorizer. See option
7090 ftree-vect-loop-version for more information.
7092 @item vect-max-version-for-alias-checks
7093 The maximum number of runtime checks that can be performed when
7094 doing loop versioning for alias in the vectorizer. See option
7095 ftree-vect-loop-version for more information.
7097 @item max-iterations-to-track
7099 The maximum number of iterations of a loop the brute force algorithm
7100 for analysis of # of iterations of the loop tries to evaluate.
7102 @item hot-bb-count-fraction
7103 Select fraction of the maximal count of repetitions of basic block in program
7104 given basic block needs to have to be considered hot.
7106 @item hot-bb-frequency-fraction
7107 Select fraction of the maximal frequency of executions of basic block in
7108 function given basic block needs to have to be considered hot
7110 @item max-predicted-iterations
7111 The maximum number of loop iterations we predict statically. This is useful
7112 in cases where function contain single loop with known bound and other loop
7113 with unknown. We predict the known number of iterations correctly, while
7114 the unknown number of iterations average to roughly 10. This means that the
7115 loop without bounds would appear artificially cold relative to the other one.
7117 @item align-threshold
7119 Select fraction of the maximal frequency of executions of basic block in
7120 function given basic block will get aligned.
7122 @item align-loop-iterations
7124 A loop expected to iterate at lest the selected number of iterations will get
7127 @item tracer-dynamic-coverage
7128 @itemx tracer-dynamic-coverage-feedback
7130 This value is used to limit superblock formation once the given percentage of
7131 executed instructions is covered. This limits unnecessary code size
7134 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7135 feedback is available. The real profiles (as opposed to statically estimated
7136 ones) are much less balanced allowing the threshold to be larger value.
7138 @item tracer-max-code-growth
7139 Stop tail duplication once code growth has reached given percentage. This is
7140 rather hokey argument, as most of the duplicates will be eliminated later in
7141 cross jumping, so it may be set to much higher values than is the desired code
7144 @item tracer-min-branch-ratio
7146 Stop reverse growth when the reverse probability of best edge is less than this
7147 threshold (in percent).
7149 @item tracer-min-branch-ratio
7150 @itemx tracer-min-branch-ratio-feedback
7152 Stop forward growth if the best edge do have probability lower than this
7155 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7156 compilation for profile feedback and one for compilation without. The value
7157 for compilation with profile feedback needs to be more conservative (higher) in
7158 order to make tracer effective.
7160 @item max-cse-path-length
7162 Maximum number of basic blocks on path that cse considers. The default is 10.
7165 The maximum instructions CSE process before flushing. The default is 1000.
7167 @item max-aliased-vops
7169 Maximum number of virtual operands per function allowed to represent
7170 aliases before triggering the alias partitioning heuristic. Alias
7171 partitioning reduces compile times and memory consumption needed for
7172 aliasing at the expense of precision loss in alias information. The
7173 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7176 Notice that if a function contains more memory statements than the
7177 value of this parameter, it is not really possible to achieve this
7178 reduction. In this case, the compiler will use the number of memory
7179 statements as the value for @option{max-aliased-vops}.
7181 @item avg-aliased-vops
7183 Average number of virtual operands per statement allowed to represent
7184 aliases before triggering the alias partitioning heuristic. This
7185 works in conjunction with @option{max-aliased-vops}. If a function
7186 contains more than @option{max-aliased-vops} virtual operators, then
7187 memory symbols will be grouped into memory partitions until either the
7188 total number of virtual operators is below @option{max-aliased-vops}
7189 or the average number of virtual operators per memory statement is
7190 below @option{avg-aliased-vops}. The default value for this parameter
7191 is 1 for -O1 and -O2, and 3 for -O3.
7193 @item ggc-min-expand
7195 GCC uses a garbage collector to manage its own memory allocation. This
7196 parameter specifies the minimum percentage by which the garbage
7197 collector's heap should be allowed to expand between collections.
7198 Tuning this may improve compilation speed; it has no effect on code
7201 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7202 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7203 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7204 GCC is not able to calculate RAM on a particular platform, the lower
7205 bound of 30% is used. Setting this parameter and
7206 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7207 every opportunity. This is extremely slow, but can be useful for
7210 @item ggc-min-heapsize
7212 Minimum size of the garbage collector's heap before it begins bothering
7213 to collect garbage. The first collection occurs after the heap expands
7214 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7215 tuning this may improve compilation speed, and has no effect on code
7218 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7219 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7220 with a lower bound of 4096 (four megabytes) and an upper bound of
7221 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7222 particular platform, the lower bound is used. Setting this parameter
7223 very large effectively disables garbage collection. Setting this
7224 parameter and @option{ggc-min-expand} to zero causes a full collection
7225 to occur at every opportunity.
7227 @item max-reload-search-insns
7228 The maximum number of instruction reload should look backward for equivalent
7229 register. Increasing values mean more aggressive optimization, making the
7230 compile time increase with probably slightly better performance. The default
7233 @item max-cselib-memory-locations
7234 The maximum number of memory locations cselib should take into account.
7235 Increasing values mean more aggressive optimization, making the compile time
7236 increase with probably slightly better performance. The default value is 500.
7238 @item reorder-blocks-duplicate
7239 @itemx reorder-blocks-duplicate-feedback
7241 Used by basic block reordering pass to decide whether to use unconditional
7242 branch or duplicate the code on its destination. Code is duplicated when its
7243 estimated size is smaller than this value multiplied by the estimated size of
7244 unconditional jump in the hot spots of the program.
7246 The @option{reorder-block-duplicate-feedback} is used only when profile
7247 feedback is available and may be set to higher values than
7248 @option{reorder-block-duplicate} since information about the hot spots is more
7251 @item max-sched-ready-insns
7252 The maximum number of instructions ready to be issued the scheduler should
7253 consider at any given time during the first scheduling pass. Increasing
7254 values mean more thorough searches, making the compilation time increase
7255 with probably little benefit. The default value is 100.
7257 @item max-sched-region-blocks
7258 The maximum number of blocks in a region to be considered for
7259 interblock scheduling. The default value is 10.
7261 @item max-sched-region-insns
7262 The maximum number of insns in a region to be considered for
7263 interblock scheduling. The default value is 100.
7266 The minimum probability (in percents) of reaching a source block
7267 for interblock speculative scheduling. The default value is 40.
7269 @item max-sched-extend-regions-iters
7270 The maximum number of iterations through CFG to extend regions.
7271 0 - disable region extension,
7272 N - do at most N iterations.
7273 The default value is 0.
7275 @item max-sched-insn-conflict-delay
7276 The maximum conflict delay for an insn to be considered for speculative motion.
7277 The default value is 3.
7279 @item sched-spec-prob-cutoff
7280 The minimal probability of speculation success (in percents), so that
7281 speculative insn will be scheduled.
7282 The default value is 40.
7284 @item max-last-value-rtl
7286 The maximum size measured as number of RTLs that can be recorded in an expression
7287 in combiner for a pseudo register as last known value of that register. The default
7290 @item integer-share-limit
7291 Small integer constants can use a shared data structure, reducing the
7292 compiler's memory usage and increasing its speed. This sets the maximum
7293 value of a shared integer constant. The default value is 256.
7295 @item min-virtual-mappings
7296 Specifies the minimum number of virtual mappings in the incremental
7297 SSA updater that should be registered to trigger the virtual mappings
7298 heuristic defined by virtual-mappings-ratio. The default value is
7301 @item virtual-mappings-ratio
7302 If the number of virtual mappings is virtual-mappings-ratio bigger
7303 than the number of virtual symbols to be updated, then the incremental
7304 SSA updater switches to a full update for those symbols. The default
7307 @item ssp-buffer-size
7308 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
7309 protection when @option{-fstack-protection} is used.
7311 @item max-jump-thread-duplication-stmts
7312 Maximum number of statements allowed in a block that needs to be
7313 duplicated when threading jumps.
7315 @item max-fields-for-field-sensitive
7316 Maximum number of fields in a structure we will treat in
7317 a field sensitive manner during pointer analysis. The default is zero
7318 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
7320 @item prefetch-latency
7321 Estimate on average number of instructions that are executed before
7322 prefetch finishes. The distance we prefetch ahead is proportional
7323 to this constant. Increasing this number may also lead to less
7324 streams being prefetched (see @option{simultaneous-prefetches}).
7326 @item simultaneous-prefetches
7327 Maximum number of prefetches that can run at the same time.
7329 @item l1-cache-line-size
7330 The size of cache line in L1 cache, in bytes.
7333 The size of L1 cache, in kilobytes.
7336 The size of L2 cache, in kilobytes.
7338 @item use-canonical-types
7339 Whether the compiler should use the ``canonical'' type system. By
7340 default, this should always be 1, which uses a more efficient internal
7341 mechanism for comparing types in C++ and Objective-C++. However, if
7342 bugs in the canonical type system are causing compilation failures,
7343 set this value to 0 to disable canonical types.
7345 @item switch-conversion-max-branch-ratio
7346 Switch initialization conversion will refuse to create arrays that are
7347 bigger than @option{switch-conversion-max-branch-ratio} times the number of
7348 branches in the switch.
7350 @item max-partial-antic-length
7351 Maximum length of the partial antic set computed during the tree
7352 partial redundancy elimination optimization (@option{-ftree-pre}) when
7353 optimizing at @option{-O3} and above. For some sorts of source code
7354 the enhanced partial redundancy elimination optimization can run away,
7355 consuming all of the memory available on the host machine. This
7356 parameter sets a limit on the length of the sets that are computed,
7357 which prevents the runaway behaviour. Setting a value of 0 for
7358 this paramter will allow an unlimited set length.
7360 @item sccvn-max-scc-size
7361 Maximum size of a strongly connected component (SCC) during SCCVN
7362 processing. If this limit is hit, SCCVN processing for the whole
7363 function will not be done and optimizations depending on it will
7364 be disabled. The default maximum SCC size is 10000.
7369 @node Preprocessor Options
7370 @section Options Controlling the Preprocessor
7371 @cindex preprocessor options
7372 @cindex options, preprocessor
7374 These options control the C preprocessor, which is run on each C source
7375 file before actual compilation.
7377 If you use the @option{-E} option, nothing is done except preprocessing.
7378 Some of these options make sense only together with @option{-E} because
7379 they cause the preprocessor output to be unsuitable for actual
7384 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7385 and pass @var{option} directly through to the preprocessor. If
7386 @var{option} contains commas, it is split into multiple options at the
7387 commas. However, many options are modified, translated or interpreted
7388 by the compiler driver before being passed to the preprocessor, and
7389 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7390 interface is undocumented and subject to change, so whenever possible
7391 you should avoid using @option{-Wp} and let the driver handle the
7394 @item -Xpreprocessor @var{option}
7395 @opindex preprocessor
7396 Pass @var{option} as an option to the preprocessor. You can use this to
7397 supply system-specific preprocessor options which GCC does not know how to
7400 If you want to pass an option that takes an argument, you must use
7401 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7404 @include cppopts.texi
7406 @node Assembler Options
7407 @section Passing Options to the Assembler
7409 @c prevent bad page break with this line
7410 You can pass options to the assembler.
7413 @item -Wa,@var{option}
7415 Pass @var{option} as an option to the assembler. If @var{option}
7416 contains commas, it is split into multiple options at the commas.
7418 @item -Xassembler @var{option}
7420 Pass @var{option} as an option to the assembler. You can use this to
7421 supply system-specific assembler options which GCC does not know how to
7424 If you want to pass an option that takes an argument, you must use
7425 @option{-Xassembler} twice, once for the option and once for the argument.
7430 @section Options for Linking
7431 @cindex link options
7432 @cindex options, linking
7434 These options come into play when the compiler links object files into
7435 an executable output file. They are meaningless if the compiler is
7436 not doing a link step.
7440 @item @var{object-file-name}
7441 A file name that does not end in a special recognized suffix is
7442 considered to name an object file or library. (Object files are
7443 distinguished from libraries by the linker according to the file
7444 contents.) If linking is done, these object files are used as input
7453 If any of these options is used, then the linker is not run, and
7454 object file names should not be used as arguments. @xref{Overall
7458 @item -l@var{library}
7459 @itemx -l @var{library}
7461 Search the library named @var{library} when linking. (The second
7462 alternative with the library as a separate argument is only for
7463 POSIX compliance and is not recommended.)
7465 It makes a difference where in the command you write this option; the
7466 linker searches and processes libraries and object files in the order they
7467 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7468 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7469 to functions in @samp{z}, those functions may not be loaded.
7471 The linker searches a standard list of directories for the library,
7472 which is actually a file named @file{lib@var{library}.a}. The linker
7473 then uses this file as if it had been specified precisely by name.
7475 The directories searched include several standard system directories
7476 plus any that you specify with @option{-L}.
7478 Normally the files found this way are library files---archive files
7479 whose members are object files. The linker handles an archive file by
7480 scanning through it for members which define symbols that have so far
7481 been referenced but not defined. But if the file that is found is an
7482 ordinary object file, it is linked in the usual fashion. The only
7483 difference between using an @option{-l} option and specifying a file name
7484 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7485 and searches several directories.
7489 You need this special case of the @option{-l} option in order to
7490 link an Objective-C or Objective-C++ program.
7493 @opindex nostartfiles
7494 Do not use the standard system startup files when linking.
7495 The standard system libraries are used normally, unless @option{-nostdlib}
7496 or @option{-nodefaultlibs} is used.
7498 @item -nodefaultlibs
7499 @opindex nodefaultlibs
7500 Do not use the standard system libraries when linking.
7501 Only the libraries you specify will be passed to the linker.
7502 The standard startup files are used normally, unless @option{-nostartfiles}
7503 is used. The compiler may generate calls to @code{memcmp},
7504 @code{memset}, @code{memcpy} and @code{memmove}.
7505 These entries are usually resolved by entries in
7506 libc. These entry points should be supplied through some other
7507 mechanism when this option is specified.
7511 Do not use the standard system startup files or libraries when linking.
7512 No startup files and only the libraries you specify will be passed to
7513 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7514 @code{memcpy} and @code{memmove}.
7515 These entries are usually resolved by entries in
7516 libc. These entry points should be supplied through some other
7517 mechanism when this option is specified.
7519 @cindex @option{-lgcc}, use with @option{-nostdlib}
7520 @cindex @option{-nostdlib} and unresolved references
7521 @cindex unresolved references and @option{-nostdlib}
7522 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7523 @cindex @option{-nodefaultlibs} and unresolved references
7524 @cindex unresolved references and @option{-nodefaultlibs}
7525 One of the standard libraries bypassed by @option{-nostdlib} and
7526 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7527 that GCC uses to overcome shortcomings of particular machines, or special
7528 needs for some languages.
7529 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7530 Collection (GCC) Internals},
7531 for more discussion of @file{libgcc.a}.)
7532 In most cases, you need @file{libgcc.a} even when you want to avoid
7533 other standard libraries. In other words, when you specify @option{-nostdlib}
7534 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7535 This ensures that you have no unresolved references to internal GCC
7536 library subroutines. (For example, @samp{__main}, used to ensure C++
7537 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7538 GNU Compiler Collection (GCC) Internals}.)
7542 Produce a position independent executable on targets which support it.
7543 For predictable results, you must also specify the same set of options
7544 that were used to generate code (@option{-fpie}, @option{-fPIE},
7545 or model suboptions) when you specify this option.
7549 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7550 that support it. This instructs the linker to add all symbols, not
7551 only used ones, to the dynamic symbol table. This option is needed
7552 for some uses of @code{dlopen} or to allow obtaining backtraces
7553 from within a program.
7557 Remove all symbol table and relocation information from the executable.
7561 On systems that support dynamic linking, this prevents linking with the shared
7562 libraries. On other systems, this option has no effect.
7566 Produce a shared object which can then be linked with other objects to
7567 form an executable. Not all systems support this option. For predictable
7568 results, you must also specify the same set of options that were used to
7569 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7570 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7571 needs to build supplementary stub code for constructors to work. On
7572 multi-libbed systems, @samp{gcc -shared} must select the correct support
7573 libraries to link against. Failing to supply the correct flags may lead
7574 to subtle defects. Supplying them in cases where they are not necessary
7577 @item -shared-libgcc
7578 @itemx -static-libgcc
7579 @opindex shared-libgcc
7580 @opindex static-libgcc
7581 On systems that provide @file{libgcc} as a shared library, these options
7582 force the use of either the shared or static version respectively.
7583 If no shared version of @file{libgcc} was built when the compiler was
7584 configured, these options have no effect.
7586 There are several situations in which an application should use the
7587 shared @file{libgcc} instead of the static version. The most common
7588 of these is when the application wishes to throw and catch exceptions
7589 across different shared libraries. In that case, each of the libraries
7590 as well as the application itself should use the shared @file{libgcc}.
7592 Therefore, the G++ and GCJ drivers automatically add
7593 @option{-shared-libgcc} whenever you build a shared library or a main
7594 executable, because C++ and Java programs typically use exceptions, so
7595 this is the right thing to do.
7597 If, instead, you use the GCC driver to create shared libraries, you may
7598 find that they will not always be linked with the shared @file{libgcc}.
7599 If GCC finds, at its configuration time, that you have a non-GNU linker
7600 or a GNU linker that does not support option @option{--eh-frame-hdr},
7601 it will link the shared version of @file{libgcc} into shared libraries
7602 by default. Otherwise, it will take advantage of the linker and optimize
7603 away the linking with the shared version of @file{libgcc}, linking with
7604 the static version of libgcc by default. This allows exceptions to
7605 propagate through such shared libraries, without incurring relocation
7606 costs at library load time.
7608 However, if a library or main executable is supposed to throw or catch
7609 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7610 for the languages used in the program, or using the option
7611 @option{-shared-libgcc}, such that it is linked with the shared
7616 Bind references to global symbols when building a shared object. Warn
7617 about any unresolved references (unless overridden by the link editor
7618 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7621 @item -Xlinker @var{option}
7623 Pass @var{option} as an option to the linker. You can use this to
7624 supply system-specific linker options which GCC does not know how to
7627 If you want to pass an option that takes an argument, you must use
7628 @option{-Xlinker} twice, once for the option and once for the argument.
7629 For example, to pass @option{-assert definitions}, you must write
7630 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7631 @option{-Xlinker "-assert definitions"}, because this passes the entire
7632 string as a single argument, which is not what the linker expects.
7634 @item -Wl,@var{option}
7636 Pass @var{option} as an option to the linker. If @var{option} contains
7637 commas, it is split into multiple options at the commas.
7639 @item -u @var{symbol}
7641 Pretend the symbol @var{symbol} is undefined, to force linking of
7642 library modules to define it. You can use @option{-u} multiple times with
7643 different symbols to force loading of additional library modules.
7646 @node Directory Options
7647 @section Options for Directory Search
7648 @cindex directory options
7649 @cindex options, directory search
7652 These options specify directories to search for header files, for
7653 libraries and for parts of the compiler:
7658 Add the directory @var{dir} to the head of the list of directories to be
7659 searched for header files. This can be used to override a system header
7660 file, substituting your own version, since these directories are
7661 searched before the system header file directories. However, you should
7662 not use this option to add directories that contain vendor-supplied
7663 system header files (use @option{-isystem} for that). If you use more than
7664 one @option{-I} option, the directories are scanned in left-to-right
7665 order; the standard system directories come after.
7667 If a standard system include directory, or a directory specified with
7668 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7669 option will be ignored. The directory will still be searched but as a
7670 system directory at its normal position in the system include chain.
7671 This is to ensure that GCC's procedure to fix buggy system headers and
7672 the ordering for the include_next directive are not inadvertently changed.
7673 If you really need to change the search order for system directories,
7674 use the @option{-nostdinc} and/or @option{-isystem} options.
7676 @item -iquote@var{dir}
7678 Add the directory @var{dir} to the head of the list of directories to
7679 be searched for header files only for the case of @samp{#include
7680 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7681 otherwise just like @option{-I}.
7685 Add directory @var{dir} to the list of directories to be searched
7688 @item -B@var{prefix}
7690 This option specifies where to find the executables, libraries,
7691 include files, and data files of the compiler itself.
7693 The compiler driver program runs one or more of the subprograms
7694 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7695 @var{prefix} as a prefix for each program it tries to run, both with and
7696 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7698 For each subprogram to be run, the compiler driver first tries the
7699 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7700 was not specified, the driver tries two standard prefixes, which are
7701 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7702 those results in a file name that is found, the unmodified program
7703 name is searched for using the directories specified in your
7704 @env{PATH} environment variable.
7706 The compiler will check to see if the path provided by the @option{-B}
7707 refers to a directory, and if necessary it will add a directory
7708 separator character at the end of the path.
7710 @option{-B} prefixes that effectively specify directory names also apply
7711 to libraries in the linker, because the compiler translates these
7712 options into @option{-L} options for the linker. They also apply to
7713 includes files in the preprocessor, because the compiler translates these
7714 options into @option{-isystem} options for the preprocessor. In this case,
7715 the compiler appends @samp{include} to the prefix.
7717 The run-time support file @file{libgcc.a} can also be searched for using
7718 the @option{-B} prefix, if needed. If it is not found there, the two
7719 standard prefixes above are tried, and that is all. The file is left
7720 out of the link if it is not found by those means.
7722 Another way to specify a prefix much like the @option{-B} prefix is to use
7723 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7726 As a special kludge, if the path provided by @option{-B} is
7727 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7728 9, then it will be replaced by @file{[dir/]include}. This is to help
7729 with boot-strapping the compiler.
7731 @item -specs=@var{file}
7733 Process @var{file} after the compiler reads in the standard @file{specs}
7734 file, in order to override the defaults that the @file{gcc} driver
7735 program uses when determining what switches to pass to @file{cc1},
7736 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7737 @option{-specs=@var{file}} can be specified on the command line, and they
7738 are processed in order, from left to right.
7740 @item --sysroot=@var{dir}
7742 Use @var{dir} as the logical root directory for headers and libraries.
7743 For example, if the compiler would normally search for headers in
7744 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7745 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7747 If you use both this option and the @option{-isysroot} option, then
7748 the @option{--sysroot} option will apply to libraries, but the
7749 @option{-isysroot} option will apply to header files.
7751 The GNU linker (beginning with version 2.16) has the necessary support
7752 for this option. If your linker does not support this option, the
7753 header file aspect of @option{--sysroot} will still work, but the
7754 library aspect will not.
7758 This option has been deprecated. Please use @option{-iquote} instead for
7759 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7760 Any directories you specify with @option{-I} options before the @option{-I-}
7761 option are searched only for the case of @samp{#include "@var{file}"};
7762 they are not searched for @samp{#include <@var{file}>}.
7764 If additional directories are specified with @option{-I} options after
7765 the @option{-I-}, these directories are searched for all @samp{#include}
7766 directives. (Ordinarily @emph{all} @option{-I} directories are used
7769 In addition, the @option{-I-} option inhibits the use of the current
7770 directory (where the current input file came from) as the first search
7771 directory for @samp{#include "@var{file}"}. There is no way to
7772 override this effect of @option{-I-}. With @option{-I.} you can specify
7773 searching the directory which was current when the compiler was
7774 invoked. That is not exactly the same as what the preprocessor does
7775 by default, but it is often satisfactory.
7777 @option{-I-} does not inhibit the use of the standard system directories
7778 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7785 @section Specifying subprocesses and the switches to pass to them
7788 @command{gcc} is a driver program. It performs its job by invoking a
7789 sequence of other programs to do the work of compiling, assembling and
7790 linking. GCC interprets its command-line parameters and uses these to
7791 deduce which programs it should invoke, and which command-line options
7792 it ought to place on their command lines. This behavior is controlled
7793 by @dfn{spec strings}. In most cases there is one spec string for each
7794 program that GCC can invoke, but a few programs have multiple spec
7795 strings to control their behavior. The spec strings built into GCC can
7796 be overridden by using the @option{-specs=} command-line switch to specify
7799 @dfn{Spec files} are plaintext files that are used to construct spec
7800 strings. They consist of a sequence of directives separated by blank
7801 lines. The type of directive is determined by the first non-whitespace
7802 character on the line and it can be one of the following:
7805 @item %@var{command}
7806 Issues a @var{command} to the spec file processor. The commands that can
7810 @item %include <@var{file}>
7812 Search for @var{file} and insert its text at the current point in the
7815 @item %include_noerr <@var{file}>
7816 @cindex %include_noerr
7817 Just like @samp{%include}, but do not generate an error message if the include
7818 file cannot be found.
7820 @item %rename @var{old_name} @var{new_name}
7822 Rename the spec string @var{old_name} to @var{new_name}.
7826 @item *[@var{spec_name}]:
7827 This tells the compiler to create, override or delete the named spec
7828 string. All lines after this directive up to the next directive or
7829 blank line are considered to be the text for the spec string. If this
7830 results in an empty string then the spec will be deleted. (Or, if the
7831 spec did not exist, then nothing will happened.) Otherwise, if the spec
7832 does not currently exist a new spec will be created. If the spec does
7833 exist then its contents will be overridden by the text of this
7834 directive, unless the first character of that text is the @samp{+}
7835 character, in which case the text will be appended to the spec.
7837 @item [@var{suffix}]:
7838 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7839 and up to the next directive or blank line are considered to make up the
7840 spec string for the indicated suffix. When the compiler encounters an
7841 input file with the named suffix, it will processes the spec string in
7842 order to work out how to compile that file. For example:
7849 This says that any input file whose name ends in @samp{.ZZ} should be
7850 passed to the program @samp{z-compile}, which should be invoked with the
7851 command-line switch @option{-input} and with the result of performing the
7852 @samp{%i} substitution. (See below.)
7854 As an alternative to providing a spec string, the text that follows a
7855 suffix directive can be one of the following:
7858 @item @@@var{language}
7859 This says that the suffix is an alias for a known @var{language}. This is
7860 similar to using the @option{-x} command-line switch to GCC to specify a
7861 language explicitly. For example:
7868 Says that .ZZ files are, in fact, C++ source files.
7871 This causes an error messages saying:
7874 @var{name} compiler not installed on this system.
7878 GCC already has an extensive list of suffixes built into it.
7879 This directive will add an entry to the end of the list of suffixes, but
7880 since the list is searched from the end backwards, it is effectively
7881 possible to override earlier entries using this technique.
7885 GCC has the following spec strings built into it. Spec files can
7886 override these strings or create their own. Note that individual
7887 targets can also add their own spec strings to this list.
7890 asm Options to pass to the assembler
7891 asm_final Options to pass to the assembler post-processor
7892 cpp Options to pass to the C preprocessor
7893 cc1 Options to pass to the C compiler
7894 cc1plus Options to pass to the C++ compiler
7895 endfile Object files to include at the end of the link
7896 link Options to pass to the linker
7897 lib Libraries to include on the command line to the linker
7898 libgcc Decides which GCC support library to pass to the linker
7899 linker Sets the name of the linker
7900 predefines Defines to be passed to the C preprocessor
7901 signed_char Defines to pass to CPP to say whether @code{char} is signed
7903 startfile Object files to include at the start of the link
7906 Here is a small example of a spec file:
7912 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7915 This example renames the spec called @samp{lib} to @samp{old_lib} and
7916 then overrides the previous definition of @samp{lib} with a new one.
7917 The new definition adds in some extra command-line options before
7918 including the text of the old definition.
7920 @dfn{Spec strings} are a list of command-line options to be passed to their
7921 corresponding program. In addition, the spec strings can contain
7922 @samp{%}-prefixed sequences to substitute variable text or to
7923 conditionally insert text into the command line. Using these constructs
7924 it is possible to generate quite complex command lines.
7926 Here is a table of all defined @samp{%}-sequences for spec
7927 strings. Note that spaces are not generated automatically around the
7928 results of expanding these sequences. Therefore you can concatenate them
7929 together or combine them with constant text in a single argument.
7933 Substitute one @samp{%} into the program name or argument.
7936 Substitute the name of the input file being processed.
7939 Substitute the basename of the input file being processed.
7940 This is the substring up to (and not including) the last period
7941 and not including the directory.
7944 This is the same as @samp{%b}, but include the file suffix (text after
7948 Marks the argument containing or following the @samp{%d} as a
7949 temporary file name, so that that file will be deleted if GCC exits
7950 successfully. Unlike @samp{%g}, this contributes no text to the
7953 @item %g@var{suffix}
7954 Substitute a file name that has suffix @var{suffix} and is chosen
7955 once per compilation, and mark the argument in the same way as
7956 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7957 name is now chosen in a way that is hard to predict even when previously
7958 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7959 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7960 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7961 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7962 was simply substituted with a file name chosen once per compilation,
7963 without regard to any appended suffix (which was therefore treated
7964 just like ordinary text), making such attacks more likely to succeed.
7966 @item %u@var{suffix}
7967 Like @samp{%g}, but generates a new temporary file name even if
7968 @samp{%u@var{suffix}} was already seen.
7970 @item %U@var{suffix}
7971 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7972 new one if there is no such last file name. In the absence of any
7973 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7974 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7975 would involve the generation of two distinct file names, one
7976 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7977 simply substituted with a file name chosen for the previous @samp{%u},
7978 without regard to any appended suffix.
7980 @item %j@var{suffix}
7981 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7982 writable, and if save-temps is off; otherwise, substitute the name
7983 of a temporary file, just like @samp{%u}. This temporary file is not
7984 meant for communication between processes, but rather as a junk
7987 @item %|@var{suffix}
7988 @itemx %m@var{suffix}
7989 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7990 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7991 all. These are the two most common ways to instruct a program that it
7992 should read from standard input or write to standard output. If you
7993 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7994 construct: see for example @file{f/lang-specs.h}.
7996 @item %.@var{SUFFIX}
7997 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7998 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7999 terminated by the next space or %.
8002 Marks the argument containing or following the @samp{%w} as the
8003 designated output file of this compilation. This puts the argument
8004 into the sequence of arguments that @samp{%o} will substitute later.
8007 Substitutes the names of all the output files, with spaces
8008 automatically placed around them. You should write spaces
8009 around the @samp{%o} as well or the results are undefined.
8010 @samp{%o} is for use in the specs for running the linker.
8011 Input files whose names have no recognized suffix are not compiled
8012 at all, but they are included among the output files, so they will
8016 Substitutes the suffix for object files. Note that this is
8017 handled specially when it immediately follows @samp{%g, %u, or %U},
8018 because of the need for those to form complete file names. The
8019 handling is such that @samp{%O} is treated exactly as if it had already
8020 been substituted, except that @samp{%g, %u, and %U} do not currently
8021 support additional @var{suffix} characters following @samp{%O} as they would
8022 following, for example, @samp{.o}.
8025 Substitutes the standard macro predefinitions for the
8026 current target machine. Use this when running @code{cpp}.
8029 Like @samp{%p}, but puts @samp{__} before and after the name of each
8030 predefined macro, except for macros that start with @samp{__} or with
8031 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
8035 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
8036 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
8037 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
8038 and @option{-imultilib} as necessary.
8041 Current argument is the name of a library or startup file of some sort.
8042 Search for that file in a standard list of directories and substitute
8043 the full name found.
8046 Print @var{str} as an error message. @var{str} is terminated by a newline.
8047 Use this when inconsistent options are detected.
8050 Substitute the contents of spec string @var{name} at this point.
8053 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
8055 @item %x@{@var{option}@}
8056 Accumulate an option for @samp{%X}.
8059 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
8063 Output the accumulated assembler options specified by @option{-Wa}.
8066 Output the accumulated preprocessor options specified by @option{-Wp}.
8069 Process the @code{asm} spec. This is used to compute the
8070 switches to be passed to the assembler.
8073 Process the @code{asm_final} spec. This is a spec string for
8074 passing switches to an assembler post-processor, if such a program is
8078 Process the @code{link} spec. This is the spec for computing the
8079 command line passed to the linker. Typically it will make use of the
8080 @samp{%L %G %S %D and %E} sequences.
8083 Dump out a @option{-L} option for each directory that GCC believes might
8084 contain startup files. If the target supports multilibs then the
8085 current multilib directory will be prepended to each of these paths.
8088 Process the @code{lib} spec. This is a spec string for deciding which
8089 libraries should be included on the command line to the linker.
8092 Process the @code{libgcc} spec. This is a spec string for deciding
8093 which GCC support library should be included on the command line to the linker.
8096 Process the @code{startfile} spec. This is a spec for deciding which
8097 object files should be the first ones passed to the linker. Typically
8098 this might be a file named @file{crt0.o}.
8101 Process the @code{endfile} spec. This is a spec string that specifies
8102 the last object files that will be passed to the linker.
8105 Process the @code{cpp} spec. This is used to construct the arguments
8106 to be passed to the C preprocessor.
8109 Process the @code{cc1} spec. This is used to construct the options to be
8110 passed to the actual C compiler (@samp{cc1}).
