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 -mips64 @gol
633 -mips16 -mno-mips16 -mflip-mips16 @gol
634 -minterlink-mips16 -mno-interlink-mips16 @gol
635 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
636 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
637 -mfp32 -mfp64 -mhard-float -msoft-float @gol
638 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
639 -msmartmips -mno-smartmips @gol
640 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
641 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
642 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
643 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
644 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
645 -membedded-data -mno-embedded-data @gol
646 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
647 -mcode-readable=@var{setting} @gol
648 -msplit-addresses -mno-split-addresses @gol
649 -mexplicit-relocs -mno-explicit-relocs @gol
650 -mcheck-zero-division -mno-check-zero-division @gol
651 -mdivide-traps -mdivide-breaks @gol
652 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
653 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
654 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
655 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 -mno-fix-vr4130 @gol
656 -mfix-sb1 -mno-fix-sb1 @gol
657 -mflush-func=@var{func} -mno-flush-func @gol
658 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
659 -mfp-exceptions -mno-fp-exceptions @gol
660 -mvr4130-align -mno-vr4130-align}
663 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
664 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
665 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
666 -mno-base-addresses -msingle-exit -mno-single-exit}
668 @emph{MN10300 Options}
669 @gccoptlist{-mmult-bug -mno-mult-bug @gol
670 -mam33 -mno-am33 @gol
671 -mam33-2 -mno-am33-2 @gol
672 -mreturn-pointer-on-d0 @gol
675 @emph{PDP-11 Options}
676 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
677 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
678 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
679 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
680 -mbranch-expensive -mbranch-cheap @gol
681 -msplit -mno-split -munix-asm -mdec-asm}
683 @emph{PowerPC Options}
684 See RS/6000 and PowerPC Options.
686 @emph{RS/6000 and PowerPC Options}
687 @gccoptlist{-mcpu=@var{cpu-type} @gol
688 -mtune=@var{cpu-type} @gol
689 -mpower -mno-power -mpower2 -mno-power2 @gol
690 -mpowerpc -mpowerpc64 -mno-powerpc @gol
691 -maltivec -mno-altivec @gol
692 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
693 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
694 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
695 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
696 -mnew-mnemonics -mold-mnemonics @gol
697 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
698 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
699 -malign-power -malign-natural @gol
700 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
701 -mstring -mno-string -mupdate -mno-update @gol
702 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
703 -mstrict-align -mno-strict-align -mrelocatable @gol
704 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
705 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
706 -mdynamic-no-pic -maltivec -mswdiv @gol
707 -mprioritize-restricted-insns=@var{priority} @gol
708 -msched-costly-dep=@var{dependence_type} @gol
709 -minsert-sched-nops=@var{scheme} @gol
710 -mcall-sysv -mcall-netbsd @gol
711 -maix-struct-return -msvr4-struct-return @gol
712 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
713 -misel -mno-isel @gol
714 -misel=yes -misel=no @gol
716 -mspe=yes -mspe=no @gol
718 -mvrsave -mno-vrsave @gol
719 -mmulhw -mno-mulhw @gol
720 -mdlmzb -mno-dlmzb @gol
721 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
722 -mprototype -mno-prototype @gol
723 -msim -mmvme -mads -myellowknife -memb -msdata @gol
724 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
726 @emph{S/390 and zSeries Options}
727 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
728 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
729 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
730 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
731 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
732 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
733 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
736 @gccoptlist{-meb -mel @gol
740 -mscore5 -mscore5u -mscore7 -mscore7d}
743 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
744 -m4-nofpu -m4-single-only -m4-single -m4 @gol
745 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
746 -m5-64media -m5-64media-nofpu @gol
747 -m5-32media -m5-32media-nofpu @gol
748 -m5-compact -m5-compact-nofpu @gol
749 -mb -ml -mdalign -mrelax @gol
750 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
751 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
752 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
753 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
754 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
758 @gccoptlist{-mcpu=@var{cpu-type} @gol
759 -mtune=@var{cpu-type} @gol
760 -mcmodel=@var{code-model} @gol
761 -m32 -m64 -mapp-regs -mno-app-regs @gol
762 -mfaster-structs -mno-faster-structs @gol
763 -mfpu -mno-fpu -mhard-float -msoft-float @gol
764 -mhard-quad-float -msoft-quad-float @gol
765 -mimpure-text -mno-impure-text -mlittle-endian @gol
766 -mstack-bias -mno-stack-bias @gol
767 -munaligned-doubles -mno-unaligned-doubles @gol
768 -mv8plus -mno-v8plus -mvis -mno-vis
769 -threads -pthreads -pthread}
772 @gccoptlist{-mwarn-reloc -merror-reloc @gol
773 -msafe-dma -munsafe-dma @gol
775 -msmall-mem -mlarge-mem -mstdmain @gol
776 -mfixed-range=@var{register-range}}
778 @emph{System V Options}
779 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
782 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
783 -mprolog-function -mno-prolog-function -mspace @gol
784 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
785 -mapp-regs -mno-app-regs @gol
786 -mdisable-callt -mno-disable-callt @gol
792 @gccoptlist{-mg -mgnu -munix}
794 @emph{VxWorks Options}
795 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
796 -Xbind-lazy -Xbind-now}
798 @emph{x86-64 Options}
799 See i386 and x86-64 Options.
801 @emph{Xstormy16 Options}
804 @emph{Xtensa Options}
805 @gccoptlist{-mconst16 -mno-const16 @gol
806 -mfused-madd -mno-fused-madd @gol
807 -mserialize-volatile -mno-serialize-volatile @gol
808 -mtext-section-literals -mno-text-section-literals @gol
809 -mtarget-align -mno-target-align @gol
810 -mlongcalls -mno-longcalls}
812 @emph{zSeries Options}
813 See S/390 and zSeries Options.
815 @item Code Generation Options
816 @xref{Code Gen Options,,Options for Code Generation Conventions}.
817 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
818 -ffixed-@var{reg} -fexceptions @gol
819 -fnon-call-exceptions -funwind-tables @gol
820 -fasynchronous-unwind-tables @gol
821 -finhibit-size-directive -finstrument-functions @gol
822 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
823 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
824 -fno-common -fno-ident @gol
825 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
826 -fno-jump-tables @gol
827 -frecord-gcc-switches @gol
828 -freg-struct-return -fshort-enums @gol
829 -fshort-double -fshort-wchar @gol
830 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
831 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
832 -fno-stack-limit -fargument-alias -fargument-noalias @gol
833 -fargument-noalias-global -fargument-noalias-anything @gol
834 -fleading-underscore -ftls-model=@var{model} @gol
835 -ftrapv -fwrapv -fbounds-check @gol
840 * Overall Options:: Controlling the kind of output:
841 an executable, object files, assembler files,
842 or preprocessed source.
843 * C Dialect Options:: Controlling the variant of C language compiled.
844 * C++ Dialect Options:: Variations on C++.
845 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
847 * Language Independent Options:: Controlling how diagnostics should be
849 * Warning Options:: How picky should the compiler be?
850 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
851 * Optimize Options:: How much optimization?
852 * Preprocessor Options:: Controlling header files and macro definitions.
853 Also, getting dependency information for Make.
854 * Assembler Options:: Passing options to the assembler.
855 * Link Options:: Specifying libraries and so on.
856 * Directory Options:: Where to find header files and libraries.
857 Where to find the compiler executable files.
858 * Spec Files:: How to pass switches to sub-processes.
859 * Target Options:: Running a cross-compiler, or an old version of GCC.
862 @node Overall Options
863 @section Options Controlling the Kind of Output
865 Compilation can involve up to four stages: preprocessing, compilation
866 proper, assembly and linking, always in that order. GCC is capable of
867 preprocessing and compiling several files either into several
868 assembler input files, or into one assembler input file; then each
869 assembler input file produces an object file, and linking combines all
870 the object files (those newly compiled, and those specified as input)
871 into an executable file.
873 @cindex file name suffix
874 For any given input file, the file name suffix determines what kind of
879 C source code which must be preprocessed.
882 C source code which should not be preprocessed.
885 C++ source code which should not be preprocessed.
888 Objective-C source code. Note that you must link with the @file{libobjc}
889 library to make an Objective-C program work.
892 Objective-C source code which should not be preprocessed.
896 Objective-C++ source code. Note that you must link with the @file{libobjc}
897 library to make an Objective-C++ program work. Note that @samp{.M} refers
898 to a literal capital M@.
901 Objective-C++ source code which should not be preprocessed.
904 C, C++, Objective-C or Objective-C++ header file to be turned into a
909 @itemx @var{file}.cxx
910 @itemx @var{file}.cpp
911 @itemx @var{file}.CPP
912 @itemx @var{file}.c++
914 C++ source code which must be preprocessed. Note that in @samp{.cxx},
915 the last two letters must both be literally @samp{x}. Likewise,
916 @samp{.C} refers to a literal capital C@.
920 Objective-C++ source code which must be preprocessed.
923 Objective-C++ source code which should not be preprocessed.
928 @itemx @var{file}.hxx
929 @itemx @var{file}.hpp
930 @itemx @var{file}.HPP
931 @itemx @var{file}.h++
932 @itemx @var{file}.tcc
933 C++ header file to be turned into a precompiled header.
936 @itemx @var{file}.for
937 @itemx @var{file}.ftn
938 Fixed form Fortran source code which should not be preprocessed.
941 @itemx @var{file}.FOR
942 @itemx @var{file}.fpp
943 @itemx @var{file}.FPP
944 @itemx @var{file}.FTN
945 Fixed form Fortran source code which must be preprocessed (with the traditional
949 @itemx @var{file}.f95
950 @itemx @var{file}.f03
951 @itemx @var{file}.f08
952 Free form Fortran source code which should not be preprocessed.
955 @itemx @var{file}.F95
956 @itemx @var{file}.F03
957 @itemx @var{file}.F08
958 Free form Fortran source code which must be preprocessed (with the
959 traditional preprocessor).
961 @c FIXME: Descriptions of Java file types.
968 Ada source code file which contains a library unit declaration (a
969 declaration of a package, subprogram, or generic, or a generic
970 instantiation), or a library unit renaming declaration (a package,
971 generic, or subprogram renaming declaration). Such files are also
975 Ada source code file containing a library unit body (a subprogram or
976 package body). Such files are also called @dfn{bodies}.
978 @c GCC also knows about some suffixes for languages not yet included:
990 Assembler code which must be preprocessed.
993 An object file to be fed straight into linking.
994 Any file name with no recognized suffix is treated this way.
998 You can specify the input language explicitly with the @option{-x} option:
1001 @item -x @var{language}
1002 Specify explicitly the @var{language} for the following input files
1003 (rather than letting the compiler choose a default based on the file
1004 name suffix). This option applies to all following input files until
1005 the next @option{-x} option. Possible values for @var{language} are:
1007 c c-header c-cpp-output
1008 c++ c++-header c++-cpp-output
1009 objective-c objective-c-header objective-c-cpp-output
1010 objective-c++ objective-c++-header objective-c++-cpp-output
1011 assembler assembler-with-cpp
1013 f77 f77-cpp-input f95 f95-cpp-input
1018 Turn off any specification of a language, so that subsequent files are
1019 handled according to their file name suffixes (as they are if @option{-x}
1020 has not been used at all).
1022 @item -pass-exit-codes
1023 @opindex pass-exit-codes
1024 Normally the @command{gcc} program will exit with the code of 1 if any
1025 phase of the compiler returns a non-success return code. If you specify
1026 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1027 numerically highest error produced by any phase that returned an error
1028 indication. The C, C++, and Fortran frontends return 4, if an internal
1029 compiler error is encountered.
1032 If you only want some of the stages of compilation, you can use
1033 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1034 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1035 @command{gcc} is to stop. Note that some combinations (for example,
1036 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1041 Compile or assemble the source files, but do not link. The linking
1042 stage simply is not done. The ultimate output is in the form of an
1043 object file for each source file.
1045 By default, the object file name for a source file is made by replacing
1046 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1048 Unrecognized input files, not requiring compilation or assembly, are
1053 Stop after the stage of compilation proper; do not assemble. The output
1054 is in the form of an assembler code file for each non-assembler input
1057 By default, the assembler file name for a source file is made by
1058 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1060 Input files that don't require compilation are ignored.
1064 Stop after the preprocessing stage; do not run the compiler proper. The
1065 output is in the form of preprocessed source code, which is sent to the
1068 Input files which don't require preprocessing are ignored.
1070 @cindex output file option
1073 Place output in file @var{file}. This applies regardless to whatever
1074 sort of output is being produced, whether it be an executable file,
1075 an object file, an assembler file or preprocessed C code.
1077 If @option{-o} is not specified, the default is to put an executable
1078 file in @file{a.out}, the object file for
1079 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1080 assembler file in @file{@var{source}.s}, a precompiled header file in
1081 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1086 Print (on standard error output) the commands executed to run the stages
1087 of compilation. Also print the version number of the compiler driver
1088 program and of the preprocessor and the compiler proper.
1092 Like @option{-v} except the commands are not executed and all command
1093 arguments are quoted. This is useful for shell scripts to capture the
1094 driver-generated command lines.
1098 Use pipes rather than temporary files for communication between the
1099 various stages of compilation. This fails to work on some systems where
1100 the assembler is unable to read from a pipe; but the GNU assembler has
1105 If you are compiling multiple source files, this option tells the driver
1106 to pass all the source files to the compiler at once (for those
1107 languages for which the compiler can handle this). This will allow
1108 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1109 language for which this is supported is C@. If you pass source files for
1110 multiple languages to the driver, using this option, the driver will invoke
1111 the compiler(s) that support IMA once each, passing each compiler all the
1112 source files appropriate for it. For those languages that do not support
1113 IMA this option will be ignored, and the compiler will be invoked once for
1114 each source file in that language. If you use this option in conjunction
1115 with @option{-save-temps}, the compiler will generate multiple
1117 (one for each source file), but only one (combined) @file{.o} or
1122 Print (on the standard output) a description of the command line options
1123 understood by @command{gcc}. If the @option{-v} option is also specified
1124 then @option{--help} will also be passed on to the various processes
1125 invoked by @command{gcc}, so that they can display the command line options
1126 they accept. If the @option{-Wextra} option has also been specified
1127 (prior to the @option{--help} option), then command line options which
1128 have no documentation associated with them will also be displayed.
1131 @opindex target-help
1132 Print (on the standard output) a description of target-specific command
1133 line options for each tool. For some targets extra target-specific
1134 information may also be printed.
1136 @item --help=@var{class}@r{[},@var{qualifier}@r{]}
1137 Print (on the standard output) a description of the command line
1138 options understood by the compiler that fit into a specific class.
1139 The class can be one of @samp{optimizers}, @samp{warnings}, @samp{target},
1140 @samp{params}, or @var{language}:
1143 @item @samp{optimizers}
1144 This will display all of the optimization options supported by the
1147 @item @samp{warnings}
1148 This will display all of the options controlling warning messages
1149 produced by the compiler.
1152 This will display target-specific options. Unlike the
1153 @option{--target-help} option however, target-specific options of the
1154 linker and assembler will not be displayed. This is because those
1155 tools do not currently support the extended @option{--help=} syntax.
1158 This will display the values recognized by the @option{--param}
1161 @item @var{language}
1162 This will display the options supported for @var{language}, where
1163 @var{language} is the name of one of the languages supported in this
1167 This will display the options that are common to all languages.
1170 It is possible to further refine the output of the @option{--help=}
1171 option by adding a comma separated list of qualifiers after the
1172 class. These can be any from the following list:
1175 @item @samp{undocumented}
1176 Display only those options which are undocumented.
1179 Display options which take an argument that appears after an equal
1180 sign in the same continuous piece of text, such as:
1181 @samp{--help=target}.
1183 @item @samp{separate}
1184 Display options which take an argument that appears as a separate word
1185 following the original option, such as: @samp{-o output-file}.
1188 Thus for example to display all the undocumented target-specific
1189 switches supported by the compiler the following can be used:
1192 --help=target,undocumented
1195 The sense of a qualifier can be inverted by prefixing it with the
1196 @var{^} character, so for example to display all binary warning
1197 options (i.e., ones that are either on or off and that do not take an
1198 argument), which have a description the following can be used:
1201 --help=warnings,^joined,^undocumented
1204 A class can also be used as a qualifier, although this usually
1205 restricts the output by so much that there is nothing to display. One
1206 case where it does work however is when one of the classes is
1207 @var{target}. So for example to display all the target-specific
1208 optimization options the following can be used:
1211 --help=target,optimizers
1214 The @option{--help=} option can be repeated on the command line. Each
1215 successive use will display its requested class of options, skipping
1216 those that have already been displayed.
1218 If the @option{-Q} option appears on the command line before the
1219 @option{--help=} option, then the descriptive text displayed by
1220 @option{--help=} is changed. Instead of describing the displayed
1221 options, an indication is given as to whether the option is enabled,
1222 disabled or set to a specific value (assuming that the compiler
1223 knows this at the point where the @option{--help=} option is used).
1225 Here is a truncated example from the ARM port of @command{gcc}:
1228 % gcc -Q -mabi=2 --help=target -c
1229 The following options are target specific:
1231 -mabort-on-noreturn [disabled]
1235 The output is sensitive to the effects of previous command line
1236 options, so for example it is possible to find out which optimizations
1237 are enabled at @option{-O2} by using:
1240 -O2 --help=optimizers
1243 Alternatively you can discover which binary optimizations are enabled
1244 by @option{-O3} by using:
1247 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1248 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1249 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1254 Display the version number and copyrights of the invoked GCC@.
1258 Invoke all subcommands under a wrapper program. It takes a single
1259 comma separated list as an argument, which will be used to invoke
1263 gcc -c t.c -wrapper gdb,--args
1266 This will invoke all subprograms of gcc under "gdb --args",
1267 thus cc1 invocation will be "gdb --args cc1 ...".
1269 @include @value{srcdir}/../libiberty/at-file.texi
1273 @section Compiling C++ Programs
1275 @cindex suffixes for C++ source
1276 @cindex C++ source file suffixes
1277 C++ source files conventionally use one of the suffixes @samp{.C},
1278 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1279 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1280 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1281 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1282 files with these names and compiles them as C++ programs even if you
1283 call the compiler the same way as for compiling C programs (usually
1284 with the name @command{gcc}).
1288 However, the use of @command{gcc} does not add the C++ library.
1289 @command{g++} is a program that calls GCC and treats @samp{.c},
1290 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1291 files unless @option{-x} is used, and automatically specifies linking
1292 against the C++ library. This program is also useful when
1293 precompiling a C header file with a @samp{.h} extension for use in C++
1294 compilations. On many systems, @command{g++} is also installed with
1295 the name @command{c++}.
1297 @cindex invoking @command{g++}
1298 When you compile C++ programs, you may specify many of the same
1299 command-line options that you use for compiling programs in any
1300 language; or command-line options meaningful for C and related
1301 languages; or options that are meaningful only for C++ programs.
1302 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1303 explanations of options for languages related to C@.
1304 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1305 explanations of options that are meaningful only for C++ programs.
1307 @node C Dialect Options
1308 @section Options Controlling C Dialect
1309 @cindex dialect options
1310 @cindex language dialect options
1311 @cindex options, dialect
1313 The following options control the dialect of C (or languages derived
1314 from C, such as C++, Objective-C and Objective-C++) that the compiler
1318 @cindex ANSI support
1322 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1323 equivalent to @samp{-std=c++98}.
1325 This turns off certain features of GCC that are incompatible with ISO
1326 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1327 such as the @code{asm} and @code{typeof} keywords, and
1328 predefined macros such as @code{unix} and @code{vax} that identify the
1329 type of system you are using. It also enables the undesirable and
1330 rarely used ISO trigraph feature. For the C compiler,
1331 it disables recognition of C++ style @samp{//} comments as well as
1332 the @code{inline} keyword.
1334 The alternate keywords @code{__asm__}, @code{__extension__},
1335 @code{__inline__} and @code{__typeof__} continue to work despite
1336 @option{-ansi}. You would not want to use them in an ISO C program, of
1337 course, but it is useful to put them in header files that might be included
1338 in compilations done with @option{-ansi}. Alternate predefined macros
1339 such as @code{__unix__} and @code{__vax__} are also available, with or
1340 without @option{-ansi}.
1342 The @option{-ansi} option does not cause non-ISO programs to be
1343 rejected gratuitously. For that, @option{-pedantic} is required in
1344 addition to @option{-ansi}. @xref{Warning Options}.
1346 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1347 option is used. Some header files may notice this macro and refrain
1348 from declaring certain functions or defining certain macros that the
1349 ISO standard doesn't call for; this is to avoid interfering with any
1350 programs that might use these names for other things.
1352 Functions that would normally be built in but do not have semantics
1353 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1354 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1355 built-in functions provided by GCC}, for details of the functions
1360 Determine the language standard. @xref{Standards,,Language Standards
1361 Supported by GCC}, for details of these standard versions. This option
1362 is currently only supported when compiling C or C++.
1364 The compiler can accept several base standards, such as @samp{c89} or
1365 @samp{c++98}, and GNU dialects of those standards, such as
1366 @samp{gnu89} or @samp{gnu++98}. By specifing a base standard, the
1367 compiler will accept all programs following that standard and those
1368 using GNU extensions that do not contradict it. For example,
1369 @samp{-std=c89} turns off certain features of GCC that are
1370 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1371 keywords, but not other GNU extensions that do not have a meaning in
1372 ISO C90, such as omitting the middle term of a @code{?:}
1373 expression. On the other hand, by specifing a GNU dialect of a
1374 standard, all features the compiler support are enabled, even when
1375 those features change the meaning of the base standard and some
1376 strict-conforming programs may be rejected. The particular standard
1377 is used by @option{-pedantic} to identify which features are GNU
1378 extensions given that version of the standard. For example
1379 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1380 comments, while @samp{-std=gnu99 -pedantic} would not.
1382 A value for this option must be provided; possible values are
1387 Support all ISO C90 programs (certain GNU extensions that conflict
1388 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1390 @item iso9899:199409
1391 ISO C90 as modified in amendment 1.
1397 ISO C99. Note that this standard is not yet fully supported; see
1398 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1399 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1402 GNU dialect of ISO C90 (including some C99 features). This
1403 is the default for C code.
1407 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1408 this will become the default. The name @samp{gnu9x} is deprecated.
1411 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1415 GNU dialect of @option{-std=c++98}. This is the default for
1419 The working draft of the upcoming ISO C++0x standard. This option
1420 enables experimental features that are likely to be included in
1421 C++0x. The working draft is constantly changing, and any feature that is
1422 enabled by this flag may be removed from future versions of GCC if it is
1423 not part of the C++0x standard.
1426 GNU dialect of @option{-std=c++0x}. This option enables
1427 experimental features that may be removed in future versions of GCC.
1430 @item -fgnu89-inline
1431 @opindex fgnu89-inline
1432 The option @option{-fgnu89-inline} tells GCC to use the traditional
1433 GNU semantics for @code{inline} functions when in C99 mode.
1434 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1435 is accepted and ignored by GCC versions 4.1.3 up to but not including
1436 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1437 C99 mode. Using this option is roughly equivalent to adding the
1438 @code{gnu_inline} function attribute to all inline functions
1439 (@pxref{Function Attributes}).
1441 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1442 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1443 specifies the default behavior). This option was first supported in
1444 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1446 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1447 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1448 in effect for @code{inline} functions. @xref{Common Predefined
1449 Macros,,,cpp,The C Preprocessor}.
1451 @item -aux-info @var{filename}
1453 Output to the given filename prototyped declarations for all functions
1454 declared and/or defined in a translation unit, including those in header
1455 files. This option is silently ignored in any language other than C@.
1457 Besides declarations, the file indicates, in comments, the origin of
1458 each declaration (source file and line), whether the declaration was
1459 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1460 @samp{O} for old, respectively, in the first character after the line
1461 number and the colon), and whether it came from a declaration or a
1462 definition (@samp{C} or @samp{F}, respectively, in the following
1463 character). In the case of function definitions, a K&R-style list of
1464 arguments followed by their declarations is also provided, inside
1465 comments, after the declaration.
1469 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1470 keyword, so that code can use these words as identifiers. You can use
1471 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1472 instead. @option{-ansi} implies @option{-fno-asm}.
1474 In C++, this switch only affects the @code{typeof} keyword, since
1475 @code{asm} and @code{inline} are standard keywords. You may want to
1476 use the @option{-fno-gnu-keywords} flag instead, which has the same
1477 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1478 switch only affects the @code{asm} and @code{typeof} keywords, since
1479 @code{inline} is a standard keyword in ISO C99.
1482 @itemx -fno-builtin-@var{function}
1483 @opindex fno-builtin
1484 @cindex built-in functions
1485 Don't recognize built-in functions that do not begin with
1486 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1487 functions provided by GCC}, for details of the functions affected,
1488 including those which are not built-in functions when @option{-ansi} or
1489 @option{-std} options for strict ISO C conformance are used because they
1490 do not have an ISO standard meaning.
1492 GCC normally generates special code to handle certain built-in functions
1493 more efficiently; for instance, calls to @code{alloca} may become single
1494 instructions that adjust the stack directly, and calls to @code{memcpy}
1495 may become inline copy loops. The resulting code is often both smaller
1496 and faster, but since the function calls no longer appear as such, you
1497 cannot set a breakpoint on those calls, nor can you change the behavior
1498 of the functions by linking with a different library. In addition,
1499 when a function is recognized as a built-in function, GCC may use
1500 information about that function to warn about problems with calls to
1501 that function, or to generate more efficient code, even if the
1502 resulting code still contains calls to that function. For example,
1503 warnings are given with @option{-Wformat} for bad calls to
1504 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1505 known not to modify global memory.
1507 With the @option{-fno-builtin-@var{function}} option
1508 only the built-in function @var{function} is
1509 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1510 function is named this is not built-in in this version of GCC, this
1511 option is ignored. There is no corresponding
1512 @option{-fbuiltin-@var{function}} option; if you wish to enable
1513 built-in functions selectively when using @option{-fno-builtin} or
1514 @option{-ffreestanding}, you may define macros such as:
1517 #define abs(n) __builtin_abs ((n))
1518 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1523 @cindex hosted environment
1525 Assert that compilation takes place in a hosted environment. This implies
1526 @option{-fbuiltin}. A hosted environment is one in which the
1527 entire standard library is available, and in which @code{main} has a return
1528 type of @code{int}. Examples are nearly everything except a kernel.
1529 This is equivalent to @option{-fno-freestanding}.
1531 @item -ffreestanding
1532 @opindex ffreestanding
1533 @cindex hosted environment
1535 Assert that compilation takes place in a freestanding environment. This
1536 implies @option{-fno-builtin}. A freestanding environment
1537 is one in which the standard library may not exist, and program startup may
1538 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1539 This is equivalent to @option{-fno-hosted}.
1541 @xref{Standards,,Language Standards Supported by GCC}, for details of
1542 freestanding and hosted environments.
1546 @cindex openmp parallel
1547 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1548 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1549 compiler generates parallel code according to the OpenMP Application
1550 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}. This option
1551 implies @option{-pthread}, and thus is only supported on targets that
1552 have support for @option{-pthread}.
1554 @item -fms-extensions
1555 @opindex fms-extensions
1556 Accept some non-standard constructs used in Microsoft header files.
1558 Some cases of unnamed fields in structures and unions are only
1559 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1560 fields within structs/unions}, for details.
1564 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1565 options for strict ISO C conformance) implies @option{-trigraphs}.
1567 @item -no-integrated-cpp
1568 @opindex no-integrated-cpp
1569 Performs a compilation in two passes: preprocessing and compiling. This
1570 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1571 @option{-B} option. The user supplied compilation step can then add in
1572 an additional preprocessing step after normal preprocessing but before
1573 compiling. The default is to use the integrated cpp (internal cpp)
1575 The semantics of this option will change if "cc1", "cc1plus", and
1576 "cc1obj" are merged.
1578 @cindex traditional C language
1579 @cindex C language, traditional
1581 @itemx -traditional-cpp
1582 @opindex traditional-cpp
1583 @opindex traditional
1584 Formerly, these options caused GCC to attempt to emulate a pre-standard
1585 C compiler. They are now only supported with the @option{-E} switch.
1586 The preprocessor continues to support a pre-standard mode. See the GNU
1587 CPP manual for details.
1589 @item -fcond-mismatch
1590 @opindex fcond-mismatch
1591 Allow conditional expressions with mismatched types in the second and
1592 third arguments. The value of such an expression is void. This option
1593 is not supported for C++.
1595 @item -flax-vector-conversions
1596 @opindex flax-vector-conversions
1597 Allow implicit conversions between vectors with differing numbers of
1598 elements and/or incompatible element types. This option should not be
1601 @item -funsigned-char
1602 @opindex funsigned-char
1603 Let the type @code{char} be unsigned, like @code{unsigned char}.
1605 Each kind of machine has a default for what @code{char} should
1606 be. It is either like @code{unsigned char} by default or like
1607 @code{signed char} by default.
1609 Ideally, a portable program should always use @code{signed char} or
1610 @code{unsigned char} when it depends on the signedness of an object.
1611 But many programs have been written to use plain @code{char} and
1612 expect it to be signed, or expect it to be unsigned, depending on the
1613 machines they were written for. This option, and its inverse, let you
1614 make such a program work with the opposite default.