8113 Process the @code{cc1plus} spec. This is used to construct the options to be
8114 passed to the actual C++ compiler (@samp{cc1plus}).
8117 Substitute the variable part of a matched option. See below.
8118 Note that each comma in the substituted string is replaced by
8122 Remove all occurrences of @code{-S} from the command line. Note---this
8123 command is position dependent. @samp{%} commands in the spec string
8124 before this one will see @code{-S}, @samp{%} commands in the spec string
8125 after this one will not.
8127 @item %:@var{function}(@var{args})
8128 Call the named function @var{function}, passing it @var{args}.
8129 @var{args} is first processed as a nested spec string, then split
8130 into an argument vector in the usual fashion. The function returns
8131 a string which is processed as if it had appeared literally as part
8132 of the current spec.
8134 The following built-in spec functions are provided:
8138 The @code{getenv} spec function takes two arguments: an environment
8139 variable name and a string. If the environment variable is not
8140 defined, a fatal error is issued. Otherwise, the return value is the
8141 value of the environment variable concatenated with the string. For
8142 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8145 %:getenv(TOPDIR /include)
8148 expands to @file{/path/to/top/include}.
8150 @item @code{if-exists}
8151 The @code{if-exists} spec function takes one argument, an absolute
8152 pathname to a file. If the file exists, @code{if-exists} returns the
8153 pathname. Here is a small example of its usage:
8157 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
8160 @item @code{if-exists-else}
8161 The @code{if-exists-else} spec function is similar to the @code{if-exists}
8162 spec function, except that it takes two arguments. The first argument is
8163 an absolute pathname to a file. If the file exists, @code{if-exists-else}
8164 returns the pathname. If it does not exist, it returns the second argument.
8165 This way, @code{if-exists-else} can be used to select one file or another,
8166 based on the existence of the first. Here is a small example of its usage:
8170 crt0%O%s %:if-exists(crti%O%s) \
8171 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8174 @item @code{replace-outfile}
8175 The @code{replace-outfile} spec function takes two arguments. It looks for the
8176 first argument in the outfiles array and replaces it with the second argument. Here
8177 is a small example of its usage:
8180 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8183 @item @code{print-asm-header}
8184 The @code{print-asm-header} function takes no arguments and simply
8185 prints a banner like:
8191 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8194 It is used to separate compiler options from assembler options
8195 in the @option{--target-help} output.
8199 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8200 If that switch was not specified, this substitutes nothing. Note that
8201 the leading dash is omitted when specifying this option, and it is
8202 automatically inserted if the substitution is performed. Thus the spec
8203 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8204 and would output the command line option @option{-foo}.
8206 @item %W@{@code{S}@}
8207 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8210 @item %@{@code{S}*@}
8211 Substitutes all the switches specified to GCC whose names start
8212 with @code{-S}, but which also take an argument. This is used for
8213 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8214 GCC considers @option{-o foo} as being
8215 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
8216 text, including the space. Thus two arguments would be generated.
8218 @item %@{@code{S}*&@code{T}*@}
8219 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8220 (the order of @code{S} and @code{T} in the spec is not significant).
8221 There can be any number of ampersand-separated variables; for each the
8222 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8224 @item %@{@code{S}:@code{X}@}
8225 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8227 @item %@{!@code{S}:@code{X}@}
8228 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8230 @item %@{@code{S}*:@code{X}@}
8231 Substitutes @code{X} if one or more switches whose names start with
8232 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8233 once, no matter how many such switches appeared. However, if @code{%*}
8234 appears somewhere in @code{X}, then @code{X} will be substituted once
8235 for each matching switch, with the @code{%*} replaced by the part of
8236 that switch that matched the @code{*}.
8238 @item %@{.@code{S}:@code{X}@}
8239 Substitutes @code{X}, if processing a file with suffix @code{S}.
8241 @item %@{!.@code{S}:@code{X}@}
8242 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8244 @item %@{,@code{S}:@code{X}@}
8245 Substitutes @code{X}, if processing a file for language @code{S}.
8247 @item %@{!,@code{S}:@code{X}@}
8248 Substitutes @code{X}, if not processing a file for language @code{S}.
8250 @item %@{@code{S}|@code{P}:@code{X}@}
8251 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8252 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8253 @code{*} sequences as well, although they have a stronger binding than
8254 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8255 alternatives must be starred, and only the first matching alternative
8258 For example, a spec string like this:
8261 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8264 will output the following command-line options from the following input
8265 command-line options:
8270 -d fred.c -foo -baz -boggle
8271 -d jim.d -bar -baz -boggle
8274 @item %@{S:X; T:Y; :D@}
8276 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8277 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8278 be as many clauses as you need. This may be combined with @code{.},
8279 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8284 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8285 construct may contain other nested @samp{%} constructs or spaces, or
8286 even newlines. They are processed as usual, as described above.
8287 Trailing white space in @code{X} is ignored. White space may also
8288 appear anywhere on the left side of the colon in these constructs,
8289 except between @code{.} or @code{*} and the corresponding word.
8291 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8292 handled specifically in these constructs. If another value of
8293 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8294 @option{-W} switch is found later in the command line, the earlier
8295 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8296 just one letter, which passes all matching options.
8298 The character @samp{|} at the beginning of the predicate text is used to
8299 indicate that a command should be piped to the following command, but
8300 only if @option{-pipe} is specified.
8302 It is built into GCC which switches take arguments and which do not.
8303 (You might think it would be useful to generalize this to allow each
8304 compiler's spec to say which switches take arguments. But this cannot
8305 be done in a consistent fashion. GCC cannot even decide which input
8306 files have been specified without knowing which switches take arguments,
8307 and it must know which input files to compile in order to tell which
8310 GCC also knows implicitly that arguments starting in @option{-l} are to be
8311 treated as compiler output files, and passed to the linker in their
8312 proper position among the other output files.
8314 @c man begin OPTIONS
8316 @node Target Options
8317 @section Specifying Target Machine and Compiler Version
8318 @cindex target options
8319 @cindex cross compiling
8320 @cindex specifying machine version
8321 @cindex specifying compiler version and target machine
8322 @cindex compiler version, specifying
8323 @cindex target machine, specifying
8325 The usual way to run GCC is to run the executable called @file{gcc}, or
8326 @file{<machine>-gcc} when cross-compiling, or
8327 @file{<machine>-gcc-<version>} to run a version other than the one that
8328 was installed last. Sometimes this is inconvenient, so GCC provides
8329 options that will switch to another cross-compiler or version.
8332 @item -b @var{machine}
8334 The argument @var{machine} specifies the target machine for compilation.
8336 The value to use for @var{machine} is the same as was specified as the
8337 machine type when configuring GCC as a cross-compiler. For
8338 example, if a cross-compiler was configured with @samp{configure
8339 arm-elf}, meaning to compile for an arm processor with elf binaries,
8340 then you would specify @option{-b arm-elf} to run that cross compiler.
8341 Because there are other options beginning with @option{-b}, the
8342 configuration must contain a hyphen.
8344 @item -V @var{version}
8346 The argument @var{version} specifies which version of GCC to run.
8347 This is useful when multiple versions are installed. For example,
8348 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8351 The @option{-V} and @option{-b} options work by running the
8352 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8353 use them if you can just run that directly.
8355 @node Submodel Options
8356 @section Hardware Models and Configurations
8357 @cindex submodel options
8358 @cindex specifying hardware config
8359 @cindex hardware models and configurations, specifying
8360 @cindex machine dependent options
8362 Earlier we discussed the standard option @option{-b} which chooses among
8363 different installed compilers for completely different target
8364 machines, such as VAX vs.@: 68000 vs.@: 80386.
8366 In addition, each of these target machine types can have its own
8367 special options, starting with @samp{-m}, to choose among various
8368 hardware models or configurations---for example, 68010 vs 68020,
8369 floating coprocessor or none. A single installed version of the
8370 compiler can compile for any model or configuration, according to the
8373 Some configurations of the compiler also support additional special
8374 options, usually for compatibility with other compilers on the same
8377 @c This list is ordered alphanumerically by subsection name.
8378 @c It should be the same order and spelling as these options are listed
8379 @c in Machine Dependent Options
8385 * Blackfin Options::
8389 * DEC Alpha Options::
8390 * DEC Alpha/VMS Options::
8392 * GNU/Linux Options::
8395 * i386 and x86-64 Options::
8407 * RS/6000 and PowerPC Options::
8408 * S/390 and zSeries Options::
8413 * System V Options::
8418 * Xstormy16 Options::
8424 @subsection ARC Options
8427 These options are defined for ARC implementations:
8432 Compile code for little endian mode. This is the default.
8436 Compile code for big endian mode.
8439 @opindex mmangle-cpu
8440 Prepend the name of the cpu to all public symbol names.
8441 In multiple-processor systems, there are many ARC variants with different
8442 instruction and register set characteristics. This flag prevents code
8443 compiled for one cpu to be linked with code compiled for another.
8444 No facility exists for handling variants that are ``almost identical''.
8445 This is an all or nothing option.
8447 @item -mcpu=@var{cpu}
8449 Compile code for ARC variant @var{cpu}.
8450 Which variants are supported depend on the configuration.
8451 All variants support @option{-mcpu=base}, this is the default.
8453 @item -mtext=@var{text-section}
8454 @itemx -mdata=@var{data-section}
8455 @itemx -mrodata=@var{readonly-data-section}
8459 Put functions, data, and readonly data in @var{text-section},
8460 @var{data-section}, and @var{readonly-data-section} respectively
8461 by default. This can be overridden with the @code{section} attribute.
8462 @xref{Variable Attributes}.
8467 @subsection ARM Options
8470 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8474 @item -mabi=@var{name}
8476 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8477 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8480 @opindex mapcs-frame
8481 Generate a stack frame that is compliant with the ARM Procedure Call
8482 Standard for all functions, even if this is not strictly necessary for
8483 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8484 with this option will cause the stack frames not to be generated for
8485 leaf functions. The default is @option{-mno-apcs-frame}.
8489 This is a synonym for @option{-mapcs-frame}.
8492 @c not currently implemented
8493 @item -mapcs-stack-check
8494 @opindex mapcs-stack-check
8495 Generate code to check the amount of stack space available upon entry to
8496 every function (that actually uses some stack space). If there is
8497 insufficient space available then either the function
8498 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8499 called, depending upon the amount of stack space required. The run time
8500 system is required to provide these functions. The default is
8501 @option{-mno-apcs-stack-check}, since this produces smaller code.
8503 @c not currently implemented
8505 @opindex mapcs-float
8506 Pass floating point arguments using the float point registers. This is
8507 one of the variants of the APCS@. This option is recommended if the
8508 target hardware has a floating point unit or if a lot of floating point
8509 arithmetic is going to be performed by the code. The default is
8510 @option{-mno-apcs-float}, since integer only code is slightly increased in
8511 size if @option{-mapcs-float} is used.
8513 @c not currently implemented
8514 @item -mapcs-reentrant
8515 @opindex mapcs-reentrant
8516 Generate reentrant, position independent code. The default is
8517 @option{-mno-apcs-reentrant}.
8520 @item -mthumb-interwork
8521 @opindex mthumb-interwork
8522 Generate code which supports calling between the ARM and Thumb
8523 instruction sets. Without this option the two instruction sets cannot
8524 be reliably used inside one program. The default is
8525 @option{-mno-thumb-interwork}, since slightly larger code is generated
8526 when @option{-mthumb-interwork} is specified.
8528 @item -mno-sched-prolog
8529 @opindex mno-sched-prolog
8530 Prevent the reordering of instructions in the function prolog, or the
8531 merging of those instruction with the instructions in the function's
8532 body. This means that all functions will start with a recognizable set
8533 of instructions (or in fact one of a choice from a small set of
8534 different function prologues), and this information can be used to
8535 locate the start if functions inside an executable piece of code. The
8536 default is @option{-msched-prolog}.
8539 @opindex mhard-float
8540 Generate output containing floating point instructions. This is the
8544 @opindex msoft-float
8545 Generate output containing library calls for floating point.
8546 @strong{Warning:} the requisite libraries are not available for all ARM
8547 targets. Normally the facilities of the machine's usual C compiler are
8548 used, but this cannot be done directly in cross-compilation. You must make
8549 your own arrangements to provide suitable library functions for
8552 @option{-msoft-float} changes the calling convention in the output file;
8553 therefore, it is only useful if you compile @emph{all} of a program with
8554 this option. In particular, you need to compile @file{libgcc.a}, the
8555 library that comes with GCC, with @option{-msoft-float} in order for
8558 @item -mfloat-abi=@var{name}
8560 Specifies which ABI to use for floating point values. Permissible values
8561 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8563 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8564 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8565 of floating point instructions, but still uses the soft-float calling
8568 @item -mlittle-endian
8569 @opindex mlittle-endian
8570 Generate code for a processor running in little-endian mode. This is
8571 the default for all standard configurations.
8574 @opindex mbig-endian
8575 Generate code for a processor running in big-endian mode; the default is
8576 to compile code for a little-endian processor.
8578 @item -mwords-little-endian
8579 @opindex mwords-little-endian
8580 This option only applies when generating code for big-endian processors.
8581 Generate code for a little-endian word order but a big-endian byte
8582 order. That is, a byte order of the form @samp{32107654}. Note: this
8583 option should only be used if you require compatibility with code for
8584 big-endian ARM processors generated by versions of the compiler prior to
8587 @item -mcpu=@var{name}
8589 This specifies the name of the target ARM processor. GCC uses this name
8590 to determine what kind of instructions it can emit when generating
8591 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8592 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8593 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8594 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8595 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8596 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8597 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8598 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8599 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8600 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8601 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8602 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8603 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8604 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8605 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
8607 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8609 @item -mtune=@var{name}
8611 This option is very similar to the @option{-mcpu=} option, except that
8612 instead of specifying the actual target processor type, and hence
8613 restricting which instructions can be used, it specifies that GCC should
8614 tune the performance of the code as if the target were of the type
8615 specified in this option, but still choosing the instructions that it
8616 will generate based on the cpu specified by a @option{-mcpu=} option.
8617 For some ARM implementations better performance can be obtained by using
8620 @item -march=@var{name}
8622 This specifies the name of the target ARM architecture. GCC uses this
8623 name to determine what kind of instructions it can emit when generating
8624 assembly code. This option can be used in conjunction with or instead
8625 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8626 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8627 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8628 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
8629 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
8630 @samp{iwmmxt}, @samp{ep9312}.
8632 @item -mfpu=@var{name}
8633 @itemx -mfpe=@var{number}
8634 @itemx -mfp=@var{number}
8638 This specifies what floating point hardware (or hardware emulation) is
8639 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8640 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-d16} and
8641 @samp{neon}. @option{-mfp} and @option{-mfpe}
8642 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8643 with older versions of GCC@.
8645 If @option{-msoft-float} is specified this specifies the format of
8646 floating point values.
8648 @item -mstructure-size-boundary=@var{n}
8649 @opindex mstructure-size-boundary
8650 The size of all structures and unions will be rounded up to a multiple
8651 of the number of bits set by this option. Permissible values are 8, 32
8652 and 64. The default value varies for different toolchains. For the COFF
8653 targeted toolchain the default value is 8. A value of 64 is only allowed
8654 if the underlying ABI supports it.
8656 Specifying the larger number can produce faster, more efficient code, but
8657 can also increase the size of the program. Different values are potentially
8658 incompatible. Code compiled with one value cannot necessarily expect to
8659 work with code or libraries compiled with another value, if they exchange
8660 information using structures or unions.
8662 @item -mabort-on-noreturn
8663 @opindex mabort-on-noreturn
8664 Generate a call to the function @code{abort} at the end of a
8665 @code{noreturn} function. It will be executed if the function tries to
8669 @itemx -mno-long-calls
8670 @opindex mlong-calls
8671 @opindex mno-long-calls
8672 Tells the compiler to perform function calls by first loading the
8673 address of the function into a register and then performing a subroutine
8674 call on this register. This switch is needed if the target function
8675 will lie outside of the 64 megabyte addressing range of the offset based
8676 version of subroutine call instruction.
8678 Even if this switch is enabled, not all function calls will be turned
8679 into long calls. The heuristic is that static functions, functions
8680 which have the @samp{short-call} attribute, functions that are inside
8681 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8682 definitions have already been compiled within the current compilation
8683 unit, will not be turned into long calls. The exception to this rule is
8684 that weak function definitions, functions with the @samp{long-call}
8685 attribute or the @samp{section} attribute, and functions that are within
8686 the scope of a @samp{#pragma long_calls} directive, will always be
8687 turned into long calls.
8689 This feature is not enabled by default. Specifying
8690 @option{-mno-long-calls} will restore the default behavior, as will
8691 placing the function calls within the scope of a @samp{#pragma
8692 long_calls_off} directive. Note these switches have no effect on how
8693 the compiler generates code to handle function calls via function
8696 @item -mnop-fun-dllimport
8697 @opindex mnop-fun-dllimport
8698 Disable support for the @code{dllimport} attribute.
8700 @item -msingle-pic-base
8701 @opindex msingle-pic-base
8702 Treat the register used for PIC addressing as read-only, rather than
8703 loading it in the prologue for each function. The run-time system is
8704 responsible for initializing this register with an appropriate value
8705 before execution begins.
8707 @item -mpic-register=@var{reg}
8708 @opindex mpic-register
8709 Specify the register to be used for PIC addressing. The default is R10
8710 unless stack-checking is enabled, when R9 is used.
8712 @item -mcirrus-fix-invalid-insns
8713 @opindex mcirrus-fix-invalid-insns
8714 @opindex mno-cirrus-fix-invalid-insns
8715 Insert NOPs into the instruction stream to in order to work around
8716 problems with invalid Maverick instruction combinations. This option
8717 is only valid if the @option{-mcpu=ep9312} option has been used to
8718 enable generation of instructions for the Cirrus Maverick floating
8719 point co-processor. This option is not enabled by default, since the
8720 problem is only present in older Maverick implementations. The default
8721 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8724 @item -mpoke-function-name
8725 @opindex mpoke-function-name
8726 Write the name of each function into the text section, directly
8727 preceding the function prologue. The generated code is similar to this:
8731 .ascii "arm_poke_function_name", 0
8734 .word 0xff000000 + (t1 - t0)
8735 arm_poke_function_name
8737 stmfd sp!, @{fp, ip, lr, pc@}
8741 When performing a stack backtrace, code can inspect the value of
8742 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8743 location @code{pc - 12} and the top 8 bits are set, then we know that
8744 there is a function name embedded immediately preceding this location
8745 and has length @code{((pc[-3]) & 0xff000000)}.
8749 Generate code for the Thumb instruction set. The default is to
8750 use the 32-bit ARM instruction set.
8751 This option automatically enables either 16-bit Thumb-1 or
8752 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8753 and @option{-march=@var{name}} options.
8756 @opindex mtpcs-frame
8757 Generate a stack frame that is compliant with the Thumb Procedure Call
8758 Standard for all non-leaf functions. (A leaf function is one that does
8759 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8761 @item -mtpcs-leaf-frame
8762 @opindex mtpcs-leaf-frame
8763 Generate a stack frame that is compliant with the Thumb Procedure Call
8764 Standard for all leaf functions. (A leaf function is one that does
8765 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8767 @item -mcallee-super-interworking
8768 @opindex mcallee-super-interworking
8769 Gives all externally visible functions in the file being compiled an ARM
8770 instruction set header which switches to Thumb mode before executing the
8771 rest of the function. This allows these functions to be called from
8772 non-interworking code.
8774 @item -mcaller-super-interworking
8775 @opindex mcaller-super-interworking
8776 Allows calls via function pointers (including virtual functions) to
8777 execute correctly regardless of whether the target code has been
8778 compiled for interworking or not. There is a small overhead in the cost
8779 of executing a function pointer if this option is enabled.
8781 @item -mtp=@var{name}
8783 Specify the access model for the thread local storage pointer. The valid
8784 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8785 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8786 (supported in the arm6k architecture), and @option{auto}, which uses the
8787 best available method for the selected processor. The default setting is
8793 @subsection AVR Options
8796 These options are defined for AVR implementations:
8799 @item -mmcu=@var{mcu}
8801 Specify ATMEL AVR instruction set or MCU type.
8803 Instruction set avr1 is for the minimal AVR core, not supported by the C
8804 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8805 attiny11, attiny12, attiny15, attiny28).
8807 Instruction set avr2 (default) is for the classic AVR core with up to
8808 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8809 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8810 at90c8534, at90s8535).
8812 Instruction set avr3 is for the classic AVR core with up to 128K program
8813 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8815 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8816 memory space (MCU types: atmega8, atmega83, atmega85).
8818 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8819 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8820 atmega64, atmega128, at43usb355, at94k).
8824 Output instruction sizes to the asm file.
8826 @item -minit-stack=@var{N}
8827 @opindex minit-stack
8828 Specify the initial stack address, which may be a symbol or numeric value,
8829 @samp{__stack} is the default.
8831 @item -mno-interrupts
8832 @opindex mno-interrupts
8833 Generated code is not compatible with hardware interrupts.
8834 Code size will be smaller.
8836 @item -mcall-prologues
8837 @opindex mcall-prologues
8838 Functions prologues/epilogues expanded as call to appropriate
8839 subroutines. Code size will be smaller.
8841 @item -mno-tablejump
8842 @opindex mno-tablejump
8843 Do not generate tablejump insns which sometimes increase code size.
8846 @opindex mtiny-stack
8847 Change only the low 8 bits of the stack pointer.
8851 Assume int to be 8 bit integer. This affects the sizes of all types: A
8852 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8853 and long long will be 4 bytes. Please note that this option does not
8854 comply to the C standards, but it will provide you with smaller code
8858 @node Blackfin Options
8859 @subsection Blackfin Options
8860 @cindex Blackfin Options
8863 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
8865 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8866 can be one of @samp{bf522}, @samp{bf523}, @samp{bf524},
8867 @samp{bf525}, @samp{bf526}, @samp{bf527},
8868 @samp{bf531}, @samp{bf532}, @samp{bf533}, @samp{bf534},
8869 @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
8870 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
8872 The optional @var{sirevision} specifies the silicon revision of the target
8873 Blackfin processor. Any workarounds available for the targeted silicon revision
8874 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
8875 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
8876 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
8877 hexadecimal digits representing the major and minor numbers in the silicon
8878 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
8879 is not defined. If @var{sirevision} is @samp{any}, the
8880 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
8881 If this optional @var{sirevision} is not used, GCC assumes the latest known
8882 silicon revision of the targeted Blackfin processor.
8884 Support for @samp{bf561} is incomplete. For @samp{bf561},
8885 Only the processor macro is defined.
8886 Without this option, @samp{bf532} is used as the processor by default.
8887 The corresponding predefined processor macros for @var{cpu} is to
8888 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
8889 provided by libgloss to be linked in if @option{-msim} is not given.
8893 Specifies that the program will be run on the simulator. This causes
8894 the simulator BSP provided by libgloss to be linked in. This option
8895 has effect only for @samp{bfin-elf} toolchain.
8896 Certain other options, such as @option{-mid-shared-library} and
8897 @option{-mfdpic}, imply @option{-msim}.
8899 @item -momit-leaf-frame-pointer
8900 @opindex momit-leaf-frame-pointer
8901 Don't keep the frame pointer in a register for leaf functions. This
8902 avoids the instructions to save, set up and restore frame pointers and
8903 makes an extra register available in leaf functions. The option
8904 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8905 which might make debugging harder.
8907 @item -mspecld-anomaly
8908 @opindex mspecld-anomaly
8909 When enabled, the compiler will ensure that the generated code does not
8910 contain speculative loads after jump instructions. If this option is used,
8911 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
8913 @item -mno-specld-anomaly
8914 @opindex mno-specld-anomaly
8915 Don't generate extra code to prevent speculative loads from occurring.
8917 @item -mcsync-anomaly
8918 @opindex mcsync-anomaly
8919 When enabled, the compiler will ensure that the generated code does not
8920 contain CSYNC or SSYNC instructions too soon after conditional branches.
8921 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
8923 @item -mno-csync-anomaly
8924 @opindex mno-csync-anomaly
8925 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8926 occurring too soon after a conditional branch.
8930 When enabled, the compiler is free to take advantage of the knowledge that
8931 the entire program fits into the low 64k of memory.
8934 @opindex mno-low-64k
8935 Assume that the program is arbitrarily large. This is the default.
8937 @item -mstack-check-l1
8938 @opindex mstack-check-l1
8939 Do stack checking using information placed into L1 scratchpad memory by the
8942 @item -mid-shared-library
8943 @opindex mid-shared-library
8944 Generate code that supports shared libraries via the library ID method.
8945 This allows for execute in place and shared libraries in an environment
8946 without virtual memory management. This option implies @option{-fPIC}.
8947 With a @samp{bfin-elf} target, this option implies @option{-msim}.
8949 @item -mno-id-shared-library
8950 @opindex mno-id-shared-library
8951 Generate code that doesn't assume ID based shared libraries are being used.
8952 This is the default.
8954 @item -mleaf-id-shared-library
8955 @opindex mleaf-id-shared-library
8956 Generate code that supports shared libraries via the library ID method,
8957 but assumes that this library or executable won't link against any other
8958 ID shared libraries. That allows the compiler to use faster code for jumps
8961 @item -mno-leaf-id-shared-library
8962 @opindex mno-leaf-id-shared-library
8963 Do not assume that the code being compiled won't link against any ID shared
8964 libraries. Slower code will be generated for jump and call insns.
8966 @item -mshared-library-id=n
8967 @opindex mshared-library-id
8968 Specified the identification number of the ID based shared library being
8969 compiled. Specifying a value of 0 will generate more compact code, specifying
8970 other values will force the allocation of that number to the current
8971 library but is no more space or time efficient than omitting this option.
8975 Generate code that allows the data segment to be located in a different
8976 area of memory from the text segment. This allows for execute in place in
8977 an environment without virtual memory management by eliminating relocations
8978 against the text section.
8981 @opindex mno-sep-data
8982 Generate code that assumes that the data segment follows the text segment.
8983 This is the default.
8986 @itemx -mno-long-calls
8987 @opindex mlong-calls
8988 @opindex mno-long-calls
8989 Tells the compiler to perform function calls by first loading the
8990 address of the function into a register and then performing a subroutine
8991 call on this register. This switch is needed if the target function
8992 will lie outside of the 24 bit addressing range of the offset based
8993 version of subroutine call instruction.
8995 This feature is not enabled by default. Specifying
8996 @option{-mno-long-calls} will restore the default behavior. Note these
8997 switches have no effect on how the compiler generates code to handle
8998 function calls via function pointers.
9002 Link with the fast floating-point library. This library relaxes some of
9003 the IEEE floating-point standard's rules for checking inputs against
9004 Not-a-Number (NAN), in the interest of performance.
9007 @opindex minline-plt
9008 Enable inlining of PLT entries in function calls to functions that are
9009 not known to bind locally. It has no effect without @option{-mfdpic}.
9013 Build standalone application for multicore Blackfin processor. Proper
9014 start files and link scripts will be used to support multicore.
9015 This option defines @code{__BFIN_MULTICORE}. It can only be used with
9016 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
9017 @option{-mcorea} or @option{-mcoreb}. If it's used without
9018 @option{-mcorea} or @option{-mcoreb}, single application/dual core
9019 programming model is used. In this model, the main function of Core B
9020 should be named as coreb_main. If it's used with @option{-mcorea} or
9021 @option{-mcoreb}, one application per core programming model is used.
9022 If this option is not used, single core application programming
9027 Build standalone application for Core A of BF561 when using
9028 one application per core programming model. Proper start files
9029 and link scripts will be used to support Core A. This option
9030 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
9034 Build standalone application for Core B of BF561 when using
9035 one application per core programming model. Proper start files
9036 and link scripts will be used to support Core B. This option
9037 defines @code{__BFIN_COREB}. When this option is used, coreb_main
9038 should be used instead of main. It must be used with
9039 @option{-mmulticore}.
9043 Build standalone application for SDRAM. Proper start files and
9044 link scripts will be used to put the application into SDRAM.
9045 Loader should initialize SDRAM before loading the application
9046 into SDRAM. This option defines @code{__BFIN_SDRAM}.
9050 @subsection CRIS Options
9051 @cindex CRIS Options
9053 These options are defined specifically for the CRIS ports.
9056 @item -march=@var{architecture-type}
9057 @itemx -mcpu=@var{architecture-type}
9060 Generate code for the specified architecture. The choices for
9061 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9062 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
9063 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9066 @item -mtune=@var{architecture-type}
9068 Tune to @var{architecture-type} everything applicable about the generated
9069 code, except for the ABI and the set of available instructions. The
9070 choices for @var{architecture-type} are the same as for
9071 @option{-march=@var{architecture-type}}.
9073 @item -mmax-stack-frame=@var{n}
9074 @opindex mmax-stack-frame
9075 Warn when the stack frame of a function exceeds @var{n} bytes.
9081 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9082 @option{-march=v3} and @option{-march=v8} respectively.
9084 @item -mmul-bug-workaround
9085 @itemx -mno-mul-bug-workaround
9086 @opindex mmul-bug-workaround
9087 @opindex mno-mul-bug-workaround
9088 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
9089 models where it applies. This option is active by default.
9093 Enable CRIS-specific verbose debug-related information in the assembly
9094 code. This option also has the effect to turn off the @samp{#NO_APP}
9095 formatted-code indicator to the assembler at the beginning of the
9100 Do not use condition-code results from previous instruction; always emit
9101 compare and test instructions before use of condition codes.
9103 @item -mno-side-effects
9104 @opindex mno-side-effects
9105 Do not emit instructions with side-effects in addressing modes other than
9109 @itemx -mno-stack-align
9111 @itemx -mno-data-align
9112 @itemx -mconst-align
9113 @itemx -mno-const-align
9114 @opindex mstack-align
9115 @opindex mno-stack-align
9116 @opindex mdata-align
9117 @opindex mno-data-align
9118 @opindex mconst-align
9119 @opindex mno-const-align
9120 These options (no-options) arranges (eliminate arrangements) for the
9121 stack-frame, individual data and constants to be aligned for the maximum
9122 single data access size for the chosen CPU model. The default is to
9123 arrange for 32-bit alignment. ABI details such as structure layout are
9124 not affected by these options.
9132 Similar to the stack- data- and const-align options above, these options
9133 arrange for stack-frame, writable data and constants to all be 32-bit,
9134 16-bit or 8-bit aligned. The default is 32-bit alignment.
9136 @item -mno-prologue-epilogue
9137 @itemx -mprologue-epilogue
9138 @opindex mno-prologue-epilogue
9139 @opindex mprologue-epilogue
9140 With @option{-mno-prologue-epilogue}, the normal function prologue and
9141 epilogue that sets up the stack-frame are omitted and no return
9142 instructions or return sequences are generated in the code. Use this
9143 option only together with visual inspection of the compiled code: no
9144 warnings or errors are generated when call-saved registers must be saved,
9145 or storage for local variable needs to be allocated.
9151 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9152 instruction sequences that load addresses for functions from the PLT part
9153 of the GOT rather than (traditional on other architectures) calls to the
9154 PLT@. The default is @option{-mgotplt}.
9158 Legacy no-op option only recognized with the cris-axis-elf and
9159 cris-axis-linux-gnu targets.
9163 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9167 This option, recognized for the cris-axis-elf arranges
9168 to link with input-output functions from a simulator library. Code,
9169 initialized data and zero-initialized data are allocated consecutively.
9173 Like @option{-sim}, but pass linker options to locate initialized data at
9174 0x40000000 and zero-initialized data at 0x80000000.
9178 @subsection CRX Options
9181 These options are defined specifically for the CRX ports.
9187 Enable the use of multiply-accumulate instructions. Disabled by default.
9191 Push instructions will be used to pass outgoing arguments when functions
9192 are called. Enabled by default.
9195 @node Darwin Options
9196 @subsection Darwin Options
9197 @cindex Darwin options
9199 These options are defined for all architectures running the Darwin operating
9202 FSF GCC on Darwin does not create ``fat'' object files; it will create
9203 an object file for the single architecture that it was built to
9204 target. Apple's GCC on Darwin does create ``fat'' files if multiple
9205 @option{-arch} options are used; it does so by running the compiler or
9206 linker multiple times and joining the results together with
9209 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9210 @samp{i686}) is determined by the flags that specify the ISA
9211 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
9212 @option{-force_cpusubtype_ALL} option can be used to override this.
9214 The Darwin tools vary in their behavior when presented with an ISA
9215 mismatch. The assembler, @file{as}, will only permit instructions to
9216 be used that are valid for the subtype of the file it is generating,
9217 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9218 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9219 and print an error if asked to create a shared library with a less
9220 restrictive subtype than its input files (for instance, trying to put
9221 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
9222 for executables, @file{ld}, will quietly give the executable the most
9223 restrictive subtype of any of its input files.