1616 The type @code{char} is always a distinct type from each of
1617 @code{signed char} or @code{unsigned char}, even though its behavior
1618 is always just like one of those two.
1621 @opindex fsigned-char
1622 Let the type @code{char} be signed, like @code{signed char}.
1624 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1625 the negative form of @option{-funsigned-char}. Likewise, the option
1626 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1628 @item -fsigned-bitfields
1629 @itemx -funsigned-bitfields
1630 @itemx -fno-signed-bitfields
1631 @itemx -fno-unsigned-bitfields
1632 @opindex fsigned-bitfields
1633 @opindex funsigned-bitfields
1634 @opindex fno-signed-bitfields
1635 @opindex fno-unsigned-bitfields
1636 These options control whether a bit-field is signed or unsigned, when the
1637 declaration does not use either @code{signed} or @code{unsigned}. By
1638 default, such a bit-field is signed, because this is consistent: the
1639 basic integer types such as @code{int} are signed types.
1642 @node C++ Dialect Options
1643 @section Options Controlling C++ Dialect
1645 @cindex compiler options, C++
1646 @cindex C++ options, command line
1647 @cindex options, C++
1648 This section describes the command-line options that are only meaningful
1649 for C++ programs; but you can also use most of the GNU compiler options
1650 regardless of what language your program is in. For example, you
1651 might compile a file @code{firstClass.C} like this:
1654 g++ -g -frepo -O -c firstClass.C
1658 In this example, only @option{-frepo} is an option meant
1659 only for C++ programs; you can use the other options with any
1660 language supported by GCC@.
1662 Here is a list of options that are @emph{only} for compiling C++ programs:
1666 @item -fabi-version=@var{n}
1667 @opindex fabi-version
1668 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1669 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1670 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1671 the version that conforms most closely to the C++ ABI specification.
1672 Therefore, the ABI obtained using version 0 will change as ABI bugs
1675 The default is version 2.
1677 @item -fno-access-control
1678 @opindex fno-access-control
1679 Turn off all access checking. This switch is mainly useful for working
1680 around bugs in the access control code.
1684 Check that the pointer returned by @code{operator new} is non-null
1685 before attempting to modify the storage allocated. This check is
1686 normally unnecessary because the C++ standard specifies that
1687 @code{operator new} will only return @code{0} if it is declared
1688 @samp{throw()}, in which case the compiler will always check the
1689 return value even without this option. In all other cases, when
1690 @code{operator new} has a non-empty exception specification, memory
1691 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1692 @samp{new (nothrow)}.
1694 @item -fconserve-space
1695 @opindex fconserve-space
1696 Put uninitialized or runtime-initialized global variables into the
1697 common segment, as C does. This saves space in the executable at the
1698 cost of not diagnosing duplicate definitions. If you compile with this
1699 flag and your program mysteriously crashes after @code{main()} has
1700 completed, you may have an object that is being destroyed twice because
1701 two definitions were merged.
1703 This option is no longer useful on most targets, now that support has
1704 been added for putting variables into BSS without making them common.
1706 @item -ffriend-injection
1707 @opindex ffriend-injection
1708 Inject friend functions into the enclosing namespace, so that they are
1709 visible outside the scope of the class in which they are declared.
1710 Friend functions were documented to work this way in the old Annotated
1711 C++ Reference Manual, and versions of G++ before 4.1 always worked
1712 that way. However, in ISO C++ a friend function which is not declared
1713 in an enclosing scope can only be found using argument dependent
1714 lookup. This option causes friends to be injected as they were in
1717 This option is for compatibility, and may be removed in a future
1720 @item -fno-elide-constructors
1721 @opindex fno-elide-constructors
1722 The C++ standard allows an implementation to omit creating a temporary
1723 which is only used to initialize another object of the same type.
1724 Specifying this option disables that optimization, and forces G++ to
1725 call the copy constructor in all cases.
1727 @item -fno-enforce-eh-specs
1728 @opindex fno-enforce-eh-specs
1729 Don't generate code to check for violation of exception specifications
1730 at runtime. This option violates the C++ standard, but may be useful
1731 for reducing code size in production builds, much like defining
1732 @samp{NDEBUG}. This does not give user code permission to throw
1733 exceptions in violation of the exception specifications; the compiler
1734 will still optimize based on the specifications, so throwing an
1735 unexpected exception will result in undefined behavior.
1738 @itemx -fno-for-scope
1740 @opindex fno-for-scope
1741 If @option{-ffor-scope} is specified, the scope of variables declared in
1742 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1743 as specified by the C++ standard.
1744 If @option{-fno-for-scope} is specified, the scope of variables declared in
1745 a @i{for-init-statement} extends to the end of the enclosing scope,
1746 as was the case in old versions of G++, and other (traditional)
1747 implementations of C++.
1749 The default if neither flag is given to follow the standard,
1750 but to allow and give a warning for old-style code that would
1751 otherwise be invalid, or have different behavior.
1753 @item -fno-gnu-keywords
1754 @opindex fno-gnu-keywords
1755 Do not recognize @code{typeof} as a keyword, so that code can use this
1756 word as an identifier. You can use the keyword @code{__typeof__} instead.
1757 @option{-ansi} implies @option{-fno-gnu-keywords}.
1759 @item -fno-implicit-templates
1760 @opindex fno-implicit-templates
1761 Never emit code for non-inline templates which are instantiated
1762 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1763 @xref{Template Instantiation}, for more information.
1765 @item -fno-implicit-inline-templates
1766 @opindex fno-implicit-inline-templates
1767 Don't emit code for implicit instantiations of inline templates, either.
1768 The default is to handle inlines differently so that compiles with and
1769 without optimization will need the same set of explicit instantiations.
1771 @item -fno-implement-inlines
1772 @opindex fno-implement-inlines
1773 To save space, do not emit out-of-line copies of inline functions
1774 controlled by @samp{#pragma implementation}. This will cause linker
1775 errors if these functions are not inlined everywhere they are called.
1777 @item -fms-extensions
1778 @opindex fms-extensions
1779 Disable pedantic warnings about constructs used in MFC, such as implicit
1780 int and getting a pointer to member function via non-standard syntax.
1782 @item -fno-nonansi-builtins
1783 @opindex fno-nonansi-builtins
1784 Disable built-in declarations of functions that are not mandated by
1785 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1786 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1788 @item -fno-operator-names
1789 @opindex fno-operator-names
1790 Do not treat the operator name keywords @code{and}, @code{bitand},
1791 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1792 synonyms as keywords.
1794 @item -fno-optional-diags
1795 @opindex fno-optional-diags
1796 Disable diagnostics that the standard says a compiler does not need to
1797 issue. Currently, the only such diagnostic issued by G++ is the one for
1798 a name having multiple meanings within a class.
1801 @opindex fpermissive
1802 Downgrade some diagnostics about nonconformant code from errors to
1803 warnings. Thus, using @option{-fpermissive} will allow some
1804 nonconforming code to compile.
1808 Enable automatic template instantiation at link time. This option also
1809 implies @option{-fno-implicit-templates}. @xref{Template
1810 Instantiation}, for more information.
1814 Disable generation of information about every class with virtual
1815 functions for use by the C++ runtime type identification features
1816 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1817 of the language, you can save some space by using this flag. Note that
1818 exception handling uses the same information, but it will generate it as
1819 needed. The @samp{dynamic_cast} operator can still be used for casts that
1820 do not require runtime type information, i.e.@: casts to @code{void *} or to
1821 unambiguous base classes.
1825 Emit statistics about front-end processing at the end of the compilation.
1826 This information is generally only useful to the G++ development team.
1828 @item -ftemplate-depth-@var{n}
1829 @opindex ftemplate-depth
1830 Set the maximum instantiation depth for template classes to @var{n}.
1831 A limit on the template instantiation depth is needed to detect
1832 endless recursions during template class instantiation. ANSI/ISO C++
1833 conforming programs must not rely on a maximum depth greater than 17.
1835 @item -fno-threadsafe-statics
1836 @opindex fno-threadsafe-statics
1837 Do not emit the extra code to use the routines specified in the C++
1838 ABI for thread-safe initialization of local statics. You can use this
1839 option to reduce code size slightly in code that doesn't need to be
1842 @item -fuse-cxa-atexit
1843 @opindex fuse-cxa-atexit
1844 Register destructors for objects with static storage duration with the
1845 @code{__cxa_atexit} function rather than the @code{atexit} function.
1846 This option is required for fully standards-compliant handling of static
1847 destructors, but will only work if your C library supports
1848 @code{__cxa_atexit}.
1850 @item -fno-use-cxa-get-exception-ptr
1851 @opindex fno-use-cxa-get-exception-ptr
1852 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1853 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1854 if the runtime routine is not available.
1856 @item -fvisibility-inlines-hidden
1857 @opindex fvisibility-inlines-hidden
1858 This switch declares that the user does not attempt to compare
1859 pointers to inline methods where the addresses of the two functions
1860 were taken in different shared objects.
1862 The effect of this is that GCC may, effectively, mark inline methods with
1863 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1864 appear in the export table of a DSO and do not require a PLT indirection
1865 when used within the DSO@. Enabling this option can have a dramatic effect
1866 on load and link times of a DSO as it massively reduces the size of the
1867 dynamic export table when the library makes heavy use of templates.
1869 The behavior of this switch is not quite the same as marking the
1870 methods as hidden directly, because it does not affect static variables
1871 local to the function or cause the compiler to deduce that
1872 the function is defined in only one shared object.
1874 You may mark a method as having a visibility explicitly to negate the
1875 effect of the switch for that method. For example, if you do want to
1876 compare pointers to a particular inline method, you might mark it as
1877 having default visibility. Marking the enclosing class with explicit
1878 visibility will have no effect.
1880 Explicitly instantiated inline methods are unaffected by this option
1881 as their linkage might otherwise cross a shared library boundary.
1882 @xref{Template Instantiation}.
1884 @item -fvisibility-ms-compat
1885 @opindex fvisibility-ms-compat
1886 This flag attempts to use visibility settings to make GCC's C++
1887 linkage model compatible with that of Microsoft Visual Studio.
1889 The flag makes these changes to GCC's linkage model:
1893 It sets the default visibility to @code{hidden}, like
1894 @option{-fvisibility=hidden}.
1897 Types, but not their members, are not hidden by default.
1900 The One Definition Rule is relaxed for types without explicit
1901 visibility specifications which are defined in more than one different
1902 shared object: those declarations are permitted if they would have
1903 been permitted when this option was not used.
1906 In new code it is better to use @option{-fvisibility=hidden} and
1907 export those classes which are intended to be externally visible.
1908 Unfortunately it is possible for code to rely, perhaps accidentally,
1909 on the Visual Studio behavior.
1911 Among the consequences of these changes are that static data members
1912 of the same type with the same name but defined in different shared
1913 objects will be different, so changing one will not change the other;
1914 and that pointers to function members defined in different shared
1915 objects may not compare equal. When this flag is given, it is a
1916 violation of the ODR to define types with the same name differently.
1920 Do not use weak symbol support, even if it is provided by the linker.
1921 By default, G++ will use weak symbols if they are available. This
1922 option exists only for testing, and should not be used by end-users;
1923 it will result in inferior code and has no benefits. This option may
1924 be removed in a future release of G++.
1928 Do not search for header files in the standard directories specific to
1929 C++, but do still search the other standard directories. (This option
1930 is used when building the C++ library.)
1933 In addition, these optimization, warning, and code generation options
1934 have meanings only for C++ programs:
1937 @item -fno-default-inline
1938 @opindex fno-default-inline
1939 Do not assume @samp{inline} for functions defined inside a class scope.
1940 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1941 functions will have linkage like inline functions; they just won't be
1944 @item -Wabi @r{(C++ and Objective-C++ only)}
1947 Warn when G++ generates code that is probably not compatible with the
1948 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1949 all such cases, there are probably some cases that are not warned about,
1950 even though G++ is generating incompatible code. There may also be
1951 cases where warnings are emitted even though the code that is generated
1954 You should rewrite your code to avoid these warnings if you are
1955 concerned about the fact that code generated by G++ may not be binary
1956 compatible with code generated by other compilers.
1958 The known incompatibilities at this point include:
1963 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1964 pack data into the same byte as a base class. For example:
1967 struct A @{ virtual void f(); int f1 : 1; @};
1968 struct B : public A @{ int f2 : 1; @};
1972 In this case, G++ will place @code{B::f2} into the same byte
1973 as@code{A::f1}; other compilers will not. You can avoid this problem
1974 by explicitly padding @code{A} so that its size is a multiple of the
1975 byte size on your platform; that will cause G++ and other compilers to
1976 layout @code{B} identically.
1979 Incorrect handling of tail-padding for virtual bases. G++ does not use
1980 tail padding when laying out virtual bases. For example:
1983 struct A @{ virtual void f(); char c1; @};
1984 struct B @{ B(); char c2; @};
1985 struct C : public A, public virtual B @{@};
1989 In this case, G++ will not place @code{B} into the tail-padding for
1990 @code{A}; other compilers will. You can avoid this problem by
1991 explicitly padding @code{A} so that its size is a multiple of its
1992 alignment (ignoring virtual base classes); that will cause G++ and other
1993 compilers to layout @code{C} identically.
1996 Incorrect handling of bit-fields with declared widths greater than that
1997 of their underlying types, when the bit-fields appear in a union. For
2001 union U @{ int i : 4096; @};
2005 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2006 union too small by the number of bits in an @code{int}.
2009 Empty classes can be placed at incorrect offsets. For example:
2019 struct C : public B, public A @{@};
2023 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2024 it should be placed at offset zero. G++ mistakenly believes that the
2025 @code{A} data member of @code{B} is already at offset zero.
2028 Names of template functions whose types involve @code{typename} or
2029 template template parameters can be mangled incorrectly.
2032 template <typename Q>
2033 void f(typename Q::X) @{@}
2035 template <template <typename> class Q>
2036 void f(typename Q<int>::X) @{@}
2040 Instantiations of these templates may be mangled incorrectly.
2044 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2045 @opindex Wctor-dtor-privacy
2046 @opindex Wno-ctor-dtor-privacy
2047 Warn when a class seems unusable because all the constructors or
2048 destructors in that class are private, and it has neither friends nor
2049 public static member functions.
2051 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2052 @opindex Wnon-virtual-dtor
2053 @opindex Wno-non-virtual-dtor
2054 Warn when a class has virtual functions and accessible non-virtual
2055 destructor, in which case it would be possible but unsafe to delete
2056 an instance of a derived class through a pointer to the base class.
2057 This warning is also enabled if -Weffc++ is specified.
2059 @item -Wreorder @r{(C++ and Objective-C++ only)}
2061 @opindex Wno-reorder
2062 @cindex reordering, warning
2063 @cindex warning for reordering of member initializers
2064 Warn when the order of member initializers given in the code does not
2065 match the order in which they must be executed. For instance:
2071 A(): j (0), i (1) @{ @}
2075 The compiler will rearrange the member initializers for @samp{i}
2076 and @samp{j} to match the declaration order of the members, emitting
2077 a warning to that effect. This warning is enabled by @option{-Wall}.
2080 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2083 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2086 Warn about violations of the following style guidelines from Scott Meyers'
2087 @cite{Effective C++} book:
2091 Item 11: Define a copy constructor and an assignment operator for classes
2092 with dynamically allocated memory.
2095 Item 12: Prefer initialization to assignment in constructors.
2098 Item 14: Make destructors virtual in base classes.
2101 Item 15: Have @code{operator=} return a reference to @code{*this}.
2104 Item 23: Don't try to return a reference when you must return an object.
2108 Also warn about violations of the following style guidelines from
2109 Scott Meyers' @cite{More Effective C++} book:
2113 Item 6: Distinguish between prefix and postfix forms of increment and
2114 decrement operators.
2117 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2121 When selecting this option, be aware that the standard library
2122 headers do not obey all of these guidelines; use @samp{grep -v}
2123 to filter out those warnings.
2125 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2126 @opindex Wstrict-null-sentinel
2127 @opindex Wno-strict-null-sentinel
2128 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2129 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2130 to @code{__null}. Although it is a null pointer constant not a null pointer,
2131 it is guaranteed to of the same size as a pointer. But this use is
2132 not portable across different compilers.
2134 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2135 @opindex Wno-non-template-friend
2136 @opindex Wnon-template-friend
2137 Disable warnings when non-templatized friend functions are declared
2138 within a template. Since the advent of explicit template specification
2139 support in G++, if the name of the friend is an unqualified-id (i.e.,
2140 @samp{friend foo(int)}), the C++ language specification demands that the
2141 friend declare or define an ordinary, nontemplate function. (Section
2142 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2143 could be interpreted as a particular specialization of a templatized
2144 function. Because this non-conforming behavior is no longer the default
2145 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2146 check existing code for potential trouble spots and is on by default.
2147 This new compiler behavior can be turned off with
2148 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2149 but disables the helpful warning.
2151 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2152 @opindex Wold-style-cast
2153 @opindex Wno-old-style-cast
2154 Warn if an old-style (C-style) cast to a non-void type is used within
2155 a C++ program. The new-style casts (@samp{dynamic_cast},
2156 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2157 less vulnerable to unintended effects and much easier to search for.
2159 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2160 @opindex Woverloaded-virtual
2161 @opindex Wno-overloaded-virtual
2162 @cindex overloaded virtual fn, warning
2163 @cindex warning for overloaded virtual fn
2164 Warn when a function declaration hides virtual functions from a
2165 base class. For example, in:
2172 struct B: public A @{
2177 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2185 will fail to compile.
2187 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2188 @opindex Wno-pmf-conversions
2189 @opindex Wpmf-conversions
2190 Disable the diagnostic for converting a bound pointer to member function
2193 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2194 @opindex Wsign-promo
2195 @opindex Wno-sign-promo
2196 Warn when overload resolution chooses a promotion from unsigned or
2197 enumerated type to a signed type, over a conversion to an unsigned type of
2198 the same size. Previous versions of G++ would try to preserve
2199 unsignedness, but the standard mandates the current behavior.
2204 A& operator = (int);
2214 In this example, G++ will synthesize a default @samp{A& operator =
2215 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2218 @node Objective-C and Objective-C++ Dialect Options
2219 @section Options Controlling Objective-C and Objective-C++ Dialects
2221 @cindex compiler options, Objective-C and Objective-C++
2222 @cindex Objective-C and Objective-C++ options, command line
2223 @cindex options, Objective-C and Objective-C++
2224 (NOTE: This manual does not describe the Objective-C and Objective-C++
2225 languages themselves. See @xref{Standards,,Language Standards
2226 Supported by GCC}, for references.)
2228 This section describes the command-line options that are only meaningful
2229 for Objective-C and Objective-C++ programs, but you can also use most of
2230 the language-independent GNU compiler options.
2231 For example, you might compile a file @code{some_class.m} like this:
2234 gcc -g -fgnu-runtime -O -c some_class.m
2238 In this example, @option{-fgnu-runtime} is an option meant only for
2239 Objective-C and Objective-C++ programs; you can use the other options with
2240 any language supported by GCC@.
2242 Note that since Objective-C is an extension of the C language, Objective-C
2243 compilations may also use options specific to the C front-end (e.g.,
2244 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2245 C++-specific options (e.g., @option{-Wabi}).
2247 Here is a list of options that are @emph{only} for compiling Objective-C
2248 and Objective-C++ programs:
2251 @item -fconstant-string-class=@var{class-name}
2252 @opindex fconstant-string-class
2253 Use @var{class-name} as the name of the class to instantiate for each
2254 literal string specified with the syntax @code{@@"@dots{}"}. The default
2255 class name is @code{NXConstantString} if the GNU runtime is being used, and
2256 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2257 @option{-fconstant-cfstrings} option, if also present, will override the
2258 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2259 to be laid out as constant CoreFoundation strings.
2262 @opindex fgnu-runtime
2263 Generate object code compatible with the standard GNU Objective-C
2264 runtime. This is the default for most types of systems.
2266 @item -fnext-runtime
2267 @opindex fnext-runtime
2268 Generate output compatible with the NeXT runtime. This is the default
2269 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2270 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2273 @item -fno-nil-receivers
2274 @opindex fno-nil-receivers
2275 Assume that all Objective-C message dispatches (e.g.,
2276 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2277 is not @code{nil}. This allows for more efficient entry points in the runtime
2278 to be used. Currently, this option is only available in conjunction with
2279 the NeXT runtime on Mac OS X 10.3 and later.
2281 @item -fobjc-call-cxx-cdtors
2282 @opindex fobjc-call-cxx-cdtors
2283 For each Objective-C class, check if any of its instance variables is a
2284 C++ object with a non-trivial default constructor. If so, synthesize a
2285 special @code{- (id) .cxx_construct} instance method that will run
2286 non-trivial default constructors on any such instance variables, in order,
2287 and then return @code{self}. Similarly, check if any instance variable
2288 is a C++ object with a non-trivial destructor, and if so, synthesize a
2289 special @code{- (void) .cxx_destruct} method that will run
2290 all such default destructors, in reverse order.
2292 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2293 thusly generated will only operate on instance variables declared in the
2294 current Objective-C class, and not those inherited from superclasses. It
2295 is the responsibility of the Objective-C runtime to invoke all such methods
2296 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2297 will be invoked by the runtime immediately after a new object
2298 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2299 be invoked immediately before the runtime deallocates an object instance.
2301 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2302 support for invoking the @code{- (id) .cxx_construct} and
2303 @code{- (void) .cxx_destruct} methods.
2305 @item -fobjc-direct-dispatch
2306 @opindex fobjc-direct-dispatch
2307 Allow fast jumps to the message dispatcher. On Darwin this is
2308 accomplished via the comm page.
2310 @item -fobjc-exceptions
2311 @opindex fobjc-exceptions
2312 Enable syntactic support for structured exception handling in Objective-C,
2313 similar to what is offered by C++ and Java. This option is
2314 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2323 @@catch (AnObjCClass *exc) @{
2330 @@catch (AnotherClass *exc) @{
2333 @@catch (id allOthers) @{
2343 The @code{@@throw} statement may appear anywhere in an Objective-C or
2344 Objective-C++ program; when used inside of a @code{@@catch} block, the
2345 @code{@@throw} may appear without an argument (as shown above), in which case
2346 the object caught by the @code{@@catch} will be rethrown.
2348 Note that only (pointers to) Objective-C objects may be thrown and
2349 caught using this scheme. When an object is thrown, it will be caught
2350 by the nearest @code{@@catch} clause capable of handling objects of that type,
2351 analogously to how @code{catch} blocks work in C++ and Java. A
2352 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2353 any and all Objective-C exceptions not caught by previous @code{@@catch}
2356 The @code{@@finally} clause, if present, will be executed upon exit from the
2357 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2358 regardless of whether any exceptions are thrown, caught or rethrown
2359 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2360 of the @code{finally} clause in Java.
2362 There are several caveats to using the new exception mechanism:
2366 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2367 idioms provided by the @code{NSException} class, the new
2368 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2369 systems, due to additional functionality needed in the (NeXT) Objective-C
2373 As mentioned above, the new exceptions do not support handling
2374 types other than Objective-C objects. Furthermore, when used from
2375 Objective-C++, the Objective-C exception model does not interoperate with C++
2376 exceptions at this time. This means you cannot @code{@@throw} an exception
2377 from Objective-C and @code{catch} it in C++, or vice versa
2378 (i.e., @code{throw @dots{} @@catch}).
2381 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2382 blocks for thread-safe execution:
2385 @@synchronized (ObjCClass *guard) @{
2390 Upon entering the @code{@@synchronized} block, a thread of execution shall
2391 first check whether a lock has been placed on the corresponding @code{guard}
2392 object by another thread. If it has, the current thread shall wait until
2393 the other thread relinquishes its lock. Once @code{guard} becomes available,
2394 the current thread will place its own lock on it, execute the code contained in
2395 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2396 making @code{guard} available to other threads).
2398 Unlike Java, Objective-C does not allow for entire methods to be marked
2399 @code{@@synchronized}. Note that throwing exceptions out of
2400 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2401 to be unlocked properly.
2405 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2407 @item -freplace-objc-classes
2408 @opindex freplace-objc-classes
2409 Emit a special marker instructing @command{ld(1)} not to statically link in
2410 the resulting object file, and allow @command{dyld(1)} to load it in at
2411 run time instead. This is used in conjunction with the Fix-and-Continue
2412 debugging mode, where the object file in question may be recompiled and
2413 dynamically reloaded in the course of program execution, without the need
2414 to restart the program itself. Currently, Fix-and-Continue functionality
2415 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2420 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2421 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2422 compile time) with static class references that get initialized at load time,
2423 which improves run-time performance. Specifying the @option{-fzero-link} flag
2424 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2425 to be retained. This is useful in Zero-Link debugging mode, since it allows
2426 for individual class implementations to be modified during program execution.
2430 Dump interface declarations for all classes seen in the source file to a
2431 file named @file{@var{sourcename}.decl}.
2433 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2434 @opindex Wassign-intercept
2435 @opindex Wno-assign-intercept
2436 Warn whenever an Objective-C assignment is being intercepted by the
2439 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2440 @opindex Wno-protocol
2442 If a class is declared to implement a protocol, a warning is issued for
2443 every method in the protocol that is not implemented by the class. The
2444 default behavior is to issue a warning for every method not explicitly
2445 implemented in the class, even if a method implementation is inherited
2446 from the superclass. If you use the @option{-Wno-protocol} option, then
2447 methods inherited from the superclass are considered to be implemented,
2448 and no warning is issued for them.
2450 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2452 @opindex Wno-selector
2453 Warn if multiple methods of different types for the same selector are
2454 found during compilation. The check is performed on the list of methods
2455 in the final stage of compilation. Additionally, a check is performed
2456 for each selector appearing in a @code{@@selector(@dots{})}
2457 expression, and a corresponding method for that selector has been found
2458 during compilation. Because these checks scan the method table only at
2459 the end of compilation, these warnings are not produced if the final
2460 stage of compilation is not reached, for example because an error is
2461 found during compilation, or because the @option{-fsyntax-only} option is
2464 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2465 @opindex Wstrict-selector-match
2466 @opindex Wno-strict-selector-match
2467 Warn if multiple methods with differing argument and/or return types are
2468 found for a given selector when attempting to send a message using this
2469 selector to a receiver of type @code{id} or @code{Class}. When this flag
2470 is off (which is the default behavior), the compiler will omit such warnings
2471 if any differences found are confined to types which share the same size
2474 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2475 @opindex Wundeclared-selector
2476 @opindex Wno-undeclared-selector
2477 Warn if a @code{@@selector(@dots{})} expression referring to an
2478 undeclared selector is found. A selector is considered undeclared if no
2479 method with that name has been declared before the
2480 @code{@@selector(@dots{})} expression, either explicitly in an
2481 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2482 an @code{@@implementation} section. This option always performs its
2483 checks as soon as a @code{@@selector(@dots{})} expression is found,
2484 while @option{-Wselector} only performs its checks in the final stage of
2485 compilation. This also enforces the coding style convention
2486 that methods and selectors must be declared before being used.
2488 @item -print-objc-runtime-info
2489 @opindex print-objc-runtime-info
2490 Generate C header describing the largest structure that is passed by
2495 @node Language Independent Options
2496 @section Options to Control Diagnostic Messages Formatting
2497 @cindex options to control diagnostics formatting
2498 @cindex diagnostic messages
2499 @cindex message formatting
2501 Traditionally, diagnostic messages have been formatted irrespective of
2502 the output device's aspect (e.g.@: its width, @dots{}). The options described
2503 below can be used to control the diagnostic messages formatting
2504 algorithm, e.g.@: how many characters per line, how often source location
2505 information should be reported. Right now, only the C++ front end can
2506 honor these options. However it is expected, in the near future, that
2507 the remaining front ends would be able to digest them correctly.
2510 @item -fmessage-length=@var{n}
2511 @opindex fmessage-length
2512 Try to format error messages so that they fit on lines of about @var{n}
2513 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2514 the front ends supported by GCC@. If @var{n} is zero, then no
2515 line-wrapping will be done; each error message will appear on a single
2518 @opindex fdiagnostics-show-location
2519 @item -fdiagnostics-show-location=once
2520 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2521 reporter to emit @emph{once} source location information; that is, in
2522 case the message is too long to fit on a single physical line and has to
2523 be wrapped, the source location won't be emitted (as prefix) again,
2524 over and over, in subsequent continuation lines. This is the default
2527 @item -fdiagnostics-show-location=every-line
2528 Only meaningful in line-wrapping mode. Instructs the diagnostic
2529 messages reporter to emit the same source location information (as
2530 prefix) for physical lines that result from the process of breaking
2531 a message which is too long to fit on a single line.
2533 @item -fdiagnostics-show-option
2534 @opindex fdiagnostics-show-option
2535 This option instructs the diagnostic machinery to add text to each
2536 diagnostic emitted, which indicates which command line option directly
2537 controls that diagnostic, when such an option is known to the
2538 diagnostic machinery.
2540 @item -Wcoverage-mismatch
2541 @opindex Wcoverage-mismatch
2542 Warn if feedback profiles do not match when using the
2543 @option{-fprofile-use} option.
2544 If a source file was changed between @option{-fprofile-gen} and
2545 @option{-fprofile-use}, the files with the profile feedback can fail
2546 to match the source file and GCC can not use the profile feedback
2547 information. By default, GCC emits an error message in this case.
2548 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2549 error. GCC does not use appropriate feedback profiles, so using this
2550 option can result in poorly optimized code. This option is useful
2551 only in the case of very minor changes such as bug fixes to an
2556 @node Warning Options
2557 @section Options to Request or Suppress Warnings
2558 @cindex options to control warnings
2559 @cindex warning messages
2560 @cindex messages, warning
2561 @cindex suppressing warnings
2563 Warnings are diagnostic messages that report constructions which
2564 are not inherently erroneous but which are risky or suggest there
2565 may have been an error.
2567 The following language-independent options do not enable specific
2568 warnings but control the kinds of diagnostics produced by GCC.
2571 @cindex syntax checking
2573 @opindex fsyntax-only
2574 Check the code for syntax errors, but don't do anything beyond that.
2578 Inhibit all warning messages.
2583 Make all warnings into errors.
2588 Make the specified warning into an error. The specifier for a warning
2589 is appended, for example @option{-Werror=switch} turns the warnings
2590 controlled by @option{-Wswitch} into errors. This switch takes a
2591 negative form, to be used to negate @option{-Werror} for specific
2592 warnings, for example @option{-Wno-error=switch} makes
2593 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2594 is in effect. You can use the @option{-fdiagnostics-show-option}
2595 option to have each controllable warning amended with the option which
2596 controls it, to determine what to use with this option.
2598 Note that specifying @option{-Werror=}@var{foo} automatically implies
2599 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2602 @item -Wfatal-errors
2603 @opindex Wfatal-errors
2604 @opindex Wno-fatal-errors
2605 This option causes the compiler to abort compilation on the first error
2606 occurred rather than trying to keep going and printing further error
2611 You can request many specific warnings with options beginning
2612 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2613 implicit declarations. Each of these specific warning options also
2614 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2615 example, @option{-Wno-implicit}. This manual lists only one of the
2616 two forms, whichever is not the default. For further,
2617 language-specific options also refer to @ref{C++ Dialect Options} and
2618 @ref{Objective-C and Objective-C++ Dialect Options}.
2623 Issue all the warnings demanded by strict ISO C and ISO C++;
2624 reject all programs that use forbidden extensions, and some other
2625 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2626 version of the ISO C standard specified by any @option{-std} option used.
2628 Valid ISO C and ISO C++ programs should compile properly with or without
2629 this option (though a rare few will require @option{-ansi} or a
2630 @option{-std} option specifying the required version of ISO C)@. However,
2631 without this option, certain GNU extensions and traditional C and C++
2632 features are supported as well. With this option, they are rejected.
2634 @option{-pedantic} does not cause warning messages for use of the
2635 alternate keywords whose names begin and end with @samp{__}. Pedantic
2636 warnings are also disabled in the expression that follows
2637 @code{__extension__}. However, only system header files should use
2638 these escape routes; application programs should avoid them.
2639 @xref{Alternate Keywords}.
2641 Some users try to use @option{-pedantic} to check programs for strict ISO
2642 C conformance. They soon find that it does not do quite what they want:
2643 it finds some non-ISO practices, but not all---only those for which
2644 ISO C @emph{requires} a diagnostic, and some others for which
2645 diagnostics have been added.
2647 A feature to report any failure to conform to ISO C might be useful in
2648 some instances, but would require considerable additional work and would
2649 be quite different from @option{-pedantic}. We don't have plans to
2650 support such a feature in the near future.
2652 Where the standard specified with @option{-std} represents a GNU
2653 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2654 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2655 extended dialect is based. Warnings from @option{-pedantic} are given
2656 where they are required by the base standard. (It would not make sense
2657 for such warnings to be given only for features not in the specified GNU
2658 C dialect, since by definition the GNU dialects of C include all
2659 features the compiler supports with the given option, and there would be
2660 nothing to warn about.)
2662 @item -pedantic-errors
2663 @opindex pedantic-errors
2664 Like @option{-pedantic}, except that errors are produced rather than
2670 This enables all the warnings about constructions that some users
2671 consider questionable, and that are easy to avoid (or modify to
2672 prevent the warning), even in conjunction with macros. This also
2673 enables some language-specific warnings described in @ref{C++ Dialect
2674 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2676 @option{-Wall} turns on the following warning flags:
2678 @gccoptlist{-Waddress @gol
2679 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2681 -Wchar-subscripts @gol
2683 -Wimplicit-function-declaration @gol
2686 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2687 -Wmissing-braces @gol
2693 -Wsequence-point @gol
2694 -Wsign-compare @r{(only in C++)} @gol
2695 -Wstrict-aliasing @gol
2696 -Wstrict-overflow=1 @gol
2699 -Wuninitialized @gol
2700 -Wunknown-pragmas @gol
2701 -Wunused-function @gol
2704 -Wunused-variable @gol
2705 -Wvolatile-register-var @gol
2708 Note that some warning flags are not implied by @option{-Wall}. Some of
2709 them warn about constructions that users generally do not consider
2710 questionable, but which occasionally you might wish to check for;
2711 others warn about constructions that are necessary or hard to avoid in
2712 some cases, and there is no simple way to modify the code to suppress
2713 the warning. Some of them are enabled by @option{-Wextra} but many of
2714 them must be enabled individually.
2720 This enables some extra warning flags that are not enabled by
2721 @option{-Wall}. (This option used to be called @option{-W}. The older
2722 name is still supported, but the newer name is more descriptive.)
2724 @gccoptlist{-Wclobbered @gol
2726 -Wignored-qualifiers @gol
2727 -Wmissing-field-initializers @gol
2728 -Wmissing-parameter-type @r{(C only)} @gol
2729 -Wold-style-declaration @r{(C only)} @gol
2730 -Woverride-init @gol
2733 -Wuninitialized @gol
2734 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2737 The option @option{-Wextra} also prints warning messages for the
2743 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2744 @samp{>}, or @samp{>=}.
2747 (C++ only) An enumerator and a non-enumerator both appear in a
2748 conditional expression.
2751 (C++ only) A non-static reference or non-static @samp{const} member
2752 appears in a class without constructors.
2755 (C++ only) Ambiguous virtual bases.
2758 (C++ only) Subscripting an array which has been declared @samp{register}.
2761 (C++ only) Taking the address of a variable which has been declared
2765 (C++ only) A base class is not initialized in a derived class' copy
2770 @item -Wchar-subscripts
2771 @opindex Wchar-subscripts
2772 @opindex Wno-char-subscripts
2773 Warn if an array subscript has type @code{char}. This is a common cause
2774 of error, as programmers often forget that this type is signed on some
2776 This warning is enabled by @option{-Wall}.
2780 @opindex Wno-comment
2781 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2782 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2783 This warning is enabled by @option{-Wall}.
2788 @opindex ffreestanding
2789 @opindex fno-builtin
2790 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2791 the arguments supplied have types appropriate to the format string
2792 specified, and that the conversions specified in the format string make
2793 sense. This includes standard functions, and others specified by format
2794 attributes (@pxref{Function Attributes}), in the @code{printf},
2795 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2796 not in the C standard) families (or other target-specific families).
2797 Which functions are checked without format attributes having been
2798 specified depends on the standard version selected, and such checks of
2799 functions without the attribute specified are disabled by
2800 @option{-ffreestanding} or @option{-fno-builtin}.
2802 The formats are checked against the format features supported by GNU
2803 libc version 2.2. These include all ISO C90 and C99 features, as well
2804 as features from the Single Unix Specification and some BSD and GNU
2805 extensions. Other library implementations may not support all these
2806 features; GCC does not support warning about features that go beyond a
2807 particular library's limitations. However, if @option{-pedantic} is used
2808 with @option{-Wformat}, warnings will be given about format features not
2809 in the selected standard version (but not for @code{strfmon} formats,
2810 since those are not in any version of the C standard). @xref{C Dialect
2811 Options,,Options Controlling C Dialect}.
2813 Since @option{-Wformat} also checks for null format arguments for
2814 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2816 @option{-Wformat} is included in @option{-Wall}. For more control over some
2817 aspects of format checking, the options @option{-Wformat-y2k},
2818 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2819 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2820 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2823 @opindex Wformat-y2k
2824 @opindex Wno-format-y2k
2825 If @option{-Wformat} is specified, also warn about @code{strftime}
2826 formats which may yield only a two-digit year.
2828 @item -Wno-format-contains-nul
2829 @opindex Wno-format-contains-nul
2830 @opindex Wformat-contains-nul
2831 If @option{-Wformat} is specified, do not warn about format strings that
2834 @item -Wno-format-extra-args
2835 @opindex Wno-format-extra-args
2836 @opindex Wformat-extra-args
2837 If @option{-Wformat} is specified, do not warn about excess arguments to a
2838 @code{printf} or @code{scanf} format function. The C standard specifies
2839 that such arguments are ignored.
2841 Where the unused arguments lie between used arguments that are
2842 specified with @samp{$} operand number specifications, normally
2843 warnings are still given, since the implementation could not know what
2844 type to pass to @code{va_arg} to skip the unused arguments. However,
2845 in the case of @code{scanf} formats, this option will suppress the
2846 warning if the unused arguments are all pointers, since the Single
2847 Unix Specification says that such unused arguments are allowed.
2849 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2850 @opindex Wno-format-zero-length
2851 @opindex Wformat-zero-length
2852 If @option{-Wformat} is specified, do not warn about zero-length formats.
2853 The C standard specifies that zero-length formats are allowed.
2855 @item -Wformat-nonliteral
2856 @opindex Wformat-nonliteral
2857 @opindex Wno-format-nonliteral
2858 If @option{-Wformat} is specified, also warn if the format string is not a
2859 string literal and so cannot be checked, unless the format function
2860 takes its format arguments as a @code{va_list}.
2862 @item -Wformat-security
2863 @opindex Wformat-security
2864 @opindex Wno-format-security
2865 If @option{-Wformat} is specified, also warn about uses of format
2866 functions that represent possible security problems. At present, this
2867 warns about calls to @code{printf} and @code{scanf} functions where the
2868 format string is not a string literal and there are no format arguments,
2869 as in @code{printf (foo);}. This may be a security hole if the format
2870 string came from untrusted input and contains @samp{%n}. (This is
2871 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2872 in future warnings may be added to @option{-Wformat-security} that are not
2873 included in @option{-Wformat-nonliteral}.)
2877 @opindex Wno-format=2
2878 Enable @option{-Wformat} plus format checks not included in
2879 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2880 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2882 @item -Wnonnull @r{(C and Objective-C only)}
2884 @opindex Wno-nonnull
2885 Warn about passing a null pointer for arguments marked as
2886 requiring a non-null value by the @code{nonnull} function attribute.
2888 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2889 can be disabled with the @option{-Wno-nonnull} option.
2891 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2893 @opindex Wno-init-self
2894 Warn about uninitialized variables which are initialized with themselves.
2895 Note this option can only be used with the @option{-Wuninitialized} option.
2897 For example, GCC will warn about @code{i} being uninitialized in the
2898 following snippet only when @option{-Winit-self} has been specified:
2909 @item -Wimplicit-int @r{(C and Objective-C only)}
2910 @opindex Wimplicit-int
2911 @opindex Wno-implicit-int
2912 Warn when a declaration does not specify a type.
2913 This warning is enabled by @option{-Wall}.
2915 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
2916 @opindex Wimplicit-function-declaration
2917 @opindex Wno-implicit-function-declaration
2918 Give a warning whenever a function is used before being declared. In
2919 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
2920 enabled by default and it is made into an error by
2921 @option{-pedantic-errors}. This warning is also enabled by
2926 @opindex Wno-implicit
2927 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2928 This warning is enabled by @option{-Wall}.
2930 @item -Wignored-qualifiers @r{(C and C++ only)}
2931 @opindex Wignored-qualifiers
2932 @opindex Wno-ignored-qualifiers
2933 Warn if the return type of a function has a type qualifier
2934 such as @code{const}. For ISO C such a type qualifier has no effect,
2935 since the value returned by a function is not an lvalue.
2936 For C++, the warning is only emitted for scalar types or @code{void}.
2937 ISO C prohibits qualified @code{void} return types on function
2938 definitions, so such return types always receive a warning
2939 even without this option.
2941 This warning is also enabled by @option{-Wextra}.
2946 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2947 function with external linkage, returning int, taking either zero
2948 arguments, two, or three arguments of appropriate types.
2949 This warning is enabled by @option{-Wall}.
2951 @item -Wmissing-braces
2952 @opindex Wmissing-braces
2953 @opindex Wno-missing-braces
2954 Warn if an aggregate or union initializer is not fully bracketed. In
2955 the following example, the initializer for @samp{a} is not fully
2956 bracketed, but that for @samp{b} is fully bracketed.
2959 int a[2][2] = @{ 0, 1, 2, 3 @};
2960 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2963 This warning is enabled by @option{-Wall}.
2965 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2966 @opindex Wmissing-include-dirs
2967 @opindex Wno-missing-include-dirs
2968 Warn if a user-supplied include directory does not exist.
2971 @opindex Wparentheses
2972 @opindex Wno-parentheses
2973 Warn if parentheses are omitted in certain contexts, such
2974 as when there is an assignment in a context where a truth value
2975 is expected, or when operators are nested whose precedence people
2976 often get confused about.
2978 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2979 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2980 interpretation from that of ordinary mathematical notation.
2982 Also warn about constructions where there may be confusion to which
2983 @code{if} statement an @code{else} branch belongs. Here is an example of
2998 In C/C++, every @code{else} branch belongs to the innermost possible
2999 @code{if} statement, which in this example is @code{if (b)}. This is
3000 often not what the programmer expected, as illustrated in the above
3001 example by indentation the programmer chose. When there is the
3002 potential for this confusion, GCC will issue a warning when this flag
3003 is specified. To eliminate the warning, add explicit braces around
3004 the innermost @code{if} statement so there is no way the @code{else}
3005 could belong to the enclosing @code{if}. The resulting code would
3022 This warning is enabled by @option{-Wall}.
3024 @item -Wsequence-point
3025 @opindex Wsequence-point
3026 @opindex Wno-sequence-point
3027 Warn about code that may have undefined semantics because of violations
3028 of sequence point rules in the C and C++ standards.
3030 The C and C++ standards defines the order in which expressions in a C/C++
3031 program are evaluated in terms of @dfn{sequence points}, which represent
3032 a partial ordering between the execution of parts of the program: those
3033 executed before the sequence point, and those executed after it. These
3034 occur after the evaluation of a full expression (one which is not part
3035 of a larger expression), after the evaluation of the first operand of a
3036 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3037 function is called (but after the evaluation of its arguments and the
3038 expression denoting the called function), and in certain other places.
3039 Other than as expressed by the sequence point rules, the order of
3040 evaluation of subexpressions of an expression is not specified. All
3041 these rules describe only a partial order rather than a total order,
3042 since, for example, if two functions are called within one expression
3043 with no sequence point between them, the order in which the functions
3044 are called is not specified. However, the standards committee have
3045 ruled that function calls do not overlap.
3047 It is not specified when between sequence points modifications to the
3048 values of objects take effect. Programs whose behavior depends on this
3049 have undefined behavior; the C and C++ standards specify that ``Between
3050 the previous and next sequence point an object shall have its stored
3051 value modified at most once by the evaluation of an expression.
3052 Furthermore, the prior value shall be read only to determine the value
3053 to be stored.''. If a program breaks these rules, the results on any
3054 particular implementation are entirely unpredictable.
3056 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3057 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3058 diagnosed by this option, and it may give an occasional false positive
3059 result, but in general it has been found fairly effective at detecting
3060 this sort of problem in programs.
3062 The standard is worded confusingly, therefore there is some debate
3063 over the precise meaning of the sequence point rules in subtle cases.
3064 Links to discussions of the problem, including proposed formal
3065 definitions, may be found on the GCC readings page, at
3066 @w{@uref{http://gcc.gnu.org/readings.html}}.
3068 This warning is enabled by @option{-Wall} for C and C++.
3071 @opindex Wreturn-type
3072 @opindex Wno-return-type
3073 Warn whenever a function is defined with a return-type that defaults
3074 to @code{int}. Also warn about any @code{return} statement with no
3075 return-value in a function whose return-type is not @code{void}
3076 (falling off the end of the function body is considered returning
3077 without a value), and about a @code{return} statement with a
3078 expression in a function whose return-type is @code{void}.
3080 For C++, a function without return type always produces a diagnostic
3081 message, even when @option{-Wno-return-type} is specified. The only
3082 exceptions are @samp{main} and functions defined in system headers.
3084 This warning is enabled by @option{-Wall}.
3089 Warn whenever a @code{switch} statement has an index of enumerated type
3090 and lacks a @code{case} for one or more of the named codes of that
3091 enumeration. (The presence of a @code{default} label prevents this
3092 warning.) @code{case} labels outside the enumeration range also
3093 provoke warnings when this option is used.
3094 This warning is enabled by @option{-Wall}.
3096 @item -Wswitch-default
3097 @opindex Wswitch-default
3098 @opindex Wno-switch-default
3099 Warn whenever a @code{switch} statement does not have a @code{default}
3103 @opindex Wswitch-enum
3104 @opindex Wno-switch-enum
3105 Warn whenever a @code{switch} statement has an index of enumerated type
3106 and lacks a @code{case} for one or more of the named codes of that
3107 enumeration. @code{case} labels outside the enumeration range also
3108 provoke warnings when this option is used.
3112 @opindex Wno-trigraphs
3113 Warn if any trigraphs are encountered that might change the meaning of
3114 the program (trigraphs within comments are not warned about).
3115 This warning is enabled by @option{-Wall}.
3117 @item -Wunused-function
3118 @opindex Wunused-function
3119 @opindex Wno-unused-function
3120 Warn whenever a static function is declared but not defined or a
3121 non-inline static function is unused.
3122 This warning is enabled by @option{-Wall}.
3124 @item -Wunused-label
3125 @opindex Wunused-label
3126 @opindex Wno-unused-label
3127 Warn whenever a label is declared but not used.
3128 This warning is enabled by @option{-Wall}.
3130 To suppress this warning use the @samp{unused} attribute
3131 (@pxref{Variable Attributes}).
3133 @item -Wunused-parameter
3134 @opindex Wunused-parameter
3135 @opindex Wno-unused-parameter
3136 Warn whenever a function parameter is unused aside from its declaration.
3138 To suppress this warning use the @samp{unused} attribute
3139 (@pxref{Variable Attributes}).
3141 @item -Wunused-variable
3142 @opindex Wunused-variable
3143 @opindex Wno-unused-variable
3144 Warn whenever a local variable or non-constant static variable is unused
3145 aside from its declaration.
3146 This warning is enabled by @option{-Wall}.
3148 To suppress this warning use the @samp{unused} attribute
3149 (@pxref{Variable Attributes}).
3151 @item -Wunused-value
3152 @opindex Wunused-value
3153 @opindex Wno-unused-value
3154 Warn whenever a statement computes a result that is explicitly not
3155 used. To suppress this warning cast the unused expression to
3156 @samp{void}. This includes an expression-statement or the left-hand
3157 side of a comma expression that contains no side effects. For example,
3158 an expression such as @samp{x[i,j]} will cause a warning, while
3159 @samp{x[(void)i,j]} will not.
3161 This warning is enabled by @option{-Wall}.
3166 All the above @option{-Wunused} options combined.
3168 In order to get a warning about an unused function parameter, you must
3169 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3170 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3172 @item -Wuninitialized
3173 @opindex Wuninitialized
3174 @opindex Wno-uninitialized
3175 Warn if an automatic variable is used without first being initialized or
3176 if a variable may be clobbered by a @code{setjmp} call.
3178 If you want to warn about code which uses the uninitialized value of the
3179 variable in its own initializer, use the @option{-Winit-self} option.
3181 These warnings occur for individual uninitialized or clobbered
3182 elements of structure, union or array variables as well as for
3183 variables which are uninitialized or clobbered as a whole. They do
3184 not occur for variables or elements declared @code{volatile}. Because
3185 these warnings depend on optimization, the exact variables or elements
3186 for which there are warnings will depend on the precise optimization
3187 options and version of GCC used.
3189 Note that there may be no warning about a variable that is used only
3190 to compute a value that itself is never used, because such
3191 computations may be deleted by data flow analysis before the warnings
3194 These warnings are made optional because GCC is not smart
3195 enough to see all the reasons why the code might be correct
3196 despite appearing to have an error. Here is one example of how
3217 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3218 always initialized, but GCC doesn't know this. Here is
3219 another common case:
3224 if (change_y) save_y = y, y = new_y;
3226 if (change_y) y = save_y;
3231 This has no bug because @code{save_y} is used only if it is set.
3233 @cindex @code{longjmp} warnings
3234 This option also warns when a non-volatile automatic variable might be
3235 changed by a call to @code{longjmp}. These warnings as well are possible
3236 only in optimizing compilation.
3238 The compiler sees only the calls to @code{setjmp}. It cannot know
3239 where @code{longjmp} will be called; in fact, a signal handler could
3240 call it at any point in the code. As a result, you may get a warning
3241 even when there is in fact no problem because @code{longjmp} cannot
3242 in fact be called at the place which would cause a problem.
3244 Some spurious warnings can be avoided if you declare all the functions
3245 you use that never return as @code{noreturn}. @xref{Function
3248 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3250 @item -Wunknown-pragmas
3251 @opindex Wunknown-pragmas
3252 @opindex Wno-unknown-pragmas
3253 @cindex warning for unknown pragmas
3254 @cindex unknown pragmas, warning
3255 @cindex pragmas, warning of unknown
3256 Warn when a #pragma directive is encountered which is not understood by
3257 GCC@. If this command line option is used, warnings will even be issued
3258 for unknown pragmas in system header files. This is not the case if
3259 the warnings were only enabled by the @option{-Wall} command line option.
3262 @opindex Wno-pragmas
3264 Do not warn about misuses of pragmas, such as incorrect parameters,
3265 invalid syntax, or conflicts between pragmas. See also
3266 @samp{-Wunknown-pragmas}.
3268 @item -Wstrict-aliasing
3269 @opindex Wstrict-aliasing
3270 @opindex Wno-strict-aliasing
3271 This option is only active when @option{-fstrict-aliasing} is active.
3272 It warns about code which might break the strict aliasing rules that the
3273 compiler is using for optimization. The warning does not catch all
3274 cases, but does attempt to catch the more common pitfalls. It is
3275 included in @option{-Wall}.
3276 It is equivalent to @option{-Wstrict-aliasing=3}
3278 @item -Wstrict-aliasing=n
3279 @opindex Wstrict-aliasing=n
3280 @opindex Wno-strict-aliasing=n
3281 This option is only active when @option{-fstrict-aliasing} is active.
3282 It warns about code which might break the strict aliasing rules that the
3283 compiler is using for optimization.
3284 Higher levels correspond to higher accuracy (fewer false positives).
3285 Higher levels also correspond to more effort, similar to the way -O works.
3286 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3289 Level 1: Most aggressive, quick, least accurate.
3290 Possibly useful when higher levels
3291 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3292 false negatives. However, it has many false positives.
3293 Warns for all pointer conversions between possibly incompatible types,
3294 even if never dereferenced. Runs in the frontend only.
3296 Level 2: Aggressive, quick, not too precise.
3297 May still have many false positives (not as many as level 1 though),
3298 and few false negatives (but possibly more than level 1).
3299 Unlike level 1, it only warns when an address is taken. Warns about
3300 incomplete types. Runs in the frontend only.
3302 Level 3 (default for @option{-Wstrict-aliasing}):
3303 Should have very few false positives and few false
3304 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3305 Takes care of the common punn+dereference pattern in the frontend:
3306 @code{*(int*)&some_float}.
3307 If optimization is enabled, it also runs in the backend, where it deals
3308 with multiple statement cases using flow-sensitive points-to information.
3309 Only warns when the converted pointer is dereferenced.
3310 Does not warn about incomplete types.
3312 @item -Wstrict-overflow
3313 @itemx -Wstrict-overflow=@var{n}
3314 @opindex Wstrict-overflow
3315 @opindex Wno-strict-overflow
3316 This option is only active when @option{-fstrict-overflow} is active.
3317 It warns about cases where the compiler optimizes based on the
3318 assumption that signed overflow does not occur. Note that it does not
3319 warn about all cases where the code might overflow: it only warns
3320 about cases where the compiler implements some optimization. Thus
3321 this warning depends on the optimization level.
3323 An optimization which assumes that signed overflow does not occur is
3324 perfectly safe if the values of the variables involved are such that
3325 overflow never does, in fact, occur. Therefore this warning can
3326 easily give a false positive: a warning about code which is not
3327 actually a problem. To help focus on important issues, several
3328 warning levels are defined. No warnings are issued for the use of
3329 undefined signed overflow when estimating how many iterations a loop
3330 will require, in particular when determining whether a loop will be
3334 @item -Wstrict-overflow=1
3335 Warn about cases which are both questionable and easy to avoid. For
3336 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3337 compiler will simplify this to @code{1}. This level of
3338 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3339 are not, and must be explicitly requested.
3341 @item -Wstrict-overflow=2
3342 Also warn about other cases where a comparison is simplified to a
3343 constant. For example: @code{abs (x) >= 0}. This can only be
3344 simplified when @option{-fstrict-overflow} is in effect, because
3345 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3346 zero. @option{-Wstrict-overflow} (with no level) is the same as
3347 @option{-Wstrict-overflow=2}.
3349 @item -Wstrict-overflow=3
3350 Also warn about other cases where a comparison is simplified. For
3351 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3353 @item -Wstrict-overflow=4
3354 Also warn about other simplifications not covered by the above cases.
3355 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3357 @item -Wstrict-overflow=5
3358 Also warn about cases where the compiler reduces the magnitude of a
3359 constant involved in a comparison. For example: @code{x + 2 > y} will
3360 be simplified to @code{x + 1 >= y}. This is reported only at the
3361 highest warning level because this simplification applies to many
3362 comparisons, so this warning level will give a very large number of
3366 @item -Warray-bounds
3367 @opindex Wno-array-bounds
3368 @opindex Warray-bounds
3369 This option is only active when @option{-ftree-vrp} is active
3370 (default for -O2 and above). It warns about subscripts to arrays
3371 that are always out of bounds. This warning is enabled by @option{-Wall}.
3373 @item -Wno-div-by-zero
3374 @opindex Wno-div-by-zero
3375 @opindex Wdiv-by-zero
3376 Do not warn about compile-time integer division by zero. Floating point
3377 division by zero is not warned about, as it can be a legitimate way of
3378 obtaining infinities and NaNs.
3380 @item -Wsystem-headers
3381 @opindex Wsystem-headers
3382 @opindex Wno-system-headers
3383 @cindex warnings from system headers
3384 @cindex system headers, warnings from
3385 Print warning messages for constructs found in system header files.
3386 Warnings from system headers are normally suppressed, on the assumption
3387 that they usually do not indicate real problems and would only make the
3388 compiler output harder to read. Using this command line option tells
3389 GCC to emit warnings from system headers as if they occurred in user
3390 code. However, note that using @option{-Wall} in conjunction with this
3391 option will @emph{not} warn about unknown pragmas in system
3392 headers---for that, @option{-Wunknown-pragmas} must also be used.
3395 @opindex Wfloat-equal
3396 @opindex Wno-float-equal
3397 Warn if floating point values are used in equality comparisons.
3399 The idea behind this is that sometimes it is convenient (for the
3400 programmer) to consider floating-point values as approximations to
3401 infinitely precise real numbers. If you are doing this, then you need
3402 to compute (by analyzing the code, or in some other way) the maximum or
3403 likely maximum error that the computation introduces, and allow for it
3404 when performing comparisons (and when producing output, but that's a
3405 different problem). In particular, instead of testing for equality, you
3406 would check to see whether the two values have ranges that overlap; and
3407 this is done with the relational operators, so equality comparisons are
3410 @item -Wtraditional @r{(C and Objective-C only)}
3411 @opindex Wtraditional
3412 @opindex Wno-traditional
3413 Warn about certain constructs that behave differently in traditional and
3414 ISO C@. Also warn about ISO C constructs that have no traditional C
3415 equivalent, and/or problematic constructs which should be avoided.
3419 Macro parameters that appear within string literals in the macro body.
3420 In traditional C macro replacement takes place within string literals,
3421 but does not in ISO C@.
3424 In traditional C, some preprocessor directives did not exist.
3425 Traditional preprocessors would only consider a line to be a directive
3426 if the @samp{#} appeared in column 1 on the line. Therefore
3427 @option{-Wtraditional} warns about directives that traditional C
3428 understands but would ignore because the @samp{#} does not appear as the
3429 first character on the line. It also suggests you hide directives like
3430 @samp{#pragma} not understood by traditional C by indenting them. Some
3431 traditional implementations would not recognize @samp{#elif}, so it
3432 suggests avoiding it altogether.
3435 A function-like macro that appears without arguments.
3438 The unary plus operator.
3441 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3442 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3443 constants.) Note, these suffixes appear in macros defined in the system
3444 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3445 Use of these macros in user code might normally lead to spurious
3446 warnings, however GCC's integrated preprocessor has enough context to
3447 avoid warning in these cases.