9228 Add the framework directory @var{dir} to the head of the list of
9229 directories to be searched for header files. These directories are
9230 interleaved with those specified by @option{-I} options and are
9231 scanned in a left-to-right order.
9233 A framework directory is a directory with frameworks in it. A
9234 framework is a directory with a @samp{"Headers"} and/or
9235 @samp{"PrivateHeaders"} directory contained directly in it that ends
9236 in @samp{".framework"}. The name of a framework is the name of this
9237 directory excluding the @samp{".framework"}. Headers associated with
9238 the framework are found in one of those two directories, with
9239 @samp{"Headers"} being searched first. A subframework is a framework
9240 directory that is in a framework's @samp{"Frameworks"} directory.
9241 Includes of subframework headers can only appear in a header of a
9242 framework that contains the subframework, or in a sibling subframework
9243 header. Two subframeworks are siblings if they occur in the same
9244 framework. A subframework should not have the same name as a
9245 framework, a warning will be issued if this is violated. Currently a
9246 subframework cannot have subframeworks, in the future, the mechanism
9247 may be extended to support this. The standard frameworks can be found
9248 in @samp{"/System/Library/Frameworks"} and
9249 @samp{"/Library/Frameworks"}. An example include looks like
9250 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9251 the name of the framework and header.h is found in the
9252 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9254 @item -iframework@var{dir}
9256 Like @option{-F} except the directory is a treated as a system
9257 directory. The main difference between this @option{-iframework} and
9258 @option{-F} is that with @option{-iframework} the compiler does not
9259 warn about constructs contained within header files found via
9260 @var{dir}. This option is valid only for the C family of languages.
9264 Emit debugging information for symbols that are used. For STABS
9265 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9266 This is by default ON@.
9270 Emit debugging information for all symbols and types.
9272 @item -mmacosx-version-min=@var{version}
9273 The earliest version of MacOS X that this executable will run on
9274 is @var{version}. Typical values of @var{version} include @code{10.1},
9275 @code{10.2}, and @code{10.3.9}.
9277 If the compiler was built to use the system's headers by default,
9278 then the default for this option is the system version on which the
9279 compiler is running, otherwise the default is to make choices which
9280 are compatible with as many systems and code bases as possible.
9284 Enable kernel development mode. The @option{-mkernel} option sets
9285 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9286 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9287 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9288 applicable. This mode also sets @option{-mno-altivec},
9289 @option{-msoft-float}, @option{-fno-builtin} and
9290 @option{-mlong-branch} for PowerPC targets.
9292 @item -mone-byte-bool
9293 @opindex mone-byte-bool
9294 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9295 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9296 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9297 option has no effect on x86.
9299 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9300 to generate code that is not binary compatible with code generated
9301 without that switch. Using this switch may require recompiling all
9302 other modules in a program, including system libraries. Use this
9303 switch to conform to a non-default data model.
9305 @item -mfix-and-continue
9306 @itemx -ffix-and-continue
9307 @itemx -findirect-data
9308 @opindex mfix-and-continue
9309 @opindex ffix-and-continue
9310 @opindex findirect-data
9311 Generate code suitable for fast turn around development. Needed to
9312 enable gdb to dynamically load @code{.o} files into already running
9313 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9314 are provided for backwards compatibility.
9318 Loads all members of static archive libraries.
9319 See man ld(1) for more information.
9321 @item -arch_errors_fatal
9322 @opindex arch_errors_fatal
9323 Cause the errors having to do with files that have the wrong architecture
9327 @opindex bind_at_load
9328 Causes the output file to be marked such that the dynamic linker will
9329 bind all undefined references when the file is loaded or launched.
9333 Produce a Mach-o bundle format file.
9334 See man ld(1) for more information.
9336 @item -bundle_loader @var{executable}
9337 @opindex bundle_loader
9338 This option specifies the @var{executable} that will be loading the build
9339 output file being linked. See man ld(1) for more information.
9343 When passed this option, GCC will produce a dynamic library instead of
9344 an executable when linking, using the Darwin @file{libtool} command.
9346 @item -force_cpusubtype_ALL
9347 @opindex force_cpusubtype_ALL
9348 This causes GCC's output file to have the @var{ALL} subtype, instead of
9349 one controlled by the @option{-mcpu} or @option{-march} option.
9351 @item -allowable_client @var{client_name}
9353 @itemx -compatibility_version
9354 @itemx -current_version
9356 @itemx -dependency-file
9358 @itemx -dylinker_install_name
9360 @itemx -exported_symbols_list
9362 @itemx -flat_namespace
9363 @itemx -force_flat_namespace
9364 @itemx -headerpad_max_install_names
9367 @itemx -install_name
9368 @itemx -keep_private_externs
9369 @itemx -multi_module
9370 @itemx -multiply_defined
9371 @itemx -multiply_defined_unused
9373 @itemx -no_dead_strip_inits_and_terms
9374 @itemx -nofixprebinding
9377 @itemx -noseglinkedit
9378 @itemx -pagezero_size
9380 @itemx -prebind_all_twolevel_modules
9381 @itemx -private_bundle
9382 @itemx -read_only_relocs
9384 @itemx -sectobjectsymbols
9388 @itemx -sectobjectsymbols
9391 @itemx -segs_read_only_addr
9392 @itemx -segs_read_write_addr
9393 @itemx -seg_addr_table
9394 @itemx -seg_addr_table_filename
9397 @itemx -segs_read_only_addr
9398 @itemx -segs_read_write_addr
9399 @itemx -single_module
9402 @itemx -sub_umbrella
9403 @itemx -twolevel_namespace
9406 @itemx -unexported_symbols_list
9407 @itemx -weak_reference_mismatches
9409 @opindex allowable_client
9410 @opindex client_name
9411 @opindex compatibility_version
9412 @opindex current_version
9414 @opindex dependency-file
9416 @opindex dylinker_install_name
9418 @opindex exported_symbols_list
9420 @opindex flat_namespace
9421 @opindex force_flat_namespace
9422 @opindex headerpad_max_install_names
9425 @opindex install_name
9426 @opindex keep_private_externs
9427 @opindex multi_module
9428 @opindex multiply_defined
9429 @opindex multiply_defined_unused
9431 @opindex no_dead_strip_inits_and_terms
9432 @opindex nofixprebinding
9433 @opindex nomultidefs
9435 @opindex noseglinkedit
9436 @opindex pagezero_size
9438 @opindex prebind_all_twolevel_modules
9439 @opindex private_bundle
9440 @opindex read_only_relocs
9442 @opindex sectobjectsymbols
9446 @opindex sectobjectsymbols
9449 @opindex segs_read_only_addr
9450 @opindex segs_read_write_addr
9451 @opindex seg_addr_table
9452 @opindex seg_addr_table_filename
9453 @opindex seglinkedit
9455 @opindex segs_read_only_addr
9456 @opindex segs_read_write_addr
9457 @opindex single_module
9459 @opindex sub_library
9460 @opindex sub_umbrella
9461 @opindex twolevel_namespace
9464 @opindex unexported_symbols_list
9465 @opindex weak_reference_mismatches
9466 @opindex whatsloaded
9467 These options are passed to the Darwin linker. The Darwin linker man page
9468 describes them in detail.
9471 @node DEC Alpha Options
9472 @subsection DEC Alpha Options
9474 These @samp{-m} options are defined for the DEC Alpha implementations:
9477 @item -mno-soft-float
9479 @opindex mno-soft-float
9480 @opindex msoft-float
9481 Use (do not use) the hardware floating-point instructions for
9482 floating-point operations. When @option{-msoft-float} is specified,
9483 functions in @file{libgcc.a} will be used to perform floating-point
9484 operations. Unless they are replaced by routines that emulate the
9485 floating-point operations, or compiled in such a way as to call such
9486 emulations routines, these routines will issue floating-point
9487 operations. If you are compiling for an Alpha without floating-point
9488 operations, you must ensure that the library is built so as not to call
9491 Note that Alpha implementations without floating-point operations are
9492 required to have floating-point registers.
9497 @opindex mno-fp-regs
9498 Generate code that uses (does not use) the floating-point register set.
9499 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9500 register set is not used, floating point operands are passed in integer
9501 registers as if they were integers and floating-point results are passed
9502 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9503 so any function with a floating-point argument or return value called by code
9504 compiled with @option{-mno-fp-regs} must also be compiled with that
9507 A typical use of this option is building a kernel that does not use,
9508 and hence need not save and restore, any floating-point registers.
9512 The Alpha architecture implements floating-point hardware optimized for
9513 maximum performance. It is mostly compliant with the IEEE floating
9514 point standard. However, for full compliance, software assistance is
9515 required. This option generates code fully IEEE compliant code
9516 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9517 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9518 defined during compilation. The resulting code is less efficient but is
9519 able to correctly support denormalized numbers and exceptional IEEE
9520 values such as not-a-number and plus/minus infinity. Other Alpha
9521 compilers call this option @option{-ieee_with_no_inexact}.
9523 @item -mieee-with-inexact
9524 @opindex mieee-with-inexact
9525 This is like @option{-mieee} except the generated code also maintains
9526 the IEEE @var{inexact-flag}. Turning on this option causes the
9527 generated code to implement fully-compliant IEEE math. In addition to
9528 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9529 macro. On some Alpha implementations the resulting code may execute
9530 significantly slower than the code generated by default. Since there is
9531 very little code that depends on the @var{inexact-flag}, you should
9532 normally not specify this option. Other Alpha compilers call this
9533 option @option{-ieee_with_inexact}.
9535 @item -mfp-trap-mode=@var{trap-mode}
9536 @opindex mfp-trap-mode
9537 This option controls what floating-point related traps are enabled.
9538 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9539 The trap mode can be set to one of four values:
9543 This is the default (normal) setting. The only traps that are enabled
9544 are the ones that cannot be disabled in software (e.g., division by zero
9548 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9552 Like @samp{u}, but the instructions are marked to be safe for software
9553 completion (see Alpha architecture manual for details).
9556 Like @samp{su}, but inexact traps are enabled as well.
9559 @item -mfp-rounding-mode=@var{rounding-mode}
9560 @opindex mfp-rounding-mode
9561 Selects the IEEE rounding mode. Other Alpha compilers call this option
9562 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9567 Normal IEEE rounding mode. Floating point numbers are rounded towards
9568 the nearest machine number or towards the even machine number in case
9572 Round towards minus infinity.
9575 Chopped rounding mode. Floating point numbers are rounded towards zero.
9578 Dynamic rounding mode. A field in the floating point control register
9579 (@var{fpcr}, see Alpha architecture reference manual) controls the
9580 rounding mode in effect. The C library initializes this register for
9581 rounding towards plus infinity. Thus, unless your program modifies the
9582 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9585 @item -mtrap-precision=@var{trap-precision}
9586 @opindex mtrap-precision
9587 In the Alpha architecture, floating point traps are imprecise. This
9588 means without software assistance it is impossible to recover from a
9589 floating trap and program execution normally needs to be terminated.
9590 GCC can generate code that can assist operating system trap handlers
9591 in determining the exact location that caused a floating point trap.
9592 Depending on the requirements of an application, different levels of
9593 precisions can be selected:
9597 Program precision. This option is the default and means a trap handler
9598 can only identify which program caused a floating point exception.
9601 Function precision. The trap handler can determine the function that
9602 caused a floating point exception.
9605 Instruction precision. The trap handler can determine the exact
9606 instruction that caused a floating point exception.
9609 Other Alpha compilers provide the equivalent options called
9610 @option{-scope_safe} and @option{-resumption_safe}.
9612 @item -mieee-conformant
9613 @opindex mieee-conformant
9614 This option marks the generated code as IEEE conformant. You must not
9615 use this option unless you also specify @option{-mtrap-precision=i} and either
9616 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9617 is to emit the line @samp{.eflag 48} in the function prologue of the
9618 generated assembly file. Under DEC Unix, this has the effect that
9619 IEEE-conformant math library routines will be linked in.
9621 @item -mbuild-constants
9622 @opindex mbuild-constants
9623 Normally GCC examines a 32- or 64-bit integer constant to
9624 see if it can construct it from smaller constants in two or three
9625 instructions. If it cannot, it will output the constant as a literal and
9626 generate code to load it from the data segment at runtime.
9628 Use this option to require GCC to construct @emph{all} integer constants
9629 using code, even if it takes more instructions (the maximum is six).
9631 You would typically use this option to build a shared library dynamic
9632 loader. Itself a shared library, it must relocate itself in memory
9633 before it can find the variables and constants in its own data segment.
9639 Select whether to generate code to be assembled by the vendor-supplied
9640 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9658 Indicate whether GCC should generate code to use the optional BWX,
9659 CIX, FIX and MAX instruction sets. The default is to use the instruction
9660 sets supported by the CPU type specified via @option{-mcpu=} option or that
9661 of the CPU on which GCC was built if none was specified.
9666 @opindex mfloat-ieee
9667 Generate code that uses (does not use) VAX F and G floating point
9668 arithmetic instead of IEEE single and double precision.
9670 @item -mexplicit-relocs
9671 @itemx -mno-explicit-relocs
9672 @opindex mexplicit-relocs
9673 @opindex mno-explicit-relocs
9674 Older Alpha assemblers provided no way to generate symbol relocations
9675 except via assembler macros. Use of these macros does not allow
9676 optimal instruction scheduling. GNU binutils as of version 2.12
9677 supports a new syntax that allows the compiler to explicitly mark
9678 which relocations should apply to which instructions. This option
9679 is mostly useful for debugging, as GCC detects the capabilities of
9680 the assembler when it is built and sets the default accordingly.
9684 @opindex msmall-data
9685 @opindex mlarge-data
9686 When @option{-mexplicit-relocs} is in effect, static data is
9687 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9688 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9689 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9690 16-bit relocations off of the @code{$gp} register. This limits the
9691 size of the small data area to 64KB, but allows the variables to be
9692 directly accessed via a single instruction.
9694 The default is @option{-mlarge-data}. With this option the data area
9695 is limited to just below 2GB@. Programs that require more than 2GB of
9696 data must use @code{malloc} or @code{mmap} to allocate the data in the
9697 heap instead of in the program's data segment.
9699 When generating code for shared libraries, @option{-fpic} implies
9700 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9704 @opindex msmall-text
9705 @opindex mlarge-text
9706 When @option{-msmall-text} is used, the compiler assumes that the
9707 code of the entire program (or shared library) fits in 4MB, and is
9708 thus reachable with a branch instruction. When @option{-msmall-data}
9709 is used, the compiler can assume that all local symbols share the
9710 same @code{$gp} value, and thus reduce the number of instructions
9711 required for a function call from 4 to 1.
9713 The default is @option{-mlarge-text}.
9715 @item -mcpu=@var{cpu_type}
9717 Set the instruction set and instruction scheduling parameters for
9718 machine type @var{cpu_type}. You can specify either the @samp{EV}
9719 style name or the corresponding chip number. GCC supports scheduling
9720 parameters for the EV4, EV5 and EV6 family of processors and will
9721 choose the default values for the instruction set from the processor
9722 you specify. If you do not specify a processor type, GCC will default
9723 to the processor on which the compiler was built.
9725 Supported values for @var{cpu_type} are
9731 Schedules as an EV4 and has no instruction set extensions.
9735 Schedules as an EV5 and has no instruction set extensions.
9739 Schedules as an EV5 and supports the BWX extension.
9744 Schedules as an EV5 and supports the BWX and MAX extensions.
9748 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9752 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9755 @item -mtune=@var{cpu_type}
9757 Set only the instruction scheduling parameters for machine type
9758 @var{cpu_type}. The instruction set is not changed.
9760 @item -mmemory-latency=@var{time}
9761 @opindex mmemory-latency
9762 Sets the latency the scheduler should assume for typical memory
9763 references as seen by the application. This number is highly
9764 dependent on the memory access patterns used by the application
9765 and the size of the external cache on the machine.
9767 Valid options for @var{time} are
9771 A decimal number representing clock cycles.
9777 The compiler contains estimates of the number of clock cycles for
9778 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9779 (also called Dcache, Scache, and Bcache), as well as to main memory.
9780 Note that L3 is only valid for EV5.
9785 @node DEC Alpha/VMS Options
9786 @subsection DEC Alpha/VMS Options
9788 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9791 @item -mvms-return-codes
9792 @opindex mvms-return-codes
9793 Return VMS condition codes from main. The default is to return POSIX
9794 style condition (e.g.@: error) codes.
9798 @subsection FRV Options
9805 Only use the first 32 general purpose registers.
9810 Use all 64 general purpose registers.
9815 Use only the first 32 floating point registers.
9820 Use all 64 floating point registers
9823 @opindex mhard-float
9825 Use hardware instructions for floating point operations.
9828 @opindex msoft-float
9830 Use library routines for floating point operations.
9835 Dynamically allocate condition code registers.
9840 Do not try to dynamically allocate condition code registers, only
9841 use @code{icc0} and @code{fcc0}.
9846 Change ABI to use double word insns.
9851 Do not use double word instructions.
9856 Use floating point double instructions.
9861 Do not use floating point double instructions.
9866 Use media instructions.
9871 Do not use media instructions.
9876 Use multiply and add/subtract instructions.
9881 Do not use multiply and add/subtract instructions.
9886 Select the FDPIC ABI, that uses function descriptors to represent
9887 pointers to functions. Without any PIC/PIE-related options, it
9888 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9889 assumes GOT entries and small data are within a 12-bit range from the
9890 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9891 are computed with 32 bits.
9892 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9895 @opindex minline-plt
9897 Enable inlining of PLT entries in function calls to functions that are
9898 not known to bind locally. It has no effect without @option{-mfdpic}.
9899 It's enabled by default if optimizing for speed and compiling for
9900 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9901 optimization option such as @option{-O3} or above is present in the
9907 Assume a large TLS segment when generating thread-local code.
9912 Do not assume a large TLS segment when generating thread-local code.
9917 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9918 that is known to be in read-only sections. It's enabled by default,
9919 except for @option{-fpic} or @option{-fpie}: even though it may help
9920 make the global offset table smaller, it trades 1 instruction for 4.
9921 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9922 one of which may be shared by multiple symbols, and it avoids the need
9923 for a GOT entry for the referenced symbol, so it's more likely to be a
9924 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9926 @item -multilib-library-pic
9927 @opindex multilib-library-pic
9929 Link with the (library, not FD) pic libraries. It's implied by
9930 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9931 @option{-fpic} without @option{-mfdpic}. You should never have to use
9937 Follow the EABI requirement of always creating a frame pointer whenever
9938 a stack frame is allocated. This option is enabled by default and can
9939 be disabled with @option{-mno-linked-fp}.
9942 @opindex mlong-calls
9944 Use indirect addressing to call functions outside the current
9945 compilation unit. This allows the functions to be placed anywhere
9946 within the 32-bit address space.
9948 @item -malign-labels
9949 @opindex malign-labels
9951 Try to align labels to an 8-byte boundary by inserting nops into the
9952 previous packet. This option only has an effect when VLIW packing
9953 is enabled. It doesn't create new packets; it merely adds nops to
9957 @opindex mlibrary-pic
9959 Generate position-independent EABI code.
9964 Use only the first four media accumulator registers.
9969 Use all eight media accumulator registers.
9974 Pack VLIW instructions.
9979 Do not pack VLIW instructions.
9984 Do not mark ABI switches in e_flags.
9989 Enable the use of conditional-move instructions (default).
9991 This switch is mainly for debugging the compiler and will likely be removed
9992 in a future version.
9994 @item -mno-cond-move
9995 @opindex mno-cond-move
9997 Disable the use of conditional-move instructions.
9999 This switch is mainly for debugging the compiler and will likely be removed
10000 in a future version.
10005 Enable the use of conditional set instructions (default).
10007 This switch is mainly for debugging the compiler and will likely be removed
10008 in a future version.
10013 Disable the use of conditional set instructions.
10015 This switch is mainly for debugging the compiler and will likely be removed
10016 in a future version.
10019 @opindex mcond-exec
10021 Enable the use of conditional execution (default).
10023 This switch is mainly for debugging the compiler and will likely be removed
10024 in a future version.
10026 @item -mno-cond-exec
10027 @opindex mno-cond-exec
10029 Disable the use of conditional execution.
10031 This switch is mainly for debugging the compiler and will likely be removed
10032 in a future version.
10034 @item -mvliw-branch
10035 @opindex mvliw-branch
10037 Run a pass to pack branches into VLIW instructions (default).
10039 This switch is mainly for debugging the compiler and will likely be removed
10040 in a future version.
10042 @item -mno-vliw-branch
10043 @opindex mno-vliw-branch
10045 Do not run a pass to pack branches into VLIW instructions.
10047 This switch is mainly for debugging the compiler and will likely be removed
10048 in a future version.
10050 @item -mmulti-cond-exec
10051 @opindex mmulti-cond-exec
10053 Enable optimization of @code{&&} and @code{||} in conditional execution
10056 This switch is mainly for debugging the compiler and will likely be removed
10057 in a future version.
10059 @item -mno-multi-cond-exec
10060 @opindex mno-multi-cond-exec
10062 Disable optimization of @code{&&} and @code{||} in conditional execution.
10064 This switch is mainly for debugging the compiler and will likely be removed
10065 in a future version.
10067 @item -mnested-cond-exec
10068 @opindex mnested-cond-exec
10070 Enable nested conditional execution optimizations (default).
10072 This switch is mainly for debugging the compiler and will likely be removed
10073 in a future version.
10075 @item -mno-nested-cond-exec
10076 @opindex mno-nested-cond-exec
10078 Disable nested conditional execution optimizations.
10080 This switch is mainly for debugging the compiler and will likely be removed
10081 in a future version.
10083 @item -moptimize-membar
10084 @opindex moptimize-membar
10086 This switch removes redundant @code{membar} instructions from the
10087 compiler generated code. It is enabled by default.
10089 @item -mno-optimize-membar
10090 @opindex mno-optimize-membar
10092 This switch disables the automatic removal of redundant @code{membar}
10093 instructions from the generated code.
10095 @item -mtomcat-stats
10096 @opindex mtomcat-stats
10098 Cause gas to print out tomcat statistics.
10100 @item -mcpu=@var{cpu}
10103 Select the processor type for which to generate code. Possible values are
10104 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
10105 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
10109 @node GNU/Linux Options
10110 @subsection GNU/Linux Options
10112 These @samp{-m} options are defined for GNU/Linux targets:
10117 Use the GNU C library instead of uClibc. This is the default except
10118 on @samp{*-*-linux-*uclibc*} targets.
10122 Use uClibc instead of the GNU C library. This is the default on
10123 @samp{*-*-linux-*uclibc*} targets.
10126 @node H8/300 Options
10127 @subsection H8/300 Options
10129 These @samp{-m} options are defined for the H8/300 implementations:
10134 Shorten some address references at link time, when possible; uses the
10135 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
10136 ld, Using ld}, for a fuller description.
10140 Generate code for the H8/300H@.
10144 Generate code for the H8S@.
10148 Generate code for the H8S and H8/300H in the normal mode. This switch
10149 must be used either with @option{-mh} or @option{-ms}.
10153 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
10157 Make @code{int} data 32 bits by default.
10160 @opindex malign-300
10161 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
10162 The default for the H8/300H and H8S is to align longs and floats on 4
10164 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
10165 This option has no effect on the H8/300.
10169 @subsection HPPA Options
10170 @cindex HPPA Options
10172 These @samp{-m} options are defined for the HPPA family of computers:
10175 @item -march=@var{architecture-type}
10177 Generate code for the specified architecture. The choices for
10178 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
10179 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
10180 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
10181 architecture option for your machine. Code compiled for lower numbered
10182 architectures will run on higher numbered architectures, but not the
10185 @item -mpa-risc-1-0
10186 @itemx -mpa-risc-1-1
10187 @itemx -mpa-risc-2-0
10188 @opindex mpa-risc-1-0
10189 @opindex mpa-risc-1-1
10190 @opindex mpa-risc-2-0
10191 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10194 @opindex mbig-switch
10195 Generate code suitable for big switch tables. Use this option only if
10196 the assembler/linker complain about out of range branches within a switch
10199 @item -mjump-in-delay
10200 @opindex mjump-in-delay
10201 Fill delay slots of function calls with unconditional jump instructions
10202 by modifying the return pointer for the function call to be the target
10203 of the conditional jump.
10205 @item -mdisable-fpregs
10206 @opindex mdisable-fpregs
10207 Prevent floating point registers from being used in any manner. This is
10208 necessary for compiling kernels which perform lazy context switching of
10209 floating point registers. If you use this option and attempt to perform
10210 floating point operations, the compiler will abort.
10212 @item -mdisable-indexing
10213 @opindex mdisable-indexing
10214 Prevent the compiler from using indexing address modes. This avoids some
10215 rather obscure problems when compiling MIG generated code under MACH@.
10217 @item -mno-space-regs
10218 @opindex mno-space-regs
10219 Generate code that assumes the target has no space registers. This allows
10220 GCC to generate faster indirect calls and use unscaled index address modes.
10222 Such code is suitable for level 0 PA systems and kernels.
10224 @item -mfast-indirect-calls
10225 @opindex mfast-indirect-calls
10226 Generate code that assumes calls never cross space boundaries. This
10227 allows GCC to emit code which performs faster indirect calls.
10229 This option will not work in the presence of shared libraries or nested
10232 @item -mfixed-range=@var{register-range}
10233 @opindex mfixed-range
10234 Generate code treating the given register range as fixed registers.
10235 A fixed register is one that the register allocator can not use. This is
10236 useful when compiling kernel code. A register range is specified as
10237 two registers separated by a dash. Multiple register ranges can be
10238 specified separated by a comma.
10240 @item -mlong-load-store
10241 @opindex mlong-load-store
10242 Generate 3-instruction load and store sequences as sometimes required by
10243 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
10246 @item -mportable-runtime
10247 @opindex mportable-runtime
10248 Use the portable calling conventions proposed by HP for ELF systems.
10252 Enable the use of assembler directives only GAS understands.
10254 @item -mschedule=@var{cpu-type}
10256 Schedule code according to the constraints for the machine type
10257 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10258 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10259 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10260 proper scheduling option for your machine. The default scheduling is
10264 @opindex mlinker-opt
10265 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10266 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10267 linkers in which they give bogus error messages when linking some programs.
10270 @opindex msoft-float
10271 Generate output containing library calls for floating point.
10272 @strong{Warning:} the requisite libraries are not available for all HPPA
10273 targets. Normally the facilities of the machine's usual C compiler are
10274 used, but this cannot be done directly in cross-compilation. You must make
10275 your own arrangements to provide suitable library functions for
10278 @option{-msoft-float} changes the calling convention in the output file;
10279 therefore, it is only useful if you compile @emph{all} of a program with
10280 this option. In particular, you need to compile @file{libgcc.a}, the
10281 library that comes with GCC, with @option{-msoft-float} in order for
10286 Generate the predefine, @code{_SIO}, for server IO@. The default is
10287 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10288 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10289 options are available under HP-UX and HI-UX@.
10293 Use GNU ld specific options. This passes @option{-shared} to ld when
10294 building a shared library. It is the default when GCC is configured,
10295 explicitly or implicitly, with the GNU linker. This option does not
10296 have any affect on which ld is called, it only changes what parameters
10297 are passed to that ld. The ld that is called is determined by the
10298 @option{--with-ld} configure option, GCC's program search path, and
10299 finally by the user's @env{PATH}. The linker used by GCC can be printed
10300 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10301 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10305 Use HP ld specific options. This passes @option{-b} to ld when building
10306 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10307 links. It is the default when GCC is configured, explicitly or
10308 implicitly, with the HP linker. This option does not have any affect on
10309 which ld is called, it only changes what parameters are passed to that
10310 ld. The ld that is called is determined by the @option{--with-ld}
10311 configure option, GCC's program search path, and finally by the user's
10312 @env{PATH}. The linker used by GCC can be printed using @samp{which
10313 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10314 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10317 @opindex mno-long-calls
10318 Generate code that uses long call sequences. This ensures that a call
10319 is always able to reach linker generated stubs. The default is to generate
10320 long calls only when the distance from the call site to the beginning
10321 of the function or translation unit, as the case may be, exceeds a
10322 predefined limit set by the branch type being used. The limits for
10323 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10324 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10327 Distances are measured from the beginning of functions when using the
10328 @option{-ffunction-sections} option, or when using the @option{-mgas}
10329 and @option{-mno-portable-runtime} options together under HP-UX with
10332 It is normally not desirable to use this option as it will degrade
10333 performance. However, it may be useful in large applications,
10334 particularly when partial linking is used to build the application.
10336 The types of long calls used depends on the capabilities of the
10337 assembler and linker, and the type of code being generated. The
10338 impact on systems that support long absolute calls, and long pic
10339 symbol-difference or pc-relative calls should be relatively small.
10340 However, an indirect call is used on 32-bit ELF systems in pic code
10341 and it is quite long.
10343 @item -munix=@var{unix-std}
10345 Generate compiler predefines and select a startfile for the specified
10346 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10347 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10348 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10349 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10350 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10353 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10354 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10355 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10356 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10357 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10358 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10360 It is @emph{important} to note that this option changes the interfaces
10361 for various library routines. It also affects the operational behavior
10362 of the C library. Thus, @emph{extreme} care is needed in using this
10365 Library code that is intended to operate with more than one UNIX
10366 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10367 as appropriate. Most GNU software doesn't provide this capability.
10371 Suppress the generation of link options to search libdld.sl when the
10372 @option{-static} option is specified on HP-UX 10 and later.
10376 The HP-UX implementation of setlocale in libc has a dependency on
10377 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10378 when the @option{-static} option is specified, special link options
10379 are needed to resolve this dependency.
10381 On HP-UX 10 and later, the GCC driver adds the necessary options to
10382 link with libdld.sl when the @option{-static} option is specified.
10383 This causes the resulting binary to be dynamic. On the 64-bit port,
10384 the linkers generate dynamic binaries by default in any case. The
10385 @option{-nolibdld} option can be used to prevent the GCC driver from
10386 adding these link options.
10390 Add support for multithreading with the @dfn{dce thread} library
10391 under HP-UX@. This option sets flags for both the preprocessor and
10395 @node i386 and x86-64 Options
10396 @subsection Intel 386 and AMD x86-64 Options
10397 @cindex i386 Options
10398 @cindex x86-64 Options
10399 @cindex Intel 386 Options
10400 @cindex AMD x86-64 Options
10402 These @samp{-m} options are defined for the i386 and x86-64 family of
10406 @item -mtune=@var{cpu-type}
10408 Tune to @var{cpu-type} everything applicable about the generated code, except
10409 for the ABI and the set of available instructions. The choices for
10410 @var{cpu-type} are:
10413 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10414 If you know the CPU on which your code will run, then you should use
10415 the corresponding @option{-mtune} option instead of
10416 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10417 of your application will have, then you should use this option.
10419 As new processors are deployed in the marketplace, the behavior of this
10420 option will change. Therefore, if you upgrade to a newer version of
10421 GCC, the code generated option will change to reflect the processors
10422 that were most common when that version of GCC was released.
10424 There is no @option{-march=generic} option because @option{-march}
10425 indicates the instruction set the compiler can use, and there is no
10426 generic instruction set applicable to all processors. In contrast,
10427 @option{-mtune} indicates the processor (or, in this case, collection of
10428 processors) for which the code is optimized.
10430 This selects the CPU to tune for at compilation time by determining
10431 the processor type of the compiling machine. Using @option{-mtune=native}
10432 will produce code optimized for the local machine under the constraints
10433 of the selected instruction set. Using @option{-march=native} will
10434 enable all instruction subsets supported by the local machine (hence
10435 the result might not run on different machines).
10437 Original Intel's i386 CPU@.
10439 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10440 @item i586, pentium
10441 Intel Pentium CPU with no MMX support.
10443 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10445 Intel PentiumPro CPU@.
10447 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10448 instruction set will be used, so the code will run on all i686 family chips.
10450 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10451 @item pentium3, pentium3m
10452 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10455 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10456 support. Used by Centrino notebooks.
10457 @item pentium4, pentium4m
10458 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10460 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10463 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10464 SSE2 and SSE3 instruction set support.
10466 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10467 instruction set support.
10469 AMD K6 CPU with MMX instruction set support.
10471 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
10472 @item athlon, athlon-tbird
10473 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
10475 @item athlon-4, athlon-xp, athlon-mp
10476 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
10477 instruction set support.
10478 @item k8, opteron, athlon64, athlon-fx
10479 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10480 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
10481 @item k8-sse3, opteron-sse3, athlon64-sse3
10482 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10483 @item amdfam10, barcelona
10484 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10485 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10486 instruction set extensions.)