3450 A function declared external in one block and then used after the end of
3454 A @code{switch} statement has an operand of type @code{long}.
3457 A non-@code{static} function declaration follows a @code{static} one.
3458 This construct is not accepted by some traditional C compilers.
3461 The ISO type of an integer constant has a different width or
3462 signedness from its traditional type. This warning is only issued if
3463 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3464 typically represent bit patterns, are not warned about.
3467 Usage of ISO string concatenation is detected.
3470 Initialization of automatic aggregates.
3473 Identifier conflicts with labels. Traditional C lacks a separate
3474 namespace for labels.
3477 Initialization of unions. If the initializer is zero, the warning is
3478 omitted. This is done under the assumption that the zero initializer in
3479 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3480 initializer warnings and relies on default initialization to zero in the
3484 Conversions by prototypes between fixed/floating point values and vice
3485 versa. The absence of these prototypes when compiling with traditional
3486 C would cause serious problems. This is a subset of the possible
3487 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3490 Use of ISO C style function definitions. This warning intentionally is
3491 @emph{not} issued for prototype declarations or variadic functions
3492 because these ISO C features will appear in your code when using
3493 libiberty's traditional C compatibility macros, @code{PARAMS} and
3494 @code{VPARAMS}. This warning is also bypassed for nested functions
3495 because that feature is already a GCC extension and thus not relevant to
3496 traditional C compatibility.
3499 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3500 @opindex Wtraditional-conversion
3501 @opindex Wno-traditional-conversion
3502 Warn if a prototype causes a type conversion that is different from what
3503 would happen to the same argument in the absence of a prototype. This
3504 includes conversions of fixed point to floating and vice versa, and
3505 conversions changing the width or signedness of a fixed point argument
3506 except when the same as the default promotion.
3508 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3509 @opindex Wdeclaration-after-statement
3510 @opindex Wno-declaration-after-statement
3511 Warn when a declaration is found after a statement in a block. This
3512 construct, known from C++, was introduced with ISO C99 and is by default
3513 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3514 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3519 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3521 @item -Wno-endif-labels
3522 @opindex Wno-endif-labels
3523 @opindex Wendif-labels
3524 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3529 Warn whenever a local variable shadows another local variable, parameter or
3530 global variable or whenever a built-in function is shadowed.
3532 @item -Wlarger-than=@var{len}
3533 @opindex Wlarger-than=@var{len}
3534 @opindex Wlarger-than-@var{len}
3535 Warn whenever an object of larger than @var{len} bytes is defined.
3537 @item -Wframe-larger-than=@var{len}
3538 @opindex Wframe-larger-than
3539 Warn if the size of a function frame is larger than @var{len} bytes.
3540 The computation done to determine the stack frame size is approximate
3541 and not conservative.
3542 The actual requirements may be somewhat greater than @var{len}
3543 even if you do not get a warning. In addition, any space allocated
3544 via @code{alloca}, variable-length arrays, or related constructs
3545 is not included by the compiler when determining
3546 whether or not to issue a warning.
3548 @item -Wunsafe-loop-optimizations
3549 @opindex Wunsafe-loop-optimizations
3550 @opindex Wno-unsafe-loop-optimizations
3551 Warn if the loop cannot be optimized because the compiler could not
3552 assume anything on the bounds of the loop indices. With
3553 @option{-funsafe-loop-optimizations} warn if the compiler made
3556 @item -Wpointer-arith
3557 @opindex Wpointer-arith
3558 @opindex Wno-pointer-arith
3559 Warn about anything that depends on the ``size of'' a function type or
3560 of @code{void}. GNU C assigns these types a size of 1, for
3561 convenience in calculations with @code{void *} pointers and pointers
3562 to functions. In C++, warn also when an arithmetic operation involves
3563 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3566 @opindex Wtype-limits
3567 @opindex Wno-type-limits
3568 Warn if a comparison is always true or always false due to the limited
3569 range of the data type, but do not warn for constant expressions. For
3570 example, warn if an unsigned variable is compared against zero with
3571 @samp{<} or @samp{>=}. This warning is also enabled by
3574 @item -Wbad-function-cast @r{(C and Objective-C only)}
3575 @opindex Wbad-function-cast
3576 @opindex Wno-bad-function-cast
3577 Warn whenever a function call is cast to a non-matching type.
3578 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3580 @item -Wc++-compat @r{(C and Objective-C only)}
3581 Warn about ISO C constructs that are outside of the common subset of
3582 ISO C and ISO C++, e.g.@: request for implicit conversion from
3583 @code{void *} to a pointer to non-@code{void} type.
3585 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3586 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3587 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3588 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3592 @opindex Wno-cast-qual
3593 Warn whenever a pointer is cast so as to remove a type qualifier from
3594 the target type. For example, warn if a @code{const char *} is cast
3595 to an ordinary @code{char *}.
3598 @opindex Wcast-align
3599 @opindex Wno-cast-align
3600 Warn whenever a pointer is cast such that the required alignment of the
3601 target is increased. For example, warn if a @code{char *} is cast to
3602 an @code{int *} on machines where integers can only be accessed at
3603 two- or four-byte boundaries.
3605 @item -Wwrite-strings
3606 @opindex Wwrite-strings
3607 @opindex Wno-write-strings
3608 When compiling C, give string constants the type @code{const
3609 char[@var{length}]} so that
3610 copying the address of one into a non-@code{const} @code{char *}
3611 pointer will get a warning; when compiling C++, warn about the
3612 deprecated conversion from string literals to @code{char *}. This
3613 warning, by default, is enabled for C++ programs.
3614 These warnings will help you find at
3615 compile time code that can try to write into a string constant, but
3616 only if you have been very careful about using @code{const} in
3617 declarations and prototypes. Otherwise, it will just be a nuisance;
3618 this is why we did not make @option{-Wall} request these warnings.
3622 @opindex Wno-clobbered
3623 Warn for variables that might be changed by @samp{longjmp} or
3624 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3627 @opindex Wconversion
3628 @opindex Wno-conversion
3629 Warn for implicit conversions that may alter a value. This includes
3630 conversions between real and integer, like @code{abs (x)} when
3631 @code{x} is @code{double}; conversions between signed and unsigned,
3632 like @code{unsigned ui = -1}; and conversions to smaller types, like
3633 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3634 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3635 changed by the conversion like in @code{abs (2.0)}. Warnings about
3636 conversions between signed and unsigned integers can be disabled by
3637 using @option{-Wno-sign-conversion}.
3639 For C++, also warn for conversions between @code{NULL} and non-pointer
3640 types; confusing overload resolution for user-defined conversions; and
3641 conversions that will never use a type conversion operator:
3642 conversions to @code{void}, the same type, a base class or a reference
3643 to them. Warnings about conversions between signed and unsigned
3644 integers are disabled by default in C++ unless
3645 @option{-Wsign-conversion} is explicitly enabled.
3648 @opindex Wempty-body
3649 @opindex Wno-empty-body
3650 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3651 while} statement. Additionally, in C++, warn when an empty body occurs
3652 in a @samp{while} or @samp{for} statement with no whitespacing before
3653 the semicolon. This warning is also enabled by @option{-Wextra}.
3655 @item -Wenum-compare @r{(C++ and Objective-C++ only)}
3656 @opindex Wenum-compare
3657 @opindex Wno-enum-compare
3658 Warn about a comparison between values of different enum types. This
3659 warning is enabled by default.
3661 @item -Wsign-compare
3662 @opindex Wsign-compare
3663 @opindex Wno-sign-compare
3664 @cindex warning for comparison of signed and unsigned values
3665 @cindex comparison of signed and unsigned values, warning
3666 @cindex signed and unsigned values, comparison warning
3667 Warn when a comparison between signed and unsigned values could produce
3668 an incorrect result when the signed value is converted to unsigned.
3669 This warning is also enabled by @option{-Wextra}; to get the other warnings
3670 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3672 @item -Wsign-conversion
3673 @opindex Wsign-conversion
3674 @opindex Wno-sign-conversion
3675 Warn for implicit conversions that may change the sign of an integer
3676 value, like assigning a signed integer expression to an unsigned
3677 integer variable. An explicit cast silences the warning. In C, this
3678 option is enabled also by @option{-Wconversion}.
3682 @opindex Wno-address
3683 Warn about suspicious uses of memory addresses. These include using
3684 the address of a function in a conditional expression, such as
3685 @code{void func(void); if (func)}, and comparisons against the memory
3686 address of a string literal, such as @code{if (x == "abc")}. Such
3687 uses typically indicate a programmer error: the address of a function
3688 always evaluates to true, so their use in a conditional usually
3689 indicate that the programmer forgot the parentheses in a function
3690 call; and comparisons against string literals result in unspecified
3691 behavior and are not portable in C, so they usually indicate that the
3692 programmer intended to use @code{strcmp}. This warning is enabled by
3696 @opindex Wlogical-op
3697 @opindex Wno-logical-op
3698 Warn about suspicious uses of logical operators in expressions.
3699 This includes using logical operators in contexts where a
3700 bit-wise operator is likely to be expected.
3702 @item -Waggregate-return
3703 @opindex Waggregate-return
3704 @opindex Wno-aggregate-return
3705 Warn if any functions that return structures or unions are defined or
3706 called. (In languages where you can return an array, this also elicits
3709 @item -Wno-attributes
3710 @opindex Wno-attributes
3711 @opindex Wattributes
3712 Do not warn if an unexpected @code{__attribute__} is used, such as
3713 unrecognized attributes, function attributes applied to variables,
3714 etc. This will not stop errors for incorrect use of supported
3717 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3718 @opindex Wstrict-prototypes
3719 @opindex Wno-strict-prototypes
3720 Warn if a function is declared or defined without specifying the
3721 argument types. (An old-style function definition is permitted without
3722 a warning if preceded by a declaration which specifies the argument
3725 @item -Wold-style-declaration @r{(C and Objective-C only)}
3726 @opindex Wold-style-declaration
3727 @opindex Wno-old-style-declaration
3728 Warn for obsolescent usages, according to the C Standard, in a
3729 declaration. For example, warn if storage-class specifiers like
3730 @code{static} are not the first things in a declaration. This warning
3731 is also enabled by @option{-Wextra}.
3733 @item -Wold-style-definition @r{(C and Objective-C only)}
3734 @opindex Wold-style-definition
3735 @opindex Wno-old-style-definition
3736 Warn if an old-style function definition is used. A warning is given
3737 even if there is a previous prototype.
3739 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3740 @opindex Wmissing-parameter-type
3741 @opindex Wno-missing-parameter-type
3742 A function parameter is declared without a type specifier in K&R-style
3749 This warning is also enabled by @option{-Wextra}.
3751 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3752 @opindex Wmissing-prototypes
3753 @opindex Wno-missing-prototypes
3754 Warn if a global function is defined without a previous prototype
3755 declaration. This warning is issued even if the definition itself
3756 provides a prototype. The aim is to detect global functions that fail
3757 to be declared in header files.
3759 @item -Wmissing-declarations
3760 @opindex Wmissing-declarations
3761 @opindex Wno-missing-declarations
3762 Warn if a global function is defined without a previous declaration.
3763 Do so even if the definition itself provides a prototype.
3764 Use this option to detect global functions that are not declared in
3765 header files. In C++, no warnings are issued for function templates,
3766 or for inline functions, or for functions in anonymous namespaces.
3768 @item -Wmissing-field-initializers
3769 @opindex Wmissing-field-initializers
3770 @opindex Wno-missing-field-initializers
3774 Warn if a structure's initializer has some fields missing. For
3775 example, the following code would cause such a warning, because
3776 @code{x.h} is implicitly zero:
3779 struct s @{ int f, g, h; @};
3780 struct s x = @{ 3, 4 @};
3783 This option does not warn about designated initializers, so the following
3784 modification would not trigger a warning:
3787 struct s @{ int f, g, h; @};
3788 struct s x = @{ .f = 3, .g = 4 @};
3791 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3792 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3794 @item -Wmissing-noreturn
3795 @opindex Wmissing-noreturn
3796 @opindex Wno-missing-noreturn
3797 Warn about functions which might be candidates for attribute @code{noreturn}.
3798 Note these are only possible candidates, not absolute ones. Care should
3799 be taken to manually verify functions actually do not ever return before
3800 adding the @code{noreturn} attribute, otherwise subtle code generation
3801 bugs could be introduced. You will not get a warning for @code{main} in
3802 hosted C environments.
3804 @item -Wmissing-format-attribute
3805 @opindex Wmissing-format-attribute
3806 @opindex Wno-missing-format-attribute
3809 Warn about function pointers which might be candidates for @code{format}
3810 attributes. Note these are only possible candidates, not absolute ones.
3811 GCC will guess that function pointers with @code{format} attributes that
3812 are used in assignment, initialization, parameter passing or return
3813 statements should have a corresponding @code{format} attribute in the
3814 resulting type. I.e.@: the left-hand side of the assignment or
3815 initialization, the type of the parameter variable, or the return type
3816 of the containing function respectively should also have a @code{format}
3817 attribute to avoid the warning.
3819 GCC will also warn about function definitions which might be
3820 candidates for @code{format} attributes. Again, these are only
3821 possible candidates. GCC will guess that @code{format} attributes
3822 might be appropriate for any function that calls a function like
3823 @code{vprintf} or @code{vscanf}, but this might not always be the
3824 case, and some functions for which @code{format} attributes are
3825 appropriate may not be detected.
3827 @item -Wno-multichar
3828 @opindex Wno-multichar
3830 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3831 Usually they indicate a typo in the user's code, as they have
3832 implementation-defined values, and should not be used in portable code.
3834 @item -Wnormalized=<none|id|nfc|nfkc>
3835 @opindex Wnormalized=
3838 @cindex character set, input normalization
3839 In ISO C and ISO C++, two identifiers are different if they are
3840 different sequences of characters. However, sometimes when characters
3841 outside the basic ASCII character set are used, you can have two
3842 different character sequences that look the same. To avoid confusion,
3843 the ISO 10646 standard sets out some @dfn{normalization rules} which
3844 when applied ensure that two sequences that look the same are turned into
3845 the same sequence. GCC can warn you if you are using identifiers which
3846 have not been normalized; this option controls that warning.
3848 There are four levels of warning that GCC supports. The default is
3849 @option{-Wnormalized=nfc}, which warns about any identifier which is
3850 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3851 recommended form for most uses.
3853 Unfortunately, there are some characters which ISO C and ISO C++ allow
3854 in identifiers that when turned into NFC aren't allowable as
3855 identifiers. That is, there's no way to use these symbols in portable
3856 ISO C or C++ and have all your identifiers in NFC@.
3857 @option{-Wnormalized=id} suppresses the warning for these characters.
3858 It is hoped that future versions of the standards involved will correct
3859 this, which is why this option is not the default.
3861 You can switch the warning off for all characters by writing
3862 @option{-Wnormalized=none}. You would only want to do this if you
3863 were using some other normalization scheme (like ``D''), because
3864 otherwise you can easily create bugs that are literally impossible to see.
3866 Some characters in ISO 10646 have distinct meanings but look identical
3867 in some fonts or display methodologies, especially once formatting has
3868 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3869 LETTER N'', will display just like a regular @code{n} which has been
3870 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3871 normalization scheme to convert all these into a standard form as
3872 well, and GCC will warn if your code is not in NFKC if you use
3873 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3874 about every identifier that contains the letter O because it might be
3875 confused with the digit 0, and so is not the default, but may be
3876 useful as a local coding convention if the programming environment is
3877 unable to be fixed to display these characters distinctly.
3879 @item -Wno-deprecated
3880 @opindex Wno-deprecated
3881 @opindex Wdeprecated
3882 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
3884 @item -Wno-deprecated-declarations
3885 @opindex Wno-deprecated-declarations
3886 @opindex Wdeprecated-declarations
3887 Do not warn about uses of functions (@pxref{Function Attributes}),
3888 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3889 Attributes}) marked as deprecated by using the @code{deprecated}
3893 @opindex Wno-overflow
3895 Do not warn about compile-time overflow in constant expressions.
3897 @item -Woverride-init @r{(C and Objective-C only)}
3898 @opindex Woverride-init
3899 @opindex Wno-override-init
3903 Warn if an initialized field without side effects is overridden when
3904 using designated initializers (@pxref{Designated Inits, , Designated
3907 This warning is included in @option{-Wextra}. To get other
3908 @option{-Wextra} warnings without this one, use @samp{-Wextra
3909 -Wno-override-init}.
3914 Warn if a structure is given the packed attribute, but the packed
3915 attribute has no effect on the layout or size of the structure.
3916 Such structures may be mis-aligned for little benefit. For
3917 instance, in this code, the variable @code{f.x} in @code{struct bar}
3918 will be misaligned even though @code{struct bar} does not itself
3919 have the packed attribute:
3926 @} __attribute__((packed));
3937 Warn if padding is included in a structure, either to align an element
3938 of the structure or to align the whole structure. Sometimes when this
3939 happens it is possible to rearrange the fields of the structure to
3940 reduce the padding and so make the structure smaller.
3942 @item -Wredundant-decls
3943 @opindex Wredundant-decls
3944 @opindex Wno-redundant-decls
3945 Warn if anything is declared more than once in the same scope, even in
3946 cases where multiple declaration is valid and changes nothing.
3948 @item -Wnested-externs @r{(C and Objective-C only)}
3949 @opindex Wnested-externs
3950 @opindex Wno-nested-externs
3951 Warn if an @code{extern} declaration is encountered within a function.
3953 @item -Wunreachable-code
3954 @opindex Wunreachable-code
3955 @opindex Wno-unreachable-code
3956 Warn if the compiler detects that code will never be executed.
3958 This option is intended to warn when the compiler detects that at
3959 least a whole line of source code will never be executed, because
3960 some condition is never satisfied or because it is after a
3961 procedure that never returns.
3963 It is possible for this option to produce a warning even though there
3964 are circumstances under which part of the affected line can be executed,
3965 so care should be taken when removing apparently-unreachable code.
3967 For instance, when a function is inlined, a warning may mean that the
3968 line is unreachable in only one inlined copy of the function.
3970 This option is not made part of @option{-Wall} because in a debugging
3971 version of a program there is often substantial code which checks
3972 correct functioning of the program and is, hopefully, unreachable
3973 because the program does work. Another common use of unreachable
3974 code is to provide behavior which is selectable at compile-time.
3979 Warn if a function can not be inlined and it was declared as inline.
3980 Even with this option, the compiler will not warn about failures to
3981 inline functions declared in system headers.
3983 The compiler uses a variety of heuristics to determine whether or not
3984 to inline a function. For example, the compiler takes into account
3985 the size of the function being inlined and the amount of inlining
3986 that has already been done in the current function. Therefore,
3987 seemingly insignificant changes in the source program can cause the
3988 warnings produced by @option{-Winline} to appear or disappear.
3990 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
3991 @opindex Wno-invalid-offsetof
3992 @opindex Winvalid-offsetof
3993 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3994 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3995 to a non-POD type is undefined. In existing C++ implementations,
3996 however, @samp{offsetof} typically gives meaningful results even when
3997 applied to certain kinds of non-POD types. (Such as a simple
3998 @samp{struct} that fails to be a POD type only by virtue of having a
3999 constructor.) This flag is for users who are aware that they are
4000 writing nonportable code and who have deliberately chosen to ignore the
4003 The restrictions on @samp{offsetof} may be relaxed in a future version
4004 of the C++ standard.
4006 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4007 @opindex Wno-int-to-pointer-cast
4008 @opindex Wint-to-pointer-cast
4009 Suppress warnings from casts to pointer type of an integer of a
4012 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4013 @opindex Wno-pointer-to-int-cast
4014 @opindex Wpointer-to-int-cast
4015 Suppress warnings from casts from a pointer to an integer type of a
4019 @opindex Winvalid-pch
4020 @opindex Wno-invalid-pch
4021 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4022 the search path but can't be used.
4026 @opindex Wno-long-long
4027 Warn if @samp{long long} type is used. This is default. To inhibit
4028 the warning messages, use @option{-Wno-long-long}. Flags
4029 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
4030 only when @option{-pedantic} flag is used.
4032 @item -Wvariadic-macros
4033 @opindex Wvariadic-macros
4034 @opindex Wno-variadic-macros
4035 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4036 alternate syntax when in pedantic ISO C99 mode. This is default.
4037 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4042 Warn if variable length array is used in the code.
4043 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4044 the variable length array.
4046 @item -Wvolatile-register-var
4047 @opindex Wvolatile-register-var
4048 @opindex Wno-volatile-register-var
4049 Warn if a register variable is declared volatile. The volatile
4050 modifier does not inhibit all optimizations that may eliminate reads
4051 and/or writes to register variables. This warning is enabled by
4054 @item -Wdisabled-optimization
4055 @opindex Wdisabled-optimization
4056 @opindex Wno-disabled-optimization
4057 Warn if a requested optimization pass is disabled. This warning does
4058 not generally indicate that there is anything wrong with your code; it
4059 merely indicates that GCC's optimizers were unable to handle the code
4060 effectively. Often, the problem is that your code is too big or too
4061 complex; GCC will refuse to optimize programs when the optimization
4062 itself is likely to take inordinate amounts of time.
4064 @item -Wpointer-sign @r{(C and Objective-C only)}
4065 @opindex Wpointer-sign
4066 @opindex Wno-pointer-sign
4067 Warn for pointer argument passing or assignment with different signedness.
4068 This option is only supported for C and Objective-C@. It is implied by
4069 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4070 @option{-Wno-pointer-sign}.
4072 @item -Wstack-protector
4073 @opindex Wstack-protector
4074 @opindex Wno-stack-protector
4075 This option is only active when @option{-fstack-protector} is active. It
4076 warns about functions that will not be protected against stack smashing.
4079 @opindex Wno-mudflap
4080 Suppress warnings about constructs that cannot be instrumented by
4083 @item -Woverlength-strings
4084 @opindex Woverlength-strings
4085 @opindex Wno-overlength-strings
4086 Warn about string constants which are longer than the ``minimum
4087 maximum'' length specified in the C standard. Modern compilers
4088 generally allow string constants which are much longer than the
4089 standard's minimum limit, but very portable programs should avoid
4090 using longer strings.
4092 The limit applies @emph{after} string constant concatenation, and does
4093 not count the trailing NUL@. In C89, the limit was 509 characters; in
4094 C99, it was raised to 4095. C++98 does not specify a normative
4095 minimum maximum, so we do not diagnose overlength strings in C++@.
4097 This option is implied by @option{-pedantic}, and can be disabled with
4098 @option{-Wno-overlength-strings}.
4100 @item -Wdisallowed-function-list=@var{sym},@var{sym},@dots{}
4101 @opindex Wdisallowed-function-list
4103 If any of @var{sym} is called, GCC will issue a warning. This can be useful
4104 in enforcing coding conventions that ban calls to certain functions, for
4105 example, @code{alloca}, @code{malloc}, etc.
4108 @node Debugging Options
4109 @section Options for Debugging Your Program or GCC
4110 @cindex options, debugging
4111 @cindex debugging information options
4113 GCC has various special options that are used for debugging
4114 either your program or GCC:
4119 Produce debugging information in the operating system's native format
4120 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4123 On most systems that use stabs format, @option{-g} enables use of extra
4124 debugging information that only GDB can use; this extra information
4125 makes debugging work better in GDB but will probably make other debuggers
4127 refuse to read the program. If you want to control for certain whether
4128 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4129 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4131 GCC allows you to use @option{-g} with
4132 @option{-O}. The shortcuts taken by optimized code may occasionally
4133 produce surprising results: some variables you declared may not exist
4134 at all; flow of control may briefly move where you did not expect it;
4135 some statements may not be executed because they compute constant
4136 results or their values were already at hand; some statements may
4137 execute in different places because they were moved out of loops.
4139 Nevertheless it proves possible to debug optimized output. This makes
4140 it reasonable to use the optimizer for programs that might have bugs.
4142 The following options are useful when GCC is generated with the
4143 capability for more than one debugging format.
4147 Produce debugging information for use by GDB@. This means to use the
4148 most expressive format available (DWARF 2, stabs, or the native format
4149 if neither of those are supported), including GDB extensions if at all
4154 Produce debugging information in stabs format (if that is supported),
4155 without GDB extensions. This is the format used by DBX on most BSD
4156 systems. On MIPS, Alpha and System V Release 4 systems this option
4157 produces stabs debugging output which is not understood by DBX or SDB@.
4158 On System V Release 4 systems this option requires the GNU assembler.
4160 @item -feliminate-unused-debug-symbols
4161 @opindex feliminate-unused-debug-symbols
4162 Produce debugging information in stabs format (if that is supported),
4163 for only symbols that are actually used.
4165 @item -femit-class-debug-always
4166 Instead of emitting debugging information for a C++ class in only one
4167 object file, emit it in all object files using the class. This option
4168 should be used only with debuggers that are unable to handle the way GCC
4169 normally emits debugging information for classes because using this
4170 option will increase the size of debugging information by as much as a
4175 Produce debugging information in stabs format (if that is supported),
4176 using GNU extensions understood only by the GNU debugger (GDB)@. The
4177 use of these extensions is likely to make other debuggers crash or
4178 refuse to read the program.
4182 Produce debugging information in COFF format (if that is supported).
4183 This is the format used by SDB on most System V systems prior to
4188 Produce debugging information in XCOFF format (if that is supported).
4189 This is the format used by the DBX debugger on IBM RS/6000 systems.
4193 Produce debugging information in XCOFF format (if that is supported),
4194 using GNU extensions understood only by the GNU debugger (GDB)@. The
4195 use of these extensions is likely to make other debuggers crash or
4196 refuse to read the program, and may cause assemblers other than the GNU
4197 assembler (GAS) to fail with an error.
4201 Produce debugging information in DWARF version 2 format (if that is
4202 supported). This is the format used by DBX on IRIX 6. With this
4203 option, GCC uses features of DWARF version 3 when they are useful;
4204 version 3 is upward compatible with version 2, but may still cause
4205 problems for older debuggers.
4209 Produce debugging information in VMS debug format (if that is
4210 supported). This is the format used by DEBUG on VMS systems.
4213 @itemx -ggdb@var{level}
4214 @itemx -gstabs@var{level}
4215 @itemx -gcoff@var{level}
4216 @itemx -gxcoff@var{level}
4217 @itemx -gvms@var{level}
4218 Request debugging information and also use @var{level} to specify how
4219 much information. The default level is 2.
4221 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4224 Level 1 produces minimal information, enough for making backtraces in
4225 parts of the program that you don't plan to debug. This includes
4226 descriptions of functions and external variables, but no information
4227 about local variables and no line numbers.
4229 Level 3 includes extra information, such as all the macro definitions
4230 present in the program. Some debuggers support macro expansion when
4231 you use @option{-g3}.
4233 @option{-gdwarf-2} does not accept a concatenated debug level, because
4234 GCC used to support an option @option{-gdwarf} that meant to generate
4235 debug information in version 1 of the DWARF format (which is very
4236 different from version 2), and it would have been too confusing. That
4237 debug format is long obsolete, but the option cannot be changed now.
4238 Instead use an additional @option{-g@var{level}} option to change the
4239 debug level for DWARF2.
4241 @item -feliminate-dwarf2-dups
4242 @opindex feliminate-dwarf2-dups
4243 Compress DWARF2 debugging information by eliminating duplicated
4244 information about each symbol. This option only makes sense when
4245 generating DWARF2 debugging information with @option{-gdwarf-2}.
4247 @item -femit-struct-debug-baseonly
4248 Emit debug information for struct-like types
4249 only when the base name of the compilation source file
4250 matches the base name of file in which the struct was defined.
4252 This option substantially reduces the size of debugging information,
4253 but at significant potential loss in type information to the debugger.
4254 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4255 See @option{-femit-struct-debug-detailed} for more detailed control.
4257 This option works only with DWARF 2.
4259 @item -femit-struct-debug-reduced
4260 Emit debug information for struct-like types
4261 only when the base name of the compilation source file
4262 matches the base name of file in which the type was defined,
4263 unless the struct is a template or defined in a system header.
4265 This option significantly reduces the size of debugging information,
4266 with some potential loss in type information to the debugger.
4267 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4268 See @option{-femit-struct-debug-detailed} for more detailed control.
4270 This option works only with DWARF 2.
4272 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4273 Specify the struct-like types
4274 for which the compiler will generate debug information.
4275 The intent is to reduce duplicate struct debug information
4276 between different object files within the same program.
4278 This option is a detailed version of
4279 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4280 which will serve for most needs.
4282 A specification has the syntax
4283 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4285 The optional first word limits the specification to
4286 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4287 A struct type is used directly when it is the type of a variable, member.
4288 Indirect uses arise through pointers to structs.
4289 That is, when use of an incomplete struct would be legal, the use is indirect.
4291 @samp{struct one direct; struct two * indirect;}.
4293 The optional second word limits the specification to
4294 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4295 Generic structs are a bit complicated to explain.
4296 For C++, these are non-explicit specializations of template classes,
4297 or non-template classes within the above.
4298 Other programming languages have generics,
4299 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4301 The third word specifies the source files for those
4302 structs for which the compiler will emit debug information.