10488 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10491 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
10492 instruction set support.
10494 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
10495 implemented for this chip.)
10497 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10498 implemented for this chip.)
10500 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10503 While picking a specific @var{cpu-type} will schedule things appropriately
10504 for that particular chip, the compiler will not generate any code that
10505 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10508 @item -march=@var{cpu-type}
10510 Generate instructions for the machine type @var{cpu-type}. The choices
10511 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10512 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10514 @item -mcpu=@var{cpu-type}
10516 A deprecated synonym for @option{-mtune}.
10518 @item -mfpmath=@var{unit}
10520 Generate floating point arithmetics for selected unit @var{unit}. The choices
10521 for @var{unit} are:
10525 Use the standard 387 floating point coprocessor present majority of chips and
10526 emulated otherwise. Code compiled with this option will run almost everywhere.
10527 The temporary results are computed in 80bit precision instead of precision
10528 specified by the type resulting in slightly different results compared to most
10529 of other chips. See @option{-ffloat-store} for more detailed description.
10531 This is the default choice for i386 compiler.
10534 Use scalar floating point instructions present in the SSE instruction set.
10535 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10536 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10537 instruction set supports only single precision arithmetics, thus the double and
10538 extended precision arithmetics is still done using 387. Later version, present
10539 only in Pentium4 and the future AMD x86-64 chips supports double precision
10542 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10543 or @option{-msse2} switches to enable SSE extensions and make this option
10544 effective. For the x86-64 compiler, these extensions are enabled by default.
10546 The resulting code should be considerably faster in the majority of cases and avoid
10547 the numerical instability problems of 387 code, but may break some existing
10548 code that expects temporaries to be 80bit.
10550 This is the default choice for the x86-64 compiler.
10555 Attempt to utilize both instruction sets at once. This effectively double the
10556 amount of available registers and on chips with separate execution units for
10557 387 and SSE the execution resources too. Use this option with care, as it is
10558 still experimental, because the GCC register allocator does not model separate
10559 functional units well resulting in instable performance.
10562 @item -masm=@var{dialect}
10563 @opindex masm=@var{dialect}
10564 Output asm instructions using selected @var{dialect}. Supported
10565 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10566 not support @samp{intel}.
10569 @itemx -mno-ieee-fp
10571 @opindex mno-ieee-fp
10572 Control whether or not the compiler uses IEEE floating point
10573 comparisons. These handle correctly the case where the result of a
10574 comparison is unordered.
10577 @opindex msoft-float
10578 Generate output containing library calls for floating point.
10579 @strong{Warning:} the requisite libraries are not part of GCC@.
10580 Normally the facilities of the machine's usual C compiler are used, but
10581 this can't be done directly in cross-compilation. You must make your
10582 own arrangements to provide suitable library functions for
10585 On machines where a function returns floating point results in the 80387
10586 register stack, some floating point opcodes may be emitted even if
10587 @option{-msoft-float} is used.
10589 @item -mno-fp-ret-in-387
10590 @opindex mno-fp-ret-in-387
10591 Do not use the FPU registers for return values of functions.
10593 The usual calling convention has functions return values of types
10594 @code{float} and @code{double} in an FPU register, even if there
10595 is no FPU@. The idea is that the operating system should emulate
10598 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10599 in ordinary CPU registers instead.
10601 @item -mno-fancy-math-387
10602 @opindex mno-fancy-math-387
10603 Some 387 emulators do not support the @code{sin}, @code{cos} and
10604 @code{sqrt} instructions for the 387. Specify this option to avoid
10605 generating those instructions. This option is the default on FreeBSD,
10606 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10607 indicates that the target cpu will always have an FPU and so the
10608 instruction will not need emulation. As of revision 2.6.1, these
10609 instructions are not generated unless you also use the
10610 @option{-funsafe-math-optimizations} switch.
10612 @item -malign-double
10613 @itemx -mno-align-double
10614 @opindex malign-double
10615 @opindex mno-align-double
10616 Control whether GCC aligns @code{double}, @code{long double}, and
10617 @code{long long} variables on a two word boundary or a one word
10618 boundary. Aligning @code{double} variables on a two word boundary will
10619 produce code that runs somewhat faster on a @samp{Pentium} at the
10620 expense of more memory.
10622 On x86-64, @option{-malign-double} is enabled by default.
10624 @strong{Warning:} if you use the @option{-malign-double} switch,
10625 structures containing the above types will be aligned differently than
10626 the published application binary interface specifications for the 386
10627 and will not be binary compatible with structures in code compiled
10628 without that switch.
10630 @item -m96bit-long-double
10631 @itemx -m128bit-long-double
10632 @opindex m96bit-long-double
10633 @opindex m128bit-long-double
10634 These switches control the size of @code{long double} type. The i386
10635 application binary interface specifies the size to be 96 bits,
10636 so @option{-m96bit-long-double} is the default in 32 bit mode.
10638 Modern architectures (Pentium and newer) would prefer @code{long double}
10639 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10640 conforming to the ABI, this would not be possible. So specifying a
10641 @option{-m128bit-long-double} will align @code{long double}
10642 to a 16 byte boundary by padding the @code{long double} with an additional
10645 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10646 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10648 Notice that neither of these options enable any extra precision over the x87
10649 standard of 80 bits for a @code{long double}.
10651 @strong{Warning:} if you override the default value for your target ABI, the
10652 structures and arrays containing @code{long double} variables will change
10653 their size as well as function calling convention for function taking
10654 @code{long double} will be modified. Hence they will not be binary
10655 compatible with arrays or structures in code compiled without that switch.
10657 @item -mmlarge-data-threshold=@var{number}
10658 @opindex mlarge-data-threshold=@var{number}
10659 When @option{-mcmodel=medium} is specified, the data greater than
10660 @var{threshold} are placed in large data section. This value must be the
10661 same across all object linked into the binary and defaults to 65535.
10665 Use a different function-calling convention, in which functions that
10666 take a fixed number of arguments return with the @code{ret} @var{num}
10667 instruction, which pops their arguments while returning. This saves one
10668 instruction in the caller since there is no need to pop the arguments
10671 You can specify that an individual function is called with this calling
10672 sequence with the function attribute @samp{stdcall}. You can also
10673 override the @option{-mrtd} option by using the function attribute
10674 @samp{cdecl}. @xref{Function Attributes}.
10676 @strong{Warning:} this calling convention is incompatible with the one
10677 normally used on Unix, so you cannot use it if you need to call
10678 libraries compiled with the Unix compiler.
10680 Also, you must provide function prototypes for all functions that
10681 take variable numbers of arguments (including @code{printf});
10682 otherwise incorrect code will be generated for calls to those
10685 In addition, seriously incorrect code will result if you call a
10686 function with too many arguments. (Normally, extra arguments are
10687 harmlessly ignored.)
10689 @item -mregparm=@var{num}
10691 Control how many registers are used to pass integer arguments. By
10692 default, no registers are used to pass arguments, and at most 3
10693 registers can be used. You can control this behavior for a specific
10694 function by using the function attribute @samp{regparm}.
10695 @xref{Function Attributes}.
10697 @strong{Warning:} if you use this switch, and
10698 @var{num} is nonzero, then you must build all modules with the same
10699 value, including any libraries. This includes the system libraries and
10703 @opindex msseregparm
10704 Use SSE register passing conventions for float and double arguments
10705 and return values. You can control this behavior for a specific
10706 function by using the function attribute @samp{sseregparm}.
10707 @xref{Function Attributes}.
10709 @strong{Warning:} if you use this switch then you must build all
10710 modules with the same value, including any libraries. This includes
10711 the system libraries and startup modules.
10720 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10721 is specified, the significands of results of floating-point operations are
10722 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10723 significands of results of floating-point operations to 53 bits (double
10724 precision) and @option{-mpc80} rounds the significands of results of
10725 floating-point operations to 64 bits (extended double precision), which is
10726 the default. When this option is used, floating-point operations in higher
10727 precisions are not available to the programmer without setting the FPU
10728 control word explicitly.
10730 Setting the rounding of floating-point operations to less than the default
10731 80 bits can speed some programs by 2% or more. Note that some mathematical
10732 libraries assume that extended precision (80 bit) floating-point operations
10733 are enabled by default; routines in such libraries could suffer significant
10734 loss of accuracy, typically through so-called "catastrophic cancellation",
10735 when this option is used to set the precision to less than extended precision.
10737 @item -mstackrealign
10738 @opindex mstackrealign
10739 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
10740 option will generate an alternate prologue and epilogue that realigns the
10741 runtime stack if necessary. This supports mixing legacy codes that keep
10742 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
10743 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
10744 applicable to individual functions.
10746 @item -mpreferred-stack-boundary=@var{num}
10747 @opindex mpreferred-stack-boundary
10748 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10749 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10750 the default is 4 (16 bytes or 128 bits).
10752 @item -mincoming-stack-boundary=@var{num}
10753 @opindex mincoming-stack-boundary
10754 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
10755 boundary. If @option{-mincoming-stack-boundary} is not specified,
10756 the one specified by @option{-mpreferred-stack-boundary} will be used.
10758 On Pentium and PentiumPro, @code{double} and @code{long double} values
10759 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10760 suffer significant run time performance penalties. On Pentium III, the
10761 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10762 properly if it is not 16 byte aligned.
10764 To ensure proper alignment of this values on the stack, the stack boundary
10765 must be as aligned as that required by any value stored on the stack.
10766 Further, every function must be generated such that it keeps the stack
10767 aligned. Thus calling a function compiled with a higher preferred
10768 stack boundary from a function compiled with a lower preferred stack
10769 boundary will most likely misalign the stack. It is recommended that
10770 libraries that use callbacks always use the default setting.
10772 This extra alignment does consume extra stack space, and generally
10773 increases code size. Code that is sensitive to stack space usage, such
10774 as embedded systems and operating system kernels, may want to reduce the
10775 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10813 These switches enable or disable the use of instructions in the MMX,
10814 SSE, SSE2, SSE3, SSSE3, SSE4.1, AES, PCLMUL, SSE4A, SSE5, ABM or
10815 3DNow!@: extended instruction sets.
10816 These extensions are also available as built-in functions: see
10817 @ref{X86 Built-in Functions}, for details of the functions enabled and
10818 disabled by these switches.
10820 To have SSE/SSE2 instructions generated automatically from floating-point
10821 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10823 These options will enable GCC to use these extended instructions in
10824 generated code, even without @option{-mfpmath=sse}. Applications which
10825 perform runtime CPU detection must compile separate files for each
10826 supported architecture, using the appropriate flags. In particular,
10827 the file containing the CPU detection code should be compiled without
10832 This option instructs GCC to emit a @code{cld} instruction in the prologue
10833 of functions that use string instructions. String instructions depend on
10834 the DF flag to select between autoincrement or autodecrement mode. While the
10835 ABI specifies the DF flag to be cleared on function entry, some operating
10836 systems violate this specification by not clearing the DF flag in their
10837 exception dispatchers. The exception handler can be invoked with the DF flag
10838 set which leads to wrong direction mode, when string instructions are used.
10839 This option can be enabled by default on 32-bit x86 targets by configuring
10840 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
10841 instructions can be suppressed with the @option{-mno-cld} compiler option
10846 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10847 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10848 data types. This is useful for high resolution counters that could be updated
10849 by multiple processors (or cores). This instruction is generated as part of
10850 atomic built-in functions: see @ref{Atomic Builtins} for details.
10854 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10855 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10856 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10857 SAHF are load and store instructions, respectively, for certain status flags.
10858 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10859 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10863 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10864 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Rhapson step
10865 to increase precision instead of DIVSS and SQRTSS (and their vectorized
10866 variants) for single precision floating point arguments. These instructions
10867 are generated only when @option{-funsafe-math-optimizations} is enabled
10868 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
10869 Note that while the throughput of the sequence is higher than the throughput
10870 of the non-reciprocal instruction, the precision of the sequence can be
10871 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
10873 @item -mveclibabi=@var{type}
10874 @opindex mveclibabi
10875 Specifies the ABI type to use for vectorizing intrinsics using an
10876 external library. Supported types are @code{svml} for the Intel short
10877 vector math library and @code{acml} for the AMD math core library style
10878 of interfacing. GCC will currently emit calls to @code{vmldExp2},
10879 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
10880 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
10881 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
10882 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
10883 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
10884 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
10885 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
10886 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
10887 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
10888 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
10889 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
10890 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
10891 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
10892 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
10893 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
10894 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
10895 compatible library will have to be specified at link time.
10898 @itemx -mno-push-args
10899 @opindex mpush-args
10900 @opindex mno-push-args
10901 Use PUSH operations to store outgoing parameters. This method is shorter
10902 and usually equally fast as method using SUB/MOV operations and is enabled
10903 by default. In some cases disabling it may improve performance because of
10904 improved scheduling and reduced dependencies.
10906 @item -maccumulate-outgoing-args
10907 @opindex maccumulate-outgoing-args
10908 If enabled, the maximum amount of space required for outgoing arguments will be
10909 computed in the function prologue. This is faster on most modern CPUs
10910 because of reduced dependencies, improved scheduling and reduced stack usage
10911 when preferred stack boundary is not equal to 2. The drawback is a notable
10912 increase in code size. This switch implies @option{-mno-push-args}.
10916 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10917 on thread-safe exception handling must compile and link all code with the
10918 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10919 @option{-D_MT}; when linking, it links in a special thread helper library
10920 @option{-lmingwthrd} which cleans up per thread exception handling data.
10922 @item -mno-align-stringops
10923 @opindex mno-align-stringops
10924 Do not align destination of inlined string operations. This switch reduces
10925 code size and improves performance in case the destination is already aligned,
10926 but GCC doesn't know about it.
10928 @item -minline-all-stringops
10929 @opindex minline-all-stringops
10930 By default GCC inlines string operations only when destination is known to be
10931 aligned at least to 4 byte boundary. This enables more inlining, increase code
10932 size, but may improve performance of code that depends on fast memcpy, strlen
10933 and memset for short lengths.
10935 @item -minline-stringops-dynamically
10936 @opindex minline-stringops-dynamically
10937 For string operation of unknown size, inline runtime checks so for small
10938 blocks inline code is used, while for large blocks library call is used.
10940 @item -mstringop-strategy=@var{alg}
10941 @opindex mstringop-strategy=@var{alg}
10942 Overwrite internal decision heuristic about particular algorithm to inline
10943 string operation with. The allowed values are @code{rep_byte},
10944 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10945 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10946 expanding inline loop, @code{libcall} for always expanding library call.
10948 @item -momit-leaf-frame-pointer
10949 @opindex momit-leaf-frame-pointer
10950 Don't keep the frame pointer in a register for leaf functions. This
10951 avoids the instructions to save, set up and restore frame pointers and
10952 makes an extra register available in leaf functions. The option
10953 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10954 which might make debugging harder.
10956 @item -mtls-direct-seg-refs
10957 @itemx -mno-tls-direct-seg-refs
10958 @opindex mtls-direct-seg-refs
10959 Controls whether TLS variables may be accessed with offsets from the
10960 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10961 or whether the thread base pointer must be added. Whether or not this
10962 is legal depends on the operating system, and whether it maps the
10963 segment to cover the entire TLS area.
10965 For systems that use GNU libc, the default is on.
10968 @itemx -mno-fused-madd
10969 @opindex mfused-madd
10970 Enable automatic generation of fused floating point multiply-add instructions
10971 if the ISA supports such instructions. The -mfused-madd option is on by
10972 default. The fused multiply-add instructions have a different
10973 rounding behavior compared to executing a multiply followed by an add.
10976 These @samp{-m} switches are supported in addition to the above
10977 on AMD x86-64 processors in 64-bit environments.
10984 Generate code for a 32-bit or 64-bit environment.
10985 The 32-bit environment sets int, long and pointer to 32 bits and
10986 generates code that runs on any i386 system.
10987 The 64-bit environment sets int to 32 bits and long and pointer
10988 to 64 bits and generates code for AMD's x86-64 architecture. For
10989 darwin only the -m64 option turns off the @option{-fno-pic} and
10990 @option{-mdynamic-no-pic} options.
10992 @item -mno-red-zone
10993 @opindex no-red-zone
10994 Do not use a so called red zone for x86-64 code. The red zone is mandated
10995 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10996 stack pointer that will not be modified by signal or interrupt handlers
10997 and therefore can be used for temporary data without adjusting the stack
10998 pointer. The flag @option{-mno-red-zone} disables this red zone.
11000 @item -mcmodel=small
11001 @opindex mcmodel=small
11002 Generate code for the small code model: the program and its symbols must
11003 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
11004 Programs can be statically or dynamically linked. This is the default
11007 @item -mcmodel=kernel
11008 @opindex mcmodel=kernel
11009 Generate code for the kernel code model. The kernel runs in the
11010 negative 2 GB of the address space.
11011 This model has to be used for Linux kernel code.
11013 @item -mcmodel=medium
11014 @opindex mcmodel=medium
11015 Generate code for the medium model: The program is linked in the lower 2
11016 GB of the address space but symbols can be located anywhere in the
11017 address space. Programs can be statically or dynamically linked, but
11018 building of shared libraries are not supported with the medium model.
11020 @item -mcmodel=large
11021 @opindex mcmodel=large
11022 Generate code for the large model: This model makes no assumptions
11023 about addresses and sizes of sections.
11026 @node IA-64 Options
11027 @subsection IA-64 Options
11028 @cindex IA-64 Options
11030 These are the @samp{-m} options defined for the Intel IA-64 architecture.
11034 @opindex mbig-endian
11035 Generate code for a big endian target. This is the default for HP-UX@.
11037 @item -mlittle-endian
11038 @opindex mlittle-endian
11039 Generate code for a little endian target. This is the default for AIX5
11045 @opindex mno-gnu-as
11046 Generate (or don't) code for the GNU assembler. This is the default.
11047 @c Also, this is the default if the configure option @option{--with-gnu-as}
11053 @opindex mno-gnu-ld
11054 Generate (or don't) code for the GNU linker. This is the default.
11055 @c Also, this is the default if the configure option @option{--with-gnu-ld}
11060 Generate code that does not use a global pointer register. The result
11061 is not position independent code, and violates the IA-64 ABI@.
11063 @item -mvolatile-asm-stop
11064 @itemx -mno-volatile-asm-stop
11065 @opindex mvolatile-asm-stop
11066 @opindex mno-volatile-asm-stop
11067 Generate (or don't) a stop bit immediately before and after volatile asm
11070 @item -mregister-names
11071 @itemx -mno-register-names
11072 @opindex mregister-names
11073 @opindex mno-register-names
11074 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
11075 the stacked registers. This may make assembler output more readable.
11081 Disable (or enable) optimizations that use the small data section. This may
11082 be useful for working around optimizer bugs.
11084 @item -mconstant-gp
11085 @opindex mconstant-gp
11086 Generate code that uses a single constant global pointer value. This is
11087 useful when compiling kernel code.
11091 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
11092 This is useful when compiling firmware code.
11094 @item -minline-float-divide-min-latency
11095 @opindex minline-float-divide-min-latency
11096 Generate code for inline divides of floating point values
11097 using the minimum latency algorithm.
11099 @item -minline-float-divide-max-throughput
11100 @opindex minline-float-divide-max-throughput
11101 Generate code for inline divides of floating point values
11102 using the maximum throughput algorithm.
11104 @item -minline-int-divide-min-latency
11105 @opindex minline-int-divide-min-latency
11106 Generate code for inline divides of integer values
11107 using the minimum latency algorithm.
11109 @item -minline-int-divide-max-throughput
11110 @opindex minline-int-divide-max-throughput
11111 Generate code for inline divides of integer values
11112 using the maximum throughput algorithm.
11114 @item -minline-sqrt-min-latency
11115 @opindex minline-sqrt-min-latency
11116 Generate code for inline square roots
11117 using the minimum latency algorithm.
11119 @item -minline-sqrt-max-throughput
11120 @opindex minline-sqrt-max-throughput
11121 Generate code for inline square roots
11122 using the maximum throughput algorithm.
11124 @item -mno-dwarf2-asm
11125 @itemx -mdwarf2-asm
11126 @opindex mno-dwarf2-asm
11127 @opindex mdwarf2-asm
11128 Don't (or do) generate assembler code for the DWARF2 line number debugging
11129 info. This may be useful when not using the GNU assembler.
11131 @item -mearly-stop-bits
11132 @itemx -mno-early-stop-bits
11133 @opindex mearly-stop-bits
11134 @opindex mno-early-stop-bits
11135 Allow stop bits to be placed earlier than immediately preceding the
11136 instruction that triggered the stop bit. This can improve instruction
11137 scheduling, but does not always do so.
11139 @item -mfixed-range=@var{register-range}
11140 @opindex mfixed-range
11141 Generate code treating the given register range as fixed registers.
11142 A fixed register is one that the register allocator can not use. This is
11143 useful when compiling kernel code. A register range is specified as
11144 two registers separated by a dash. Multiple register ranges can be
11145 specified separated by a comma.
11147 @item -mtls-size=@var{tls-size}
11149 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
11152 @item -mtune=@var{cpu-type}
11154 Tune the instruction scheduling for a particular CPU, Valid values are
11155 itanium, itanium1, merced, itanium2, and mckinley.
11161 Add support for multithreading using the POSIX threads library. This
11162 option sets flags for both the preprocessor and linker. It does
11163 not affect the thread safety of object code produced by the compiler or
11164 that of libraries supplied with it. These are HP-UX specific flags.
11170 Generate code for a 32-bit or 64-bit environment.
11171 The 32-bit environment sets int, long and pointer to 32 bits.
11172 The 64-bit environment sets int to 32 bits and long and pointer
11173 to 64 bits. These are HP-UX specific flags.
11175 @item -mno-sched-br-data-spec
11176 @itemx -msched-br-data-spec
11177 @opindex mno-sched-br-data-spec
11178 @opindex msched-br-data-spec
11179 (Dis/En)able data speculative scheduling before reload.
11180 This will result in generation of the ld.a instructions and
11181 the corresponding check instructions (ld.c / chk.a).
11182 The default is 'disable'.
11184 @item -msched-ar-data-spec
11185 @itemx -mno-sched-ar-data-spec
11186 @opindex msched-ar-data-spec
11187 @opindex mno-sched-ar-data-spec
11188 (En/Dis)able data speculative scheduling after reload.
11189 This will result in generation of the ld.a instructions and
11190 the corresponding check instructions (ld.c / chk.a).
11191 The default is 'enable'.
11193 @item -mno-sched-control-spec
11194 @itemx -msched-control-spec
11195 @opindex mno-sched-control-spec
11196 @opindex msched-control-spec
11197 (Dis/En)able control speculative scheduling. This feature is
11198 available only during region scheduling (i.e.@: before reload).
11199 This will result in generation of the ld.s instructions and
11200 the corresponding check instructions chk.s .
11201 The default is 'disable'.
11203 @item -msched-br-in-data-spec
11204 @itemx -mno-sched-br-in-data-spec
11205 @opindex msched-br-in-data-spec
11206 @opindex mno-sched-br-in-data-spec
11207 (En/Dis)able speculative scheduling of the instructions that
11208 are dependent on the data speculative loads before reload.
11209 This is effective only with @option{-msched-br-data-spec} enabled.
11210 The default is 'enable'.
11212 @item -msched-ar-in-data-spec
11213 @itemx -mno-sched-ar-in-data-spec
11214 @opindex msched-ar-in-data-spec
11215 @opindex mno-sched-ar-in-data-spec
11216 (En/Dis)able speculative scheduling of the instructions that
11217 are dependent on the data speculative loads after reload.
11218 This is effective only with @option{-msched-ar-data-spec} enabled.
11219 The default is 'enable'.
11221 @item -msched-in-control-spec
11222 @itemx -mno-sched-in-control-spec
11223 @opindex msched-in-control-spec
11224 @opindex mno-sched-in-control-spec
11225 (En/Dis)able speculative scheduling of the instructions that
11226 are dependent on the control speculative loads.
11227 This is effective only with @option{-msched-control-spec} enabled.
11228 The default is 'enable'.
11231 @itemx -mno-sched-ldc
11232 @opindex msched-ldc
11233 @opindex mno-sched-ldc
11234 (En/Dis)able use of simple data speculation checks ld.c .
11235 If disabled, only chk.a instructions will be emitted to check
11236 data speculative loads.
11237 The default is 'enable'.
11239 @item -mno-sched-control-ldc
11240 @itemx -msched-control-ldc
11241 @opindex mno-sched-control-ldc
11242 @opindex msched-control-ldc
11243 (Dis/En)able use of ld.c instructions to check control speculative loads.
11244 If enabled, in case of control speculative load with no speculatively
11245 scheduled dependent instructions this load will be emitted as ld.sa and
11246 ld.c will be used to check it.
11247 The default is 'disable'.
11249 @item -mno-sched-spec-verbose
11250 @itemx -msched-spec-verbose
11251 @opindex mno-sched-spec-verbose
11252 @opindex msched-spec-verbose
11253 (Dis/En)able printing of the information about speculative motions.
11255 @item -mno-sched-prefer-non-data-spec-insns
11256 @itemx -msched-prefer-non-data-spec-insns
11257 @opindex mno-sched-prefer-non-data-spec-insns
11258 @opindex msched-prefer-non-data-spec-insns
11259 If enabled, data speculative instructions will be chosen for schedule
11260 only if there are no other choices at the moment. This will make
11261 the use of the data speculation much more conservative.
11262 The default is 'disable'.
11264 @item -mno-sched-prefer-non-control-spec-insns
11265 @itemx -msched-prefer-non-control-spec-insns
11266 @opindex mno-sched-prefer-non-control-spec-insns
11267 @opindex msched-prefer-non-control-spec-insns
11268 If enabled, control speculative instructions will be chosen for schedule
11269 only if there are no other choices at the moment. This will make
11270 the use of the control speculation much more conservative.
11271 The default is 'disable'.
11273 @item -mno-sched-count-spec-in-critical-path
11274 @itemx -msched-count-spec-in-critical-path
11275 @opindex mno-sched-count-spec-in-critical-path
11276 @opindex msched-count-spec-in-critical-path
11277 If enabled, speculative dependencies will be considered during
11278 computation of the instructions priorities. This will make the use of the
11279 speculation a bit more conservative.
11280 The default is 'disable'.
11285 @subsection M32C Options
11286 @cindex M32C options
11289 @item -mcpu=@var{name}
11291 Select the CPU for which code is generated. @var{name} may be one of
11292 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11293 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11294 the M32C/80 series.
11298 Specifies that the program will be run on the simulator. This causes
11299 an alternate runtime library to be linked in which supports, for
11300 example, file I/O@. You must not use this option when generating
11301 programs that will run on real hardware; you must provide your own
11302 runtime library for whatever I/O functions are needed.
11304 @item -memregs=@var{number}
11306 Specifies the number of memory-based pseudo-registers GCC will use
11307 during code generation. These pseudo-registers will be used like real
11308 registers, so there is a tradeoff between GCC's ability to fit the
11309 code into available registers, and the performance penalty of using
11310 memory instead of registers. Note that all modules in a program must
11311 be compiled with the same value for this option. Because of that, you
11312 must not use this option with the default runtime libraries gcc
11317 @node M32R/D Options
11318 @subsection M32R/D Options
11319 @cindex M32R/D options
11321 These @option{-m} options are defined for Renesas M32R/D architectures:
11326 Generate code for the M32R/2@.
11330 Generate code for the M32R/X@.
11334 Generate code for the M32R@. This is the default.
11336 @item -mmodel=small
11337 @opindex mmodel=small
11338 Assume all objects live in the lower 16MB of memory (so that their addresses
11339 can be loaded with the @code{ld24} instruction), and assume all subroutines
11340 are reachable with the @code{bl} instruction.
11341 This is the default.
11343 The addressability of a particular object can be set with the
11344 @code{model} attribute.
11346 @item -mmodel=medium
11347 @opindex mmodel=medium
11348 Assume objects may be anywhere in the 32-bit address space (the compiler
11349 will generate @code{seth/add3} instructions to load their addresses), and
11350 assume all subroutines are reachable with the @code{bl} instruction.
11352 @item -mmodel=large
11353 @opindex mmodel=large
11354 Assume objects may be anywhere in the 32-bit address space (the compiler
11355 will generate @code{seth/add3} instructions to load their addresses), and
11356 assume subroutines may not be reachable with the @code{bl} instruction
11357 (the compiler will generate the much slower @code{seth/add3/jl}
11358 instruction sequence).
11361 @opindex msdata=none
11362 Disable use of the small data area. Variables will be put into
11363 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11364 @code{section} attribute has been specified).
11365 This is the default.
11367 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11368 Objects may be explicitly put in the small data area with the
11369 @code{section} attribute using one of these sections.
11371 @item -msdata=sdata
11372 @opindex msdata=sdata
11373 Put small global and static data in the small data area, but do not
11374 generate special code to reference them.
11377 @opindex msdata=use
11378 Put small global and static data in the small data area, and generate
11379 special instructions to reference them.
11383 @cindex smaller data references
11384 Put global and static objects less than or equal to @var{num} bytes
11385 into the small data or bss sections instead of the normal data or bss
11386 sections. The default value of @var{num} is 8.
11387 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11388 for this option to have any effect.
11390 All modules should be compiled with the same @option{-G @var{num}} value.
11391 Compiling with different values of @var{num} may or may not work; if it
11392 doesn't the linker will give an error message---incorrect code will not be
11397 Makes the M32R specific code in the compiler display some statistics
11398 that might help in debugging programs.
11400 @item -malign-loops
11401 @opindex malign-loops
11402 Align all loops to a 32-byte boundary.
11404 @item -mno-align-loops
11405 @opindex mno-align-loops
11406 Do not enforce a 32-byte alignment for loops. This is the default.
11408 @item -missue-rate=@var{number}
11409 @opindex missue-rate=@var{number}
11410 Issue @var{number} instructions per cycle. @var{number} can only be 1
11413 @item -mbranch-cost=@var{number}
11414 @opindex mbranch-cost=@var{number}
11415 @var{number} can only be 1 or 2. If it is 1 then branches will be
11416 preferred over conditional code, if it is 2, then the opposite will
11419 @item -mflush-trap=@var{number}
11420 @opindex mflush-trap=@var{number}
11421 Specifies the trap number to use to flush the cache. The default is
11422 12. Valid numbers are between 0 and 15 inclusive.
11424 @item -mno-flush-trap
11425 @opindex mno-flush-trap
11426 Specifies that the cache cannot be flushed by using a trap.
11428 @item -mflush-func=@var{name}
11429 @opindex mflush-func=@var{name}
11430 Specifies the name of the operating system function to call to flush
11431 the cache. The default is @emph{_flush_cache}, but a function call
11432 will only be used if a trap is not available.
11434 @item -mno-flush-func
11435 @opindex mno-flush-func
11436 Indicates that there is no OS function for flushing the cache.
11440 @node M680x0 Options
11441 @subsection M680x0 Options
11442 @cindex M680x0 options
11444 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11445 The default settings depend on which architecture was selected when
11446 the compiler was configured; the defaults for the most common choices
11450 @item -march=@var{arch}
11452 Generate code for a specific M680x0 or ColdFire instruction set
11453 architecture. Permissible values of @var{arch} for M680x0
11454 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11455 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11456 architectures are selected according to Freescale's ISA classification
11457 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11458 @samp{isab} and @samp{isac}.
11460 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11461 code for a ColdFire target. The @var{arch} in this macro is one of the
11462 @option{-march} arguments given above.
11464 When used together, @option{-march} and @option{-mtune} select code
11465 that runs on a family of similar processors but that is optimized
11466 for a particular microarchitecture.
11468 @item -mcpu=@var{cpu}
11470 Generate code for a specific M680x0 or ColdFire processor.
11471 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11472 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11473 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11474 below, which also classifies the CPUs into families:
11476 @multitable @columnfractions 0.20 0.80
11477 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11478 @item @samp{51qe} @tab @samp{51qe}
11479 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11480 @item @samp{5206e} @tab @samp{5206e}
11481 @item @samp{5208} @tab @samp{5207} @samp{5208}
11482 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11483 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11484 @item @samp{5216} @tab @samp{5214} @samp{5216}
11485 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11486 @item @samp{5225} @tab @samp{5224} @samp{5225}
11487 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11488 @item @samp{5249} @tab @samp{5249}
11489 @item @samp{5250} @tab @samp{5250}
11490 @item @samp{5271} @tab @samp{5270} @samp{5271}
11491 @item @samp{5272} @tab @samp{5272}
11492 @item @samp{5275} @tab @samp{5274} @samp{5275}
11493 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11494 @item @samp{5307} @tab @samp{5307}
11495 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11496 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11497 @item @samp{5407} @tab @samp{5407}
11498 @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}
11501 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11502 @var{arch} is compatible with @var{cpu}. Other combinations of
11503 @option{-mcpu} and @option{-march} are rejected.