4303 The values @samp{none} and @samp{any} have the normal meaning.
4304 The value @samp{base} means that
4305 the base of name of the file in which the type declaration appears
4306 must match the base of the name of the main compilation file.
4307 In practice, this means that
4308 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4309 but types declared in other header will not.
4310 The value @samp{sys} means those types satisfying @samp{base}
4311 or declared in system or compiler headers.
4313 You may need to experiment to determine the best settings for your application.
4315 The default is @samp{-femit-struct-debug-detailed=all}.
4317 This option works only with DWARF 2.
4319 @item -fno-merge-debug-strings
4320 @opindex fmerge-debug-strings
4321 @opindex fno-merge-debug-strings
4322 Direct the linker to merge together strings which are identical in
4323 different object files. This is not supported by all assemblers or
4324 linker. This decreases the size of the debug information in the
4325 output file at the cost of increasing link processing time. This is
4328 @item -fdebug-prefix-map=@var{old}=@var{new}
4329 @opindex fdebug-prefix-map
4330 When compiling files in directory @file{@var{old}}, record debugging
4331 information describing them as in @file{@var{new}} instead.
4333 @cindex @command{prof}
4336 Generate extra code to write profile information suitable for the
4337 analysis program @command{prof}. You must use this option when compiling
4338 the source files you want data about, and you must also use it when
4341 @cindex @command{gprof}
4344 Generate extra code to write profile information suitable for the
4345 analysis program @command{gprof}. You must use this option when compiling
4346 the source files you want data about, and you must also use it when
4351 Makes the compiler print out each function name as it is compiled, and
4352 print some statistics about each pass when it finishes.
4355 @opindex ftime-report
4356 Makes the compiler print some statistics about the time consumed by each
4357 pass when it finishes.
4360 @opindex fmem-report
4361 Makes the compiler print some statistics about permanent memory
4362 allocation when it finishes.
4364 @item -fpre-ipa-mem-report
4365 @opindex fpre-ipa-mem-report
4366 @item -fpost-ipa-mem-report
4367 @opindex fpost-ipa-mem-report
4368 Makes the compiler print some statistics about permanent memory
4369 allocation before or after interprocedural optimization.
4371 @item -fprofile-arcs
4372 @opindex fprofile-arcs
4373 Add code so that program flow @dfn{arcs} are instrumented. During
4374 execution the program records how many times each branch and call is
4375 executed and how many times it is taken or returns. When the compiled
4376 program exits it saves this data to a file called
4377 @file{@var{auxname}.gcda} for each source file. The data may be used for
4378 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4379 test coverage analysis (@option{-ftest-coverage}). Each object file's
4380 @var{auxname} is generated from the name of the output file, if
4381 explicitly specified and it is not the final executable, otherwise it is
4382 the basename of the source file. In both cases any suffix is removed
4383 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4384 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4385 @xref{Cross-profiling}.
4387 @cindex @command{gcov}
4391 This option is used to compile and link code instrumented for coverage
4392 analysis. The option is a synonym for @option{-fprofile-arcs}
4393 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4394 linking). See the documentation for those options for more details.
4399 Compile the source files with @option{-fprofile-arcs} plus optimization
4400 and code generation options. For test coverage analysis, use the
4401 additional @option{-ftest-coverage} option. You do not need to profile
4402 every source file in a program.
4405 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4406 (the latter implies the former).
4409 Run the program on a representative workload to generate the arc profile
4410 information. This may be repeated any number of times. You can run
4411 concurrent instances of your program, and provided that the file system
4412 supports locking, the data files will be correctly updated. Also
4413 @code{fork} calls are detected and correctly handled (double counting
4417 For profile-directed optimizations, compile the source files again with
4418 the same optimization and code generation options plus
4419 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4420 Control Optimization}).
4423 For test coverage analysis, use @command{gcov} to produce human readable
4424 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4425 @command{gcov} documentation for further information.
4429 With @option{-fprofile-arcs}, for each function of your program GCC
4430 creates a program flow graph, then finds a spanning tree for the graph.
4431 Only arcs that are not on the spanning tree have to be instrumented: the
4432 compiler adds code to count the number of times that these arcs are
4433 executed. When an arc is the only exit or only entrance to a block, the
4434 instrumentation code can be added to the block; otherwise, a new basic
4435 block must be created to hold the instrumentation code.
4438 @item -ftest-coverage
4439 @opindex ftest-coverage
4440 Produce a notes file that the @command{gcov} code-coverage utility
4441 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4442 show program coverage. Each source file's note file is called
4443 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4444 above for a description of @var{auxname} and instructions on how to
4445 generate test coverage data. Coverage data will match the source files
4446 more closely, if you do not optimize.
4448 @item -fdbg-cnt-list
4449 @opindex fdbg-cnt-list
4450 Print the name and the counter upperbound for all debug counters.
4452 @item -fdbg-cnt=@var{counter-value-list}
4454 Set the internal debug counter upperbound. @var{counter-value-list}
4455 is a comma-separated list of @var{name}:@var{value} pairs
4456 which sets the upperbound of each debug counter @var{name} to @var{value}.
4457 All debug counters have the initial upperbound of @var{UINT_MAX},
4458 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4459 e.g. With -fdbg-cnt=dce:10,tail_call:0
4460 dbg_cnt(dce) will return true only for first 10 invocations
4461 and dbg_cnt(tail_call) will return false always.
4463 @item -d@var{letters}
4464 @itemx -fdump-rtl-@var{pass}
4466 Says to make debugging dumps during compilation at times specified by
4467 @var{letters}. This is used for debugging the RTL-based passes of the
4468 compiler. The file names for most of the dumps are made by appending a
4469 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
4470 from the name of the output file, if explicitly specified and it is not
4471 an executable, otherwise it is the basename of the source file. These
4472 switches may have different effects when @option{-E} is used for
4475 Most debug dumps can be enabled either passing a letter to the @option{-d}
4476 option, or with a long @option{-fdump-rtl} switch; here are the possible
4477 letters for use in @var{letters} and @var{pass}, and their meanings:
4482 Annotate the assembler output with miscellaneous debugging information.
4484 @item -fdump-rtl-bbro
4485 @opindex fdump-rtl-bbro
4486 Dump after block reordering, to @file{@var{file}.148r.bbro}.
4488 @item -fdump-rtl-combine
4489 @opindex fdump-rtl-combine
4490 Dump after the RTL instruction combination pass, to the file
4491 @file{@var{file}.129r.combine}.
4493 @item -fdump-rtl-ce1
4494 @itemx -fdump-rtl-ce2
4495 @opindex fdump-rtl-ce1
4496 @opindex fdump-rtl-ce2
4497 @option{-fdump-rtl-ce1} enable dumping after the
4498 first if conversion, to the file @file{@var{file}.117r.ce1}.
4499 @option{-fdump-rtl-ce2} enable dumping after the second if
4500 conversion, to the file @file{@var{file}.130r.ce2}.
4502 @item -fdump-rtl-btl
4503 @itemx -fdump-rtl-dbr
4504 @opindex fdump-rtl-btl
4505 @opindex fdump-rtl-dbr
4506 @option{-fdump-rtl-btl} enable dumping after branch
4507 target load optimization, to @file{@var{file}.31.btl}.
4508 @option{-fdump-rtl-dbr} enable dumping after delayed branch
4509 scheduling, to @file{@var{file}.36.dbr}.
4513 Dump all macro definitions, at the end of preprocessing, in addition to
4516 @item -fdump-rtl-ce3
4517 @opindex fdump-rtl-ce3
4518 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
4520 @item -fdump-rtl-cfg
4521 @itemx -fdump-rtl-life
4522 @opindex fdump-rtl-cfg
4523 @opindex fdump-rtl-life
4524 @option{-fdump-rtl-cfg} enable dumping after control
4525 and data flow analysis, to @file{@var{file}.116r.cfg}.
4526 @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
4527 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
4529 @item -fdump-rtl-greg
4530 @opindex fdump-rtl-greg
4531 Dump after global register allocation, to @file{@var{file}.139r.greg}.
4533 @item -fdump-rtl-gcse
4534 @itemx -fdump-rtl-bypass
4535 @opindex fdump-rtl-gcse
4536 @opindex fdump-rtl-bypass
4537 @option{-fdump-rtl-gcse} enable dumping after GCSE, to
4538 @file{@var{file}.114r.gcse}. @option{-fdump-rtl-bypass}
4539 enable dumping after jump bypassing and control flow optimizations, to
4540 @file{@var{file}.115r.bypass}.
4543 @opindex fdump-rtl-eh
4544 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
4546 @item -fdump-rtl-sibling
4547 @opindex fdump-rtl-sibling
4548 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
4550 @item -fdump-rtl-jump
4551 @opindex fdump-rtl-jump
4552 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
4554 @item -fdump-rtl-stack
4555 @opindex fdump-rtl-stack
4556 Dump after conversion from GCC's "flat register file" registers to the
4557 x87's stack-like registers, to @file{@var{file}.152r.stack}.
4559 @item -fdump-rtl-lreg
4560 @opindex fdump-rtl-lreg
4561 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
4563 @item -fdump-rtl-loop2
4564 @opindex fdump-rtl-loop2
4565 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
4566 loop optimization pass, to @file{@var{file}.119r.loop2},
4567 @file{@var{file}.120r.loop2_init},
4568 @file{@var{file}.121r.loop2_invariant}, and
4569 @file{@var{file}.125r.loop2_done}.
4571 @item -fdump-rtl-sms
4572 @opindex fdump-rtl-sms
4573 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
4575 @item -fdump-rtl-mach
4576 @opindex fdump-rtl-mach
4577 Dump after performing the machine dependent reorganization pass, to
4578 @file{@var{file}.155r.mach} if that pass exists.
4580 @item -fdump-rtl-rnreg
4581 @opindex fdump-rtl-rnreg
4582 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
4584 @item -fdump-rtl-regmove
4585 @opindex fdump-rtl-regmove
4586 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
4588 @item -fdump-rtl-postreload
4589 @opindex fdump-rtl-postreload
4590 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
4592 @item -fdump-rtl-expand
4593 @opindex fdump-rtl-expand
4594 Dump after RTL generation, to @file{@var{file}.104r.expand}.
4596 @item -fdump-rtl-sched2
4597 @opindex fdump-rtl-sched2
4598 Dump after the second scheduling pass, to @file{@var{file}.149r.sched2}.
4600 @item -fdump-rtl-cse
4601 @opindex fdump-rtl-cse
4602 Dump after CSE (including the jump optimization that sometimes follows
4603 CSE), to @file{@var{file}.113r.cse}.
4605 @item -fdump-rtl-sched1
4606 @opindex fdump-rtl-sched1
4607 Dump after the first scheduling pass, to @file{@var{file}.136r.sched1}.
4609 @item -fdump-rtl-cse2
4610 @opindex fdump-rtl-cse2
4611 Dump after the second CSE pass (including the jump optimization that
4612 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
4614 @item -fdump-rtl-tracer
4615 @opindex fdump-rtl-tracer
4616 Dump after running tracer, to @file{@var{file}.118r.tracer}.
4618 @item -fdump-rtl-vpt
4619 @itemx -fdump-rtl-vartrack
4620 @opindex fdump-rtl-vpt
4621 @opindex fdump-rtl-vartrack
4622 @option{-fdump-rtl-vpt} enable dumping after the value
4623 profile transformations, to @file{@var{file}.10.vpt}.
4624 @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
4625 to @file{@var{file}.154r.vartrack}.
4627 @item -fdump-rtl-flow2
4628 @opindex fdump-rtl-flow2
4629 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
4631 @item -fdump-rtl-peephole2
4632 @opindex fdump-rtl-peephole2
4633 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
4635 @item -fdump-rtl-web
4636 @opindex fdump-rtl-web
4637 Dump after live range splitting, to @file{@var{file}.126r.web}.
4639 @item -fdump-rtl-all
4640 @opindex fdump-rtl-all
4641 Produce all the dumps listed above.
4645 Produce a core dump whenever an error occurs.
4649 Print statistics on memory usage, at the end of the run, to
4654 Annotate the assembler output with a comment indicating which
4655 pattern and alternative was used. The length of each instruction is
4660 Dump the RTL in the assembler output as a comment before each instruction.
4661 Also turns on @option{-dp} annotation.
4665 For each of the other indicated dump files (either with @option{-d} or
4666 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4667 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4671 Just generate RTL for a function instead of compiling it. Usually used
4672 with @samp{r} (@option{-fdump-rtl-expand}).
4676 Dump debugging information during parsing, to standard error.
4680 @opindex fdump-noaddr
4681 When doing debugging dumps (see @option{-d} option above), suppress
4682 address output. This makes it more feasible to use diff on debugging
4683 dumps for compiler invocations with different compiler binaries and/or
4684 different text / bss / data / heap / stack / dso start locations.
4686 @item -fdump-unnumbered
4687 @opindex fdump-unnumbered
4688 When doing debugging dumps (see @option{-d} option above), suppress instruction
4689 numbers and address output. This makes it more feasible to
4690 use diff on debugging dumps for compiler invocations with different
4691 options, in particular with and without @option{-g}.
4693 @item -fdump-translation-unit @r{(C++ only)}
4694 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4695 @opindex fdump-translation-unit
4696 Dump a representation of the tree structure for the entire translation
4697 unit to a file. The file name is made by appending @file{.tu} to the
4698 source file name. If the @samp{-@var{options}} form is used, @var{options}
4699 controls the details of the dump as described for the
4700 @option{-fdump-tree} options.
4702 @item -fdump-class-hierarchy @r{(C++ only)}
4703 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4704 @opindex fdump-class-hierarchy
4705 Dump a representation of each class's hierarchy and virtual function
4706 table layout to a file. The file name is made by appending @file{.class}
4707 to the source file name. If the @samp{-@var{options}} form is used,
4708 @var{options} controls the details of the dump as described for the
4709 @option{-fdump-tree} options.
4711 @item -fdump-ipa-@var{switch}
4713 Control the dumping at various stages of inter-procedural analysis
4714 language tree to a file. The file name is generated by appending a switch
4715 specific suffix to the source file name. The following dumps are possible:
4719 Enables all inter-procedural analysis dumps.
4722 Dumps information about call-graph optimization, unused function removal,
4723 and inlining decisions.
4726 Dump after function inlining.
4730 @item -fdump-statistics-@var{option}
4731 @opindex -fdump-statistics
4732 Enable and control dumping of pass statistics in a separate file. The
4733 file name is generated by appending a suffix ending in @samp{.statistics}
4734 to the source file name. If the @samp{-@var{option}} form is used,
4735 @samp{-stats} will cause counters to be summed over the whole compilation unit
4736 while @samp{-details} will dump every event as the passes generate them.
4737 The default with no option is to sum counters for each function compiled.
4739 @item -fdump-tree-@var{switch}
4740 @itemx -fdump-tree-@var{switch}-@var{options}
4742 Control the dumping at various stages of processing the intermediate
4743 language tree to a file. The file name is generated by appending a switch
4744 specific suffix to the source file name. If the @samp{-@var{options}}
4745 form is used, @var{options} is a list of @samp{-} separated options that
4746 control the details of the dump. Not all options are applicable to all
4747 dumps, those which are not meaningful will be ignored. The following
4748 options are available
4752 Print the address of each node. Usually this is not meaningful as it
4753 changes according to the environment and source file. Its primary use
4754 is for tying up a dump file with a debug environment.
4756 Inhibit dumping of members of a scope or body of a function merely
4757 because that scope has been reached. Only dump such items when they
4758 are directly reachable by some other path. When dumping pretty-printed
4759 trees, this option inhibits dumping the bodies of control structures.
4761 Print a raw representation of the tree. By default, trees are
4762 pretty-printed into a C-like representation.
4764 Enable more detailed dumps (not honored by every dump option).
4766 Enable dumping various statistics about the pass (not honored by every dump
4769 Enable showing basic block boundaries (disabled in raw dumps).
4771 Enable showing virtual operands for every statement.
4773 Enable showing line numbers for statements.
4775 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4777 Enable showing the tree dump for each statement.
4779 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
4780 and @option{lineno}.
4783 The following tree dumps are possible:
4787 Dump before any tree based optimization, to @file{@var{file}.original}.
4790 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4793 @opindex fdump-tree-gimple
4794 Dump each function before and after the gimplification pass to a file. The
4795 file name is made by appending @file{.gimple} to the source file name.
4798 @opindex fdump-tree-cfg
4799 Dump the control flow graph of each function to a file. The file name is
4800 made by appending @file{.cfg} to the source file name.
4803 @opindex fdump-tree-vcg
4804 Dump the control flow graph of each function to a file in VCG format. The
4805 file name is made by appending @file{.vcg} to the source file name. Note
4806 that if the file contains more than one function, the generated file cannot
4807 be used directly by VCG@. You will need to cut and paste each function's
4808 graph into its own separate file first.
4811 @opindex fdump-tree-ch
4812 Dump each function after copying loop headers. The file name is made by
4813 appending @file{.ch} to the source file name.
4816 @opindex fdump-tree-ssa
4817 Dump SSA related information to a file. The file name is made by appending
4818 @file{.ssa} to the source file name.
4821 @opindex fdump-tree-alias
4822 Dump aliasing information for each function. The file name is made by
4823 appending @file{.alias} to the source file name.
4826 @opindex fdump-tree-ccp
4827 Dump each function after CCP@. The file name is made by appending
4828 @file{.ccp} to the source file name.
4831 @opindex fdump-tree-storeccp
4832 Dump each function after STORE-CCP@. The file name is made by appending
4833 @file{.storeccp} to the source file name.
4836 @opindex fdump-tree-pre
4837 Dump trees after partial redundancy elimination. The file name is made
4838 by appending @file{.pre} to the source file name.
4841 @opindex fdump-tree-fre
4842 Dump trees after full redundancy elimination. The file name is made
4843 by appending @file{.fre} to the source file name.
4846 @opindex fdump-tree-copyprop
4847 Dump trees after copy propagation. The file name is made
4848 by appending @file{.copyprop} to the source file name.
4850 @item store_copyprop
4851 @opindex fdump-tree-store_copyprop
4852 Dump trees after store copy-propagation. The file name is made
4853 by appending @file{.store_copyprop} to the source file name.
4856 @opindex fdump-tree-dce
4857 Dump each function after dead code elimination. The file name is made by
4858 appending @file{.dce} to the source file name.
4861 @opindex fdump-tree-mudflap
4862 Dump each function after adding mudflap instrumentation. The file name is
4863 made by appending @file{.mudflap} to the source file name.
4866 @opindex fdump-tree-sra
4867 Dump each function after performing scalar replacement of aggregates. The
4868 file name is made by appending @file{.sra} to the source file name.
4871 @opindex fdump-tree-sink
4872 Dump each function after performing code sinking. The file name is made
4873 by appending @file{.sink} to the source file name.
4876 @opindex fdump-tree-dom
4877 Dump each function after applying dominator tree optimizations. The file
4878 name is made by appending @file{.dom} to the source file name.
4881 @opindex fdump-tree-dse
4882 Dump each function after applying dead store elimination. The file
4883 name is made by appending @file{.dse} to the source file name.
4886 @opindex fdump-tree-phiopt
4887 Dump each function after optimizing PHI nodes into straightline code. The file
4888 name is made by appending @file{.phiopt} to the source file name.
4891 @opindex fdump-tree-forwprop
4892 Dump each function after forward propagating single use variables. The file
4893 name is made by appending @file{.forwprop} to the source file name.
4896 @opindex fdump-tree-copyrename
4897 Dump each function after applying the copy rename optimization. The file
4898 name is made by appending @file{.copyrename} to the source file name.
4901 @opindex fdump-tree-nrv
4902 Dump each function after applying the named return value optimization on
4903 generic trees. The file name is made by appending @file{.nrv} to the source
4907 @opindex fdump-tree-vect
4908 Dump each function after applying vectorization of loops. The file name is
4909 made by appending @file{.vect} to the source file name.
4912 @opindex fdump-tree-vrp
4913 Dump each function after Value Range Propagation (VRP). The file name
4914 is made by appending @file{.vrp} to the source file name.
4917 @opindex fdump-tree-all
4918 Enable all the available tree dumps with the flags provided in this option.
4921 @item -ftree-vectorizer-verbose=@var{n}
4922 @opindex ftree-vectorizer-verbose
4923 This option controls the amount of debugging output the vectorizer prints.
4924 This information is written to standard error, unless
4925 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4926 in which case it is output to the usual dump listing file, @file{.vect}.
4927 For @var{n}=0 no diagnostic information is reported.
4928 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4929 and the total number of loops that got vectorized.
4930 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4931 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
4932 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4933 level that @option{-fdump-tree-vect-stats} uses.
4934 Higher verbosity levels mean either more information dumped for each
4935 reported loop, or same amount of information reported for more loops:
4936 If @var{n}=3, alignment related information is added to the reports.
4937 If @var{n}=4, data-references related information (e.g.@: memory dependences,
4938 memory access-patterns) is added to the reports.
4939 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4940 that did not pass the first analysis phase (i.e., may not be countable, or
4941 may have complicated control-flow).
4942 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4943 For @var{n}=7, all the information the vectorizer generates during its
4944 analysis and transformation is reported. This is the same verbosity level
4945 that @option{-fdump-tree-vect-details} uses.
4947 @item -frandom-seed=@var{string}
4948 @opindex frandom-string
4949 This option provides a seed that GCC uses when it would otherwise use
4950 random numbers. It is used to generate certain symbol names
4951 that have to be different in every compiled file. It is also used to
4952 place unique stamps in coverage data files and the object files that
4953 produce them. You can use the @option{-frandom-seed} option to produce
4954 reproducibly identical object files.
4956 The @var{string} should be different for every file you compile.
4958 @item -fsched-verbose=@var{n}
4959 @opindex fsched-verbose
4960 On targets that use instruction scheduling, this option controls the
4961 amount of debugging output the scheduler prints. This information is
4962 written to standard error, unless @option{-dS} or @option{-dR} is
4963 specified, in which case it is output to the usual dump
4964 listing file, @file{.sched} or @file{.sched2} respectively. However
4965 for @var{n} greater than nine, the output is always printed to standard
4968 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4969 same information as @option{-dRS}. For @var{n} greater than one, it
4970 also output basic block probabilities, detailed ready list information
4971 and unit/insn info. For @var{n} greater than two, it includes RTL
4972 at abort point, control-flow and regions info. And for @var{n} over
4973 four, @option{-fsched-verbose} also includes dependence info.
4977 Store the usual ``temporary'' intermediate files permanently; place them
4978 in the current directory and name them based on the source file. Thus,
4979 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4980 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4981 preprocessed @file{foo.i} output file even though the compiler now
4982 normally uses an integrated preprocessor.
4984 When used in combination with the @option{-x} command line option,
4985 @option{-save-temps} is sensible enough to avoid over writing an
4986 input source file with the same extension as an intermediate file.
4987 The corresponding intermediate file may be obtained by renaming the
4988 source file before using @option{-save-temps}.
4992 Report the CPU time taken by each subprocess in the compilation
4993 sequence. For C source files, this is the compiler proper and assembler
4994 (plus the linker if linking is done). The output looks like this:
5001 The first number on each line is the ``user time'', that is time spent
5002 executing the program itself. The second number is ``system time'',
5003 time spent executing operating system routines on behalf of the program.
5004 Both numbers are in seconds.
5006 @item -fvar-tracking
5007 @opindex fvar-tracking
5008 Run variable tracking pass. It computes where variables are stored at each
5009 position in code. Better debugging information is then generated
5010 (if the debugging information format supports this information).
5012 It is enabled by default when compiling with optimization (@option{-Os},
5013 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5014 the debug info format supports it.
5016 @item -print-file-name=@var{library}
5017 @opindex print-file-name
5018 Print the full absolute name of the library file @var{library} that
5019 would be used when linking---and don't do anything else. With this
5020 option, GCC does not compile or link anything; it just prints the
5023 @item -print-multi-directory
5024 @opindex print-multi-directory
5025 Print the directory name corresponding to the multilib selected by any
5026 other switches present in the command line. This directory is supposed
5027 to exist in @env{GCC_EXEC_PREFIX}.
5029 @item -print-multi-lib
5030 @opindex print-multi-lib
5031 Print the mapping from multilib directory names to compiler switches
5032 that enable them. The directory name is separated from the switches by
5033 @samp{;}, and each switch starts with an @samp{@@} instead of the
5034 @samp{-}, without spaces between multiple switches. This is supposed to
5035 ease shell-processing.
5037 @item -print-prog-name=@var{program}
5038 @opindex print-prog-name
5039 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5041 @item -print-libgcc-file-name
5042 @opindex print-libgcc-file-name
5043 Same as @option{-print-file-name=libgcc.a}.
5045 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5046 but you do want to link with @file{libgcc.a}. You can do
5049 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5052 @item -print-search-dirs
5053 @opindex print-search-dirs
5054 Print the name of the configured installation directory and a list of
5055 program and library directories @command{gcc} will search---and don't do anything else.
5057 This is useful when @command{gcc} prints the error message
5058 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5059 To resolve this you either need to put @file{cpp0} and the other compiler
5060 components where @command{gcc} expects to find them, or you can set the environment
5061 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5062 Don't forget the trailing @samp{/}.
5063 @xref{Environment Variables}.
5065 @item -print-sysroot
5066 @opindex print-sysroot
5067 Print the target sysroot directory that will be used during
5068 compilation. This is the target sysroot specified either at configure
5069 time or or using the @option{--sysroot} option, possibly with an extra
5070 suffix that depends on compilation options. If no target sysroot is
5071 specified, the option prints nothing.
5073 @item -print-sysroot-headers-suffix
5074 @opindex print-sysroot-headers-suffix
5075 Print the suffix added to the target sysroot when searching for
5076 headers, or give an error if the compiler is not configured with such
5077 a suffix---and don't do anything else.
5080 @opindex dumpmachine
5081 Print the compiler's target machine (for example,
5082 @samp{i686-pc-linux-gnu})---and don't do anything else.
5085 @opindex dumpversion
5086 Print the compiler version (for example, @samp{3.0})---and don't do
5091 Print the compiler's built-in specs---and don't do anything else. (This
5092 is used when GCC itself is being built.) @xref{Spec Files}.
5094 @item -feliminate-unused-debug-types
5095 @opindex feliminate-unused-debug-types
5096 Normally, when producing DWARF2 output, GCC will emit debugging
5097 information for all types declared in a compilation
5098 unit, regardless of whether or not they are actually used
5099 in that compilation unit. Sometimes this is useful, such as
5100 if, in the debugger, you want to cast a value to a type that is
5101 not actually used in your program (but is declared). More often,
5102 however, this results in a significant amount of wasted space.
5103 With this option, GCC will avoid producing debug symbol output
5104 for types that are nowhere used in the source file being compiled.
5107 @node Optimize Options
5108 @section Options That Control Optimization
5109 @cindex optimize options
5110 @cindex options, optimization
5112 These options control various sorts of optimizations.
5114 Without any optimization option, the compiler's goal is to reduce the
5115 cost of compilation and to make debugging produce the expected
5116 results. Statements are independent: if you stop the program with a
5117 breakpoint between statements, you can then assign a new value to any
5118 variable or change the program counter to any other statement in the
5119 function and get exactly the results you would expect from the source
5122 Turning on optimization flags makes the compiler attempt to improve
5123 the performance and/or code size at the expense of compilation time
5124 and possibly the ability to debug the program.
5126 The compiler performs optimization based on the knowledge it has of the
5127 program. Compiling multiple files at once to a single output file mode allows
5128 the compiler to use information gained from all of the files when compiling
5131 Not all optimizations are controlled directly by a flag. Only
5132 optimizations that have a flag are listed.
5139 Optimize. Optimizing compilation takes somewhat more time, and a lot
5140 more memory for a large function.
5142 With @option{-O}, the compiler tries to reduce code size and execution
5143 time, without performing any optimizations that take a great deal of
5146 @option{-O} turns on the following optimization flags:
5149 -fcprop-registers @gol
5152 -fdelayed-branch @gol
5154 -fguess-branch-probability @gol
5155 -fif-conversion2 @gol
5156 -fif-conversion @gol
5157 -finline-small-functions @gol
5158 -fipa-pure-const @gol
5159 -fipa-reference @gol
5161 -fsplit-wide-types @gol
5162 -ftree-builtin-call-dce @gol
5165 -ftree-copyrename @gol
5167 -ftree-dominator-opts @gol
5174 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5175 where doing so does not interfere with debugging.
5179 Optimize even more. GCC performs nearly all supported optimizations
5180 that do not involve a space-speed tradeoff. The compiler does not
5181 perform loop unrolling or function inlining when you specify @option{-O2}.
5182 As compared to @option{-O}, this option increases both compilation time
5183 and the performance of the generated code.
5185 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5186 also turns on the following optimization flags:
5187 @gccoptlist{-fthread-jumps @gol
5188 -falign-functions -falign-jumps @gol
5189 -falign-loops -falign-labels @gol
5192 -fcse-follow-jumps -fcse-skip-blocks @gol
5193 -fdelete-null-pointer-checks @gol
5194 -fexpensive-optimizations @gol
5195 -fgcse -fgcse-lm @gol
5196 -findirect-inlining @gol
5197 -foptimize-sibling-calls @gol
5200 -freorder-blocks -freorder-functions @gol
5201 -frerun-cse-after-loop @gol
5202 -fsched-interblock -fsched-spec @gol
5203 -fschedule-insns -fschedule-insns2 @gol
5204 -fstrict-aliasing -fstrict-overflow @gol
5205 -ftree-switch-conversion @gol
5209 Please note the warning under @option{-fgcse} about
5210 invoking @option{-O2} on programs that use computed gotos.
5214 Optimize yet more. @option{-O3} turns on all optimizations specified
5215 by @option{-O2} and also turns on the @option{-finline-functions},
5216 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5217 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5221 Reduce compilation time and make debugging produce the expected
5222 results. This is the default.
5226 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5227 do not typically increase code size. It also performs further
5228 optimizations designed to reduce code size.
5230 @option{-Os} disables the following optimization flags:
5231 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5232 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5233 -fprefetch-loop-arrays -ftree-vect-loop-version}
5235 If you use multiple @option{-O} options, with or without level numbers,
5236 the last such option is the one that is effective.
5239 Options of the form @option{-f@var{flag}} specify machine-independent
5240 flags. Most flags have both positive and negative forms; the negative
5241 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5242 below, only one of the forms is listed---the one you typically will
5243 use. You can figure out the other form by either removing @samp{no-}
5246 The following options control specific optimizations. They are either
5247 activated by @option{-O} options or are related to ones that are. You
5248 can use the following flags in the rare cases when ``fine-tuning'' of
5249 optimizations to be performed is desired.
5252 @item -fno-default-inline
5253 @opindex fno-default-inline
5254 Do not make member functions inline by default merely because they are
5255 defined inside the class scope (C++ only). Otherwise, when you specify
5256 @w{@option{-O}}, member functions defined inside class scope are compiled
5257 inline by default; i.e., you don't need to add @samp{inline} in front of
5258 the member function name.
5260 @item -fno-defer-pop
5261 @opindex fno-defer-pop
5262 Always pop the arguments to each function call as soon as that function
5263 returns. For machines which must pop arguments after a function call,
5264 the compiler normally lets arguments accumulate on the stack for several
5265 function calls and pops them all at once.
5267 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5269 @item -fforward-propagate
5270 @opindex fforward-propagate
5271 Perform a forward propagation pass on RTL@. The pass tries to combine two
5272 instructions and checks if the result can be simplified. If loop unrolling
5273 is active, two passes are performed and the second is scheduled after
5276 This option is enabled by default at optimization levels @option{-O2},
5277 @option{-O3}, @option{-Os}.
5279 @item -fomit-frame-pointer
5280 @opindex fomit-frame-pointer
5281 Don't keep the frame pointer in a register for functions that
5282 don't need one. This avoids the instructions to save, set up and
5283 restore frame pointers; it also makes an extra register available
5284 in many functions. @strong{It also makes debugging impossible on
5287 On some machines, such as the VAX, this flag has no effect, because
5288 the standard calling sequence automatically handles the frame pointer
5289 and nothing is saved by pretending it doesn't exist. The
5290 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5291 whether a target machine supports this flag. @xref{Registers,,Register
5292 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5294 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5296 @item -foptimize-sibling-calls
5297 @opindex foptimize-sibling-calls
5298 Optimize sibling and tail recursive calls.
5300 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5304 Don't pay attention to the @code{inline} keyword. Normally this option
5305 is used to keep the compiler from expanding any functions inline.
5306 Note that if you are not optimizing, no functions can be expanded inline.
5308 @item -finline-small-functions
5309 @opindex finline-small-functions
5310 Integrate functions into their callers when their body is smaller than expected
5311 function call code (so overall size of program gets smaller). The compiler
5312 heuristically decides which functions are simple enough to be worth integrating
5315 Enabled at level @option{-O2}.
5317 @item -findirect-inlining
5318 @opindex findirect-inlining
5319 Inline also indirect calls that are discovered to be known at compile
5320 time thanks to previous inlining. This option has any effect only
5321 when inlining itself is turned on by the @option{-finline-functions}
5322 or @option{-finline-small-functions} options.
5324 Enabled at level @option{-O2}.
5326 @item -finline-functions
5327 @opindex finline-functions
5328 Integrate all simple functions into their callers. The compiler
5329 heuristically decides which functions are simple enough to be worth
5330 integrating in this way.
5332 If all calls to a given function are integrated, and the function is
5333 declared @code{static}, then the function is normally not output as
5334 assembler code in its own right.
5336 Enabled at level @option{-O3}.
5338 @item -finline-functions-called-once
5339 @opindex finline-functions-called-once
5340 Consider all @code{static} functions called once for inlining into their
5341 caller even if they are not marked @code{inline}. If a call to a given
5342 function is integrated, then the function is not output as assembler code
5345 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5347 @item -fearly-inlining
5348 @opindex fearly-inlining
5349 Inline functions marked by @code{always_inline} and functions whose body seems
5350 smaller than the function call overhead early before doing
5351 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5352 makes profiling significantly cheaper and usually inlining faster on programs
5353 having large chains of nested wrapper functions.
5357 @item -finline-limit=@var{n}
5358 @opindex finline-limit
5359 By default, GCC limits the size of functions that can be inlined. This flag
5360 allows coarse control of this limit. @var{n} is the size of functions that
5361 can be inlined in number of pseudo instructions.
5363 Inlining is actually controlled by a number of parameters, which may be
5364 specified individually by using @option{--param @var{name}=@var{value}}.
5365 The @option{-finline-limit=@var{n}} option sets some of these parameters
5369 @item max-inline-insns-single
5370 is set to @var{n}/2.
5371 @item max-inline-insns-auto
5372 is set to @var{n}/2.
5375 See below for a documentation of the individual
5376 parameters controlling inlining and for the defaults of these parameters.
5378 @emph{Note:} there may be no value to @option{-finline-limit} that results
5379 in default behavior.
5381 @emph{Note:} pseudo instruction represents, in this particular context, an
5382 abstract measurement of function's size. In no way does it represent a count
5383 of assembly instructions and as such its exact meaning might change from one
5384 release to an another.
5386 @item -fkeep-inline-functions
5387 @opindex fkeep-inline-functions
5388 In C, emit @code{static} functions that are declared @code{inline}
5389 into the object file, even if the function has been inlined into all
5390 of its callers. This switch does not affect functions using the
5391 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5392 inline functions into the object file.
5394 @item -fkeep-static-consts
5395 @opindex fkeep-static-consts
5396 Emit variables declared @code{static const} when optimization isn't turned
5397 on, even if the variables aren't referenced.
5399 GCC enables this option by default. If you want to force the compiler to
5400 check if the variable was referenced, regardless of whether or not
5401 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5403 @item -fmerge-constants
5404 @opindex fmerge-constants
5405 Attempt to merge identical constants (string constants and floating point
5406 constants) across compilation units.
5408 This option is the default for optimized compilation if the assembler and
5409 linker support it. Use @option{-fno-merge-constants} to inhibit this
5412 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5414 @item -fmerge-all-constants
5415 @opindex fmerge-all-constants
5416 Attempt to merge identical constants and identical variables.
5418 This option implies @option{-fmerge-constants}. In addition to
5419 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5420 arrays or initialized constant variables with integral or floating point
5421 types. Languages like C or C++ require each non-automatic variable to
5422 have distinct location, so using this option will result in non-conforming
5425 @item -fmodulo-sched
5426 @opindex fmodulo-sched
5427 Perform swing modulo scheduling immediately before the first scheduling
5428 pass. This pass looks at innermost loops and reorders their
5429 instructions by overlapping different iterations.
5431 @item -fmodulo-sched-allow-regmoves
5432 @opindex fmodulo-sched-allow-regmoves
5433 Perform more aggressive SMS based modulo scheduling with register moves
5434 allowed. By setting this flag certain anti-dependences edges will be
5435 deleted which will trigger the generation of reg-moves based on the
5436 life-range analysis. This option is effective only with
5437 @option{-fmodulo-sched} enabled.
5439 @item -fno-branch-count-reg
5440 @opindex fno-branch-count-reg
5441 Do not use ``decrement and branch'' instructions on a count register,
5442 but instead generate a sequence of instructions that decrement a
5443 register, compare it against zero, then branch based upon the result.
5444 This option is only meaningful on architectures that support such
5445 instructions, which include x86, PowerPC, IA-64 and S/390.
5447 The default is @option{-fbranch-count-reg}.
5449 @item -fno-function-cse
5450 @opindex fno-function-cse
5451 Do not put function addresses in registers; make each instruction that
5452 calls a constant function contain the function's address explicitly.
5454 This option results in less efficient code, but some strange hacks
5455 that alter the assembler output may be confused by the optimizations
5456 performed when this option is not used.
5458 The default is @option{-ffunction-cse}
5460 @item -fno-zero-initialized-in-bss
5461 @opindex fno-zero-initialized-in-bss
5462 If the target supports a BSS section, GCC by default puts variables that
5463 are initialized to zero into BSS@. This can save space in the resulting
5466 This option turns off this behavior because some programs explicitly
5467 rely on variables going to the data section. E.g., so that the
5468 resulting executable can find the beginning of that section and/or make
5469 assumptions based on that.
5471 The default is @option{-fzero-initialized-in-bss}.
5473 @item -fmudflap -fmudflapth -fmudflapir
5477 @cindex bounds checking
5479 For front-ends that support it (C and C++), instrument all risky
5480 pointer/array dereferencing operations, some standard library
5481 string/heap functions, and some other associated constructs with
5482 range/validity tests. Modules so instrumented should be immune to
5483 buffer overflows, invalid heap use, and some other classes of C/C++
5484 programming errors. The instrumentation relies on a separate runtime
5485 library (@file{libmudflap}), which will be linked into a program if
5486 @option{-fmudflap} is given at link time. Run-time behavior of the
5487 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5488 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5491 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5492 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5493 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5494 instrumentation should ignore pointer reads. This produces less
5495 instrumentation (and therefore faster execution) and still provides
5496 some protection against outright memory corrupting writes, but allows
5497 erroneously read data to propagate within a program.
5499 @item -fthread-jumps
5500 @opindex fthread-jumps
5501 Perform optimizations where we check to see if a jump branches to a
5502 location where another comparison subsumed by the first is found. If
5503 so, the first branch is redirected to either the destination of the
5504 second branch or a point immediately following it, depending on whether
5505 the condition is known to be true or false.
5507 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5509 @item -fsplit-wide-types
5510 @opindex fsplit-wide-types
5511 When using a type that occupies multiple registers, such as @code{long
5512 long} on a 32-bit system, split the registers apart and allocate them
5513 independently. This normally generates better code for those types,
5514 but may make debugging more difficult.
5516 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
5519 @item -fcse-follow-jumps
5520 @opindex fcse-follow-jumps
5521 In common subexpression elimination (CSE), scan through jump instructions
5522 when the target of the jump is not reached by any other path. For
5523 example, when CSE encounters an @code{if} statement with an
5524 @code{else} clause, CSE will follow the jump when the condition
5527 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5529 @item -fcse-skip-blocks
5530 @opindex fcse-skip-blocks
5531 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
5532 follow jumps which conditionally skip over blocks. When CSE
5533 encounters a simple @code{if} statement with no else clause,
5534 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
5535 body of the @code{if}.
5537 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5539 @item -frerun-cse-after-loop
5540 @opindex frerun-cse-after-loop
5541 Re-run common subexpression elimination after loop optimizations has been
5544 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5548 Perform a global common subexpression elimination pass.
5549 This pass also performs global constant and copy propagation.
5551 @emph{Note:} When compiling a program using computed gotos, a GCC
5552 extension, you may get better runtime performance if you disable
5553 the global common subexpression elimination pass by adding
5554 @option{-fno-gcse} to the command line.
5556 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5560 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
5561 attempt to move loads which are only killed by stores into themselves. This
5562 allows a loop containing a load/store sequence to be changed to a load outside
5563 the loop, and a copy/store within the loop.
5565 Enabled by default when gcse is enabled.
5569 When @option{-fgcse-sm} is enabled, a store motion pass is run after
5570 global common subexpression elimination. This pass will attempt to move
5571 stores out of loops. When used in conjunction with @option{-fgcse-lm},
5572 loops containing a load/store sequence can be changed to a load before
5573 the loop and a store after the loop.
5575 Not enabled at any optimization level.
5579 When @option{-fgcse-las} is enabled, the global common subexpression
5580 elimination pass eliminates redundant loads that come after stores to the
5581 same memory location (both partial and full redundancies).
5583 Not enabled at any optimization level.
5585 @item -fgcse-after-reload
5586 @opindex fgcse-after-reload
5587 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
5588 pass is performed after reload. The purpose of this pass is to cleanup
5591 @item -funsafe-loop-optimizations
5592 @opindex funsafe-loop-optimizations
5593 If given, the loop optimizer will assume that loop indices do not
5594 overflow, and that the loops with nontrivial exit condition are not
5595 infinite. This enables a wider range of loop optimizations even if
5596 the loop optimizer itself cannot prove that these assumptions are valid.
5597 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
5598 if it finds this kind of loop.
5600 @item -fcrossjumping
5601 @opindex fcrossjumping
5602 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
5603 resulting code may or may not perform better than without cross-jumping.
5605 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5607 @item -fauto-inc-dec
5608 @opindex fauto-inc-dec
5609 Combine increments or decrements of addresses with memory accesses.
5610 This pass is always skipped on architectures that do not have
5611 instructions to support this. Enabled by default at @option{-O} and
5612 higher on architectures that support this.
5616 Perform dead code elimination (DCE) on RTL@.
5617 Enabled by default at @option{-O} and higher.
5621 Perform dead store elimination (DSE) on RTL@.
5622 Enabled by default at @option{-O} and higher.
5624 @item -fif-conversion
5625 @opindex fif-conversion
5626 Attempt to transform conditional jumps into branch-less equivalents. This
5627 include use of conditional moves, min, max, set flags and abs instructions, and
5628 some tricks doable by standard arithmetics. The use of conditional execution
5629 on chips where it is available is controlled by @code{if-conversion2}.
5631 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5633 @item -fif-conversion2
5634 @opindex fif-conversion2
5635 Use conditional execution (where available) to transform conditional jumps into
5636 branch-less equivalents.
5638 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5640 @item -fdelete-null-pointer-checks
5641 @opindex fdelete-null-pointer-checks
5642 Use global dataflow analysis to identify and eliminate useless checks
5643 for null pointers. The compiler assumes that dereferencing a null
5644 pointer would have halted the program. If a pointer is checked after
5645 it has already been dereferenced, it cannot be null.
5647 In some environments, this assumption is not true, and programs can
5648 safely dereference null pointers. Use
5649 @option{-fno-delete-null-pointer-checks} to disable this optimization
5650 for programs which depend on that behavior.
5652 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5654 @item -fexpensive-optimizations
5655 @opindex fexpensive-optimizations
5656 Perform a number of minor optimizations that are relatively expensive.
5658 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5660 @item -foptimize-register-move
5662 @opindex foptimize-register-move
5664 Attempt to reassign register numbers in move instructions and as
5665 operands of other simple instructions in order to maximize the amount of
5666 register tying. This is especially helpful on machines with two-operand
5669 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
5672 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5674 @item -fdelayed-branch
5675 @opindex fdelayed-branch
5676 If supported for the target machine, attempt to reorder instructions
5677 to exploit instruction slots available after delayed branch
5680 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5682 @item -fschedule-insns
5683 @opindex fschedule-insns
5684 If supported for the target machine, attempt to reorder instructions to
5685 eliminate execution stalls due to required data being unavailable. This
5686 helps machines that have slow floating point or memory load instructions
5687 by allowing other instructions to be issued until the result of the load
5688 or floating point instruction is required.
5690 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5692 @item -fschedule-insns2
5693 @opindex fschedule-insns2
5694 Similar to @option{-fschedule-insns}, but requests an additional pass of
5695 instruction scheduling after register allocation has been done. This is
5696 especially useful on machines with a relatively small number of
5697 registers and where memory load instructions take more than one cycle.
5699 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5701 @item -fno-sched-interblock
5702 @opindex fno-sched-interblock
5703 Don't schedule instructions across basic blocks. This is normally
5704 enabled by default when scheduling before register allocation, i.e.@:
5705 with @option{-fschedule-insns} or at @option{-O2} or higher.
5707 @item -fno-sched-spec
5708 @opindex fno-sched-spec
5709 Don't allow speculative motion of non-load instructions. This is normally
5710 enabled by default when scheduling before register allocation, i.e.@:
5711 with @option{-fschedule-insns} or at @option{-O2} or higher.
5713 @item -fsched-spec-load
5714 @opindex fsched-spec-load
5715 Allow speculative motion of some load instructions. This only makes
5716 sense when scheduling before register allocation, i.e.@: with
5717 @option{-fschedule-insns} or at @option{-O2} or higher.
5719 @item -fsched-spec-load-dangerous
5720 @opindex fsched-spec-load-dangerous
5721 Allow speculative motion of more load instructions. This only makes
5722 sense when scheduling before register allocation, i.e.@: with
5723 @option{-fschedule-insns} or at @option{-O2} or higher.
5725 @item -fsched-stalled-insns
5726 @itemx -fsched-stalled-insns=@var{n}
5727 @opindex fsched-stalled-insns
5728 Define how many insns (if any) can be moved prematurely from the queue
5729 of stalled insns into the ready list, during the second scheduling pass.
5730 @option{-fno-sched-stalled-insns} means that no insns will be moved
5731 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
5732 on how many queued insns can be moved prematurely.
5733 @option{-fsched-stalled-insns} without a value is equivalent to
5734 @option{-fsched-stalled-insns=1}.
5736 @item -fsched-stalled-insns-dep
5737 @itemx -fsched-stalled-insns-dep=@var{n}
5738 @opindex fsched-stalled-insns-dep
5739 Define how many insn groups (cycles) will be examined for a dependency
5740 on a stalled insn that is candidate for premature removal from the queue
5741 of stalled insns. This has an effect only during the second scheduling pass,
5742 and only if @option{-fsched-stalled-insns} is used.
5743 @option{-fno-sched-stalled-insns-dep} is equivalent to
5744 @option{-fsched-stalled-insns-dep=0}.
5745 @option{-fsched-stalled-insns-dep} without a value is equivalent to
5746 @option{-fsched-stalled-insns-dep=1}.
5748 @item -fsched2-use-superblocks
5749 @opindex fsched2-use-superblocks
5750 When scheduling after register allocation, do use superblock scheduling
5751 algorithm. Superblock scheduling allows motion across basic block boundaries
5752 resulting on faster schedules. This option is experimental, as not all machine
5753 descriptions used by GCC model the CPU closely enough to avoid unreliable
5754 results from the algorithm.
5756 This only makes sense when scheduling after register allocation, i.e.@: with
5757 @option{-fschedule-insns2} or at @option{-O2} or higher.
5759 @item -fsched2-use-traces
5760 @opindex fsched2-use-traces
5761 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5762 allocation and additionally perform code duplication in order to increase the
5763 size of superblocks using tracer pass. See @option{-ftracer} for details on
5766 This mode should produce faster but significantly longer programs. Also
5767 without @option{-fbranch-probabilities} the traces constructed may not
5768 match the reality and hurt the performance. This only makes
5769 sense when scheduling after register allocation, i.e.@: with
5770 @option{-fschedule-insns2} or at @option{-O2} or higher.
5774 Eliminate redundant sign extension instructions and move the non-redundant
5775 ones to optimal placement using lazy code motion (LCM).
5777 @item -freschedule-modulo-scheduled-loops
5778 @opindex freschedule-modulo-scheduled-loops
5779 The modulo scheduling comes before the traditional scheduling, if a loop
5780 was modulo scheduled we may want to prevent the later scheduling passes
5781 from changing its schedule, we use this option to control that.
5783 @item -fcaller-saves
5784 @opindex fcaller-saves
5785 Enable values to be allocated in registers that will be clobbered by
5786 function calls, by emitting extra instructions to save and restore the
5787 registers around such calls. Such allocation is done only when it
5788 seems to result in better code than would otherwise be produced.
5790 This option is always enabled by default on certain machines, usually
5791 those which have no call-preserved registers to use instead.
5793 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5795 @item -ftree-reassoc
5796 @opindex ftree-reassoc
5797 Perform reassociation on trees. This flag is enabled by default
5798 at @option{-O} and higher.
5802 Perform partial redundancy elimination (PRE) on trees. This flag is
5803 enabled by default at @option{-O2} and @option{-O3}.
5807 Perform full redundancy elimination (FRE) on trees. The difference
5808 between FRE and PRE is that FRE only considers expressions
5809 that are computed on all paths leading to the redundant computation.
5810 This analysis is faster than PRE, though it exposes fewer redundancies.
5811 This flag is enabled by default at @option{-O} and higher.
5813 @item -ftree-copy-prop
5814 @opindex ftree-copy-prop
5815 Perform copy propagation on trees. This pass eliminates unnecessary
5816 copy operations. This flag is enabled by default at @option{-O} and
5819 @item -fipa-pure-const
5820 @opindex fipa-pure-const
5821 Discover which functions are pure or constant.
5822 Enabled by default at @option{-O} and higher.
5824 @item -fipa-reference
5825 @opindex fipa-reference
5826 Discover which static variables do not escape cannot escape the
5828 Enabled by default at @option{-O} and higher.
5830 @item -fipa-struct-reorg
5831 @opindex fipa-struct-reorg
5832 Perform structure reorganization optimization, that change C-like structures
5833 layout in order to better utilize spatial locality. This transformation is
5834 affective for programs containing arrays of structures. Available in two
5835 compilation modes: profile-based (enabled with @option{-fprofile-generate})
5836 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
5837 to provide the safety of this transformation. It works only in whole program
5838 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
5839 enabled. Structures considered @samp{cold} by this transformation are not
5840 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
5842 With this flag, the program debug info reflects a new structure layout.
5846 Perform interprocedural pointer analysis.
5850 Perform interprocedural constant propagation.
5851 This optimization analyzes the program to determine when values passed
5852 to functions are constants and then optimizes accordingly.
5853 This optimization can substantially increase performance
5854 if the application has constants passed to functions, but
5855 because this optimization can create multiple copies of functions,
5856 it may significantly increase code size.
5858 @item -fipa-matrix-reorg
5859 @opindex fipa-matrix-reorg
5860 Perform matrix flattening and transposing.
5861 Matrix flattening tries to replace a m-dimensional matrix
5862 with its equivalent n-dimensional matrix, where n < m.
5863 This reduces the level of indirection needed for accessing the elements
5864 of the matrix. The second optimization is matrix transposing that
5865 attemps to change the order of the matrix's dimensions in order to
5866 improve cache locality.
5867 Both optimizations need fwhole-program flag.
5868 Transposing is enabled only if profiling information is avaliable.
5873 Perform forward store motion on trees. This flag is
5874 enabled by default at @option{-O} and higher.
5878 Perform sparse conditional constant propagation (CCP) on trees. This
5879 pass only operates on local scalar variables and is enabled by default
5880 at @option{-O} and higher.
5882 @item -ftree-store-ccp
5883 @opindex ftree-store-ccp
5884 Perform sparse conditional constant propagation (CCP) on trees. This
5885 pass operates on both local scalar variables and memory stores and
5886 loads (global variables, structures, arrays, etc). This flag is
5887 enabled by default at @option{-O2} and higher.
5889 @item -ftree-switch-conversion
5890 Perform conversion of simple initializations in a switch to
5891 initializations from a scalar array. This flag is enabled by default
5892 at @option{-O2} and higher.
5896 Perform dead code elimination (DCE) on trees. This flag is enabled by
5897 default at @option{-O} and higher.
5899 @item -ftree-builtin-call-dce
5900 @opindex ftree-builtin-call-dce
5901 Perform conditional dead code elimination (DCE) for calls to builtin functions
5902 that may set @code{errno} but are otherwise side-effect free. This flag is
5903 enabled by default at @option{-O2} and higher if @option{-Os} is not also
5906 @item -ftree-dominator-opts
5907 @opindex ftree-dominator-opts
5908 Perform a variety of simple scalar cleanups (constant/copy
5909 propagation, redundancy elimination, range propagation and expression
5910 simplification) based on a dominator tree traversal. This also
5911 performs jump threading (to reduce jumps to jumps). This flag is
5912 enabled by default at @option{-O} and higher.
5916 Perform dead store elimination (DSE) on trees. A dead store is a store into
5917 a memory location which will later be overwritten by another store without
5918 any intervening loads. In this case the earlier store can be deleted. This
5919 flag is enabled by default at @option{-O} and higher.
5923 Perform loop header copying on trees. This is beneficial since it increases
5924 effectiveness of code motion optimizations. It also saves one jump. This flag
5925 is enabled by default at @option{-O} and higher. It is not enabled
5926 for @option{-Os}, since it usually increases code size.
5928 @item -ftree-loop-optimize
5929 @opindex ftree-loop-optimize
5930 Perform loop optimizations on trees. This flag is enabled by default
5931 at @option{-O} and higher.
5933 @item -ftree-loop-linear
5934 @opindex ftree-loop-linear
5935 Perform linear loop transformations on tree. This flag can improve cache
5936 performance and allow further loop optimizations to take place.
5938 @item -fcheck-data-deps
5939 @opindex fcheck-data-deps
5940 Compare the results of several data dependence analyzers. This option
5941 is used for debugging the data dependence analyzers.
5943 @item -ftree-loop-distribution
5944 Perform loop distribution. This flag can improve cache performance on
5945 big loop bodies and allow further loop optimizations, like
5946 parallelization or vectorization, to take place. For example, the loop
5963 @item -ftree-loop-im
5964 @opindex ftree-loop-im
5965 Perform loop invariant motion on trees. This pass moves only invariants that
5966 would be hard to handle at RTL level (function calls, operations that expand to
5967 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5968 operands of conditions that are invariant out of the loop, so that we can use
5969 just trivial invariantness analysis in loop unswitching. The pass also includes
5972 @item -ftree-loop-ivcanon
5973 @opindex ftree-loop-ivcanon
5974 Create a canonical counter for number of iterations in the loop for that
5975 determining number of iterations requires complicated analysis. Later
5976 optimizations then may determine the number easily. Useful especially
5977 in connection with unrolling.
5981 Perform induction variable optimizations (strength reduction, induction
5982 variable merging and induction variable elimination) on trees.
5984 @item -ftree-parallelize-loops=n
5985 @opindex ftree-parallelize-loops
5986 Parallelize loops, i.e., split their iteration space to run in n threads.
5987 This is only possible for loops whose iterations are independent
5988 and can be arbitrarily reordered. The optimization is only
5989 profitable on multiprocessor machines, for loops that are CPU-intensive,
5990 rather than constrained e.g.@: by memory bandwidth. This option
5991 implies @option{-pthread}, and thus is only supported on targets
5992 that have support for @option{-pthread}.
5996 Perform scalar replacement of aggregates. This pass replaces structure
5997 references with scalars to prevent committing structures to memory too
5998 early. This flag is enabled by default at @option{-O} and higher.
6000 @item -ftree-copyrename
6001 @opindex ftree-copyrename
6002 Perform copy renaming on trees. This pass attempts to rename compiler
6003 temporaries to other variables at copy locations, usually resulting in
6004 variable names which more closely resemble the original variables. This flag
6005 is enabled by default at @option{-O} and higher.
6009 Perform temporary expression replacement during the SSA->normal phase. Single
6010 use/single def temporaries are replaced at their use location with their
6011 defining expression. This results in non-GIMPLE code, but gives the expanders
6012 much more complex trees to work on resulting in better RTL generation. This is
6013 enabled by default at @option{-O} and higher.
6015 @item -ftree-vectorize
6016 @opindex ftree-vectorize
6017 Perform loop vectorization on trees. This flag is enabled by default at
6020 @item -ftree-vect-loop-version
6021 @opindex ftree-vect-loop-version
6022 Perform loop versioning when doing loop vectorization on trees. When a loop
6023 appears to be vectorizable except that data alignment or data dependence cannot
6024 be determined at compile time then vectorized and non-vectorized versions of
6025 the loop are generated along with runtime checks for alignment or dependence
6026 to control which version is executed. This option is enabled by default
6027 except at level @option{-Os} where it is disabled.
6029 @item -fvect-cost-model
6030 @opindex fvect-cost-model
6031 Enable cost model for vectorization.
6035 Perform Value Range Propagation on trees. This is similar to the
6036 constant propagation pass, but instead of values, ranges of values are
6037 propagated. This allows the optimizers to remove unnecessary range
6038 checks like array bound checks and null pointer checks. This is
6039 enabled by default at @option{-O2} and higher. Null pointer check
6040 elimination is only done if @option{-fdelete-null-pointer-checks} is
6045 Perform tail duplication to enlarge superblock size. This transformation
6046 simplifies the control flow of the function allowing other optimizations to do
6049 @item -funroll-loops
6050 @opindex funroll-loops
6051 Unroll loops whose number of iterations can be determined at compile
6052 time or upon entry to the loop. @option{-funroll-loops} implies
6053 @option{-frerun-cse-after-loop}. This option makes code larger,
6054 and may or may not make it run faster.
6056 @item -funroll-all-loops
6057 @opindex funroll-all-loops
6058 Unroll all loops, even if their number of iterations is uncertain when
6059 the loop is entered. This usually makes programs run more slowly.