11505 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11506 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11507 where the value of @var{family} is given by the table above.
11509 @item -mtune=@var{tune}
11511 Tune the code for a particular microarchitecture, within the
11512 constraints set by @option{-march} and @option{-mcpu}.
11513 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11514 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11515 and @samp{cpu32}. The ColdFire microarchitectures
11516 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11518 You can also use @option{-mtune=68020-40} for code that needs
11519 to run relatively well on 68020, 68030 and 68040 targets.
11520 @option{-mtune=68020-60} is similar but includes 68060 targets
11521 as well. These two options select the same tuning decisions as
11522 @option{-m68020-40} and @option{-m68020-60} respectively.
11524 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11525 when tuning for 680x0 architecture @var{arch}. It also defines
11526 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11527 option is used. If gcc is tuning for a range of architectures,
11528 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11529 it defines the macros for every architecture in the range.
11531 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11532 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11533 of the arguments given above.
11539 Generate output for a 68000. This is the default
11540 when the compiler is configured for 68000-based systems.
11541 It is equivalent to @option{-march=68000}.
11543 Use this option for microcontrollers with a 68000 or EC000 core,
11544 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11548 Generate output for a 68010. This is the default
11549 when the compiler is configured for 68010-based systems.
11550 It is equivalent to @option{-march=68010}.
11556 Generate output for a 68020. This is the default
11557 when the compiler is configured for 68020-based systems.
11558 It is equivalent to @option{-march=68020}.
11562 Generate output for a 68030. This is the default when the compiler is
11563 configured for 68030-based systems. It is equivalent to
11564 @option{-march=68030}.
11568 Generate output for a 68040. This is the default when the compiler is
11569 configured for 68040-based systems. It is equivalent to
11570 @option{-march=68040}.
11572 This option inhibits the use of 68881/68882 instructions that have to be
11573 emulated by software on the 68040. Use this option if your 68040 does not
11574 have code to emulate those instructions.
11578 Generate output for a 68060. This is the default when the compiler is
11579 configured for 68060-based systems. It is equivalent to
11580 @option{-march=68060}.
11582 This option inhibits the use of 68020 and 68881/68882 instructions that
11583 have to be emulated by software on the 68060. Use this option if your 68060
11584 does not have code to emulate those instructions.
11588 Generate output for a CPU32. This is the default
11589 when the compiler is configured for CPU32-based systems.
11590 It is equivalent to @option{-march=cpu32}.
11592 Use this option for microcontrollers with a
11593 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11594 68336, 68340, 68341, 68349 and 68360.
11598 Generate output for a 520X ColdFire CPU@. This is the default
11599 when the compiler is configured for 520X-based systems.
11600 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11601 in favor of that option.
11603 Use this option for microcontroller with a 5200 core, including
11604 the MCF5202, MCF5203, MCF5204 and MCF5206.
11608 Generate output for a 5206e ColdFire CPU@. The option is now
11609 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11613 Generate output for a member of the ColdFire 528X family.
11614 The option is now deprecated in favor of the equivalent
11615 @option{-mcpu=528x}.
11619 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
11620 in favor of the equivalent @option{-mcpu=5307}.
11624 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
11625 in favor of the equivalent @option{-mcpu=5407}.
11629 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11630 This includes use of hardware floating point instructions.
11631 The option is equivalent to @option{-mcpu=547x}, and is now
11632 deprecated in favor of that option.
11636 Generate output for a 68040, without using any of the new instructions.
11637 This results in code which can run relatively efficiently on either a
11638 68020/68881 or a 68030 or a 68040. The generated code does use the
11639 68881 instructions that are emulated on the 68040.
11641 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11645 Generate output for a 68060, without using any of the new instructions.
11646 This results in code which can run relatively efficiently on either a
11647 68020/68881 or a 68030 or a 68040. The generated code does use the
11648 68881 instructions that are emulated on the 68060.
11650 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11654 @opindex mhard-float
11656 Generate floating-point instructions. This is the default for 68020
11657 and above, and for ColdFire devices that have an FPU@. It defines the
11658 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11659 on ColdFire targets.
11662 @opindex msoft-float
11663 Do not generate floating-point instructions; use library calls instead.
11664 This is the default for 68000, 68010, and 68832 targets. It is also
11665 the default for ColdFire devices that have no FPU.
11671 Generate (do not generate) ColdFire hardware divide and remainder
11672 instructions. If @option{-march} is used without @option{-mcpu},
11673 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11674 architectures. Otherwise, the default is taken from the target CPU
11675 (either the default CPU, or the one specified by @option{-mcpu}). For
11676 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11677 @option{-mcpu=5206e}.
11679 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11683 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11684 Additionally, parameters passed on the stack are also aligned to a
11685 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11689 Do not consider type @code{int} to be 16 bits wide. This is the default.
11692 @itemx -mno-bitfield
11693 @opindex mnobitfield
11694 @opindex mno-bitfield
11695 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11696 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11700 Do use the bit-field instructions. The @option{-m68020} option implies
11701 @option{-mbitfield}. This is the default if you use a configuration
11702 designed for a 68020.
11706 Use a different function-calling convention, in which functions
11707 that take a fixed number of arguments return with the @code{rtd}
11708 instruction, which pops their arguments while returning. This
11709 saves one instruction in the caller since there is no need to pop
11710 the arguments there.
11712 This calling convention is incompatible with the one normally
11713 used on Unix, so you cannot use it if you need to call libraries
11714 compiled with the Unix compiler.
11716 Also, you must provide function prototypes for all functions that
11717 take variable numbers of arguments (including @code{printf});
11718 otherwise incorrect code will be generated for calls to those
11721 In addition, seriously incorrect code will result if you call a
11722 function with too many arguments. (Normally, extra arguments are
11723 harmlessly ignored.)
11725 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11726 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11730 Do not use the calling conventions selected by @option{-mrtd}.
11731 This is the default.
11734 @itemx -mno-align-int
11735 @opindex malign-int
11736 @opindex mno-align-int
11737 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11738 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11739 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11740 Aligning variables on 32-bit boundaries produces code that runs somewhat
11741 faster on processors with 32-bit busses at the expense of more memory.
11743 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11744 align structures containing the above types differently than
11745 most published application binary interface specifications for the m68k.
11749 Use the pc-relative addressing mode of the 68000 directly, instead of
11750 using a global offset table. At present, this option implies @option{-fpic},
11751 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11752 not presently supported with @option{-mpcrel}, though this could be supported for
11753 68020 and higher processors.
11755 @item -mno-strict-align
11756 @itemx -mstrict-align
11757 @opindex mno-strict-align
11758 @opindex mstrict-align
11759 Do not (do) assume that unaligned memory references will be handled by
11763 Generate code that allows the data segment to be located in a different
11764 area of memory from the text segment. This allows for execute in place in
11765 an environment without virtual memory management. This option implies
11768 @item -mno-sep-data
11769 Generate code that assumes that the data segment follows the text segment.
11770 This is the default.
11772 @item -mid-shared-library
11773 Generate code that supports shared libraries via the library ID method.
11774 This allows for execute in place and shared libraries in an environment
11775 without virtual memory management. This option implies @option{-fPIC}.
11777 @item -mno-id-shared-library
11778 Generate code that doesn't assume ID based shared libraries are being used.
11779 This is the default.
11781 @item -mshared-library-id=n
11782 Specified the identification number of the ID based shared library being
11783 compiled. Specifying a value of 0 will generate more compact code, specifying
11784 other values will force the allocation of that number to the current
11785 library but is no more space or time efficient than omitting this option.
11791 When generating position-independent code for ColdFire, generate code
11792 that works if the GOT has more than 8192 entries. This code is
11793 larger and slower than code generated without this option. On M680x0
11794 processors, this option is not needed; @option{-fPIC} suffices.
11796 GCC normally uses a single instruction to load values from the GOT@.
11797 While this is relatively efficient, it only works if the GOT
11798 is smaller than about 64k. Anything larger causes the linker
11799 to report an error such as:
11801 @cindex relocation truncated to fit (ColdFire)
11803 relocation truncated to fit: R_68K_GOT16O foobar
11806 If this happens, you should recompile your code with @option{-mxgot}.
11807 It should then work with very large GOTs. However, code generated with
11808 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
11809 the value of a global symbol.
11811 Note that some linkers, including newer versions of the GNU linker,
11812 can create multiple GOTs and sort GOT entries. If you have such a linker,
11813 you should only need to use @option{-mxgot} when compiling a single
11814 object file that accesses more than 8192 GOT entries. Very few do.
11816 These options have no effect unless GCC is generating
11817 position-independent code.
11821 @node M68hc1x Options
11822 @subsection M68hc1x Options
11823 @cindex M68hc1x options
11825 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11826 microcontrollers. The default values for these options depends on
11827 which style of microcontroller was selected when the compiler was configured;
11828 the defaults for the most common choices are given below.
11835 Generate output for a 68HC11. This is the default
11836 when the compiler is configured for 68HC11-based systems.
11842 Generate output for a 68HC12. This is the default
11843 when the compiler is configured for 68HC12-based systems.
11849 Generate output for a 68HCS12.
11851 @item -mauto-incdec
11852 @opindex mauto-incdec
11853 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11860 Enable the use of 68HC12 min and max instructions.
11863 @itemx -mno-long-calls
11864 @opindex mlong-calls
11865 @opindex mno-long-calls
11866 Treat all calls as being far away (near). If calls are assumed to be
11867 far away, the compiler will use the @code{call} instruction to
11868 call a function and the @code{rtc} instruction for returning.
11872 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11874 @item -msoft-reg-count=@var{count}
11875 @opindex msoft-reg-count
11876 Specify the number of pseudo-soft registers which are used for the
11877 code generation. The maximum number is 32. Using more pseudo-soft
11878 register may or may not result in better code depending on the program.
11879 The default is 4 for 68HC11 and 2 for 68HC12.
11883 @node MCore Options
11884 @subsection MCore Options
11885 @cindex MCore options
11887 These are the @samp{-m} options defined for the Motorola M*Core
11893 @itemx -mno-hardlit
11895 @opindex mno-hardlit
11896 Inline constants into the code stream if it can be done in two
11897 instructions or less.
11903 Use the divide instruction. (Enabled by default).
11905 @item -mrelax-immediate
11906 @itemx -mno-relax-immediate
11907 @opindex mrelax-immediate
11908 @opindex mno-relax-immediate
11909 Allow arbitrary sized immediates in bit operations.
11911 @item -mwide-bitfields
11912 @itemx -mno-wide-bitfields
11913 @opindex mwide-bitfields
11914 @opindex mno-wide-bitfields
11915 Always treat bit-fields as int-sized.
11917 @item -m4byte-functions
11918 @itemx -mno-4byte-functions
11919 @opindex m4byte-functions
11920 @opindex mno-4byte-functions
11921 Force all functions to be aligned to a four byte boundary.
11923 @item -mcallgraph-data
11924 @itemx -mno-callgraph-data
11925 @opindex mcallgraph-data
11926 @opindex mno-callgraph-data
11927 Emit callgraph information.
11930 @itemx -mno-slow-bytes
11931 @opindex mslow-bytes
11932 @opindex mno-slow-bytes
11933 Prefer word access when reading byte quantities.
11935 @item -mlittle-endian
11936 @itemx -mbig-endian
11937 @opindex mlittle-endian
11938 @opindex mbig-endian
11939 Generate code for a little endian target.
11945 Generate code for the 210 processor.
11949 @subsection MIPS Options
11950 @cindex MIPS options
11956 Generate big-endian code.
11960 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11963 @item -march=@var{arch}
11965 Generate code that will run on @var{arch}, which can be the name of a
11966 generic MIPS ISA, or the name of a particular processor.
11968 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11969 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
11970 The processor names are:
11971 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11972 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11973 @samp{5kc}, @samp{5kf},
11975 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11976 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11977 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11978 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11979 @samp{loongson2e}, @samp{loongson2f},
11982 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11983 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11984 @samp{rm7000}, @samp{rm9000},
11987 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11988 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
11990 The special value @samp{from-abi} selects the
11991 most compatible architecture for the selected ABI (that is,
11992 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11994 Native Linux/GNU toolchains also support the value @samp{native},
11995 which selects the best architecture option for the host processor.
11996 @option{-march=native} has no effect if GCC does not recognize
11999 In processor names, a final @samp{000} can be abbreviated as @samp{k}
12000 (for example, @samp{-march=r2k}). Prefixes are optional, and
12001 @samp{vr} may be written @samp{r}.
12003 Names of the form @samp{@var{n}f2_1} refer to processors with
12004 FPUs clocked at half the rate of the core, names of the form
12005 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
12006 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
12007 processors with FPUs clocked a ratio of 3:2 with respect to the core.
12008 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
12009 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
12010 accepted as synonyms for @samp{@var{n}f1_1}.
12012 GCC defines two macros based on the value of this option. The first
12013 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
12014 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
12015 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
12016 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
12017 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
12019 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
12020 above. In other words, it will have the full prefix and will not
12021 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
12022 the macro names the resolved architecture (either @samp{"mips1"} or
12023 @samp{"mips3"}). It names the default architecture when no
12024 @option{-march} option is given.
12026 @item -mtune=@var{arch}
12028 Optimize for @var{arch}. Among other things, this option controls
12029 the way instructions are scheduled, and the perceived cost of arithmetic
12030 operations. The list of @var{arch} values is the same as for
12033 When this option is not used, GCC will optimize for the processor
12034 specified by @option{-march}. By using @option{-march} and
12035 @option{-mtune} together, it is possible to generate code that will
12036 run on a family of processors, but optimize the code for one
12037 particular member of that family.
12039 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
12040 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
12041 @samp{-march} ones described above.
12045 Equivalent to @samp{-march=mips1}.
12049 Equivalent to @samp{-march=mips2}.
12053 Equivalent to @samp{-march=mips3}.
12057 Equivalent to @samp{-march=mips4}.
12061 Equivalent to @samp{-march=mips32}.
12065 Equivalent to @samp{-march=mips32r2}.
12069 Equivalent to @samp{-march=mips64}.
12073 Equivalent to @samp{-march=mips64r2}.
12078 @opindex mno-mips16
12079 Generate (do not generate) MIPS16 code. If GCC is targetting a
12080 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
12082 MIPS16 code generation can also be controlled on a per-function basis
12083 by means of @code{mips16} and @code{nomips16} attributes.
12084 @xref{Function Attributes}, for more information.
12086 @item -mflip-mips16
12087 @opindex mflip-mips16
12088 Generate MIPS16 code on alternating functions. This option is provided
12089 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
12090 not intended for ordinary use in compiling user code.
12092 @item -minterlink-mips16
12093 @itemx -mno-interlink-mips16
12094 @opindex minterlink-mips16
12095 @opindex mno-interlink-mips16
12096 Require (do not require) that non-MIPS16 code be link-compatible with
12099 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
12100 it must either use a call or an indirect jump. @option{-minterlink-mips16}
12101 therefore disables direct jumps unless GCC knows that the target of the
12102 jump is not MIPS16.
12114 Generate code for the given ABI@.
12116 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
12117 generates 64-bit code when you select a 64-bit architecture, but you
12118 can use @option{-mgp32} to get 32-bit code instead.
12120 For information about the O64 ABI, see
12121 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
12123 GCC supports a variant of the o32 ABI in which floating-point registers
12124 are 64 rather than 32 bits wide. You can select this combination with
12125 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
12126 and @samp{mfhc1} instructions and is therefore only supported for
12127 MIPS32R2 processors.
12129 The register assignments for arguments and return values remain the
12130 same, but each scalar value is passed in a single 64-bit register
12131 rather than a pair of 32-bit registers. For example, scalar
12132 floating-point values are returned in @samp{$f0} only, not a
12133 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
12134 remains the same, but all 64 bits are saved.
12137 @itemx -mno-abicalls
12139 @opindex mno-abicalls
12140 Generate (do not generate) code that is suitable for SVR4-style
12141 dynamic objects. @option{-mabicalls} is the default for SVR4-based
12146 Generate (do not generate) code that is fully position-independent,
12147 and that can therefore be linked into shared libraries. This option
12148 only affects @option{-mabicalls}.
12150 All @option{-mabicalls} code has traditionally been position-independent,
12151 regardless of options like @option{-fPIC} and @option{-fpic}. However,
12152 as an extension, the GNU toolchain allows executables to use absolute
12153 accesses for locally-binding symbols. It can also use shorter GP
12154 initialization sequences and generate direct calls to locally-defined
12155 functions. This mode is selected by @option{-mno-shared}.
12157 @option{-mno-shared} depends on binutils 2.16 or higher and generates
12158 objects that can only be linked by the GNU linker. However, the option
12159 does not affect the ABI of the final executable; it only affects the ABI
12160 of relocatable objects. Using @option{-mno-shared} will generally make
12161 executables both smaller and quicker.
12163 @option{-mshared} is the default.
12169 Lift (do not lift) the usual restrictions on the size of the global
12172 GCC normally uses a single instruction to load values from the GOT@.
12173 While this is relatively efficient, it will only work if the GOT
12174 is smaller than about 64k. Anything larger will cause the linker
12175 to report an error such as:
12177 @cindex relocation truncated to fit (MIPS)
12179 relocation truncated to fit: R_MIPS_GOT16 foobar
12182 If this happens, you should recompile your code with @option{-mxgot}.
12183 It should then work with very large GOTs, although it will also be
12184 less efficient, since it will take three instructions to fetch the
12185 value of a global symbol.
12187 Note that some linkers can create multiple GOTs. If you have such a
12188 linker, you should only need to use @option{-mxgot} when a single object
12189 file accesses more than 64k's worth of GOT entries. Very few do.
12191 These options have no effect unless GCC is generating position
12196 Assume that general-purpose registers are 32 bits wide.
12200 Assume that general-purpose registers are 64 bits wide.
12204 Assume that floating-point registers are 32 bits wide.
12208 Assume that floating-point registers are 64 bits wide.
12211 @opindex mhard-float
12212 Use floating-point coprocessor instructions.
12215 @opindex msoft-float
12216 Do not use floating-point coprocessor instructions. Implement
12217 floating-point calculations using library calls instead.
12219 @item -msingle-float
12220 @opindex msingle-float
12221 Assume that the floating-point coprocessor only supports single-precision
12224 @item -mdouble-float
12225 @opindex mdouble-float
12226 Assume that the floating-point coprocessor supports double-precision
12227 operations. This is the default.
12233 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
12234 implement atomic memory built-in functions. When neither option is
12235 specified, GCC will use the instructions if the target architecture
12238 @option{-mllsc} is useful if the runtime environment can emulate the
12239 instructions and @option{-mno-llsc} can be useful when compiling for
12240 nonstandard ISAs. You can make either option the default by
12241 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
12242 respectively. @option{--with-llsc} is the default for some
12243 configurations; see the installation documentation for details.
12249 Use (do not use) revision 1 of the MIPS DSP ASE@.
12250 @xref{MIPS DSP Built-in Functions}. This option defines the
12251 preprocessor macro @samp{__mips_dsp}. It also defines
12252 @samp{__mips_dsp_rev} to 1.
12258 Use (do not use) revision 2 of the MIPS DSP ASE@.
12259 @xref{MIPS DSP Built-in Functions}. This option defines the
12260 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
12261 It also defines @samp{__mips_dsp_rev} to 2.
12264 @itemx -mno-smartmips
12265 @opindex msmartmips
12266 @opindex mno-smartmips
12267 Use (do not use) the MIPS SmartMIPS ASE.
12269 @item -mpaired-single
12270 @itemx -mno-paired-single
12271 @opindex mpaired-single
12272 @opindex mno-paired-single
12273 Use (do not use) paired-single floating-point instructions.
12274 @xref{MIPS Paired-Single Support}. This option requires
12275 hardware floating-point support to be enabled.
12281 Use (do not use) MIPS Digital Media Extension instructions.
12282 This option can only be used when generating 64-bit code and requires
12283 hardware floating-point support to be enabled.
12288 @opindex mno-mips3d
12289 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
12290 The option @option{-mips3d} implies @option{-mpaired-single}.
12296 Use (do not use) MT Multithreading instructions.
12300 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
12301 an explanation of the default and the way that the pointer size is
12306 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12308 The default size of @code{int}s, @code{long}s and pointers depends on
12309 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
12310 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12311 32-bit @code{long}s. Pointers are the same size as @code{long}s,
12312 or the same size as integer registers, whichever is smaller.
12318 Assume (do not assume) that all symbols have 32-bit values, regardless
12319 of the selected ABI@. This option is useful in combination with
12320 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12321 to generate shorter and faster references to symbolic addresses.
12325 Put definitions of externally-visible data in a small data section
12326 if that data is no bigger than @var{num} bytes. GCC can then access
12327 the data more efficiently; see @option{-mgpopt} for details.
12329 The default @option{-G} option depends on the configuration.
12331 @item -mlocal-sdata
12332 @itemx -mno-local-sdata
12333 @opindex mlocal-sdata
12334 @opindex mno-local-sdata
12335 Extend (do not extend) the @option{-G} behavior to local data too,
12336 such as to static variables in C@. @option{-mlocal-sdata} is the
12337 default for all configurations.
12339 If the linker complains that an application is using too much small data,
12340 you might want to try rebuilding the less performance-critical parts with
12341 @option{-mno-local-sdata}. You might also want to build large
12342 libraries with @option{-mno-local-sdata}, so that the libraries leave
12343 more room for the main program.
12345 @item -mextern-sdata
12346 @itemx -mno-extern-sdata
12347 @opindex mextern-sdata
12348 @opindex mno-extern-sdata
12349 Assume (do not assume) that externally-defined data will be in
12350 a small data section if that data is within the @option{-G} limit.
12351 @option{-mextern-sdata} is the default for all configurations.
12353 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12354 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12355 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12356 is placed in a small data section. If @var{Var} is defined by another
12357 module, you must either compile that module with a high-enough
12358 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12359 definition. If @var{Var} is common, you must link the application
12360 with a high-enough @option{-G} setting.
12362 The easiest way of satisfying these restrictions is to compile
12363 and link every module with the same @option{-G} option. However,
12364 you may wish to build a library that supports several different
12365 small data limits. You can do this by compiling the library with
12366 the highest supported @option{-G} setting and additionally using
12367 @option{-mno-extern-sdata} to stop the library from making assumptions
12368 about externally-defined data.
12374 Use (do not use) GP-relative accesses for symbols that are known to be
12375 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12376 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12379 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12380 might not hold the value of @code{_gp}. For example, if the code is
12381 part of a library that might be used in a boot monitor, programs that
12382 call boot monitor routines will pass an unknown value in @code{$gp}.
12383 (In such situations, the boot monitor itself would usually be compiled
12384 with @option{-G0}.)
12386 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12387 @option{-mno-extern-sdata}.
12389 @item -membedded-data
12390 @itemx -mno-embedded-data
12391 @opindex membedded-data
12392 @opindex mno-embedded-data
12393 Allocate variables to the read-only data section first if possible, then
12394 next in the small data section if possible, otherwise in data. This gives
12395 slightly slower code than the default, but reduces the amount of RAM required
12396 when executing, and thus may be preferred for some embedded systems.
12398 @item -muninit-const-in-rodata
12399 @itemx -mno-uninit-const-in-rodata
12400 @opindex muninit-const-in-rodata
12401 @opindex mno-uninit-const-in-rodata
12402 Put uninitialized @code{const} variables in the read-only data section.
12403 This option is only meaningful in conjunction with @option{-membedded-data}.
12405 @item -mcode-readable=@var{setting}
12406 @opindex mcode-readable
12407 Specify whether GCC may generate code that reads from executable sections.
12408 There are three possible settings:
12411 @item -mcode-readable=yes
12412 Instructions may freely access executable sections. This is the
12415 @item -mcode-readable=pcrel
12416 MIPS16 PC-relative load instructions can access executable sections,
12417 but other instructions must not do so. This option is useful on 4KSc
12418 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12419 It is also useful on processors that can be configured to have a dual
12420 instruction/data SRAM interface and that, like the M4K, automatically
12421 redirect PC-relative loads to the instruction RAM.
12423 @item -mcode-readable=no
12424 Instructions must not access executable sections. This option can be
12425 useful on targets that are configured to have a dual instruction/data
12426 SRAM interface but that (unlike the M4K) do not automatically redirect
12427 PC-relative loads to the instruction RAM.
12430 @item -msplit-addresses
12431 @itemx -mno-split-addresses
12432 @opindex msplit-addresses
12433 @opindex mno-split-addresses
12434 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12435 relocation operators. This option has been superseded by
12436 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12438 @item -mexplicit-relocs
12439 @itemx -mno-explicit-relocs
12440 @opindex mexplicit-relocs
12441 @opindex mno-explicit-relocs
12442 Use (do not use) assembler relocation operators when dealing with symbolic
12443 addresses. The alternative, selected by @option{-mno-explicit-relocs},
12444 is to use assembler macros instead.
12446 @option{-mexplicit-relocs} is the default if GCC was configured
12447 to use an assembler that supports relocation operators.
12449 @item -mcheck-zero-division
12450 @itemx -mno-check-zero-division
12451 @opindex mcheck-zero-division
12452 @opindex mno-check-zero-division
12453 Trap (do not trap) on integer division by zero.
12455 The default is @option{-mcheck-zero-division}.
12457 @item -mdivide-traps
12458 @itemx -mdivide-breaks
12459 @opindex mdivide-traps
12460 @opindex mdivide-breaks
12461 MIPS systems check for division by zero by generating either a
12462 conditional trap or a break instruction. Using traps results in
12463 smaller code, but is only supported on MIPS II and later. Also, some
12464 versions of the Linux kernel have a bug that prevents trap from
12465 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
12466 allow conditional traps on architectures that support them and
12467 @option{-mdivide-breaks} to force the use of breaks.
12469 The default is usually @option{-mdivide-traps}, but this can be
12470 overridden at configure time using @option{--with-divide=breaks}.
12471 Divide-by-zero checks can be completely disabled using
12472 @option{-mno-check-zero-division}.
12477 @opindex mno-memcpy
12478 Force (do not force) the use of @code{memcpy()} for non-trivial block
12479 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
12480 most constant-sized copies.
12483 @itemx -mno-long-calls
12484 @opindex mlong-calls
12485 @opindex mno-long-calls
12486 Disable (do not disable) use of the @code{jal} instruction. Calling
12487 functions using @code{jal} is more efficient but requires the caller
12488 and callee to be in the same 256 megabyte segment.
12490 This option has no effect on abicalls code. The default is
12491 @option{-mno-long-calls}.
12497 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
12498 instructions, as provided by the R4650 ISA@.
12501 @itemx -mno-fused-madd
12502 @opindex mfused-madd
12503 @opindex mno-fused-madd
12504 Enable (disable) use of the floating point multiply-accumulate
12505 instructions, when they are available. The default is
12506 @option{-mfused-madd}.
12508 When multiply-accumulate instructions are used, the intermediate
12509 product is calculated to infinite precision and is not subject to
12510 the FCSR Flush to Zero bit. This may be undesirable in some
12515 Tell the MIPS assembler to not run its preprocessor over user
12516 assembler files (with a @samp{.s} suffix) when assembling them.
12519 @itemx -mno-fix-r4000
12520 @opindex mfix-r4000
12521 @opindex mno-fix-r4000
12522 Work around certain R4000 CPU errata:
12525 A double-word or a variable shift may give an incorrect result if executed
12526 immediately after starting an integer division.
12528 A double-word or a variable shift may give an incorrect result if executed
12529 while an integer multiplication is in progress.
12531 An integer division may give an incorrect result if started in a delay slot
12532 of a taken branch or a jump.
12536 @itemx -mno-fix-r4400
12537 @opindex mfix-r4400
12538 @opindex mno-fix-r4400
12539 Work around certain R4400 CPU errata:
12542 A double-word or a variable shift may give an incorrect result if executed
12543 immediately after starting an integer division.
12547 @itemx -mno-fix-vr4120
12548 @opindex mfix-vr4120
12549 Work around certain VR4120 errata:
12552 @code{dmultu} does not always produce the correct result.
12554 @code{div} and @code{ddiv} do not always produce the correct result if one
12555 of the operands is negative.
12557 The workarounds for the division errata rely on special functions in
12558 @file{libgcc.a}. At present, these functions are only provided by
12559 the @code{mips64vr*-elf} configurations.
12561 Other VR4120 errata require a nop to be inserted between certain pairs of
12562 instructions. These errata are handled by the assembler, not by GCC itself.
12565 @opindex mfix-vr4130
12566 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
12567 workarounds are implemented by the assembler rather than by GCC,
12568 although GCC will avoid using @code{mflo} and @code{mfhi} if the
12569 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
12570 instructions are available instead.
12573 @itemx -mno-fix-sb1
12575 Work around certain SB-1 CPU core errata.
12576 (This flag currently works around the SB-1 revision 2
12577 ``F1'' and ``F2'' floating point errata.)
12579 @item -mflush-func=@var{func}
12580 @itemx -mno-flush-func
12581 @opindex mflush-func
12582 Specifies the function to call to flush the I and D caches, or to not
12583 call any such function. If called, the function must take the same
12584 arguments as the common @code{_flush_func()}, that is, the address of the
12585 memory range for which the cache is being flushed, the size of the
12586 memory range, and the number 3 (to flush both caches). The default
12587 depends on the target GCC was configured for, but commonly is either
12588 @samp{_flush_func} or @samp{__cpu_flush}.
12590 @item mbranch-cost=@var{num}
12591 @opindex mbranch-cost
12592 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12593 This cost is only a heuristic and is not guaranteed to produce
12594 consistent results across releases. A zero cost redundantly selects
12595 the default, which is based on the @option{-mtune} setting.
12597 @item -mbranch-likely
12598 @itemx -mno-branch-likely
12599 @opindex mbranch-likely
12600 @opindex mno-branch-likely
12601 Enable or disable use of Branch Likely instructions, regardless of the
12602 default for the selected architecture. By default, Branch Likely
12603 instructions may be generated if they are supported by the selected
12604 architecture. An exception is for the MIPS32 and MIPS64 architectures
12605 and processors which implement those architectures; for those, Branch
12606 Likely instructions will not be generated by default because the MIPS32
12607 and MIPS64 architectures specifically deprecate their use.
12609 @item -mfp-exceptions
12610 @itemx -mno-fp-exceptions
12611 @opindex mfp-exceptions
12612 Specifies whether FP exceptions are enabled. This affects how we schedule
12613 FP instructions for some processors. The default is that FP exceptions are
12616 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12617 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12620 @item -mvr4130-align
12621 @itemx -mno-vr4130-align
12622 @opindex mvr4130-align
12623 The VR4130 pipeline is two-way superscalar, but can only issue two
12624 instructions together if the first one is 8-byte aligned. When this
12625 option is enabled, GCC will align pairs of instructions that it
12626 thinks should execute in parallel.
12628 This option only has an effect when optimizing for the VR4130.
12629 It normally makes code faster, but at the expense of making it bigger.
12630 It is enabled by default at optimization level @option{-O3}.
12634 @subsection MMIX Options
12635 @cindex MMIX Options
12637 These options are defined for the MMIX:
12641 @itemx -mno-libfuncs
12643 @opindex mno-libfuncs
12644 Specify that intrinsic library functions are being compiled, passing all
12645 values in registers, no matter the size.
12648 @itemx -mno-epsilon
12650 @opindex mno-epsilon
12651 Generate floating-point comparison instructions that compare with respect
12652 to the @code{rE} epsilon register.
12654 @item -mabi=mmixware
12656 @opindex mabi-mmixware
12658 Generate code that passes function parameters and return values that (in
12659 the called function) are seen as registers @code{$0} and up, as opposed to
12660 the GNU ABI which uses global registers @code{$231} and up.
12662 @item -mzero-extend
12663 @itemx -mno-zero-extend
12664 @opindex mzero-extend
12665 @opindex mno-zero-extend
12666 When reading data from memory in sizes shorter than 64 bits, use (do not
12667 use) zero-extending load instructions by default, rather than
12668 sign-extending ones.
12671 @itemx -mno-knuthdiv
12673 @opindex mno-knuthdiv
12674 Make the result of a division yielding a remainder have the same sign as
12675 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12676 remainder follows the sign of the dividend. Both methods are
12677 arithmetically valid, the latter being almost exclusively used.
12679 @item -mtoplevel-symbols
12680 @itemx -mno-toplevel-symbols
12681 @opindex mtoplevel-symbols
12682 @opindex mno-toplevel-symbols
12683 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12684 code can be used with the @code{PREFIX} assembly directive.