6060 @option{-funroll-all-loops} implies the same options as
6061 @option{-funroll-loops},
6063 @item -fsplit-ivs-in-unroller
6064 @opindex fsplit-ivs-in-unroller
6065 Enables expressing of values of induction variables in later iterations
6066 of the unrolled loop using the value in the first iteration. This breaks
6067 long dependency chains, thus improving efficiency of the scheduling passes.
6069 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6070 same effect. However in cases the loop body is more complicated than
6071 a single basic block, this is not reliable. It also does not work at all
6072 on some of the architectures due to restrictions in the CSE pass.
6074 This optimization is enabled by default.
6076 @item -fvariable-expansion-in-unroller
6077 @opindex fvariable-expansion-in-unroller
6078 With this option, the compiler will create multiple copies of some
6079 local variables when unrolling a loop which can result in superior code.
6081 @item -fpredictive-commoning
6082 @opindex fpredictive-commoning
6083 Perform predictive commoning optimization, i.e., reusing computations
6084 (especially memory loads and stores) performed in previous
6085 iterations of loops.
6087 This option is enabled at level @option{-O3}.
6089 @item -fprefetch-loop-arrays
6090 @opindex fprefetch-loop-arrays
6091 If supported by the target machine, generate instructions to prefetch
6092 memory to improve the performance of loops that access large arrays.
6094 This option may generate better or worse code; results are highly
6095 dependent on the structure of loops within the source code.
6097 Disabled at level @option{-Os}.
6100 @itemx -fno-peephole2
6101 @opindex fno-peephole
6102 @opindex fno-peephole2
6103 Disable any machine-specific peephole optimizations. The difference
6104 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6105 are implemented in the compiler; some targets use one, some use the
6106 other, a few use both.
6108 @option{-fpeephole} is enabled by default.
6109 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6111 @item -fno-guess-branch-probability
6112 @opindex fno-guess-branch-probability
6113 Do not guess branch probabilities using heuristics.
6115 GCC will use heuristics to guess branch probabilities if they are
6116 not provided by profiling feedback (@option{-fprofile-arcs}). These
6117 heuristics are based on the control flow graph. If some branch probabilities
6118 are specified by @samp{__builtin_expect}, then the heuristics will be
6119 used to guess branch probabilities for the rest of the control flow graph,
6120 taking the @samp{__builtin_expect} info into account. The interactions
6121 between the heuristics and @samp{__builtin_expect} can be complex, and in
6122 some cases, it may be useful to disable the heuristics so that the effects
6123 of @samp{__builtin_expect} are easier to understand.
6125 The default is @option{-fguess-branch-probability} at levels
6126 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6128 @item -freorder-blocks
6129 @opindex freorder-blocks
6130 Reorder basic blocks in the compiled function in order to reduce number of
6131 taken branches and improve code locality.
6133 Enabled at levels @option{-O2}, @option{-O3}.
6135 @item -freorder-blocks-and-partition
6136 @opindex freorder-blocks-and-partition
6137 In addition to reordering basic blocks in the compiled function, in order
6138 to reduce number of taken branches, partitions hot and cold basic blocks
6139 into separate sections of the assembly and .o files, to improve
6140 paging and cache locality performance.
6142 This optimization is automatically turned off in the presence of
6143 exception handling, for linkonce sections, for functions with a user-defined
6144 section attribute and on any architecture that does not support named
6147 @item -freorder-functions
6148 @opindex freorder-functions
6149 Reorder functions in the object file in order to
6150 improve code locality. This is implemented by using special
6151 subsections @code{.text.hot} for most frequently executed functions and
6152 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6153 the linker so object file format must support named sections and linker must
6154 place them in a reasonable way.
6156 Also profile feedback must be available in to make this option effective. See
6157 @option{-fprofile-arcs} for details.
6159 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6161 @item -fstrict-aliasing
6162 @opindex fstrict-aliasing
6163 Allows the compiler to assume the strictest aliasing rules applicable to
6164 the language being compiled. For C (and C++), this activates
6165 optimizations based on the type of expressions. In particular, an
6166 object of one type is assumed never to reside at the same address as an
6167 object of a different type, unless the types are almost the same. For
6168 example, an @code{unsigned int} can alias an @code{int}, but not a
6169 @code{void*} or a @code{double}. A character type may alias any other
6172 @anchor{Type-punning}Pay special attention to code like this:
6185 The practice of reading from a different union member than the one most
6186 recently written to (called ``type-punning'') is common. Even with
6187 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6188 is accessed through the union type. So, the code above will work as
6189 expected. @xref{Structures unions enumerations and bit-fields
6190 implementation}. However, this code might not:
6201 Similarly, access by taking the address, casting the resulting pointer
6202 and dereferencing the result has undefined behavior, even if the cast
6203 uses a union type, e.g.:
6207 return ((union a_union *) &d)->i;
6211 The @option{-fstrict-aliasing} option is enabled at levels
6212 @option{-O2}, @option{-O3}, @option{-Os}.
6214 @item -fstrict-overflow
6215 @opindex fstrict-overflow
6216 Allow the compiler to assume strict signed overflow rules, depending
6217 on the language being compiled. For C (and C++) this means that
6218 overflow when doing arithmetic with signed numbers is undefined, which
6219 means that the compiler may assume that it will not happen. This
6220 permits various optimizations. For example, the compiler will assume
6221 that an expression like @code{i + 10 > i} will always be true for
6222 signed @code{i}. This assumption is only valid if signed overflow is
6223 undefined, as the expression is false if @code{i + 10} overflows when
6224 using twos complement arithmetic. When this option is in effect any
6225 attempt to determine whether an operation on signed numbers will
6226 overflow must be written carefully to not actually involve overflow.
6228 This option also allows the compiler to assume strict pointer
6229 semantics: given a pointer to an object, if adding an offset to that
6230 pointer does not produce a pointer to the same object, the addition is
6231 undefined. This permits the compiler to conclude that @code{p + u >
6232 p} is always true for a pointer @code{p} and unsigned integer
6233 @code{u}. This assumption is only valid because pointer wraparound is
6234 undefined, as the expression is false if @code{p + u} overflows using
6235 twos complement arithmetic.
6237 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6238 that integer signed overflow is fully defined: it wraps. When
6239 @option{-fwrapv} is used, there is no difference between
6240 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
6241 integers. With @option{-fwrapv} certain types of overflow are
6242 permitted. For example, if the compiler gets an overflow when doing
6243 arithmetic on constants, the overflowed value can still be used with
6244 @option{-fwrapv}, but not otherwise.
6246 The @option{-fstrict-overflow} option is enabled at levels
6247 @option{-O2}, @option{-O3}, @option{-Os}.
6249 @item -falign-functions
6250 @itemx -falign-functions=@var{n}
6251 @opindex falign-functions
6252 Align the start of functions to the next power-of-two greater than
6253 @var{n}, skipping up to @var{n} bytes. For instance,
6254 @option{-falign-functions=32} aligns functions to the next 32-byte
6255 boundary, but @option{-falign-functions=24} would align to the next
6256 32-byte boundary only if this can be done by skipping 23 bytes or less.
6258 @option{-fno-align-functions} and @option{-falign-functions=1} are
6259 equivalent and mean that functions will not be aligned.
6261 Some assemblers only support this flag when @var{n} is a power of two;
6262 in that case, it is rounded up.
6264 If @var{n} is not specified or is zero, use a machine-dependent default.
6266 Enabled at levels @option{-O2}, @option{-O3}.
6268 @item -falign-labels
6269 @itemx -falign-labels=@var{n}
6270 @opindex falign-labels
6271 Align all branch targets to a power-of-two boundary, skipping up to
6272 @var{n} bytes like @option{-falign-functions}. This option can easily
6273 make code slower, because it must insert dummy operations for when the
6274 branch target is reached in the usual flow of the code.
6276 @option{-fno-align-labels} and @option{-falign-labels=1} are
6277 equivalent and mean that labels will not be aligned.
6279 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
6280 are greater than this value, then their values are used instead.
6282 If @var{n} is not specified or is zero, use a machine-dependent default
6283 which is very likely to be @samp{1}, meaning no alignment.
6285 Enabled at levels @option{-O2}, @option{-O3}.
6288 @itemx -falign-loops=@var{n}
6289 @opindex falign-loops
6290 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
6291 like @option{-falign-functions}. The hope is that the loop will be
6292 executed many times, which will make up for any execution of the dummy
6295 @option{-fno-align-loops} and @option{-falign-loops=1} are
6296 equivalent and mean that loops will not be aligned.
6298 If @var{n} is not specified or is zero, use a machine-dependent default.
6300 Enabled at levels @option{-O2}, @option{-O3}.
6303 @itemx -falign-jumps=@var{n}
6304 @opindex falign-jumps
6305 Align branch targets to a power-of-two boundary, for branch targets
6306 where the targets can only be reached by jumping, skipping up to @var{n}
6307 bytes like @option{-falign-functions}. In this case, no dummy operations
6310 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
6311 equivalent and mean that loops will not be aligned.
6313 If @var{n} is not specified or is zero, use a machine-dependent default.
6315 Enabled at levels @option{-O2}, @option{-O3}.
6317 @item -funit-at-a-time
6318 @opindex funit-at-a-time
6319 This option is left for compatibility reasons. @option{-funit-at-a-time}
6320 has no effect, while @option{-fno-unit-at-a-time} implies
6321 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
6325 @item -fno-toplevel-reorder
6326 @opindex fno-toplevel-reorder
6327 Do not reorder top-level functions, variables, and @code{asm}
6328 statements. Output them in the same order that they appear in the
6329 input file. When this option is used, unreferenced static variables
6330 will not be removed. This option is intended to support existing code
6331 which relies on a particular ordering. For new code, it is better to
6334 Enabled at level @option{-O0}. When disabled explicitly, it also imply
6335 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
6340 Constructs webs as commonly used for register allocation purposes and assign
6341 each web individual pseudo register. This allows the register allocation pass
6342 to operate on pseudos directly, but also strengthens several other optimization
6343 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
6344 however, make debugging impossible, since variables will no longer stay in a
6347 Enabled by default with @option{-funroll-loops}.
6349 @item -fwhole-program
6350 @opindex fwhole-program
6351 Assume that the current compilation unit represents whole program being
6352 compiled. All public functions and variables with the exception of @code{main}
6353 and those merged by attribute @code{externally_visible} become static functions
6354 and in a affect gets more aggressively optimized by interprocedural optimizers.
6355 While this option is equivalent to proper use of @code{static} keyword for
6356 programs consisting of single file, in combination with option
6357 @option{--combine} this flag can be used to compile most of smaller scale C
6358 programs since the functions and variables become local for the whole combined
6359 compilation unit, not for the single source file itself.
6361 This option is not supported for Fortran programs.
6363 @item -fcprop-registers
6364 @opindex fcprop-registers
6365 After register allocation and post-register allocation instruction splitting,
6366 we perform a copy-propagation pass to try to reduce scheduling dependencies
6367 and occasionally eliminate the copy.
6369 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6371 @item -fprofile-dir=@var{path}
6372 @opindex fprofile-dir
6374 Set the directory to search the profile data files in to @var{path}.
6375 This option affects only the profile data generated by
6376 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
6377 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
6378 and its related options.
6379 By default, GCC will use the current directory as @var{path}
6380 thus the profile data file will appear in the same directory as the object file.
6382 @item -fprofile-generate
6383 @itemx -fprofile-generate=@var{path}
6384 @opindex fprofile-generate
6386 Enable options usually used for instrumenting application to produce
6387 profile useful for later recompilation with profile feedback based
6388 optimization. You must use @option{-fprofile-generate} both when
6389 compiling and when linking your program.
6391 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
6393 If @var{path} is specified, GCC will look at the @var{path} to find
6394 the profile feeedback data files. See @option{-fprofile-dir}.
6397 @itemx -fprofile-use=@var{path}
6398 @opindex fprofile-use
6399 Enable profile feedback directed optimizations, and optimizations
6400 generally profitable only with profile feedback available.
6402 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
6403 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
6405 By default, GCC emits an error message if the feedback profiles do not
6406 match the source code. This error can be turned into a warning by using
6407 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
6410 If @var{path} is specified, GCC will look at the @var{path} to find
6411 the profile feedback data files. See @option{-fprofile-dir}.
6414 The following options control compiler behavior regarding floating
6415 point arithmetic. These options trade off between speed and
6416 correctness. All must be specifically enabled.
6420 @opindex ffloat-store
6421 Do not store floating point variables in registers, and inhibit other
6422 options that might change whether a floating point value is taken from a
6425 @cindex floating point precision
6426 This option prevents undesirable excess precision on machines such as
6427 the 68000 where the floating registers (of the 68881) keep more
6428 precision than a @code{double} is supposed to have. Similarly for the
6429 x86 architecture. For most programs, the excess precision does only
6430 good, but a few programs rely on the precise definition of IEEE floating
6431 point. Use @option{-ffloat-store} for such programs, after modifying
6432 them to store all pertinent intermediate computations into variables.
6436 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
6437 @option{-ffinite-math-only}, @option{-fno-rounding-math},
6438 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
6440 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
6442 This option is not turned on by any @option{-O} option since
6443 it can result in incorrect output for programs which depend on
6444 an exact implementation of IEEE or ISO rules/specifications for
6445 math functions. It may, however, yield faster code for programs
6446 that do not require the guarantees of these specifications.
6448 @item -fno-math-errno
6449 @opindex fno-math-errno
6450 Do not set ERRNO after calling math functions that are executed
6451 with a single instruction, e.g., sqrt. A program that relies on
6452 IEEE exceptions for math error handling may want to use this flag
6453 for speed while maintaining IEEE arithmetic compatibility.
6455 This option is not turned on by any @option{-O} option since
6456 it can result in incorrect output for programs which depend on
6457 an exact implementation of IEEE or ISO rules/specifications for
6458 math functions. It may, however, yield faster code for programs
6459 that do not require the guarantees of these specifications.
6461 The default is @option{-fmath-errno}.
6463 On Darwin systems, the math library never sets @code{errno}. There is
6464 therefore no reason for the compiler to consider the possibility that
6465 it might, and @option{-fno-math-errno} is the default.
6467 @item -funsafe-math-optimizations
6468 @opindex funsafe-math-optimizations
6470 Allow optimizations for floating-point arithmetic that (a) assume
6471 that arguments and results are valid and (b) may violate IEEE or
6472 ANSI standards. When used at link-time, it may include libraries
6473 or startup files that change the default FPU control word or other
6474 similar optimizations.
6476 This option is not turned on by any @option{-O} option since
6477 it can result in incorrect output for programs which depend on
6478 an exact implementation of IEEE or ISO rules/specifications for
6479 math functions. It may, however, yield faster code for programs
6480 that do not require the guarantees of these specifications.
6481 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
6482 @option{-fassociative-math} and @option{-freciprocal-math}.
6484 The default is @option{-fno-unsafe-math-optimizations}.
6486 @item -fassociative-math
6487 @opindex fassociative-math
6489 Allow re-association of operands in series of floating-point operations.
6490 This violates the ISO C and C++ language standard by possibly changing
6491 computation result. NOTE: re-ordering may change the sign of zero as
6492 well as ignore NaNs and inhibit or create underflow or overflow (and
6493 thus cannot be used on a code which relies on rounding behavior like
6494 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
6495 and thus may not be used when ordered comparisons are required.
6496 This option requires that both @option{-fno-signed-zeros} and
6497 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
6498 much sense with @option{-frounding-math}.
6500 The default is @option{-fno-associative-math}.
6502 @item -freciprocal-math
6503 @opindex freciprocal-math
6505 Allow the reciprocal of a value to be used instead of dividing by
6506 the value if this enables optimizations. For example @code{x / y}
6507 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
6508 is subject to common subexpression elimination. Note that this loses
6509 precision and increases the number of flops operating on the value.
6511 The default is @option{-fno-reciprocal-math}.
6513 @item -ffinite-math-only
6514 @opindex ffinite-math-only
6515 Allow optimizations for floating-point arithmetic that assume
6516 that arguments and results are not NaNs or +-Infs.
6518 This option is not turned on by any @option{-O} option since
6519 it can result in incorrect output for programs which depend on
6520 an exact implementation of IEEE or ISO rules/specifications for
6521 math functions. It may, however, yield faster code for programs
6522 that do not require the guarantees of these specifications.
6524 The default is @option{-fno-finite-math-only}.
6526 @item -fno-signed-zeros
6527 @opindex fno-signed-zeros
6528 Allow optimizations for floating point arithmetic that ignore the
6529 signedness of zero. IEEE arithmetic specifies the behavior of
6530 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
6531 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
6532 This option implies that the sign of a zero result isn't significant.
6534 The default is @option{-fsigned-zeros}.
6536 @item -fno-trapping-math
6537 @opindex fno-trapping-math
6538 Compile code assuming that floating-point operations cannot generate
6539 user-visible traps. These traps include division by zero, overflow,
6540 underflow, inexact result and invalid operation. This option requires
6541 that @option{-fno-signaling-nans} be in effect. Setting this option may
6542 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
6544 This option should never be turned on by any @option{-O} option since
6545 it can result in incorrect output for programs which depend on
6546 an exact implementation of IEEE or ISO rules/specifications for
6549 The default is @option{-ftrapping-math}.
6551 @item -frounding-math
6552 @opindex frounding-math
6553 Disable transformations and optimizations that assume default floating
6554 point rounding behavior. This is round-to-zero for all floating point
6555 to integer conversions, and round-to-nearest for all other arithmetic
6556 truncations. This option should be specified for programs that change
6557 the FP rounding mode dynamically, or that may be executed with a
6558 non-default rounding mode. This option disables constant folding of
6559 floating point expressions at compile-time (which may be affected by
6560 rounding mode) and arithmetic transformations that are unsafe in the
6561 presence of sign-dependent rounding modes.
6563 The default is @option{-fno-rounding-math}.
6565 This option is experimental and does not currently guarantee to
6566 disable all GCC optimizations that are affected by rounding mode.
6567 Future versions of GCC may provide finer control of this setting
6568 using C99's @code{FENV_ACCESS} pragma. This command line option
6569 will be used to specify the default state for @code{FENV_ACCESS}.
6571 @item -frtl-abstract-sequences
6572 @opindex frtl-abstract-sequences
6573 It is a size optimization method. This option is to find identical
6574 sequences of code, which can be turned into pseudo-procedures and
6575 then replace all occurrences with calls to the newly created
6576 subroutine. It is kind of an opposite of @option{-finline-functions}.
6577 This optimization runs at RTL level.
6579 @item -fsignaling-nans
6580 @opindex fsignaling-nans
6581 Compile code assuming that IEEE signaling NaNs may generate user-visible
6582 traps during floating-point operations. Setting this option disables
6583 optimizations that may change the number of exceptions visible with
6584 signaling NaNs. This option implies @option{-ftrapping-math}.
6586 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
6589 The default is @option{-fno-signaling-nans}.
6591 This option is experimental and does not currently guarantee to
6592 disable all GCC optimizations that affect signaling NaN behavior.
6594 @item -fsingle-precision-constant
6595 @opindex fsingle-precision-constant
6596 Treat floating point constant as single precision constant instead of
6597 implicitly converting it to double precision constant.
6599 @item -fcx-limited-range
6600 @opindex fcx-limited-range
6601 When enabled, this option states that a range reduction step is not
6602 needed when performing complex division. Also, there is no checking
6603 whether the result of a complex multiplication or division is @code{NaN
6604 + I*NaN}, with an attempt to rescue the situation in that case. The
6605 default is @option{-fno-cx-limited-range}, but is enabled by
6606 @option{-ffast-math}.
6608 This option controls the default setting of the ISO C99
6609 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
6612 @item -fcx-fortran-rules
6613 @opindex fcx-fortran-rules
6614 Complex multiplication and division follow Fortran rules. Range
6615 reduction is done as part of complex division, but there is no checking
6616 whether the result of a complex multiplication or division is @code{NaN
6617 + I*NaN}, with an attempt to rescue the situation in that case.
6619 The default is @option{-fno-cx-fortran-rules}.
6623 The following options control optimizations that may improve
6624 performance, but are not enabled by any @option{-O} options. This
6625 section includes experimental options that may produce broken code.
6628 @item -fbranch-probabilities
6629 @opindex fbranch-probabilities
6630 After running a program compiled with @option{-fprofile-arcs}
6631 (@pxref{Debugging Options,, Options for Debugging Your Program or
6632 @command{gcc}}), you can compile it a second time using
6633 @option{-fbranch-probabilities}, to improve optimizations based on
6634 the number of times each branch was taken. When the program
6635 compiled with @option{-fprofile-arcs} exits it saves arc execution
6636 counts to a file called @file{@var{sourcename}.gcda} for each source
6637 file. The information in this data file is very dependent on the
6638 structure of the generated code, so you must use the same source code
6639 and the same optimization options for both compilations.
6641 With @option{-fbranch-probabilities}, GCC puts a
6642 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
6643 These can be used to improve optimization. Currently, they are only
6644 used in one place: in @file{reorg.c}, instead of guessing which path a
6645 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
6646 exactly determine which path is taken more often.
6648 @item -fprofile-values
6649 @opindex fprofile-values
6650 If combined with @option{-fprofile-arcs}, it adds code so that some
6651 data about values of expressions in the program is gathered.
6653 With @option{-fbranch-probabilities}, it reads back the data gathered
6654 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
6655 notes to instructions for their later usage in optimizations.
6657 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
6661 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
6662 a code to gather information about values of expressions.
6664 With @option{-fbranch-probabilities}, it reads back the data gathered
6665 and actually performs the optimizations based on them.
6666 Currently the optimizations include specialization of division operation
6667 using the knowledge about the value of the denominator.
6669 @item -frename-registers
6670 @opindex frename-registers
6671 Attempt to avoid false dependencies in scheduled code by making use
6672 of registers left over after register allocation. This optimization
6673 will most benefit processors with lots of registers. Depending on the
6674 debug information format adopted by the target, however, it can
6675 make debugging impossible, since variables will no longer stay in
6676 a ``home register''.
6678 Enabled by default with @option{-funroll-loops}.
6682 Perform tail duplication to enlarge superblock size. This transformation
6683 simplifies the control flow of the function allowing other optimizations to do
6686 Enabled with @option{-fprofile-use}.
6688 @item -funroll-loops
6689 @opindex funroll-loops
6690 Unroll loops whose number of iterations can be determined at compile time or
6691 upon entry to the loop. @option{-funroll-loops} implies
6692 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
6693 It also turns on complete loop peeling (i.e.@: complete removal of loops with
6694 small constant number of iterations). This option makes code larger, and may
6695 or may not make it run faster.
6697 Enabled with @option{-fprofile-use}.
6699 @item -funroll-all-loops
6700 @opindex funroll-all-loops
6701 Unroll all loops, even if their number of iterations is uncertain when
6702 the loop is entered. This usually makes programs run more slowly.
6703 @option{-funroll-all-loops} implies the same options as
6704 @option{-funroll-loops}.
6707 @opindex fpeel-loops
6708 Peels the loops for that there is enough information that they do not
6709 roll much (from profile feedback). It also turns on complete loop peeling
6710 (i.e.@: complete removal of loops with small constant number of iterations).
6712 Enabled with @option{-fprofile-use}.
6714 @item -fmove-loop-invariants
6715 @opindex fmove-loop-invariants
6716 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
6717 at level @option{-O1}
6719 @item -funswitch-loops
6720 @opindex funswitch-loops
6721 Move branches with loop invariant conditions out of the loop, with duplicates
6722 of the loop on both branches (modified according to result of the condition).
6724 @item -ffunction-sections
6725 @itemx -fdata-sections
6726 @opindex ffunction-sections
6727 @opindex fdata-sections
6728 Place each function or data item into its own section in the output
6729 file if the target supports arbitrary sections. The name of the
6730 function or the name of the data item determines the section's name
6733 Use these options on systems where the linker can perform optimizations
6734 to improve locality of reference in the instruction space. Most systems
6735 using the ELF object format and SPARC processors running Solaris 2 have
6736 linkers with such optimizations. AIX may have these optimizations in
6739 Only use these options when there are significant benefits from doing
6740 so. When you specify these options, the assembler and linker will
6741 create larger object and executable files and will also be slower.
6742 You will not be able to use @code{gprof} on all systems if you
6743 specify this option and you may have problems with debugging if
6744 you specify both this option and @option{-g}.
6746 @item -fbranch-target-load-optimize
6747 @opindex fbranch-target-load-optimize
6748 Perform branch target register load optimization before prologue / epilogue
6750 The use of target registers can typically be exposed only during reload,
6751 thus hoisting loads out of loops and doing inter-block scheduling needs
6752 a separate optimization pass.
6754 @item -fbranch-target-load-optimize2
6755 @opindex fbranch-target-load-optimize2
6756 Perform branch target register load optimization after prologue / epilogue
6759 @item -fbtr-bb-exclusive
6760 @opindex fbtr-bb-exclusive
6761 When performing branch target register load optimization, don't reuse
6762 branch target registers in within any basic block.
6764 @item -fstack-protector
6765 @opindex fstack-protector
6766 Emit extra code to check for buffer overflows, such as stack smashing
6767 attacks. This is done by adding a guard variable to functions with
6768 vulnerable objects. This includes functions that call alloca, and
6769 functions with buffers larger than 8 bytes. The guards are initialized
6770 when a function is entered and then checked when the function exits.
6771 If a guard check fails, an error message is printed and the program exits.
6773 @item -fstack-protector-all
6774 @opindex fstack-protector-all
6775 Like @option{-fstack-protector} except that all functions are protected.
6777 @item -fsection-anchors
6778 @opindex fsection-anchors
6779 Try to reduce the number of symbolic address calculations by using
6780 shared ``anchor'' symbols to address nearby objects. This transformation
6781 can help to reduce the number of GOT entries and GOT accesses on some
6784 For example, the implementation of the following function @code{foo}:
6788 int foo (void) @{ return a + b + c; @}
6791 would usually calculate the addresses of all three variables, but if you
6792 compile it with @option{-fsection-anchors}, it will access the variables
6793 from a common anchor point instead. The effect is similar to the
6794 following pseudocode (which isn't valid C):
6799 register int *xr = &x;
6800 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
6804 Not all targets support this option.
6806 @item --param @var{name}=@var{value}
6808 In some places, GCC uses various constants to control the amount of
6809 optimization that is done. For example, GCC will not inline functions
6810 that contain more that a certain number of instructions. You can
6811 control some of these constants on the command-line using the
6812 @option{--param} option.
6814 The names of specific parameters, and the meaning of the values, are
6815 tied to the internals of the compiler, and are subject to change
6816 without notice in future releases.
6818 In each case, the @var{value} is an integer. The allowable choices for
6819 @var{name} are given in the following table:
6822 @item sra-max-structure-size
6823 The maximum structure size, in bytes, at which the scalar replacement
6824 of aggregates (SRA) optimization will perform block copies. The
6825 default value, 0, implies that GCC will select the most appropriate
6828 @item sra-field-structure-ratio
6829 The threshold ratio (as a percentage) between instantiated fields and
6830 the complete structure size. We say that if the ratio of the number
6831 of bytes in instantiated fields to the number of bytes in the complete
6832 structure exceeds this parameter, then block copies are not used. The
6835 @item struct-reorg-cold-struct-ratio
6836 The threshold ratio (as a percentage) between a structure frequency
6837 and the frequency of the hottest structure in the program. This parameter
6838 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
6839 We say that if the ratio of a structure frequency, calculated by profiling,
6840 to the hottest structure frequency in the program is less than this
6841 parameter, then structure reorganization is not applied to this structure.
6844 @item max-crossjump-edges
6845 The maximum number of incoming edges to consider for crossjumping.
6846 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
6847 the number of edges incoming to each block. Increasing values mean
6848 more aggressive optimization, making the compile time increase with
6849 probably small improvement in executable size.
6851 @item min-crossjump-insns
6852 The minimum number of instructions which must be matched at the end
6853 of two blocks before crossjumping will be performed on them. This
6854 value is ignored in the case where all instructions in the block being
6855 crossjumped from are matched. The default value is 5.
6857 @item max-grow-copy-bb-insns
6858 The maximum code size expansion factor when copying basic blocks
6859 instead of jumping. The expansion is relative to a jump instruction.
6860 The default value is 8.
6862 @item max-goto-duplication-insns
6863 The maximum number of instructions to duplicate to a block that jumps
6864 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
6865 passes, GCC factors computed gotos early in the compilation process,
6866 and unfactors them as late as possible. Only computed jumps at the
6867 end of a basic blocks with no more than max-goto-duplication-insns are
6868 unfactored. The default value is 8.
6870 @item max-delay-slot-insn-search
6871 The maximum number of instructions to consider when looking for an
6872 instruction to fill a delay slot. If more than this arbitrary number of
6873 instructions is searched, the time savings from filling the delay slot
6874 will be minimal so stop searching. Increasing values mean more
6875 aggressive optimization, making the compile time increase with probably
6876 small improvement in executable run time.