12688 Generate an executable in the ELF format, rather than the default
12689 @samp{mmo} format used by the @command{mmix} simulator.
12691 @item -mbranch-predict
12692 @itemx -mno-branch-predict
12693 @opindex mbranch-predict
12694 @opindex mno-branch-predict
12695 Use (do not use) the probable-branch instructions, when static branch
12696 prediction indicates a probable branch.
12698 @item -mbase-addresses
12699 @itemx -mno-base-addresses
12700 @opindex mbase-addresses
12701 @opindex mno-base-addresses
12702 Generate (do not generate) code that uses @emph{base addresses}. Using a
12703 base address automatically generates a request (handled by the assembler
12704 and the linker) for a constant to be set up in a global register. The
12705 register is used for one or more base address requests within the range 0
12706 to 255 from the value held in the register. The generally leads to short
12707 and fast code, but the number of different data items that can be
12708 addressed is limited. This means that a program that uses lots of static
12709 data may require @option{-mno-base-addresses}.
12711 @item -msingle-exit
12712 @itemx -mno-single-exit
12713 @opindex msingle-exit
12714 @opindex mno-single-exit
12715 Force (do not force) generated code to have a single exit point in each
12719 @node MN10300 Options
12720 @subsection MN10300 Options
12721 @cindex MN10300 options
12723 These @option{-m} options are defined for Matsushita MN10300 architectures:
12728 Generate code to avoid bugs in the multiply instructions for the MN10300
12729 processors. This is the default.
12731 @item -mno-mult-bug
12732 @opindex mno-mult-bug
12733 Do not generate code to avoid bugs in the multiply instructions for the
12734 MN10300 processors.
12738 Generate code which uses features specific to the AM33 processor.
12742 Do not generate code which uses features specific to the AM33 processor. This
12745 @item -mreturn-pointer-on-d0
12746 @opindex mreturn-pointer-on-d0
12747 When generating a function which returns a pointer, return the pointer
12748 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12749 only in a0, and attempts to call such functions without a prototype
12750 would result in errors. Note that this option is on by default; use
12751 @option{-mno-return-pointer-on-d0} to disable it.
12755 Do not link in the C run-time initialization object file.
12759 Indicate to the linker that it should perform a relaxation optimization pass
12760 to shorten branches, calls and absolute memory addresses. This option only
12761 has an effect when used on the command line for the final link step.
12763 This option makes symbolic debugging impossible.
12766 @node PDP-11 Options
12767 @subsection PDP-11 Options
12768 @cindex PDP-11 Options
12770 These options are defined for the PDP-11:
12775 Use hardware FPP floating point. This is the default. (FIS floating
12776 point on the PDP-11/40 is not supported.)
12779 @opindex msoft-float
12780 Do not use hardware floating point.
12784 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12788 Return floating-point results in memory. This is the default.
12792 Generate code for a PDP-11/40.
12796 Generate code for a PDP-11/45. This is the default.
12800 Generate code for a PDP-11/10.
12802 @item -mbcopy-builtin
12803 @opindex bcopy-builtin
12804 Use inline @code{movmemhi} patterns for copying memory. This is the
12809 Do not use inline @code{movmemhi} patterns for copying memory.
12815 Use 16-bit @code{int}. This is the default.
12821 Use 32-bit @code{int}.
12824 @itemx -mno-float32
12826 @opindex mno-float32
12827 Use 64-bit @code{float}. This is the default.
12830 @itemx -mno-float64
12832 @opindex mno-float64
12833 Use 32-bit @code{float}.
12837 Use @code{abshi2} pattern. This is the default.
12841 Do not use @code{abshi2} pattern.
12843 @item -mbranch-expensive
12844 @opindex mbranch-expensive
12845 Pretend that branches are expensive. This is for experimenting with
12846 code generation only.
12848 @item -mbranch-cheap
12849 @opindex mbranch-cheap
12850 Do not pretend that branches are expensive. This is the default.
12854 Generate code for a system with split I&D@.
12858 Generate code for a system without split I&D@. This is the default.
12862 Use Unix assembler syntax. This is the default when configured for
12863 @samp{pdp11-*-bsd}.
12867 Use DEC assembler syntax. This is the default when configured for any
12868 PDP-11 target other than @samp{pdp11-*-bsd}.
12871 @node PowerPC Options
12872 @subsection PowerPC Options
12873 @cindex PowerPC options
12875 These are listed under @xref{RS/6000 and PowerPC Options}.
12877 @node RS/6000 and PowerPC Options
12878 @subsection IBM RS/6000 and PowerPC Options
12879 @cindex RS/6000 and PowerPC Options
12880 @cindex IBM RS/6000 and PowerPC Options
12882 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12889 @itemx -mno-powerpc
12890 @itemx -mpowerpc-gpopt
12891 @itemx -mno-powerpc-gpopt
12892 @itemx -mpowerpc-gfxopt
12893 @itemx -mno-powerpc-gfxopt
12895 @itemx -mno-powerpc64
12899 @itemx -mno-popcntb
12907 @itemx -mno-hard-dfp
12911 @opindex mno-power2
12913 @opindex mno-powerpc
12914 @opindex mpowerpc-gpopt
12915 @opindex mno-powerpc-gpopt
12916 @opindex mpowerpc-gfxopt
12917 @opindex mno-powerpc-gfxopt
12918 @opindex mpowerpc64
12919 @opindex mno-powerpc64
12923 @opindex mno-popcntb
12929 @opindex mno-mfpgpr
12931 @opindex mno-hard-dfp
12932 GCC supports two related instruction set architectures for the
12933 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12934 instructions supported by the @samp{rios} chip set used in the original
12935 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12936 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12937 the IBM 4xx, 6xx, and follow-on microprocessors.
12939 Neither architecture is a subset of the other. However there is a
12940 large common subset of instructions supported by both. An MQ
12941 register is included in processors supporting the POWER architecture.
12943 You use these options to specify which instructions are available on the
12944 processor you are using. The default value of these options is
12945 determined when configuring GCC@. Specifying the
12946 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12947 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12948 rather than the options listed above.
12950 The @option{-mpower} option allows GCC to generate instructions that
12951 are found only in the POWER architecture and to use the MQ register.
12952 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12953 to generate instructions that are present in the POWER2 architecture but
12954 not the original POWER architecture.
12956 The @option{-mpowerpc} option allows GCC to generate instructions that
12957 are found only in the 32-bit subset of the PowerPC architecture.
12958 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12959 GCC to use the optional PowerPC architecture instructions in the
12960 General Purpose group, including floating-point square root. Specifying
12961 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12962 use the optional PowerPC architecture instructions in the Graphics
12963 group, including floating-point select.
12965 The @option{-mmfcrf} option allows GCC to generate the move from
12966 condition register field instruction implemented on the POWER4
12967 processor and other processors that support the PowerPC V2.01
12969 The @option{-mpopcntb} option allows GCC to generate the popcount and
12970 double precision FP reciprocal estimate instruction implemented on the
12971 POWER5 processor and other processors that support the PowerPC V2.02
12973 The @option{-mfprnd} option allows GCC to generate the FP round to
12974 integer instructions implemented on the POWER5+ processor and other
12975 processors that support the PowerPC V2.03 architecture.
12976 The @option{-mcmpb} option allows GCC to generate the compare bytes
12977 instruction implemented on the POWER6 processor and other processors
12978 that support the PowerPC V2.05 architecture.
12979 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12980 general purpose register instructions implemented on the POWER6X
12981 processor and other processors that support the extended PowerPC V2.05
12983 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
12984 point instructions implemented on some POWER processors.
12986 The @option{-mpowerpc64} option allows GCC to generate the additional
12987 64-bit instructions that are found in the full PowerPC64 architecture
12988 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12989 @option{-mno-powerpc64}.
12991 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12992 will use only the instructions in the common subset of both
12993 architectures plus some special AIX common-mode calls, and will not use
12994 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12995 permits GCC to use any instruction from either architecture and to
12996 allow use of the MQ register; specify this for the Motorola MPC601.
12998 @item -mnew-mnemonics
12999 @itemx -mold-mnemonics
13000 @opindex mnew-mnemonics
13001 @opindex mold-mnemonics
13002 Select which mnemonics to use in the generated assembler code. With
13003 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
13004 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
13005 assembler mnemonics defined for the POWER architecture. Instructions
13006 defined in only one architecture have only one mnemonic; GCC uses that
13007 mnemonic irrespective of which of these options is specified.
13009 GCC defaults to the mnemonics appropriate for the architecture in
13010 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
13011 value of these option. Unless you are building a cross-compiler, you
13012 should normally not specify either @option{-mnew-mnemonics} or
13013 @option{-mold-mnemonics}, but should instead accept the default.
13015 @item -mcpu=@var{cpu_type}
13017 Set architecture type, register usage, choice of mnemonics, and
13018 instruction scheduling parameters for machine type @var{cpu_type}.
13019 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
13020 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
13021 @samp{505}, @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
13022 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
13023 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
13024 @samp{860}, @samp{970}, @samp{8540}, @samp{e300c2}, @samp{e300c3},
13025 @samp{e500mc}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
13026 @samp{power}, @samp{power2}, @samp{power3}, @samp{power4},
13027 @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x}, @samp{power7}
13028 @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
13029 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
13031 @option{-mcpu=common} selects a completely generic processor. Code
13032 generated under this option will run on any POWER or PowerPC processor.
13033 GCC will use only the instructions in the common subset of both
13034 architectures, and will not use the MQ register. GCC assumes a generic
13035 processor model for scheduling purposes.
13037 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
13038 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
13039 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
13040 types, with an appropriate, generic processor model assumed for
13041 scheduling purposes.
13043 The other options specify a specific processor. Code generated under
13044 those options will run best on that processor, and may not run at all on
13047 The @option{-mcpu} options automatically enable or disable the
13050 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
13051 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
13052 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
13054 The particular options set for any particular CPU will vary between
13055 compiler versions, depending on what setting seems to produce optimal
13056 code for that CPU; it doesn't necessarily reflect the actual hardware's
13057 capabilities. If you wish to set an individual option to a particular
13058 value, you may specify it after the @option{-mcpu} option, like
13059 @samp{-mcpu=970 -mno-altivec}.
13061 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
13062 not enabled or disabled by the @option{-mcpu} option at present because
13063 AIX does not have full support for these options. You may still
13064 enable or disable them individually if you're sure it'll work in your
13067 @item -mtune=@var{cpu_type}
13069 Set the instruction scheduling parameters for machine type
13070 @var{cpu_type}, but do not set the architecture type, register usage, or
13071 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
13072 values for @var{cpu_type} are used for @option{-mtune} as for
13073 @option{-mcpu}. If both are specified, the code generated will use the
13074 architecture, registers, and mnemonics set by @option{-mcpu}, but the
13075 scheduling parameters set by @option{-mtune}.
13081 Generate code to compute division as reciprocal estimate and iterative
13082 refinement, creating opportunities for increased throughput. This
13083 feature requires: optional PowerPC Graphics instruction set for single
13084 precision and FRE instruction for double precision, assuming divides
13085 cannot generate user-visible traps, and the domain values not include
13086 Infinities, denormals or zero denominator.
13089 @itemx -mno-altivec
13091 @opindex mno-altivec
13092 Generate code that uses (does not use) AltiVec instructions, and also
13093 enable the use of built-in functions that allow more direct access to
13094 the AltiVec instruction set. You may also need to set
13095 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
13101 @opindex mno-vrsave
13102 Generate VRSAVE instructions when generating AltiVec code.
13105 @opindex msecure-plt
13106 Generate code that allows ld and ld.so to build executables and shared
13107 libraries with non-exec .plt and .got sections. This is a PowerPC
13108 32-bit SYSV ABI option.
13112 Generate code that uses a BSS .plt section that ld.so fills in, and
13113 requires .plt and .got sections that are both writable and executable.
13114 This is a PowerPC 32-bit SYSV ABI option.
13120 This switch enables or disables the generation of ISEL instructions.
13122 @item -misel=@var{yes/no}
13123 This switch has been deprecated. Use @option{-misel} and
13124 @option{-mno-isel} instead.
13130 This switch enables or disables the generation of SPE simd
13136 @opindex mno-paired
13137 This switch enables or disables the generation of PAIRED simd
13140 @item -mspe=@var{yes/no}
13141 This option has been deprecated. Use @option{-mspe} and
13142 @option{-mno-spe} instead.
13144 @item -mfloat-gprs=@var{yes/single/double/no}
13145 @itemx -mfloat-gprs
13146 @opindex mfloat-gprs
13147 This switch enables or disables the generation of floating point
13148 operations on the general purpose registers for architectures that
13151 The argument @var{yes} or @var{single} enables the use of
13152 single-precision floating point operations.
13154 The argument @var{double} enables the use of single and
13155 double-precision floating point operations.
13157 The argument @var{no} disables floating point operations on the
13158 general purpose registers.
13160 This option is currently only available on the MPC854x.
13166 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
13167 targets (including GNU/Linux). The 32-bit environment sets int, long
13168 and pointer to 32 bits and generates code that runs on any PowerPC
13169 variant. The 64-bit environment sets int to 32 bits and long and
13170 pointer to 64 bits, and generates code for PowerPC64, as for
13171 @option{-mpowerpc64}.
13174 @itemx -mno-fp-in-toc
13175 @itemx -mno-sum-in-toc
13176 @itemx -mminimal-toc
13178 @opindex mno-fp-in-toc
13179 @opindex mno-sum-in-toc
13180 @opindex mminimal-toc
13181 Modify generation of the TOC (Table Of Contents), which is created for
13182 every executable file. The @option{-mfull-toc} option is selected by
13183 default. In that case, GCC will allocate at least one TOC entry for
13184 each unique non-automatic variable reference in your program. GCC
13185 will also place floating-point constants in the TOC@. However, only
13186 16,384 entries are available in the TOC@.
13188 If you receive a linker error message that saying you have overflowed
13189 the available TOC space, you can reduce the amount of TOC space used
13190 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
13191 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
13192 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
13193 generate code to calculate the sum of an address and a constant at
13194 run-time instead of putting that sum into the TOC@. You may specify one
13195 or both of these options. Each causes GCC to produce very slightly
13196 slower and larger code at the expense of conserving TOC space.
13198 If you still run out of space in the TOC even when you specify both of
13199 these options, specify @option{-mminimal-toc} instead. This option causes
13200 GCC to make only one TOC entry for every file. When you specify this
13201 option, GCC will produce code that is slower and larger but which
13202 uses extremely little TOC space. You may wish to use this option
13203 only on files that contain less frequently executed code.
13209 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
13210 @code{long} type, and the infrastructure needed to support them.
13211 Specifying @option{-maix64} implies @option{-mpowerpc64} and
13212 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
13213 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
13216 @itemx -mno-xl-compat
13217 @opindex mxl-compat
13218 @opindex mno-xl-compat
13219 Produce code that conforms more closely to IBM XL compiler semantics
13220 when using AIX-compatible ABI@. Pass floating-point arguments to
13221 prototyped functions beyond the register save area (RSA) on the stack
13222 in addition to argument FPRs. Do not assume that most significant
13223 double in 128-bit long double value is properly rounded when comparing
13224 values and converting to double. Use XL symbol names for long double
13227 The AIX calling convention was extended but not initially documented to
13228 handle an obscure K&R C case of calling a function that takes the
13229 address of its arguments with fewer arguments than declared. IBM XL
13230 compilers access floating point arguments which do not fit in the
13231 RSA from the stack when a subroutine is compiled without
13232 optimization. Because always storing floating-point arguments on the
13233 stack is inefficient and rarely needed, this option is not enabled by
13234 default and only is necessary when calling subroutines compiled by IBM
13235 XL compilers without optimization.
13239 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
13240 application written to use message passing with special startup code to
13241 enable the application to run. The system must have PE installed in the
13242 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13243 must be overridden with the @option{-specs=} option to specify the
13244 appropriate directory location. The Parallel Environment does not
13245 support threads, so the @option{-mpe} option and the @option{-pthread}
13246 option are incompatible.
13248 @item -malign-natural
13249 @itemx -malign-power
13250 @opindex malign-natural
13251 @opindex malign-power
13252 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13253 @option{-malign-natural} overrides the ABI-defined alignment of larger
13254 types, such as floating-point doubles, on their natural size-based boundary.
13255 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13256 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
13258 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13262 @itemx -mhard-float
13263 @opindex msoft-float
13264 @opindex mhard-float
13265 Generate code that does not use (uses) the floating-point register set.
13266 Software floating point emulation is provided if you use the
13267 @option{-msoft-float} option, and pass the option to GCC when linking.
13270 @itemx -mno-multiple
13272 @opindex mno-multiple
13273 Generate code that uses (does not use) the load multiple word
13274 instructions and the store multiple word instructions. These
13275 instructions are generated by default on POWER systems, and not
13276 generated on PowerPC systems. Do not use @option{-mmultiple} on little
13277 endian PowerPC systems, since those instructions do not work when the
13278 processor is in little endian mode. The exceptions are PPC740 and
13279 PPC750 which permit the instructions usage in little endian mode.
13284 @opindex mno-string
13285 Generate code that uses (does not use) the load string instructions
13286 and the store string word instructions to save multiple registers and
13287 do small block moves. These instructions are generated by default on
13288 POWER systems, and not generated on PowerPC systems. Do not use
13289 @option{-mstring} on little endian PowerPC systems, since those
13290 instructions do not work when the processor is in little endian mode.
13291 The exceptions are PPC740 and PPC750 which permit the instructions
13292 usage in little endian mode.
13297 @opindex mno-update
13298 Generate code that uses (does not use) the load or store instructions
13299 that update the base register to the address of the calculated memory
13300 location. These instructions are generated by default. If you use
13301 @option{-mno-update}, there is a small window between the time that the
13302 stack pointer is updated and the address of the previous frame is
13303 stored, which means code that walks the stack frame across interrupts or
13304 signals may get corrupted data.
13307 @itemx -mno-fused-madd
13308 @opindex mfused-madd
13309 @opindex mno-fused-madd
13310 Generate code that uses (does not use) the floating point multiply and
13311 accumulate instructions. These instructions are generated by default if
13312 hardware floating is used.
13318 Generate code that uses (does not use) the half-word multiply and
13319 multiply-accumulate instructions on the IBM 405, 440 and 464 processors.
13320 These instructions are generated by default when targetting those
13327 Generate code that uses (does not use) the string-search @samp{dlmzb}
13328 instruction on the IBM 405, 440 and 464 processors. This instruction is
13329 generated by default when targetting those processors.
13331 @item -mno-bit-align
13333 @opindex mno-bit-align
13334 @opindex mbit-align
13335 On System V.4 and embedded PowerPC systems do not (do) force structures
13336 and unions that contain bit-fields to be aligned to the base type of the
13339 For example, by default a structure containing nothing but 8
13340 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
13341 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
13342 the structure would be aligned to a 1 byte boundary and be one byte in
13345 @item -mno-strict-align
13346 @itemx -mstrict-align
13347 @opindex mno-strict-align
13348 @opindex mstrict-align
13349 On System V.4 and embedded PowerPC systems do not (do) assume that
13350 unaligned memory references will be handled by the system.
13352 @item -mrelocatable
13353 @itemx -mno-relocatable
13354 @opindex mrelocatable
13355 @opindex mno-relocatable
13356 On embedded PowerPC systems generate code that allows (does not allow)
13357 the program to be relocated to a different address at runtime. If you
13358 use @option{-mrelocatable} on any module, all objects linked together must
13359 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
13361 @item -mrelocatable-lib
13362 @itemx -mno-relocatable-lib
13363 @opindex mrelocatable-lib
13364 @opindex mno-relocatable-lib
13365 On embedded PowerPC systems generate code that allows (does not allow)
13366 the program to be relocated to a different address at runtime. Modules
13367 compiled with @option{-mrelocatable-lib} can be linked with either modules
13368 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
13369 with modules compiled with the @option{-mrelocatable} options.
13375 On System V.4 and embedded PowerPC systems do not (do) assume that
13376 register 2 contains a pointer to a global area pointing to the addresses
13377 used in the program.
13380 @itemx -mlittle-endian
13382 @opindex mlittle-endian
13383 On System V.4 and embedded PowerPC systems compile code for the
13384 processor in little endian mode. The @option{-mlittle-endian} option is
13385 the same as @option{-mlittle}.
13388 @itemx -mbig-endian
13390 @opindex mbig-endian
13391 On System V.4 and embedded PowerPC systems compile code for the
13392 processor in big endian mode. The @option{-mbig-endian} option is
13393 the same as @option{-mbig}.
13395 @item -mdynamic-no-pic
13396 @opindex mdynamic-no-pic
13397 On Darwin and Mac OS X systems, compile code so that it is not
13398 relocatable, but that its external references are relocatable. The
13399 resulting code is suitable for applications, but not shared
13402 @item -mprioritize-restricted-insns=@var{priority}
13403 @opindex mprioritize-restricted-insns
13404 This option controls the priority that is assigned to
13405 dispatch-slot restricted instructions during the second scheduling
13406 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
13407 @var{no/highest/second-highest} priority to dispatch slot restricted
13410 @item -msched-costly-dep=@var{dependence_type}
13411 @opindex msched-costly-dep
13412 This option controls which dependences are considered costly
13413 by the target during instruction scheduling. The argument
13414 @var{dependence_type} takes one of the following values:
13415 @var{no}: no dependence is costly,
13416 @var{all}: all dependences are costly,
13417 @var{true_store_to_load}: a true dependence from store to load is costly,
13418 @var{store_to_load}: any dependence from store to load is costly,
13419 @var{number}: any dependence which latency >= @var{number} is costly.
13421 @item -minsert-sched-nops=@var{scheme}
13422 @opindex minsert-sched-nops
13423 This option controls which nop insertion scheme will be used during
13424 the second scheduling pass. The argument @var{scheme} takes one of the
13426 @var{no}: Don't insert nops.
13427 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
13428 according to the scheduler's grouping.
13429 @var{regroup_exact}: Insert nops to force costly dependent insns into
13430 separate groups. Insert exactly as many nops as needed to force an insn
13431 to a new group, according to the estimated processor grouping.
13432 @var{number}: Insert nops to force costly dependent insns into
13433 separate groups. Insert @var{number} nops to force an insn to a new group.
13436 @opindex mcall-sysv
13437 On System V.4 and embedded PowerPC systems compile code using calling
13438 conventions that adheres to the March 1995 draft of the System V
13439 Application Binary Interface, PowerPC processor supplement. This is the
13440 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
13442 @item -mcall-sysv-eabi
13443 @opindex mcall-sysv-eabi
13444 Specify both @option{-mcall-sysv} and @option{-meabi} options.
13446 @item -mcall-sysv-noeabi
13447 @opindex mcall-sysv-noeabi
13448 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
13450 @item -mcall-solaris
13451 @opindex mcall-solaris
13452 On System V.4 and embedded PowerPC systems compile code for the Solaris
13456 @opindex mcall-linux
13457 On System V.4 and embedded PowerPC systems compile code for the
13458 Linux-based GNU system.
13462 On System V.4 and embedded PowerPC systems compile code for the
13463 Hurd-based GNU system.
13465 @item -mcall-netbsd
13466 @opindex mcall-netbsd
13467 On System V.4 and embedded PowerPC systems compile code for the
13468 NetBSD operating system.
13470 @item -maix-struct-return
13471 @opindex maix-struct-return
13472 Return all structures in memory (as specified by the AIX ABI)@.
13474 @item -msvr4-struct-return
13475 @opindex msvr4-struct-return
13476 Return structures smaller than 8 bytes in registers (as specified by the
13479 @item -mabi=@var{abi-type}
13481 Extend the current ABI with a particular extension, or remove such extension.
13482 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
13483 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
13487 Extend the current ABI with SPE ABI extensions. This does not change
13488 the default ABI, instead it adds the SPE ABI extensions to the current
13492 @opindex mabi=no-spe
13493 Disable Booke SPE ABI extensions for the current ABI@.
13495 @item -mabi=ibmlongdouble
13496 @opindex mabi=ibmlongdouble
13497 Change the current ABI to use IBM extended precision long double.
13498 This is a PowerPC 32-bit SYSV ABI option.
13500 @item -mabi=ieeelongdouble
13501 @opindex mabi=ieeelongdouble
13502 Change the current ABI to use IEEE extended precision long double.
13503 This is a PowerPC 32-bit Linux ABI option.
13506 @itemx -mno-prototype
13507 @opindex mprototype
13508 @opindex mno-prototype
13509 On System V.4 and embedded PowerPC systems assume that all calls to
13510 variable argument functions are properly prototyped. Otherwise, the
13511 compiler must insert an instruction before every non prototyped call to
13512 set or clear bit 6 of the condition code register (@var{CR}) to
13513 indicate whether floating point values were passed in the floating point
13514 registers in case the function takes a variable arguments. With
13515 @option{-mprototype}, only calls to prototyped variable argument functions
13516 will set or clear the bit.
13520 On embedded PowerPC systems, assume that the startup module is called
13521 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
13522 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
13527 On embedded PowerPC systems, assume that the startup module is called
13528 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
13533 On embedded PowerPC systems, assume that the startup module is called
13534 @file{crt0.o} and the standard C libraries are @file{libads.a} and
13537 @item -myellowknife
13538 @opindex myellowknife
13539 On embedded PowerPC systems, assume that the startup module is called
13540 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
13545 On System V.4 and embedded PowerPC systems, specify that you are
13546 compiling for a VxWorks system.
13550 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
13551 header to indicate that @samp{eabi} extended relocations are used.
13557 On System V.4 and embedded PowerPC systems do (do not) adhere to the
13558 Embedded Applications Binary Interface (eabi) which is a set of
13559 modifications to the System V.4 specifications. Selecting @option{-meabi}
13560 means that the stack is aligned to an 8 byte boundary, a function
13561 @code{__eabi} is called to from @code{main} to set up the eabi
13562 environment, and the @option{-msdata} option can use both @code{r2} and
13563 @code{r13} to point to two separate small data areas. Selecting
13564 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
13565 do not call an initialization function from @code{main}, and the
13566 @option{-msdata} option will only use @code{r13} to point to a single
13567 small data area. The @option{-meabi} option is on by default if you
13568 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
13571 @opindex msdata=eabi
13572 On System V.4 and embedded PowerPC systems, put small initialized
13573 @code{const} global and static data in the @samp{.sdata2} section, which
13574 is pointed to by register @code{r2}. Put small initialized
13575 non-@code{const} global and static data in the @samp{.sdata} section,
13576 which is pointed to by register @code{r13}. Put small uninitialized
13577 global and static data in the @samp{.sbss} section, which is adjacent to
13578 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13579 incompatible with the @option{-mrelocatable} option. The
13580 @option{-msdata=eabi} option also sets the @option{-memb} option.
13583 @opindex msdata=sysv
13584 On System V.4 and embedded PowerPC systems, put small global and static
13585 data in the @samp{.sdata} section, which is pointed to by register
13586 @code{r13}. Put small uninitialized global and static data in the
13587 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13588 The @option{-msdata=sysv} option is incompatible with the
13589 @option{-mrelocatable} option.
13591 @item -msdata=default
13593 @opindex msdata=default
13595 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13596 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13597 same as @option{-msdata=sysv}.
13600 @opindex msdata-data
13601 On System V.4 and embedded PowerPC systems, put small global
13602 data in the @samp{.sdata} section. Put small uninitialized global
13603 data in the @samp{.sbss} section. Do not use register @code{r13}
13604 to address small data however. This is the default behavior unless
13605 other @option{-msdata} options are used.
13609 @opindex msdata=none
13611 On embedded PowerPC systems, put all initialized global and static data
13612 in the @samp{.data} section, and all uninitialized data in the
13613 @samp{.bss} section.
13617 @cindex smaller data references (PowerPC)
13618 @cindex .sdata/.sdata2 references (PowerPC)
13619 On embedded PowerPC systems, put global and static items less than or
13620 equal to @var{num} bytes into the small data or bss sections instead of
13621 the normal data or bss section. By default, @var{num} is 8. The
13622 @option{-G @var{num}} switch is also passed to the linker.
13623 All modules should be compiled with the same @option{-G @var{num}} value.
13626 @itemx -mno-regnames
13628 @opindex mno-regnames
13629 On System V.4 and embedded PowerPC systems do (do not) emit register
13630 names in the assembly language output using symbolic forms.
13633 @itemx -mno-longcall
13635 @opindex mno-longcall
13636 By default assume that all calls are far away so that a longer more
13637 expensive calling sequence is required. This is required for calls
13638 further than 32 megabytes (33,554,432 bytes) from the current location.
13639 A short call will be generated if the compiler knows
13640 the call cannot be that far away. This setting can be overridden by
13641 the @code{shortcall} function attribute, or by @code{#pragma
13644 Some linkers are capable of detecting out-of-range calls and generating
13645 glue code on the fly. On these systems, long calls are unnecessary and
13646 generate slower code. As of this writing, the AIX linker can do this,
13647 as can the GNU linker for PowerPC/64. It is planned to add this feature
13648 to the GNU linker for 32-bit PowerPC systems as well.
13650 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13651 callee, L42'', plus a ``branch island'' (glue code). The two target
13652 addresses represent the callee and the ``branch island''. The
13653 Darwin/PPC linker will prefer the first address and generate a ``bl
13654 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13655 otherwise, the linker will generate ``bl L42'' to call the ``branch
13656 island''. The ``branch island'' is appended to the body of the
13657 calling function; it computes the full 32-bit address of the callee
13660 On Mach-O (Darwin) systems, this option directs the compiler emit to
13661 the glue for every direct call, and the Darwin linker decides whether
13662 to use or discard it.
13664 In the future, we may cause GCC to ignore all longcall specifications
13665 when the linker is known to generate glue.
13669 Adds support for multithreading with the @dfn{pthreads} library.
13670 This option sets flags for both the preprocessor and linker.
13674 @node S/390 and zSeries Options
13675 @subsection S/390 and zSeries Options
13676 @cindex S/390 and zSeries Options
13678 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13682 @itemx -msoft-float
13683 @opindex mhard-float
13684 @opindex msoft-float
13685 Use (do not use) the hardware floating-point instructions and registers
13686 for floating-point operations. When @option{-msoft-float} is specified,
13687 functions in @file{libgcc.a} will be used to perform floating-point
13688 operations. When @option{-mhard-float} is specified, the compiler
13689 generates IEEE floating-point instructions. This is the default.
13691 @item -mlong-double-64
13692 @itemx -mlong-double-128
13693 @opindex mlong-double-64
13694 @opindex mlong-double-128
13695 These switches control the size of @code{long double} type. A size
13696 of 64bit makes the @code{long double} type equivalent to the @code{double}
13697 type. This is the default.
13700 @itemx -mno-backchain
13701 @opindex mbackchain
13702 @opindex mno-backchain
13703 Store (do not store) the address of the caller's frame as backchain pointer
13704 into the callee's stack frame.
13705 A backchain may be needed to allow debugging using tools that do not understand
13706 DWARF-2 call frame information.
13707 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13708 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13709 the backchain is placed into the topmost word of the 96/160 byte register
13712 In general, code compiled with @option{-mbackchain} is call-compatible with
13713 code compiled with @option{-mmo-backchain}; however, use of the backchain
13714 for debugging purposes usually requires that the whole binary is built with
13715 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13716 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13717 to build a linux kernel use @option{-msoft-float}.
13719 The default is to not maintain the backchain.
13721 @item -mpacked-stack
13722 @itemx -mno-packed-stack
13723 @opindex mpacked-stack
13724 @opindex mno-packed-stack
13725 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13726 specified, the compiler uses the all fields of the 96/160 byte register save
13727 area only for their default purpose; unused fields still take up stack space.
13728 When @option{-mpacked-stack} is specified, register save slots are densely
13729 packed at the top of the register save area; unused space is reused for other
13730 purposes, allowing for more efficient use of the available stack space.
13731 However, when @option{-mbackchain} is also in effect, the topmost word of
13732 the save area is always used to store the backchain, and the return address
13733 register is always saved two words below the backchain.
13735 As long as the stack frame backchain is not used, code generated with
13736 @option{-mpacked-stack} is call-compatible with code generated with
13737 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13738 S/390 or zSeries generated code that uses the stack frame backchain at run
13739 time, not just for debugging purposes. Such code is not call-compatible
13740 with code compiled with @option{-mpacked-stack}. Also, note that the
13741 combination of @option{-mbackchain},
13742 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13743 to build a linux kernel use @option{-msoft-float}.
13745 The default is to not use the packed stack layout.
13748 @itemx -mno-small-exec
13749 @opindex msmall-exec
13750 @opindex mno-small-exec
13751 Generate (or do not generate) code using the @code{bras} instruction
13752 to do subroutine calls.
13753 This only works reliably if the total executable size does not
13754 exceed 64k. The default is to use the @code{basr} instruction instead,
13755 which does not have this limitation.
13761 When @option{-m31} is specified, generate code compliant to the
13762 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13763 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13764 particular to generate 64-bit instructions. For the @samp{s390}
13765 targets, the default is @option{-m31}, while the @samp{s390x}
13766 targets default to @option{-m64}.
13772 When @option{-mzarch} is specified, generate code using the
13773 instructions available on z/Architecture.
13774 When @option{-mesa} is specified, generate code using the
13775 instructions available on ESA/390. Note that @option{-mesa} is
13776 not possible with @option{-m64}.
13777 When generating code compliant to the GNU/Linux for S/390 ABI,
13778 the default is @option{-mesa}. When generating code compliant
13779 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13785 Generate (or do not generate) code using the @code{mvcle} instruction
13786 to perform block moves. When @option{-mno-mvcle} is specified,
13787 use a @code{mvc} loop instead. This is the default unless optimizing for
13794 Print (or do not print) additional debug information when compiling.
13795 The default is to not print debug information.
13797 @item -march=@var{cpu-type}
13799 Generate code that will run on @var{cpu-type}, which is the name of a system
13800 representing a certain processor type. Possible values for
13801 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13802 When generating code using the instructions available on z/Architecture,
13803 the default is @option{-march=z900}. Otherwise, the default is
13804 @option{-march=g5}.
13806 @item -mtune=@var{cpu-type}
13808 Tune to @var{cpu-type} everything applicable about the generated code,
13809 except for the ABI and the set of available instructions.
13810 The list of @var{cpu-type} values is the same as for @option{-march}.
13811 The default is the value used for @option{-march}.
13814 @itemx -mno-tpf-trace
13815 @opindex mtpf-trace
13816 @opindex mno-tpf-trace
13817 Generate code that adds (does not add) in TPF OS specific branches to trace
13818 routines in the operating system. This option is off by default, even
13819 when compiling for the TPF OS@.
13822 @itemx -mno-fused-madd
13823 @opindex mfused-madd
13824 @opindex mno-fused-madd
13825 Generate code that uses (does not use) the floating point multiply and
13826 accumulate instructions. These instructions are generated by default if
13827 hardware floating point is used.
13829 @item -mwarn-framesize=@var{framesize}
13830 @opindex mwarn-framesize
13831 Emit a warning if the current function exceeds the given frame size. Because
13832 this is a compile time check it doesn't need to be a real problem when the program
13833 runs. It is intended to identify functions which most probably cause
13834 a stack overflow. It is useful to be used in an environment with limited stack
13835 size e.g.@: the linux kernel.
13837 @item -mwarn-dynamicstack
13838 @opindex mwarn-dynamicstack
13839 Emit a warning if the function calls alloca or uses dynamically
13840 sized arrays. This is generally a bad idea with a limited stack size.
13842 @item -mstack-guard=@var{stack-guard}
13843 @itemx -mstack-size=@var{stack-size}
13844 @opindex mstack-guard
13845 @opindex mstack-size
13846 If these options are provided the s390 back end emits additional instructions in
13847 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13848 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13849 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13850 the frame size of the compiled function is chosen.
13851 These options are intended to be used to help debugging stack overflow problems.
13852 The additionally emitted code causes only little overhead and hence can also be
13853 used in production like systems without greater performance degradation. The given
13854 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13855 @var{stack-guard} without exceeding 64k.
13856 In order to be efficient the extra code makes the assumption that the stack starts
13857 at an address aligned to the value given by @var{stack-size}.
13858 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13861 @node Score Options
13862 @subsection Score Options
13863 @cindex Score Options
13865 These options are defined for Score implementations:
13870 Compile code for big endian mode. This is the default.
13874 Compile code for little endian mode.
13878 Disable generate bcnz instruction.
13882 Enable generate unaligned load and store instruction.
13886 Enable the use of multiply-accumulate instructions. Disabled by default.
13890 Specify the SCORE5 as the target architecture.
13894 Specify the SCORE5U of the target architecture.
13898 Specify the SCORE7 as the target architecture. This is the default.
13902 Specify the SCORE7D as the target architecture.
13906 @subsection SH Options
13908 These @samp{-m} options are defined for the SH implementations:
13913 Generate code for the SH1.
13917 Generate code for the SH2.
13920 Generate code for the SH2e.
13924 Generate code for the SH3.
13928 Generate code for the SH3e.
13932 Generate code for the SH4 without a floating-point unit.
13934 @item -m4-single-only
13935 @opindex m4-single-only
13936 Generate code for the SH4 with a floating-point unit that only
13937 supports single-precision arithmetic.
13941 Generate code for the SH4 assuming the floating-point unit is in
13942 single-precision mode by default.
13946 Generate code for the SH4.
13950 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13951 floating-point unit is not used.
13953 @item -m4a-single-only
13954 @opindex m4a-single-only
13955 Generate code for the SH4a, in such a way that no double-precision
13956 floating point operations are used.
13959 @opindex m4a-single
13960 Generate code for the SH4a assuming the floating-point unit is in
13961 single-precision mode by default.
13965 Generate code for the SH4a.
13969 Same as @option{-m4a-nofpu}, except that it implicitly passes
13970 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13971 instructions at the moment.
13975 Compile code for the processor in big endian mode.
13979 Compile code for the processor in little endian mode.
13983 Align doubles at 64-bit boundaries. Note that this changes the calling
13984 conventions, and thus some functions from the standard C library will
13985 not work unless you recompile it first with @option{-mdalign}.
13989 Shorten some address references at link time, when possible; uses the
13990 linker option @option{-relax}.
13994 Use 32-bit offsets in @code{switch} tables. The default is to use
13999 Enable the use of bit manipulation instructions on SH2A.
14003 Enable the use of the instruction @code{fmovd}.
14007 Comply with the calling conventions defined by Renesas.
14011 Comply with the calling conventions defined by Renesas.
14015 Comply with the calling conventions defined for GCC before the Renesas
14016 conventions were available. This option is the default for all
14017 targets of the SH toolchain except for @samp{sh-symbianelf}.
14020 @opindex mnomacsave
14021 Mark the @code{MAC} register as call-clobbered, even if
14022 @option{-mhitachi} is given.
14026 Increase IEEE-compliance of floating-point code.
14027 At the moment, this is equivalent to @option{-fno-finite-math-only}.
14028 When generating 16 bit SH opcodes, getting IEEE-conforming results for
14029 comparisons of NANs / infinities incurs extra overhead in every
14030 floating point comparison, therefore the default is set to
14031 @option{-ffinite-math-only}.
14033 @item -minline-ic_invalidate
14034 @opindex minline-ic_invalidate
14035 Inline code to invalidate instruction cache entries after setting up
14036 nested function trampolines.
14037 This option has no effect if -musermode is in effect and the selected
14038 code generation option (e.g. -m4) does not allow the use of the icbi
14040 If the selected code generation option does not allow the use of the icbi
14041 instruction, and -musermode is not in effect, the inlined code will
14042 manipulate the instruction cache address array directly with an associative
14043 write. This not only requires privileged mode, but it will also
14044 fail if the cache line had been mapped via the TLB and has become unmapped.
14048 Dump instruction size and location in the assembly code.
14051 @opindex mpadstruct
14052 This option is deprecated. It pads structures to multiple of 4 bytes,
14053 which is incompatible with the SH ABI@.
14057 Optimize for space instead of speed. Implied by @option{-Os}.
14060 @opindex mprefergot
14061 When generating position-independent code, emit function calls using
14062 the Global Offset Table instead of the Procedure Linkage Table.
14066 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
14067 if the inlined code would not work in user mode.
14068 This is the default when the target is @code{sh-*-linux*}.
14070 @item -multcost=@var{number}
14071 @opindex multcost=@var{number}
14072 Set the cost to assume for a multiply insn.
14074 @item -mdiv=@var{strategy}
14075 @opindex mdiv=@var{strategy}
14076 Set the division strategy to use for SHmedia code. @var{strategy} must be
14077 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
14078 inv:call2, inv:fp .
14079 "fp" performs the operation in floating point. This has a very high latency,
14080 but needs only a few instructions, so it might be a good choice if
14081 your code has enough easily exploitable ILP to allow the compiler to
14082 schedule the floating point instructions together with other instructions.
14083 Division by zero causes a floating point exception.
14084 "inv" uses integer operations to calculate the inverse of the divisor,
14085 and then multiplies the dividend with the inverse. This strategy allows
14086 cse and hoisting of the inverse calculation. Division by zero calculates
14087 an unspecified result, but does not trap.
14088 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
14089 have been found, or if the entire operation has been hoisted to the same
14090 place, the last stages of the inverse calculation are intertwined with the
14091 final multiply to reduce the overall latency, at the expense of using a few
14092 more instructions, and thus offering fewer scheduling opportunities with
14094 "call" calls a library function that usually implements the inv:minlat
14096 This gives high code density for m5-*media-nofpu compilations.
14097 "call2" uses a different entry point of the same library function, where it
14098 assumes that a pointer to a lookup table has already been set up, which
14099 exposes the pointer load to cse / code hoisting optimizations.
14100 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
14101 code generation, but if the code stays unoptimized, revert to the "call",
14102 "call2", or "fp" strategies, respectively. Note that the
14103 potentially-trapping side effect of division by zero is carried by a
14104 separate instruction, so it is possible that all the integer instructions
14105 are hoisted out, but the marker for the side effect stays where it is.
14106 A recombination to fp operations or a call is not possible in that case.
14107 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
14108 that the inverse calculation was nor separated from the multiply, they speed
14109 up division where the dividend fits into 20 bits (plus sign where applicable),
14110 by inserting a test to skip a number of operations in this case; this test
14111 slows down the case of larger dividends. inv20u assumes the case of a such
14112 a small dividend to be unlikely, and inv20l assumes it to be likely.
14114 @item -mdivsi3_libfunc=@var{name}
14115 @opindex mdivsi3_libfunc=@var{name}
14116 Set the name of the library function used for 32 bit signed division to
14117 @var{name}. This only affect the name used in the call and inv:call
14118 division strategies, and the compiler will still expect the same
14119 sets of input/output/clobbered registers as if this option was not present.
14121 @item -mfixed-range=@var{register-range}
14122 @opindex mfixed-range
14123 Generate code treating the given register range as fixed registers.
14124 A fixed register is one that the register allocator can not use. This is
14125 useful when compiling kernel code. A register range is specified as
14126 two registers separated by a dash. Multiple register ranges can be
14127 specified separated by a comma.
14129 @item -madjust-unroll
14130 @opindex madjust-unroll
14131 Throttle unrolling to avoid thrashing target registers.
14132 This option only has an effect if the gcc code base supports the
14133 TARGET_ADJUST_UNROLL_MAX target hook.
14135 @item -mindexed-addressing
14136 @opindex mindexed-addressing
14137 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
14138 This is only safe if the hardware and/or OS implement 32 bit wrap-around
14139 semantics for the indexed addressing mode. The architecture allows the
14140 implementation of processors with 64 bit MMU, which the OS could use to
14141 get 32 bit addressing, but since no current hardware implementation supports
14142 this or any other way to make the indexed addressing mode safe to use in
14143 the 32 bit ABI, the default is -mno-indexed-addressing.
14145 @item -mgettrcost=@var{number}
14146 @opindex mgettrcost=@var{number}
14147 Set the cost assumed for the gettr instruction to @var{number}.
14148 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
14152 Assume pt* instructions won't trap. This will generally generate better
14153 scheduled code, but is unsafe on current hardware. The current architecture
14154 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
14155 This has the unintentional effect of making it unsafe to schedule ptabs /
14156 ptrel before a branch, or hoist it out of a loop. For example,
14157 __do_global_ctors, a part of libgcc that runs constructors at program
14158 startup, calls functions in a list which is delimited by @minus{}1. With the
14159 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
14160 That means that all the constructors will be run a bit quicker, but when
14161 the loop comes to the end of the list, the program crashes because ptabs
14162 loads @minus{}1 into a target register. Since this option is unsafe for any
14163 hardware implementing the current architecture specification, the default
14164 is -mno-pt-fixed. Unless the user specifies a specific cost with
14165 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
14166 this deters register allocation using target registers for storing
14169 @item -minvalid-symbols
14170 @opindex minvalid-symbols
14171 Assume symbols might be invalid. Ordinary function symbols generated by
14172 the compiler will always be valid to load with movi/shori/ptabs or
14173 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
14174 to generate symbols that will cause ptabs / ptrel to trap.
14175 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
14176 It will then prevent cross-basic-block cse, hoisting and most scheduling
14177 of symbol loads. The default is @option{-mno-invalid-symbols}.
14180 @node SPARC Options
14181 @subsection SPARC Options
14182 @cindex SPARC options
14184 These @samp{-m} options are supported on the SPARC:
14187 @item -mno-app-regs
14189 @opindex mno-app-regs
14191 Specify @option{-mapp-regs} to generate output using the global registers
14192 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
14195 To be fully SVR4 ABI compliant at the cost of some performance loss,
14196 specify @option{-mno-app-regs}. You should compile libraries and system
14197 software with this option.
14200 @itemx -mhard-float
14202 @opindex mhard-float
14203 Generate output containing floating point instructions. This is the
14207 @itemx -msoft-float
14209 @opindex msoft-float
14210 Generate output containing library calls for floating point.
14211 @strong{Warning:} the requisite libraries are not available for all SPARC
14212 targets. Normally the facilities of the machine's usual C compiler are
14213 used, but this cannot be done directly in cross-compilation. You must make
14214 your own arrangements to provide suitable library functions for
14215 cross-compilation. The embedded targets @samp{sparc-*-aout} and
14216 @samp{sparclite-*-*} do provide software floating point support.
14218 @option{-msoft-float} changes the calling convention in the output file;
14219 therefore, it is only useful if you compile @emph{all} of a program with
14220 this option. In particular, you need to compile @file{libgcc.a}, the
14221 library that comes with GCC, with @option{-msoft-float} in order for
14224 @item -mhard-quad-float
14225 @opindex mhard-quad-float
14226 Generate output containing quad-word (long double) floating point
14229 @item -msoft-quad-float
14230 @opindex msoft-quad-float
14231 Generate output containing library calls for quad-word (long double)
14232 floating point instructions. The functions called are those specified
14233 in the SPARC ABI@. This is the default.
14235 As of this writing, there are no SPARC implementations that have hardware
14236 support for the quad-word floating point instructions. They all invoke
14237 a trap handler for one of these instructions, and then the trap handler
14238 emulates the effect of the instruction. Because of the trap handler overhead,
14239 this is much slower than calling the ABI library routines. Thus the
14240 @option{-msoft-quad-float} option is the default.
14242 @item -mno-unaligned-doubles
14243 @itemx -munaligned-doubles
14244 @opindex mno-unaligned-doubles
14245 @opindex munaligned-doubles
14246 Assume that doubles have 8 byte alignment. This is the default.
14248 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
14249 alignment only if they are contained in another type, or if they have an
14250 absolute address. Otherwise, it assumes they have 4 byte alignment.
14251 Specifying this option avoids some rare compatibility problems with code
14252 generated by other compilers. It is not the default because it results
14253 in a performance loss, especially for floating point code.
14255 @item -mno-faster-structs
14256 @itemx -mfaster-structs
14257 @opindex mno-faster-structs
14258 @opindex mfaster-structs
14259 With @option{-mfaster-structs}, the compiler assumes that structures
14260 should have 8 byte alignment. This enables the use of pairs of
14261 @code{ldd} and @code{std} instructions for copies in structure
14262 assignment, in place of twice as many @code{ld} and @code{st} pairs.
14263 However, the use of this changed alignment directly violates the SPARC
14264 ABI@. Thus, it's intended only for use on targets where the developer
14265 acknowledges that their resulting code will not be directly in line with
14266 the rules of the ABI@.
14268 @item -mimpure-text
14269 @opindex mimpure-text
14270 @option{-mimpure-text}, used in addition to @option{-shared}, tells
14271 the compiler to not pass @option{-z text} to the linker when linking a
14272 shared object. Using this option, you can link position-dependent
14273 code into a shared object.
14275 @option{-mimpure-text} suppresses the ``relocations remain against
14276 allocatable but non-writable sections'' linker error message.
14277 However, the necessary relocations will trigger copy-on-write, and the
14278 shared object is not actually shared across processes. Instead of
14279 using @option{-mimpure-text}, you should compile all source code with
14280 @option{-fpic} or @option{-fPIC}.
14282 This option is only available on SunOS and Solaris.
14284 @item -mcpu=@var{cpu_type}
14286 Set the instruction set, register set, and instruction scheduling parameters
14287 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
14288 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
14289 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
14290 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
14291 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
14293 Default instruction scheduling parameters are used for values that select
14294 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
14295 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
14297 Here is a list of each supported architecture and their supported
14302 v8: supersparc, hypersparc
14303 sparclite: f930, f934, sparclite86x
14305 v9: ultrasparc, ultrasparc3, niagara, niagara2
14308 By default (unless configured otherwise), GCC generates code for the V7
14309 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
14310 additionally optimizes it for the Cypress CY7C602 chip, as used in the
14311 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
14312 SPARCStation 1, 2, IPX etc.
14314 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
14315 architecture. The only difference from V7 code is that the compiler emits
14316 the integer multiply and integer divide instructions which exist in SPARC-V8
14317 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
14318 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
14321 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
14322 the SPARC architecture. This adds the integer multiply, integer divide step
14323 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
14324 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
14325 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
14326 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
14327 MB86934 chip, which is the more recent SPARClite with FPU@.
14329 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
14330 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
14331 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
14332 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
14333 optimizes it for the TEMIC SPARClet chip.
14335 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
14336 architecture. This adds 64-bit integer and floating-point move instructions,
14337 3 additional floating-point condition code registers and conditional move
14338 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
14339 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
14340 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
14341 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
14342 @option{-mcpu=niagara}, the compiler additionally optimizes it for
14343 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
14344 additionally optimizes it for Sun UltraSPARC T2 chips.
14346 @item -mtune=@var{cpu_type}
14348 Set the instruction scheduling parameters for machine type
14349 @var{cpu_type}, but do not set the instruction set or register set that the
14350 option @option{-mcpu=@var{cpu_type}} would.
14352 The same values for @option{-mcpu=@var{cpu_type}} can be used for
14353 @option{-mtune=@var{cpu_type}}, but the only useful values are those
14354 that select a particular cpu implementation. Those are @samp{cypress},
14355 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
14356 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
14357 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
14362 @opindex mno-v8plus
14363 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
14364 difference from the V8 ABI is that the global and out registers are
14365 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
14366 mode for all SPARC-V9 processors.
14372 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
14373 Visual Instruction Set extensions. The default is @option{-mno-vis}.
14376 These @samp{-m} options are supported in addition to the above
14377 on SPARC-V9 processors in 64-bit environments:
14380 @item -mlittle-endian
14381 @opindex mlittle-endian
14382 Generate code for a processor running in little-endian mode. It is only
14383 available for a few configurations and most notably not on Solaris and Linux.
14389 Generate code for a 32-bit or 64-bit environment.
14390 The 32-bit environment sets int, long and pointer to 32 bits.
14391 The 64-bit environment sets int to 32 bits and long and pointer
14394 @item -mcmodel=medlow
14395 @opindex mcmodel=medlow
14396 Generate code for the Medium/Low code model: 64-bit addresses, programs
14397 must be linked in the low 32 bits of memory. Programs can be statically
14398 or dynamically linked.
14400 @item -mcmodel=medmid
14401 @opindex mcmodel=medmid
14402 Generate code for the Medium/Middle code model: 64-bit addresses, programs
14403 must be linked in the low 44 bits of memory, the text and data segments must
14404 be less than 2GB in size and the data segment must be located within 2GB of
14407 @item -mcmodel=medany
14408 @opindex mcmodel=medany
14409 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
14410 may be linked anywhere in memory, the text and data segments must be less
14411 than 2GB in size and the data segment must be located within 2GB of the
14414 @item -mcmodel=embmedany
14415 @opindex mcmodel=embmedany
14416 Generate code for the Medium/Anywhere code model for embedded systems:
14417 64-bit addresses, the text and data segments must be less than 2GB in
14418 size, both starting anywhere in memory (determined at link time). The
14419 global register %g4 points to the base of the data segment. Programs
14420 are statically linked and PIC is not supported.
14423 @itemx -mno-stack-bias
14424 @opindex mstack-bias
14425 @opindex mno-stack-bias
14426 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
14427 frame pointer if present, are offset by @minus{}2047 which must be added back
14428 when making stack frame references. This is the default in 64-bit mode.
14429 Otherwise, assume no such offset is present.
14432 These switches are supported in addition to the above on Solaris:
14437 Add support for multithreading using the Solaris threads library. This
14438 option sets flags for both the preprocessor and linker. This option does
14439 not affect the thread safety of object code produced by the compiler or
14440 that of libraries supplied with it.
14444 Add support for multithreading using the POSIX threads library. This
14445 option sets flags for both the preprocessor and linker. This option does
14446 not affect the thread safety of object code produced by the compiler or
14447 that of libraries supplied with it.
14451 This is a synonym for @option{-pthreads}.
14455 @subsection SPU Options
14456 @cindex SPU options
14458 These @samp{-m} options are supported on the SPU:
14462 @itemx -merror-reloc
14463 @opindex mwarn-reloc
14464 @opindex merror-reloc
14466 The loader for SPU does not handle dynamic relocations. By default, GCC
14467 will give an error when it generates code that requires a dynamic
14468 relocation. @option{-mno-error-reloc} disables the error,
14469 @option{-mwarn-reloc} will generate a warning instead.
14472 @itemx -munsafe-dma
14474 @opindex munsafe-dma
14476 Instructions which initiate or test completion of DMA must not be
14477 reordered with respect to loads and stores of the memory which is being
14478 accessed. Users typically address this problem using the volatile
14479 keyword, but that can lead to inefficient code in places where the
14480 memory is known to not change. Rather than mark the memory as volatile
14481 we treat the DMA instructions as potentially effecting all memory. With
14482 @option{-munsafe-dma} users must use the volatile keyword to protect
14485 @item -mbranch-hints
14486 @opindex mbranch-hints
14488 By default, GCC will generate a branch hint instruction to avoid
14489 pipeline stalls for always taken or probably taken branches. A hint
14490 will not be generated closer than 8 instructions away from its branch.
14491 There is little reason to disable them, except for debugging purposes,
14492 or to make an object a little bit smaller.
14496 @opindex msmall-mem
14497 @opindex mlarge-mem
14499 By default, GCC generates code assuming that addresses are never larger
14500 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
14501 a full 32 bit address.
14506 By default, GCC links against startup code that assumes the SPU-style
14507 main function interface (which has an unconventional parameter list).
14508 With @option{-mstdmain}, GCC will link your program against startup
14509 code that assumes a C99-style interface to @code{main}, including a
14510 local copy of @code{argv} strings.
14512 @item -mfixed-range=@var{register-range}
14513 @opindex mfixed-range
14514 Generate code treating the given register range as fixed registers.
14515 A fixed register is one that the register allocator can not use. This is
14516 useful when compiling kernel code. A register range is specified as
14517 two registers separated by a dash. Multiple register ranges can be
14518 specified separated by a comma.
14522 @node System V Options
14523 @subsection Options for System V
14525 These additional options are available on System V Release 4 for
14526 compatibility with other compilers on those systems:
14531 Create a shared object.
14532 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
14536 Identify the versions of each tool used by the compiler, in a
14537 @code{.ident} assembler directive in the output.
14541 Refrain from adding @code{.ident} directives to the output file (this is
14544 @item -YP,@var{dirs}
14546 Search the directories @var{dirs}, and no others, for libraries
14547 specified with @option{-l}.
14549 @item -Ym,@var{dir}
14551 Look in the directory @var{dir} to find the M4 preprocessor.
14552 The assembler uses this option.
14553 @c This is supposed to go with a -Yd for predefined M4 macro files, but
14554 @c the generic assembler that comes with Solaris takes just -Ym.
14558 @subsection V850 Options
14559 @cindex V850 Options
14561 These @samp{-m} options are defined for V850 implementations:
14565 @itemx -mno-long-calls
14566 @opindex mlong-calls
14567 @opindex mno-long-calls
14568 Treat all calls as being far away (near). If calls are assumed to be
14569 far away, the compiler will always load the functions address up into a
14570 register, and call indirect through the pointer.
14576 Do not optimize (do optimize) basic blocks that use the same index
14577 pointer 4 or more times to copy pointer into the @code{ep} register, and
14578 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14579 option is on by default if you optimize.
14581 @item -mno-prolog-function
14582 @itemx -mprolog-function
14583 @opindex mno-prolog-function
14584 @opindex mprolog-function
14585 Do not use (do use) external functions to save and restore registers
14586 at the prologue and epilogue of a function. The external functions
14587 are slower, but use less code space if more than one function saves
14588 the same number of registers. The @option{-mprolog-function} option
14589 is on by default if you optimize.
14593 Try to make the code as small as possible. At present, this just turns
14594 on the @option{-mep} and @option{-mprolog-function} options.
14596 @item -mtda=@var{n}
14598 Put static or global variables whose size is @var{n} bytes or less into
14599 the tiny data area that register @code{ep} points to. The tiny data
14600 area can hold up to 256 bytes in total (128 bytes for byte references).
14602 @item -msda=@var{n}
14604 Put static or global variables whose size is @var{n} bytes or less into
14605 the small data area that register @code{gp} points to. The small data
14606 area can hold up to 64 kilobytes.
14608 @item -mzda=@var{n}
14610 Put static or global variables whose size is @var{n} bytes or less into
14611 the first 32 kilobytes of memory.
14615 Specify that the target processor is the V850.
14618 @opindex mbig-switch
14619 Generate code suitable for big switch tables. Use this option only if
14620 the assembler/linker complain about out of range branches within a switch
14625 This option will cause r2 and r5 to be used in the code generated by
14626 the compiler. This setting is the default.
14628 @item -mno-app-regs
14629 @opindex mno-app-regs
14630 This option will cause r2 and r5 to be treated as fixed registers.
14634 Specify that the target processor is the V850E1. The preprocessor
14635 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14636 this option is used.
14640 Specify that the target processor is the V850E@. The preprocessor
14641 constant @samp{__v850e__} will be defined if this option is used.
14643 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14644 are defined then a default target processor will be chosen and the
14645 relevant @samp{__v850*__} preprocessor constant will be defined.
14647 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14648 defined, regardless of which processor variant is the target.
14650 @item -mdisable-callt
14651 @opindex mdisable-callt
14652 This option will suppress generation of the CALLT instruction for the
14653 v850e and v850e1 flavors of the v850 architecture. The default is
14654 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14659 @subsection VAX Options
14660 @cindex VAX options
14662 These @samp{-m} options are defined for the VAX:
14667 Do not output certain jump instructions (@code{aobleq} and so on)
14668 that the Unix assembler for the VAX cannot handle across long
14673 Do output those jump instructions, on the assumption that you
14674 will assemble with the GNU assembler.
14678 Output code for g-format floating point numbers instead of d-format.
14681 @node VxWorks Options
14682 @subsection VxWorks Options
14683 @cindex VxWorks Options
14685 The options in this section are defined for all VxWorks targets.
14686 Options specific to the target hardware are listed with the other
14687 options for that target.
14692 GCC can generate code for both VxWorks kernels and real time processes
14693 (RTPs). This option switches from the former to the latter. It also
14694 defines the preprocessor macro @code{__RTP__}.
14697 @opindex non-static
14698 Link an RTP executable against shared libraries rather than static
14699 libraries. The options @option{-static} and @option{-shared} can
14700 also be used for RTPs (@pxref{Link Options}); @option{-static}
14707 These options are passed down to the linker. They are defined for
14708 compatibility with Diab.
14711 @opindex Xbind-lazy
14712 Enable lazy binding of function calls. This option is equivalent to
14713 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14717 Disable lazy binding of function calls. This option is the default and
14718 is defined for compatibility with Diab.
14721 @node x86-64 Options
14722 @subsection x86-64 Options
14723 @cindex x86-64 options
14725 These are listed under @xref{i386 and x86-64 Options}.
14727 @node Xstormy16 Options
14728 @subsection Xstormy16 Options
14729 @cindex Xstormy16 Options
14731 These options are defined for Xstormy16:
14736 Choose startup files and linker script suitable for the simulator.
14739 @node Xtensa Options
14740 @subsection Xtensa Options
14741 @cindex Xtensa Options
14743 These options are supported for Xtensa targets:
14747 @itemx -mno-const16
14749 @opindex mno-const16
14750 Enable or disable use of @code{CONST16} instructions for loading
14751 constant values. The @code{CONST16} instruction is currently not a
14752 standard option from Tensilica. When enabled, @code{CONST16}
14753 instructions are always used in place of the standard @code{L32R}
14754 instructions. The use of @code{CONST16} is enabled by default only if
14755 the @code{L32R} instruction is not available.
14758 @itemx -mno-fused-madd
14759 @opindex mfused-madd
14760 @opindex mno-fused-madd
14761 Enable or disable use of fused multiply/add and multiply/subtract
14762 instructions in the floating-point option. This has no effect if the
14763 floating-point option is not also enabled. Disabling fused multiply/add
14764 and multiply/subtract instructions forces the compiler to use separate
14765 instructions for the multiply and add/subtract operations. This may be
14766 desirable in some cases where strict IEEE 754-compliant results are
14767 required: the fused multiply add/subtract instructions do not round the
14768 intermediate result, thereby producing results with @emph{more} bits of
14769 precision than specified by the IEEE standard. Disabling fused multiply
14770 add/subtract instructions also ensures that the program output is not
14771 sensitive to the compiler's ability to combine multiply and add/subtract
14774 @item -mserialize-volatile
14775 @itemx -mno-serialize-volatile
14776 @opindex mserialize-volatile
14777 @opindex mno-serialize-volatile
14778 When this option is enabled, GCC inserts @code{MEMW} instructions before
14779 @code{volatile} memory references to guarantee sequential consistency.
14780 The default is @option{-mserialize-volatile}. Use
14781 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
14783 @item -mtext-section-literals
14784 @itemx -mno-text-section-literals
14785 @opindex mtext-section-literals
14786 @opindex mno-text-section-literals
14787 Control the treatment of literal pools. The default is
14788 @option{-mno-text-section-literals}, which places literals in a separate
14789 section in the output file. This allows the literal pool to be placed
14790 in a data RAM/ROM, and it also allows the linker to combine literal
14791 pools from separate object files to remove redundant literals and
14792 improve code size. With @option{-mtext-section-literals}, the literals
14793 are interspersed in the text section in order to keep them as close as
14794 possible to their references. This may be necessary for large assembly
14797 @item -mtarget-align
14798 @itemx -mno-target-align
14799 @opindex mtarget-align
14800 @opindex mno-target-align
14801 When this option is enabled, GCC instructs the assembler to
14802 automatically align instructions to reduce branch penalties at the
14803 expense of some code density. The assembler attempts to widen density
14804 instructions to align branch targets and the instructions following call
14805 instructions. If there are not enough preceding safe density
14806 instructions to align a target, no widening will be performed. The
14807 default is @option{-mtarget-align}. These options do not affect the
14808 treatment of auto-aligned instructions like @code{LOOP}, which the
14809 assembler will always align, either by widening density instructions or
14810 by inserting no-op instructions.
14813 @itemx -mno-longcalls
14814 @opindex mlongcalls
14815 @opindex mno-longcalls
14816 When this option is enabled, GCC instructs the assembler to translate
14817 direct calls to indirect calls unless it can determine that the target
14818 of a direct call is in the range allowed by the call instruction. This
14819 translation typically occurs for calls to functions in other source
14820 files. Specifically, the assembler translates a direct @code{CALL}
14821 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14822 The default is @option{-mno-longcalls}. This option should be used in
14823 programs where the call target can potentially be out of range. This
14824 option is implemented in the assembler, not the compiler, so the
14825 assembly code generated by GCC will still show direct call
14826 instructions---look at the disassembled object code to see the actual
14827 instructions. Note that the assembler will use an indirect call for
14828 every cross-file call, not just those that really will be out of range.
14831 @node zSeries Options
14832 @subsection zSeries Options
14833 @cindex zSeries options
14835 These are listed under @xref{S/390 and zSeries Options}.
14837 @node Code Gen Options
14838 @section Options for Code Generation Conventions
14839 @cindex code generation conventions
14840 @cindex options, code generation
14841 @cindex run-time options
14843 These machine-independent options control the interface conventions
14844 used in code generation.
14846 Most of them have both positive and negative forms; the negative form
14847 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14848 one of the forms is listed---the one which is not the default. You
14849 can figure out the other form by either removing @samp{no-} or adding
14853 @item -fbounds-check
14854 @opindex fbounds-check
14855 For front-ends that support it, generate additional code to check that
14856 indices used to access arrays are within the declared range. This is
14857 currently only supported by the Java and Fortran front-ends, where
14858 this option defaults to true and false respectively.
14862 This option generates traps for signed overflow on addition, subtraction,
14863 multiplication operations.
14867 This option instructs the compiler to assume that signed arithmetic
14868 overflow of addition, subtraction and multiplication wraps around
14869 using twos-complement representation. This flag enables some optimizations
14870 and disables others. This option is enabled by default for the Java
14871 front-end, as required by the Java language specification.
14874 @opindex fexceptions
14875 Enable exception handling. Generates extra code needed to propagate
14876 exceptions. For some targets, this implies GCC will generate frame
14877 unwind information for all functions, which can produce significant data
14878 size overhead, although it does not affect execution. If you do not
14879 specify this option, GCC will enable it by default for languages like
14880 C++ which normally require exception handling, and disable it for
14881 languages like C that do not normally require it. However, you may need
14882 to enable this option when compiling C code that needs to interoperate
14883 properly with exception handlers written in C++. You may also wish to
14884 disable this option if you are compiling older C++ programs that don't
14885 use exception handling.
14887 @item -fnon-call-exceptions
14888 @opindex fnon-call-exceptions
14889 Generate code that allows trapping instructions to throw exceptions.
14890 Note that this requires platform-specific runtime support that does
14891 not exist everywhere. Moreover, it only allows @emph{trapping}
14892 instructions to throw exceptions, i.e.@: memory references or floating
14893 point instructions. It does not allow exceptions to be thrown from
14894 arbitrary signal handlers such as @code{SIGALRM}.
14896 @item -funwind-tables
14897 @opindex funwind-tables
14898 Similar to @option{-fexceptions}, except that it will just generate any needed
14899 static data, but will not affect the generated code in any other way.
14900 You will normally not enable this option; instead, a language processor
14901 that needs this handling would enable it on your behalf.
14903 @item -fasynchronous-unwind-tables
14904 @opindex fasynchronous-unwind-tables
14905 Generate unwind table in dwarf2 format, if supported by target machine. The
14906 table is exact at each instruction boundary, so it can be used for stack
14907 unwinding from asynchronous events (such as debugger or garbage collector).
14909 @item -fpcc-struct-return
14910 @opindex fpcc-struct-return
14911 Return ``short'' @code{struct} and @code{union} values in memory like
14912 longer ones, rather than in registers. This convention is less
14913 efficient, but it has the advantage of allowing intercallability between
14914 GCC-compiled files and files compiled with other compilers, particularly
14915 the Portable C Compiler (pcc).
14917 The precise convention for returning structures in memory depends
14918 on the target configuration macros.
14920 Short structures and unions are those whose size and alignment match
14921 that of some integer type.
14923 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14924 switch is not binary compatible with code compiled with the
14925 @option{-freg-struct-return} switch.
14926 Use it to conform to a non-default application binary interface.
14928 @item -freg-struct-return
14929 @opindex freg-struct-return
14930 Return @code{struct} and @code{union} values in registers when possible.
14931 This is more efficient for small structures than
14932 @option{-fpcc-struct-return}.
14934 If you specify neither @option{-fpcc-struct-return} nor
14935 @option{-freg-struct-return}, GCC defaults to whichever convention is
14936 standard for the target. If there is no standard convention, GCC
14937 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14938 the principal compiler. In those cases, we can choose the standard, and
14939 we chose the more efficient register return alternative.
14941 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14942 switch is not binary compatible with code compiled with the
14943 @option{-fpcc-struct-return} switch.
14944 Use it to conform to a non-default application binary interface.
14946 @item -fshort-enums
14947 @opindex fshort-enums
14948 Allocate to an @code{enum} type only as many bytes as it needs for the
14949 declared range of possible values. Specifically, the @code{enum} type
14950 will be equivalent to the smallest integer type which has enough room.
14952 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14953 code that is not binary compatible with code generated without that switch.
14954 Use it to conform to a non-default application binary interface.
14956 @item -fshort-double
14957 @opindex fshort-double
14958 Use the same size for @code{double} as for @code{float}.
14960 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14961 code that is not binary compatible with code generated without that switch.
14962 Use it to conform to a non-default application binary interface.
14964 @item -fshort-wchar
14965 @opindex fshort-wchar
14966 Override the underlying type for @samp{wchar_t} to be @samp{short
14967 unsigned int} instead of the default for the target. This option is
14968 useful for building programs to run under WINE@.
14970 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14971 code that is not binary compatible with code generated without that switch.
14972 Use it to conform to a non-default application binary interface.
14975 @opindex fno-common
14976 In C, allocate even uninitialized global variables in the data section of the
14977 object file, rather than generating them as common blocks. This has the
14978 effect that if the same variable is declared (without @code{extern}) in
14979 two different compilations, you will get an error when you link them.
14980 The only reason this might be useful is if you wish to verify that the
14981 program will work on other systems which always work this way.
14985 Ignore the @samp{#ident} directive.
14987 @item -finhibit-size-directive
14988 @opindex finhibit-size-directive
14989 Don't output a @code{.size} assembler directive, or anything else that
14990 would cause trouble if the function is split in the middle, and the
14991 two halves are placed at locations far apart in memory. This option is
14992 used when compiling @file{crtstuff.c}; you should not need to use it
14995 @item -fverbose-asm
14996 @opindex fverbose-asm
14997 Put extra commentary information in the generated assembly code to
14998 make it more readable. This option is generally only of use to those
14999 who actually need to read the generated assembly code (perhaps while
15000 debugging the compiler itself).
15002 @option{-fno-verbose-asm}, the default, causes the
15003 extra information to be omitted and is useful when comparing two assembler
15006 @item -frecord-gcc-switches
15007 @opindex frecord-gcc-switches
15008 This switch causes the command line that was used to invoke the
15009 compiler to be recorded into the object file that is being created.
15010 This switch is only implemented on some targets and the exact format
15011 of the recording is target and binary file format dependent, but it
15012 usually takes the form of a section containing ASCII text. This
15013 switch is related to the @option{-fverbose-asm} switch, but that
15014 switch only records information in the assembler output file as
15015 comments, so it never reaches the object file.
15019 @cindex global offset table
15021 Generate position-independent code (PIC) suitable for use in a shared
15022 library, if supported for the target machine. Such code accesses all
15023 constant addresses through a global offset table (GOT)@. The dynamic
15024 loader resolves the GOT entries when the program starts (the dynamic
15025 loader is not part of GCC; it is part of the operating system). If
15026 the GOT size for the linked executable exceeds a machine-specific
15027 maximum size, you get an error message from the linker indicating that
15028 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
15029 instead. (These maximums are 8k on the SPARC and 32k
15030 on the m68k and RS/6000. The 386 has no such limit.)
15032 Position-independent code requires special support, and therefore works
15033 only on certain machines. For the 386, GCC supports PIC for System V
15034 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
15035 position-independent.
15037 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15042 If supported for the target machine, emit position-independent code,
15043 suitable for dynamic linking and avoiding any limit on the size of the
15044 global offset table. This option makes a difference on the m68k,
15045 PowerPC and SPARC@.
15047 Position-independent code requires special support, and therefore works
15048 only on certain machines.
15050 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15057 These options are similar to @option{-fpic} and @option{-fPIC}, but
15058 generated position independent code can be only linked into executables.
15059 Usually these options are used when @option{-pie} GCC option will be
15060 used during linking.
15062 @option{-fpie} and @option{-fPIE} both define the macros
15063 @code{__pie__} and @code{__PIE__}. The macros have the value 1
15064 for @option{-fpie} and 2 for @option{-fPIE}.
15066 @item -fno-jump-tables
15067 @opindex fno-jump-tables
15068 Do not use jump tables for switch statements even where it would be
15069 more efficient than other code generation strategies. This option is
15070 of use in conjunction with @option{-fpic} or @option{-fPIC} for
15071 building code which forms part of a dynamic linker and cannot
15072 reference the address of a jump table. On some targets, jump tables
15073 do not require a GOT and this option is not needed.
15075 @item -ffixed-@var{reg}
15077 Treat the register named @var{reg} as a fixed register; generated code
15078 should never refer to it (except perhaps as a stack pointer, frame
15079 pointer or in some other fixed role).
15081 @var{reg} must be the name of a register. The register names accepted
15082 are machine-specific and are defined in the @code{REGISTER_NAMES}
15083 macro in the machine description macro file.
15085 This flag does not have a negative form, because it specifies a
15088 @item -fcall-used-@var{reg}
15089 @opindex fcall-used
15090 Treat the register named @var{reg} as an allocable register that is
15091 clobbered by function calls. It may be allocated for temporaries or
15092 variables that do not live across a call. Functions compiled this way
15093 will not save and restore the register @var{reg}.
15095 It is an error to used this flag with the frame pointer or stack pointer.
15096 Use of this flag for other registers that have fixed pervasive roles in
15097 the machine's execution model will produce disastrous results.
15099 This flag does not have a negative form, because it specifies a
15102 @item -fcall-saved-@var{reg}
15103 @opindex fcall-saved
15104 Treat the register named @var{reg} as an allocable register saved by
15105 functions. It may be allocated even for temporaries or variables that
15106 live across a call. Functions compiled this way will save and restore
15107 the register @var{reg} if they use it.
15109 It is an error to used this flag with the frame pointer or stack pointer.
15110 Use of this flag for other registers that have fixed pervasive roles in
15111 the machine's execution model will produce disastrous results.
15113 A different sort of disaster will result from the use of this flag for
15114 a register in which function values may be returned.
15116 This flag does not have a negative form, because it specifies a
15119 @item -fpack-struct[=@var{n}]
15120 @opindex fpack-struct
15121 Without a value specified, pack all structure members together without
15122 holes. When a value is specified (which must be a small power of two), pack
15123 structure members according to this value, representing the maximum
15124 alignment (that is, objects with default alignment requirements larger than
15125 this will be output potentially unaligned at the next fitting location.
15127 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
15128 code that is not binary compatible with code generated without that switch.
15129 Additionally, it makes the code suboptimal.
15130 Use it to conform to a non-default application binary interface.
15132 @item -finstrument-functions
15133 @opindex finstrument-functions
15134 Generate instrumentation calls for entry and exit to functions. Just
15135 after function entry and just before function exit, the following
15136 profiling functions will be called with the address of the current
15137 function and its call site. (On some platforms,
15138 @code{__builtin_return_address} does not work beyond the current
15139 function, so the call site information may not be available to the
15140 profiling functions otherwise.)
15143 void __cyg_profile_func_enter (void *this_fn,
15145 void __cyg_profile_func_exit (void *this_fn,
15149 The first argument is the address of the start of the current function,
15150 which may be looked up exactly in the symbol table.
15152 This instrumentation is also done for functions expanded inline in other
15153 functions. The profiling calls will indicate where, conceptually, the
15154 inline function is entered and exited. This means that addressable
15155 versions of such functions must be available. If all your uses of a
15156 function are expanded inline, this may mean an additional expansion of
15157 code size. If you use @samp{extern inline} in your C code, an
15158 addressable version of such functions must be provided. (This is
15159 normally the case anyways, but if you get lucky and the optimizer always
15160 expands the functions inline, you might have gotten away without
15161 providing static copies.)
15163 A function may be given the attribute @code{no_instrument_function}, in
15164 which case this instrumentation will not be done. This can be used, for
15165 example, for the profiling functions listed above, high-priority
15166 interrupt routines, and any functions from which the profiling functions
15167 cannot safely be called (perhaps signal handlers, if the profiling
15168 routines generate output or allocate memory).
15170 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
15171 @opindex finstrument-functions-exclude-file-list
15173 Set the list of functions that are excluded from instrumentation (see
15174 the description of @code{-finstrument-functions}). If the file that
15175 contains a function definition matches with one of @var{file}, then
15176 that function is not instrumented. The match is done on substrings:
15177 if the @var{file} parameter is a substring of the file name, it is
15178 considered to be a match.
15181 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
15182 will exclude any inline function defined in files whose pathnames
15183 contain @code{/bits/stl} or @code{include/sys}.
15185 If, for some reason, you want to include letter @code{','} in one of
15186 @var{sym}, write @code{'\,'}. For example,
15187 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
15188 (note the single quote surrounding the option).
15190 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
15191 @opindex finstrument-functions-exclude-function-list
15193 This is similar to @code{-finstrument-functions-exclude-file-list},
15194 but this option sets the list of function names to be excluded from
15195 instrumentation. The function name to be matched is its user-visible
15196 name, such as @code{vector<int> blah(const vector<int> &)}, not the
15197 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
15198 match is done on substrings: if the @var{sym} parameter is a substring
15199 of the function name, it is considered to be a match.
15201 @item -fstack-check
15202 @opindex fstack-check
15203 Generate code to verify that you do not go beyond the boundary of the
15204 stack. You should specify this flag if you are running in an
15205 environment with multiple threads, but only rarely need to specify it in
15206 a single-threaded environment since stack overflow is automatically
15207 detected on nearly all systems if there is only one stack.
15209 Note that this switch does not actually cause checking to be done; the
15210 operating system must do that. The switch causes generation of code
15211 to ensure that the operating system sees the stack being extended.
15213 @item -fstack-limit-register=@var{reg}
15214 @itemx -fstack-limit-symbol=@var{sym}
15215 @itemx -fno-stack-limit
15216 @opindex fstack-limit-register
15217 @opindex fstack-limit-symbol
15218 @opindex fno-stack-limit
15219 Generate code to ensure that the stack does not grow beyond a certain value,
15220 either the value of a register or the address of a symbol. If the stack
15221 would grow beyond the value, a signal is raised. For most targets,
15222 the signal is raised before the stack overruns the boundary, so
15223 it is possible to catch the signal without taking special precautions.
15225 For instance, if the stack starts at absolute address @samp{0x80000000}
15226 and grows downwards, you can use the flags
15227 @option{-fstack-limit-symbol=__stack_limit} and
15228 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
15229 of 128KB@. Note that this may only work with the GNU linker.
15231 @cindex aliasing of parameters
15232 @cindex parameters, aliased
15233 @item -fargument-alias
15234 @itemx -fargument-noalias
15235 @itemx -fargument-noalias-global
15236 @itemx -fargument-noalias-anything
15237 @opindex fargument-alias
15238 @opindex fargument-noalias
15239 @opindex fargument-noalias-global
15240 @opindex fargument-noalias-anything
15241 Specify the possible relationships among parameters and between
15242 parameters and global data.
15244 @option{-fargument-alias} specifies that arguments (parameters) may
15245 alias each other and may alias global storage.@*
15246 @option{-fargument-noalias} specifies that arguments do not alias
15247 each other, but may alias global storage.@*
15248 @option{-fargument-noalias-global} specifies that arguments do not
15249 alias each other and do not alias global storage.
15250 @option{-fargument-noalias-anything} specifies that arguments do not
15251 alias any other storage.
15253 Each language will automatically use whatever option is required by
15254 the language standard. You should not need to use these options yourself.
15256 @item -fleading-underscore
15257 @opindex fleading-underscore
15258 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
15259 change the way C symbols are represented in the object file. One use
15260 is to help link with legacy assembly code.
15262 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
15263 generate code that is not binary compatible with code generated without that
15264 switch. Use it to conform to a non-default application binary interface.
15265 Not all targets provide complete support for this switch.
15267 @item -ftls-model=@var{model}
15268 @opindex ftls-model
15269 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
15270 The @var{model} argument should be one of @code{global-dynamic},
15271 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
15273 The default without @option{-fpic} is @code{initial-exec}; with
15274 @option{-fpic} the default is @code{global-dynamic}.
15276 @item -fvisibility=@var{default|internal|hidden|protected}
15277 @opindex fvisibility
15278 Set the default ELF image symbol visibility to the specified option---all
15279 symbols will be marked with this unless overridden within the code.
15280 Using this feature can very substantially improve linking and
15281 load times of shared object libraries, produce more optimized
15282 code, provide near-perfect API export and prevent symbol clashes.
15283 It is @strong{strongly} recommended that you use this in any shared objects
15286 Despite the nomenclature, @code{default} always means public ie;
15287 available to be linked against from outside the shared object.
15288 @code{protected} and @code{internal} are pretty useless in real-world
15289 usage so the only other commonly used option will be @code{hidden}.
15290 The default if @option{-fvisibility} isn't specified is
15291 @code{default}, i.e., make every
15292 symbol public---this causes the same behavior as previous versions of
15295 A good explanation of the benefits offered by ensuring ELF
15296 symbols have the correct visibility is given by ``How To Write
15297 Shared Libraries'' by Ulrich Drepper (which can be found at
15298 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
15299 solution made possible by this option to marking things hidden when
15300 the default is public is to make the default hidden and mark things
15301 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
15302 and @code{__attribute__ ((visibility("default")))} instead of
15303 @code{__declspec(dllexport)} you get almost identical semantics with
15304 identical syntax. This is a great boon to those working with
15305 cross-platform projects.
15307 For those adding visibility support to existing code, you may find
15308 @samp{#pragma GCC visibility} of use. This works by you enclosing
15309 the declarations you wish to set visibility for with (for example)
15310 @samp{#pragma GCC visibility push(hidden)} and
15311 @samp{#pragma GCC visibility pop}.
15312 Bear in mind that symbol visibility should be viewed @strong{as
15313 part of the API interface contract} and thus all new code should
15314 always specify visibility when it is not the default ie; declarations
15315 only for use within the local DSO should @strong{always} be marked explicitly
15316 as hidden as so to avoid PLT indirection overheads---making this
15317 abundantly clear also aids readability and self-documentation of the code.
15318 Note that due to ISO C++ specification requirements, operator new and
15319 operator delete must always be of default visibility.
15321 Be aware that headers from outside your project, in particular system
15322 headers and headers from any other library you use, may not be
15323 expecting to be compiled with visibility other than the default. You
15324 may need to explicitly say @samp{#pragma GCC visibility push(default)}
15325 before including any such headers.
15327 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
15328 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
15329 no modifications. However, this means that calls to @samp{extern}
15330 functions with no explicit visibility will use the PLT, so it is more
15331 effective to use @samp{__attribute ((visibility))} and/or
15332 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
15333 declarations should be treated as hidden.
15335 Note that @samp{-fvisibility} does affect C++ vague linkage
15336 entities. This means that, for instance, an exception class that will
15337 be thrown between DSOs must be explicitly marked with default
15338 visibility so that the @samp{type_info} nodes will be unified between
15341 An overview of these techniques, their benefits and how to use them
15342 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
15348 @node Environment Variables
15349 @section Environment Variables Affecting GCC
15350 @cindex environment variables
15352 @c man begin ENVIRONMENT
15353 This section describes several environment variables that affect how GCC
15354 operates. Some of them work by specifying directories or prefixes to use
15355 when searching for various kinds of files. Some are used to specify other
15356 aspects of the compilation environment.
15358 Note that you can also specify places to search using options such as
15359 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
15360 take precedence over places specified using environment variables, which
15361 in turn take precedence over those specified by the configuration of GCC@.
15362 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
15363 GNU Compiler Collection (GCC) Internals}.
15368 @c @itemx LC_COLLATE
15370 @c @itemx LC_MONETARY
15371 @c @itemx LC_NUMERIC
15376 @c @findex LC_COLLATE
15377 @findex LC_MESSAGES
15378 @c @findex LC_MONETARY
15379 @c @findex LC_NUMERIC
15383 These environment variables control the way that GCC uses
15384 localization information that allow GCC to work with different
15385 national conventions. GCC inspects the locale categories
15386 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
15387 so. These locale categories can be set to any value supported by your
15388 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
15389 Kingdom encoded in UTF-8.
15391 The @env{LC_CTYPE} environment variable specifies character
15392 classification. GCC uses it to determine the character boundaries in
15393 a string; this is needed for some multibyte encodings that contain quote
15394 and escape characters that would otherwise be interpreted as a string
15397 The @env{LC_MESSAGES} environment variable specifies the language to
15398 use in diagnostic messages.
15400 If the @env{LC_ALL} environment variable is set, it overrides the value
15401 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
15402 and @env{LC_MESSAGES} default to the value of the @env{LANG}
15403 environment variable. If none of these variables are set, GCC
15404 defaults to traditional C English behavior.
15408 If @env{TMPDIR} is set, it specifies the directory to use for temporary
15409 files. GCC uses temporary files to hold the output of one stage of
15410 compilation which is to be used as input to the next stage: for example,
15411 the output of the preprocessor, which is the input to the compiler
15414 @item GCC_EXEC_PREFIX
15415 @findex GCC_EXEC_PREFIX
15416 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
15417 names of the subprograms executed by the compiler. No slash is added
15418 when this prefix is combined with the name of a subprogram, but you can
15419 specify a prefix that ends with a slash if you wish.
15421 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
15422 an appropriate prefix to use based on the pathname it was invoked with.
15424 If GCC cannot find the subprogram using the specified prefix, it
15425 tries looking in the usual places for the subprogram.
15427 The default value of @env{GCC_EXEC_PREFIX} is
15428 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
15429 the installed compiler. In many cases @var{prefix} is the value
15430 of @code{prefix} when you ran the @file{configure} script.
15432 Other prefixes specified with @option{-B} take precedence over this prefix.
15434 This prefix is also used for finding files such as @file{crt0.o} that are
15437 In addition, the prefix is used in an unusual way in finding the
15438 directories to search for header files. For each of the standard
15439 directories whose name normally begins with @samp{/usr/local/lib/gcc}
15440 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
15441 replacing that beginning with the specified prefix to produce an
15442 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
15443 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
15444 These alternate directories are searched first; the standard directories
15445 come next. If a standard directory begins with the configured
15446 @var{prefix} then the value of @var{prefix} is replaced by
15447 @env{GCC_EXEC_PREFIX} when looking for header files.
15449 @item COMPILER_PATH
15450 @findex COMPILER_PATH
15451 The value of @env{COMPILER_PATH} is a colon-separated list of
15452 directories, much like @env{PATH}. GCC tries the directories thus
15453 specified when searching for subprograms, if it can't find the
15454 subprograms using @env{GCC_EXEC_PREFIX}.
15457 @findex LIBRARY_PATH
15458 The value of @env{LIBRARY_PATH} is a colon-separated list of
15459 directories, much like @env{PATH}. When configured as a native compiler,
15460 GCC tries the directories thus specified when searching for special
15461 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15462 using GCC also uses these directories when searching for ordinary
15463 libraries for the @option{-l} option (but directories specified with
15464 @option{-L} come first).
15468 @cindex locale definition
15469 This variable is used to pass locale information to the compiler. One way in
15470 which this information is used is to determine the character set to be used
15471 when character literals, string literals and comments are parsed in C and C++.
15472 When the compiler is configured to allow multibyte characters,
15473 the following values for @env{LANG} are recognized:
15477 Recognize JIS characters.
15479 Recognize SJIS characters.
15481 Recognize EUCJP characters.
15484 If @env{LANG} is not defined, or if it has some other value, then the
15485 compiler will use mblen and mbtowc as defined by the default locale to
15486 recognize and translate multibyte characters.
15490 Some additional environments variables affect the behavior of the
15493 @include cppenv.texi
15497 @node Precompiled Headers
15498 @section Using Precompiled Headers
15499 @cindex precompiled headers
15500 @cindex speed of compilation
15502 Often large projects have many header files that are included in every
15503 source file. The time the compiler takes to process these header files
15504 over and over again can account for nearly all of the time required to
15505 build the project. To make builds faster, GCC allows users to
15506 `precompile' a header file; then, if builds can use the precompiled
15507 header file they will be much faster.
15509 To create a precompiled header file, simply compile it as you would any
15510 other file, if necessary using the @option{-x} option to make the driver
15511 treat it as a C or C++ header file. You will probably want to use a
15512 tool like @command{make} to keep the precompiled header up-to-date when
15513 the headers it contains change.
15515 A precompiled header file will be searched for when @code{#include} is
15516 seen in the compilation. As it searches for the included file
15517 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15518 compiler looks for a precompiled header in each directory just before it
15519 looks for the include file in that directory. The name searched for is
15520 the name specified in the @code{#include} with @samp{.gch} appended. If
15521 the precompiled header file can't be used, it is ignored.
15523 For instance, if you have @code{#include "all.h"}, and you have
15524 @file{all.h.gch} in the same directory as @file{all.h}, then the
15525 precompiled header file will be used if possible, and the original
15526 header will be used otherwise.
15528 Alternatively, you might decide to put the precompiled header file in a
15529 directory and use @option{-I} to ensure that directory is searched
15530 before (or instead of) the directory containing the original header.
15531 Then, if you want to check that the precompiled header file is always
15532 used, you can put a file of the same name as the original header in this
15533 directory containing an @code{#error} command.
15535 This also works with @option{-include}. So yet another way to use
15536 precompiled headers, good for projects not designed with precompiled
15537 header files in mind, is to simply take most of the header files used by
15538 a project, include them from another header file, precompile that header
15539 file, and @option{-include} the precompiled header. If the header files
15540 have guards against multiple inclusion, they will be skipped because
15541 they've already been included (in the precompiled header).
15543 If you need to precompile the same header file for different
15544 languages, targets, or compiler options, you can instead make a
15545 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15546 header in the directory, perhaps using @option{-o}. It doesn't matter
15547 what you call the files in the directory, every precompiled header in
15548 the directory will be considered. The first precompiled header
15549 encountered in the directory that is valid for this compilation will
15550 be used; they're searched in no particular order.
15552 There are many other possibilities, limited only by your imagination,
15553 good sense, and the constraints of your build system.
15555 A precompiled header file can be used only when these conditions apply:
15559 Only one precompiled header can be used in a particular compilation.
15562 A precompiled header can't be used once the first C token is seen. You
15563 can have preprocessor directives before a precompiled header; you can
15564 even include a precompiled header from inside another header, so long as
15565 there are no C tokens before the @code{#include}.
15568 The precompiled header file must be produced for the same language as
15569 the current compilation. You can't use a C precompiled header for a C++
15573 The precompiled header file must have been produced by the same compiler
15574 binary as the current compilation is using.
15577 Any macros defined before the precompiled header is included must
15578 either be defined in the same way as when the precompiled header was
15579 generated, or must not affect the precompiled header, which usually
15580 means that they don't appear in the precompiled header at all.
15582 The @option{-D} option is one way to define a macro before a
15583 precompiled header is included; using a @code{#define} can also do it.
15584 There are also some options that define macros implicitly, like
15585 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15588 @item If debugging information is output when using the precompiled
15589 header, using @option{-g} or similar, the same kind of debugging information
15590 must have been output when building the precompiled header. However,
15591 a precompiled header built using @option{-g} can be used in a compilation
15592 when no debugging information is being output.
15594 @item The same @option{-m} options must generally be used when building
15595 and using the precompiled header. @xref{Submodel Options},
15596 for any cases where this rule is relaxed.
15598 @item Each of the following options must be the same when building and using
15599 the precompiled header:
15601 @gccoptlist{-fexceptions}
15604 Some other command-line options starting with @option{-f},
15605 @option{-p}, or @option{-O} must be defined in the same way as when
15606 the precompiled header was generated. At present, it's not clear
15607 which options are safe to change and which are not; the safest choice
15608 is to use exactly the same options when generating and using the
15609 precompiled header. The following are known to be safe:
15611 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15612 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15613 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15618 For all of these except the last, the compiler will automatically
15619 ignore the precompiled header if the conditions aren't met. If you
15620 find an option combination that doesn't work and doesn't cause the
15621 precompiled header to be ignored, please consider filing a bug report,
15624 If you do use differing options when generating and using the
15625 precompiled header, the actual behavior will be a mixture of the
15626 behavior for the options. For instance, if you use @option{-g} to
15627 generate the precompiled header but not when using it, you may or may
15628 not get debugging information for routines in the precompiled header.
15630 @node Running Protoize
15631 @section Running Protoize
15633 The program @code{protoize} is an optional part of GCC@. You can use
15634 it to add prototypes to a program, thus converting the program to ISO
15635 C in one respect. The companion program @code{unprotoize} does the
15636 reverse: it removes argument types from any prototypes that are found.
15638 When you run these programs, you must specify a set of source files as
15639 command line arguments. The conversion programs start out by compiling
15640 these files to see what functions they define. The information gathered
15641 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15643 After scanning comes actual conversion. The specified files are all
15644 eligible to be converted; any files they include (whether sources or
15645 just headers) are eligible as well.
15647 But not all the eligible files are converted. By default,
15648 @code{protoize} and @code{unprotoize} convert only source and header
15649 files in the current directory. You can specify additional directories
15650 whose files should be converted with the @option{-d @var{directory}}
15651 option. You can also specify particular files to exclude with the
15652 @option{-x @var{file}} option. A file is converted if it is eligible, its
15653 directory name matches one of the specified directory names, and its
15654 name within the directory has not been excluded.
15656 Basic conversion with @code{protoize} consists of rewriting most
15657 function definitions and function declarations to specify the types of
15658 the arguments. The only ones not rewritten are those for varargs
15661 @code{protoize} optionally inserts prototype declarations at the
15662 beginning of the source file, to make them available for any calls that
15663 precede the function's definition. Or it can insert prototype
15664 declarations with block scope in the blocks where undeclared functions
15667 Basic conversion with @code{unprotoize} consists of rewriting most
15668 function declarations to remove any argument types, and rewriting
15669 function definitions to the old-style pre-ISO form.
15671 Both conversion programs print a warning for any function declaration or
15672 definition that they can't convert. You can suppress these warnings
15675 The output from @code{protoize} or @code{unprotoize} replaces the
15676 original source file. The original file is renamed to a name ending
15677 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15678 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15679 for DOS) file already exists, then the source file is simply discarded.
15681 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15682 scan the program and collect information about the functions it uses.
15683 So neither of these programs will work until GCC is installed.
15685 Here is a table of the options you can use with @code{protoize} and
15686 @code{unprotoize}. Each option works with both programs unless
15690 @item -B @var{directory}
15691 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15692 usual directory (normally @file{/usr/local/lib}). This file contains
15693 prototype information about standard system functions. This option
15694 applies only to @code{protoize}.
15696 @item -c @var{compilation-options}
15697 Use @var{compilation-options} as the options when running @command{gcc} to
15698 produce the @samp{.X} files. The special option @option{-aux-info} is
15699 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15701 Note that the compilation options must be given as a single argument to
15702 @code{protoize} or @code{unprotoize}. If you want to specify several
15703 @command{gcc} options, you must quote the entire set of compilation options
15704 to make them a single word in the shell.
15706 There are certain @command{gcc} arguments that you cannot use, because they
15707 would produce the wrong kind of output. These include @option{-g},
15708 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15709 the @var{compilation-options}, they are ignored.
15712 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15713 systems) instead of @samp{.c}. This is convenient if you are converting
15714 a C program to C++. This option applies only to @code{protoize}.
15717 Add explicit global declarations. This means inserting explicit
15718 declarations at the beginning of each source file for each function
15719 that is called in the file and was not declared. These declarations
15720 precede the first function definition that contains a call to an
15721 undeclared function. This option applies only to @code{protoize}.
15723 @item -i @var{string}
15724 Indent old-style parameter declarations with the string @var{string}.
15725 This option applies only to @code{protoize}.
15727 @code{unprotoize} converts prototyped function definitions to old-style
15728 function definitions, where the arguments are declared between the
15729 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15730 uses five spaces as the indentation. If you want to indent with just
15731 one space instead, use @option{-i " "}.
15734 Keep the @samp{.X} files. Normally, they are deleted after conversion
15738 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15739 a prototype declaration for each function in each block which calls the
15740 function without any declaration. This option applies only to
15744 Make no real changes. This mode just prints information about the conversions
15745 that would have been done without @option{-n}.
15748 Make no @samp{.save} files. The original files are simply deleted.
15749 Use this option with caution.
15751 @item -p @var{program}
15752 Use the program @var{program} as the compiler. Normally, the name
15753 @file{gcc} is used.
15756 Work quietly. Most warnings are suppressed.
15759 Print the version number, just like @option{-v} for @command{gcc}.
15762 If you need special compiler options to compile one of your program's
15763 source files, then you should generate that file's @samp{.X} file
15764 specially, by running @command{gcc} on that source file with the
15765 appropriate options and the option @option{-aux-info}. Then run
15766 @code{protoize} on the entire set of files. @code{protoize} will use
15767 the existing @samp{.X} file because it is newer than the source file.
15771 gcc -Dfoo=bar file1.c -aux-info file1.X
15776 You need to include the special files along with the rest in the
15777 @code{protoize} command, even though their @samp{.X} files already
15778 exist, because otherwise they won't get converted.
15780 @xref{Protoize Caveats}, for more information on how to use
15781 @code{protoize} successfully.