6878 @item max-delay-slot-live-search
6879 When trying to fill delay slots, the maximum number of instructions to
6880 consider when searching for a block with valid live register
6881 information. Increasing this arbitrarily chosen value means more
6882 aggressive optimization, increasing the compile time. This parameter
6883 should be removed when the delay slot code is rewritten to maintain the
6886 @item max-gcse-memory
6887 The approximate maximum amount of memory that will be allocated in
6888 order to perform the global common subexpression elimination
6889 optimization. If more memory than specified is required, the
6890 optimization will not be done.
6892 @item max-gcse-passes
6893 The maximum number of passes of GCSE to run. The default is 1.
6895 @item max-pending-list-length
6896 The maximum number of pending dependencies scheduling will allow
6897 before flushing the current state and starting over. Large functions
6898 with few branches or calls can create excessively large lists which
6899 needlessly consume memory and resources.
6901 @item max-inline-insns-single
6902 Several parameters control the tree inliner used in gcc.
6903 This number sets the maximum number of instructions (counted in GCC's
6904 internal representation) in a single function that the tree inliner
6905 will consider for inlining. This only affects functions declared
6906 inline and methods implemented in a class declaration (C++).
6907 The default value is 450.
6909 @item max-inline-insns-auto
6910 When you use @option{-finline-functions} (included in @option{-O3}),
6911 a lot of functions that would otherwise not be considered for inlining
6912 by the compiler will be investigated. To those functions, a different
6913 (more restrictive) limit compared to functions declared inline can
6915 The default value is 90.
6917 @item large-function-insns
6918 The limit specifying really large functions. For functions larger than this
6919 limit after inlining inlining is constrained by
6920 @option{--param large-function-growth}. This parameter is useful primarily
6921 to avoid extreme compilation time caused by non-linear algorithms used by the
6923 The default value is 2700.
6925 @item large-function-growth
6926 Specifies maximal growth of large function caused by inlining in percents.
6927 The default value is 100 which limits large function growth to 2.0 times
6930 @item large-unit-insns
6931 The limit specifying large translation unit. Growth caused by inlining of
6932 units larger than this limit is limited by @option{--param inline-unit-growth}.
6933 For small units this might be too tight (consider unit consisting of function A
6934 that is inline and B that just calls A three time. If B is small relative to
6935 A, the growth of unit is 300\% and yet such inlining is very sane. For very
6936 large units consisting of small inlineable functions however the overall unit
6937 growth limit is needed to avoid exponential explosion of code size. Thus for
6938 smaller units, the size is increased to @option{--param large-unit-insns}
6939 before applying @option{--param inline-unit-growth}. The default is 10000
6941 @item inline-unit-growth
6942 Specifies maximal overall growth of the compilation unit caused by inlining.
6943 The default value is 30 which limits unit growth to 1.3 times the original
6946 @item large-stack-frame
6947 The limit specifying large stack frames. While inlining the algorithm is trying
6948 to not grow past this limit too much. Default value is 256 bytes.
6950 @item large-stack-frame-growth
6951 Specifies maximal growth of large stack frames caused by inlining in percents.
6952 The default value is 1000 which limits large stack frame growth to 11 times
6955 @item max-inline-insns-recursive
6956 @itemx max-inline-insns-recursive-auto
6957 Specifies maximum number of instructions out-of-line copy of self recursive inline
6958 function can grow into by performing recursive inlining.
6960 For functions declared inline @option{--param max-inline-insns-recursive} is
6961 taken into account. For function not declared inline, recursive inlining
6962 happens only when @option{-finline-functions} (included in @option{-O3}) is
6963 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
6964 default value is 450.
6966 @item max-inline-recursive-depth
6967 @itemx max-inline-recursive-depth-auto
6968 Specifies maximum recursion depth used by the recursive inlining.
6970 For functions declared inline @option{--param max-inline-recursive-depth} is
6971 taken into account. For function not declared inline, recursive inlining
6972 happens only when @option{-finline-functions} (included in @option{-O3}) is
6973 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6976 @item min-inline-recursive-probability
6977 Recursive inlining is profitable only for function having deep recursion
6978 in average and can hurt for function having little recursion depth by
6979 increasing the prologue size or complexity of function body to other
6982 When profile feedback is available (see @option{-fprofile-generate}) the actual
6983 recursion depth can be guessed from probability that function will recurse via
6984 given call expression. This parameter limits inlining only to call expression
6985 whose probability exceeds given threshold (in percents). The default value is
6988 @item inline-call-cost
6989 Specify cost of call instruction relative to simple arithmetics operations
6990 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6991 functions and at the same time increases size of leaf function that is believed to
6992 reduce function size by being inlined. In effect it increases amount of
6993 inlining for code having large abstraction penalty (many functions that just
6994 pass the arguments to other functions) and decrease inlining for code with low
6995 abstraction penalty. The default value is 12.
6997 @item min-vect-loop-bound
6998 The minimum number of iterations under which a loop will not get vectorized
6999 when @option{-ftree-vectorize} is used. The number of iterations after
7000 vectorization needs to be greater than the value specified by this option
7001 to allow vectorization. The default value is 0.
7003 @item max-unrolled-insns
7004 The maximum number of instructions that a loop should have if that loop
7005 is unrolled, and if the loop is unrolled, it determines how many times
7006 the loop code is unrolled.
7008 @item max-average-unrolled-insns
7009 The maximum number of instructions biased by probabilities of their execution
7010 that a loop should have if that loop is unrolled, and if the loop is unrolled,
7011 it determines how many times the loop code is unrolled.
7013 @item max-unroll-times
7014 The maximum number of unrollings of a single loop.
7016 @item max-peeled-insns
7017 The maximum number of instructions that a loop should have if that loop
7018 is peeled, and if the loop is peeled, it determines how many times
7019 the loop code is peeled.
7021 @item max-peel-times
7022 The maximum number of peelings of a single loop.
7024 @item max-completely-peeled-insns
7025 The maximum number of insns of a completely peeled loop.
7027 @item max-completely-peel-times
7028 The maximum number of iterations of a loop to be suitable for complete peeling.
7030 @item max-unswitch-insns
7031 The maximum number of insns of an unswitched loop.
7033 @item max-unswitch-level
7034 The maximum number of branches unswitched in a single loop.
7037 The minimum cost of an expensive expression in the loop invariant motion.
7039 @item iv-consider-all-candidates-bound
7040 Bound on number of candidates for induction variables below that
7041 all candidates are considered for each use in induction variable
7042 optimizations. Only the most relevant candidates are considered
7043 if there are more candidates, to avoid quadratic time complexity.
7045 @item iv-max-considered-uses
7046 The induction variable optimizations give up on loops that contain more
7047 induction variable uses.
7049 @item iv-always-prune-cand-set-bound
7050 If number of candidates in the set is smaller than this value,
7051 we always try to remove unnecessary ivs from the set during its
7052 optimization when a new iv is added to the set.
7054 @item scev-max-expr-size
7055 Bound on size of expressions used in the scalar evolutions analyzer.
7056 Large expressions slow the analyzer.
7058 @item omega-max-vars
7059 The maximum number of variables in an Omega constraint system.
7060 The default value is 128.
7062 @item omega-max-geqs
7063 The maximum number of inequalities in an Omega constraint system.
7064 The default value is 256.
7067 The maximum number of equalities in an Omega constraint system.
7068 The default value is 128.
7070 @item omega-max-wild-cards
7071 The maximum number of wildcard variables that the Omega solver will
7072 be able to insert. The default value is 18.
7074 @item omega-hash-table-size
7075 The size of the hash table in the Omega solver. The default value is
7078 @item omega-max-keys
7079 The maximal number of keys used by the Omega solver. The default
7082 @item omega-eliminate-redundant-constraints
7083 When set to 1, use expensive methods to eliminate all redundant
7084 constraints. The default value is 0.
7086 @item vect-max-version-for-alignment-checks
7087 The maximum number of runtime checks that can be performed when
7088 doing loop versioning for alignment in the vectorizer. See option
7089 ftree-vect-loop-version for more information.
7091 @item vect-max-version-for-alias-checks
7092 The maximum number of runtime checks that can be performed when
7093 doing loop versioning for alias in the vectorizer. See option
7094 ftree-vect-loop-version for more information.
7096 @item max-iterations-to-track
7098 The maximum number of iterations of a loop the brute force algorithm
7099 for analysis of # of iterations of the loop tries to evaluate.
7101 @item hot-bb-count-fraction
7102 Select fraction of the maximal count of repetitions of basic block in program
7103 given basic block needs to have to be considered hot.
7105 @item hot-bb-frequency-fraction
7106 Select fraction of the maximal frequency of executions of basic block in
7107 function given basic block needs to have to be considered hot
7109 @item max-predicted-iterations
7110 The maximum number of loop iterations we predict statically. This is useful
7111 in cases where function contain single loop with known bound and other loop
7112 with unknown. We predict the known number of iterations correctly, while
7113 the unknown number of iterations average to roughly 10. This means that the
7114 loop without bounds would appear artificially cold relative to the other one.
7116 @item align-threshold
7118 Select fraction of the maximal frequency of executions of basic block in
7119 function given basic block will get aligned.
7121 @item align-loop-iterations
7123 A loop expected to iterate at lest the selected number of iterations will get
7126 @item tracer-dynamic-coverage
7127 @itemx tracer-dynamic-coverage-feedback
7129 This value is used to limit superblock formation once the given percentage of
7130 executed instructions is covered. This limits unnecessary code size
7133 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7134 feedback is available. The real profiles (as opposed to statically estimated
7135 ones) are much less balanced allowing the threshold to be larger value.
7137 @item tracer-max-code-growth
7138 Stop tail duplication once code growth has reached given percentage. This is
7139 rather hokey argument, as most of the duplicates will be eliminated later in
7140 cross jumping, so it may be set to much higher values than is the desired code
7143 @item tracer-min-branch-ratio
7145 Stop reverse growth when the reverse probability of best edge is less than this
7146 threshold (in percent).
7148 @item tracer-min-branch-ratio
7149 @itemx tracer-min-branch-ratio-feedback
7151 Stop forward growth if the best edge do have probability lower than this
7154 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7155 compilation for profile feedback and one for compilation without. The value
7156 for compilation with profile feedback needs to be more conservative (higher) in
7157 order to make tracer effective.
7159 @item max-cse-path-length
7161 Maximum number of basic blocks on path that cse considers. The default is 10.
7164 The maximum instructions CSE process before flushing. The default is 1000.
7166 @item max-aliased-vops
7168 Maximum number of virtual operands per function allowed to represent
7169 aliases before triggering the alias partitioning heuristic. Alias
7170 partitioning reduces compile times and memory consumption needed for
7171 aliasing at the expense of precision loss in alias information. The
7172 default value for this parameter is 100 for -O1, 500 for -O2 and 1000
7175 Notice that if a function contains more memory statements than the
7176 value of this parameter, it is not really possible to achieve this
7177 reduction. In this case, the compiler will use the number of memory
7178 statements as the value for @option{max-aliased-vops}.
7180 @item avg-aliased-vops
7182 Average number of virtual operands per statement allowed to represent
7183 aliases before triggering the alias partitioning heuristic. This
7184 works in conjunction with @option{max-aliased-vops}. If a function
7185 contains more than @option{max-aliased-vops} virtual operators, then
7186 memory symbols will be grouped into memory partitions until either the
7187 total number of virtual operators is below @option{max-aliased-vops}
7188 or the average number of virtual operators per memory statement is
7189 below @option{avg-aliased-vops}. The default value for this parameter
7190 is 1 for -O1 and -O2, and 3 for -O3.
7192 @item ggc-min-expand
7194 GCC uses a garbage collector to manage its own memory allocation. This
7195 parameter specifies the minimum percentage by which the garbage
7196 collector's heap should be allowed to expand between collections.
7197 Tuning this may improve compilation speed; it has no effect on code
7200 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7201 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7202 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7203 GCC is not able to calculate RAM on a particular platform, the lower
7204 bound of 30% is used. Setting this parameter and
7205 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7206 every opportunity. This is extremely slow, but can be useful for
7209 @item ggc-min-heapsize
7211 Minimum size of the garbage collector's heap before it begins bothering
7212 to collect garbage. The first collection occurs after the heap expands
7213 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7214 tuning this may improve compilation speed, and has no effect on code
7217 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7218 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7219 with a lower bound of 4096 (four megabytes) and an upper bound of
7220 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7221 particular platform, the lower bound is used. Setting this parameter
7222 very large effectively disables garbage collection. Setting this
7223 parameter and @option{ggc-min-expand} to zero causes a full collection
7224 to occur at every opportunity.
7226 @item max-reload-search-insns
7227 The maximum number of instruction reload should look backward for equivalent
7228 register. Increasing values mean more aggressive optimization, making the
7229 compile time increase with probably slightly better performance. The default
7232 @item max-cselib-memory-locations
7233 The maximum number of memory locations cselib should take into account.
7234 Increasing values mean more aggressive optimization, making the compile time
7235 increase with probably slightly better performance. The default value is 500.
7237 @item reorder-blocks-duplicate
7238 @itemx reorder-blocks-duplicate-feedback
7240 Used by basic block reordering pass to decide whether to use unconditional
7241 branch or duplicate the code on its destination. Code is duplicated when its
7242 estimated size is smaller than this value multiplied by the estimated size of
7243 unconditional jump in the hot spots of the program.
7245 The @option{reorder-block-duplicate-feedback} is used only when profile
7246 feedback is available and may be set to higher values than
7247 @option{reorder-block-duplicate} since information about the hot spots is more
7250 @item max-sched-ready-insns
7251 The maximum number of instructions ready to be issued the scheduler should
7252 consider at any given time during the first scheduling pass. Increasing
7253 values mean more thorough searches, making the compilation time increase
7254 with probably little benefit. The default value is 100.
7256 @item max-sched-region-blocks
7257 The maximum number of blocks in a region to be considered for
7258 interblock scheduling. The default value is 10.
7260 @item max-sched-region-insns
7261 The maximum number of insns in a region to be considered for
7262 interblock scheduling. The default value is 100.
7265 The minimum probability (in percents) of reaching a source block
7266 for interblock speculative scheduling. The default value is 40.
7268 @item max-sched-extend-regions-iters
7269 The maximum number of iterations through CFG to extend regions.
7270 0 - disable region extension,
7271 N - do at most N iterations.
7272 The default value is 0.
7274 @item max-sched-insn-conflict-delay
7275 The maximum conflict delay for an insn to be considered for speculative motion.
7276 The default value is 3.
7278 @item sched-spec-prob-cutoff
7279 The minimal probability of speculation success (in percents), so that
7280 speculative insn will be scheduled.
7281 The default value is 40.
7283 @item max-last-value-rtl
7285 The maximum size measured as number of RTLs that can be recorded in an expression
7286 in combiner for a pseudo register as last known value of that register. The default
7289 @item integer-share-limit
7290 Small integer constants can use a shared data structure, reducing the
7291 compiler's memory usage and increasing its speed. This sets the maximum
7292 value of a shared integer constant. The default value is 256.
7294 @item min-virtual-mappings
7295 Specifies the minimum number of virtual mappings in the incremental
7296 SSA updater that should be registered to trigger the virtual mappings
7297 heuristic defined by virtual-mappings-ratio. The default value is
7300 @item virtual-mappings-ratio
7301 If the number of virtual mappings is virtual-mappings-ratio bigger
7302 than the number of virtual symbols to be updated, then the incremental
7303 SSA updater switches to a full update for those symbols. The default
7306 @item ssp-buffer-size
7307 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
7308 protection when @option{-fstack-protection} is used.
7310 @item max-jump-thread-duplication-stmts
7311 Maximum number of statements allowed in a block that needs to be
7312 duplicated when threading jumps.
7314 @item max-fields-for-field-sensitive
7315 Maximum number of fields in a structure we will treat in
7316 a field sensitive manner during pointer analysis. The default is zero
7317 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
7319 @item prefetch-latency
7320 Estimate on average number of instructions that are executed before
7321 prefetch finishes. The distance we prefetch ahead is proportional
7322 to this constant. Increasing this number may also lead to less
7323 streams being prefetched (see @option{simultaneous-prefetches}).
7325 @item simultaneous-prefetches
7326 Maximum number of prefetches that can run at the same time.
7328 @item l1-cache-line-size
7329 The size of cache line in L1 cache, in bytes.
7332 The size of L1 cache, in kilobytes.
7335 The size of L2 cache, in kilobytes.
7337 @item use-canonical-types
7338 Whether the compiler should use the ``canonical'' type system. By
7339 default, this should always be 1, which uses a more efficient internal
7340 mechanism for comparing types in C++ and Objective-C++. However, if
7341 bugs in the canonical type system are causing compilation failures,
7342 set this value to 0 to disable canonical types.
7344 @item switch-conversion-max-branch-ratio
7345 Switch initialization conversion will refuse to create arrays that are
7346 bigger than @option{switch-conversion-max-branch-ratio} times the number of
7347 branches in the switch.
7349 @item max-partial-antic-length
7350 Maximum length of the partial antic set computed during the tree
7351 partial redundancy elimination optimization (@option{-ftree-pre}) when
7352 optimizing at @option{-O3} and above. For some sorts of source code
7353 the enhanced partial redundancy elimination optimization can run away,
7354 consuming all of the memory available on the host machine. This
7355 parameter sets a limit on the length of the sets that are computed,
7356 which prevents the runaway behaviour. Setting a value of 0 for
7357 this paramter will allow an unlimited set length.
7359 @item sccvn-max-scc-size
7360 Maximum size of a strongly connected component (SCC) during SCCVN
7361 processing. If this limit is hit, SCCVN processing for the whole
7362 function will not be done and optimizations depending on it will
7363 be disabled. The default maximum SCC size is 10000.
7368 @node Preprocessor Options
7369 @section Options Controlling the Preprocessor
7370 @cindex preprocessor options
7371 @cindex options, preprocessor
7373 These options control the C preprocessor, which is run on each C source
7374 file before actual compilation.
7376 If you use the @option{-E} option, nothing is done except preprocessing.
7377 Some of these options make sense only together with @option{-E} because
7378 they cause the preprocessor output to be unsuitable for actual
7383 You can use @option{-Wp,@var{option}} to bypass the compiler driver
7384 and pass @var{option} directly through to the preprocessor. If
7385 @var{option} contains commas, it is split into multiple options at the
7386 commas. However, many options are modified, translated or interpreted
7387 by the compiler driver before being passed to the preprocessor, and
7388 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
7389 interface is undocumented and subject to change, so whenever possible
7390 you should avoid using @option{-Wp} and let the driver handle the
7393 @item -Xpreprocessor @var{option}
7394 @opindex preprocessor
7395 Pass @var{option} as an option to the preprocessor. You can use this to
7396 supply system-specific preprocessor options which GCC does not know how to
7399 If you want to pass an option that takes an argument, you must use
7400 @option{-Xpreprocessor} twice, once for the option and once for the argument.
7403 @include cppopts.texi
7405 @node Assembler Options
7406 @section Passing Options to the Assembler
7408 @c prevent bad page break with this line
7409 You can pass options to the assembler.
7412 @item -Wa,@var{option}
7414 Pass @var{option} as an option to the assembler. If @var{option}
7415 contains commas, it is split into multiple options at the commas.
7417 @item -Xassembler @var{option}
7419 Pass @var{option} as an option to the assembler. You can use this to
7420 supply system-specific assembler options which GCC does not know how to
7423 If you want to pass an option that takes an argument, you must use
7424 @option{-Xassembler} twice, once for the option and once for the argument.
7429 @section Options for Linking
7430 @cindex link options
7431 @cindex options, linking
7433 These options come into play when the compiler links object files into
7434 an executable output file. They are meaningless if the compiler is
7435 not doing a link step.
7439 @item @var{object-file-name}
7440 A file name that does not end in a special recognized suffix is
7441 considered to name an object file or library. (Object files are
7442 distinguished from libraries by the linker according to the file
7443 contents.) If linking is done, these object files are used as input
7452 If any of these options is used, then the linker is not run, and
7453 object file names should not be used as arguments. @xref{Overall
7457 @item -l@var{library}
7458 @itemx -l @var{library}
7460 Search the library named @var{library} when linking. (The second
7461 alternative with the library as a separate argument is only for
7462 POSIX compliance and is not recommended.)
7464 It makes a difference where in the command you write this option; the
7465 linker searches and processes libraries and object files in the order they
7466 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
7467 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
7468 to functions in @samp{z}, those functions may not be loaded.
7470 The linker searches a standard list of directories for the library,
7471 which is actually a file named @file{lib@var{library}.a}. The linker
7472 then uses this file as if it had been specified precisely by name.
7474 The directories searched include several standard system directories
7475 plus any that you specify with @option{-L}.
7477 Normally the files found this way are library files---archive files
7478 whose members are object files. The linker handles an archive file by
7479 scanning through it for members which define symbols that have so far
7480 been referenced but not defined. But if the file that is found is an
7481 ordinary object file, it is linked in the usual fashion. The only
7482 difference between using an @option{-l} option and specifying a file name
7483 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
7484 and searches several directories.
7488 You need this special case of the @option{-l} option in order to
7489 link an Objective-C or Objective-C++ program.
7492 @opindex nostartfiles
7493 Do not use the standard system startup files when linking.
7494 The standard system libraries are used normally, unless @option{-nostdlib}
7495 or @option{-nodefaultlibs} is used.
7497 @item -nodefaultlibs
7498 @opindex nodefaultlibs
7499 Do not use the standard system libraries when linking.
7500 Only the libraries you specify will be passed to the linker.
7501 The standard startup files are used normally, unless @option{-nostartfiles}
7502 is used. The compiler may generate calls to @code{memcmp},
7503 @code{memset}, @code{memcpy} and @code{memmove}.
7504 These entries are usually resolved by entries in
7505 libc. These entry points should be supplied through some other
7506 mechanism when this option is specified.
7510 Do not use the standard system startup files or libraries when linking.
7511 No startup files and only the libraries you specify will be passed to
7512 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
7513 @code{memcpy} and @code{memmove}.
7514 These entries are usually resolved by entries in
7515 libc. These entry points should be supplied through some other
7516 mechanism when this option is specified.
7518 @cindex @option{-lgcc}, use with @option{-nostdlib}
7519 @cindex @option{-nostdlib} and unresolved references
7520 @cindex unresolved references and @option{-nostdlib}
7521 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
7522 @cindex @option{-nodefaultlibs} and unresolved references
7523 @cindex unresolved references and @option{-nodefaultlibs}
7524 One of the standard libraries bypassed by @option{-nostdlib} and
7525 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
7526 that GCC uses to overcome shortcomings of particular machines, or special
7527 needs for some languages.
7528 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
7529 Collection (GCC) Internals},
7530 for more discussion of @file{libgcc.a}.)
7531 In most cases, you need @file{libgcc.a} even when you want to avoid
7532 other standard libraries. In other words, when you specify @option{-nostdlib}
7533 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
7534 This ensures that you have no unresolved references to internal GCC
7535 library subroutines. (For example, @samp{__main}, used to ensure C++
7536 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
7537 GNU Compiler Collection (GCC) Internals}.)
7541 Produce a position independent executable on targets which support it.
7542 For predictable results, you must also specify the same set of options
7543 that were used to generate code (@option{-fpie}, @option{-fPIE},
7544 or model suboptions) when you specify this option.
7548 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
7549 that support it. This instructs the linker to add all symbols, not
7550 only used ones, to the dynamic symbol table. This option is needed
7551 for some uses of @code{dlopen} or to allow obtaining backtraces
7552 from within a program.
7556 Remove all symbol table and relocation information from the executable.
7560 On systems that support dynamic linking, this prevents linking with the shared
7561 libraries. On other systems, this option has no effect.
7565 Produce a shared object which can then be linked with other objects to
7566 form an executable. Not all systems support this option. For predictable
7567 results, you must also specify the same set of options that were used to
7568 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
7569 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
7570 needs to build supplementary stub code for constructors to work. On
7571 multi-libbed systems, @samp{gcc -shared} must select the correct support
7572 libraries to link against. Failing to supply the correct flags may lead
7573 to subtle defects. Supplying them in cases where they are not necessary
7576 @item -shared-libgcc
7577 @itemx -static-libgcc
7578 @opindex shared-libgcc
7579 @opindex static-libgcc
7580 On systems that provide @file{libgcc} as a shared library, these options
7581 force the use of either the shared or static version respectively.
7582 If no shared version of @file{libgcc} was built when the compiler was
7583 configured, these options have no effect.
7585 There are several situations in which an application should use the
7586 shared @file{libgcc} instead of the static version. The most common
7587 of these is when the application wishes to throw and catch exceptions
7588 across different shared libraries. In that case, each of the libraries
7589 as well as the application itself should use the shared @file{libgcc}.
7591 Therefore, the G++ and GCJ drivers automatically add
7592 @option{-shared-libgcc} whenever you build a shared library or a main
7593 executable, because C++ and Java programs typically use exceptions, so
7594 this is the right thing to do.
7596 If, instead, you use the GCC driver to create shared libraries, you may
7597 find that they will not always be linked with the shared @file{libgcc}.
7598 If GCC finds, at its configuration time, that you have a non-GNU linker
7599 or a GNU linker that does not support option @option{--eh-frame-hdr},
7600 it will link the shared version of @file{libgcc} into shared libraries
7601 by default. Otherwise, it will take advantage of the linker and optimize
7602 away the linking with the shared version of @file{libgcc}, linking with
7603 the static version of libgcc by default. This allows exceptions to
7604 propagate through such shared libraries, without incurring relocation
7605 costs at library load time.
7607 However, if a library or main executable is supposed to throw or catch
7608 exceptions, you must link it using the G++ or GCJ driver, as appropriate
7609 for the languages used in the program, or using the option
7610 @option{-shared-libgcc}, such that it is linked with the shared
7615 Bind references to global symbols when building a shared object. Warn
7616 about any unresolved references (unless overridden by the link editor
7617 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
7620 @item -Xlinker @var{option}
7622 Pass @var{option} as an option to the linker. You can use this to
7623 supply system-specific linker options which GCC does not know how to
7626 If you want to pass an option that takes an argument, you must use
7627 @option{-Xlinker} twice, once for the option and once for the argument.
7628 For example, to pass @option{-assert definitions}, you must write
7629 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
7630 @option{-Xlinker "-assert definitions"}, because this passes the entire
7631 string as a single argument, which is not what the linker expects.
7633 @item -Wl,@var{option}
7635 Pass @var{option} as an option to the linker. If @var{option} contains
7636 commas, it is split into multiple options at the commas.
7638 @item -u @var{symbol}
7640 Pretend the symbol @var{symbol} is undefined, to force linking of
7641 library modules to define it. You can use @option{-u} multiple times with
7642 different symbols to force loading of additional library modules.
7645 @node Directory Options
7646 @section Options for Directory Search
7647 @cindex directory options
7648 @cindex options, directory search
7651 These options specify directories to search for header files, for
7652 libraries and for parts of the compiler:
7657 Add the directory @var{dir} to the head of the list of directories to be
7658 searched for header files. This can be used to override a system header
7659 file, substituting your own version, since these directories are
7660 searched before the system header file directories. However, you should
7661 not use this option to add directories that contain vendor-supplied
7662 system header files (use @option{-isystem} for that). If you use more than
7663 one @option{-I} option, the directories are scanned in left-to-right
7664 order; the standard system directories come after.
7666 If a standard system include directory, or a directory specified with
7667 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
7668 option will be ignored. The directory will still be searched but as a
7669 system directory at its normal position in the system include chain.
7670 This is to ensure that GCC's procedure to fix buggy system headers and
7671 the ordering for the include_next directive are not inadvertently changed.
7672 If you really need to change the search order for system directories,
7673 use the @option{-nostdinc} and/or @option{-isystem} options.
7675 @item -iquote@var{dir}
7677 Add the directory @var{dir} to the head of the list of directories to
7678 be searched for header files only for the case of @samp{#include
7679 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
7680 otherwise just like @option{-I}.
7684 Add directory @var{dir} to the list of directories to be searched
7687 @item -B@var{prefix}
7689 This option specifies where to find the executables, libraries,
7690 include files, and data files of the compiler itself.
7692 The compiler driver program runs one or more of the subprograms
7693 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
7694 @var{prefix} as a prefix for each program it tries to run, both with and
7695 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
7697 For each subprogram to be run, the compiler driver first tries the
7698 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
7699 was not specified, the driver tries two standard prefixes, which are
7700 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
7701 those results in a file name that is found, the unmodified program
7702 name is searched for using the directories specified in your
7703 @env{PATH} environment variable.
7705 The compiler will check to see if the path provided by the @option{-B}
7706 refers to a directory, and if necessary it will add a directory
7707 separator character at the end of the path.
7709 @option{-B} prefixes that effectively specify directory names also apply
7710 to libraries in the linker, because the compiler translates these
7711 options into @option{-L} options for the linker. They also apply to
7712 includes files in the preprocessor, because the compiler translates these
7713 options into @option{-isystem} options for the preprocessor. In this case,
7714 the compiler appends @samp{include} to the prefix.
7716 The run-time support file @file{libgcc.a} can also be searched for using
7717 the @option{-B} prefix, if needed. If it is not found there, the two
7718 standard prefixes above are tried, and that is all. The file is left
7719 out of the link if it is not found by those means.
7721 Another way to specify a prefix much like the @option{-B} prefix is to use
7722 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
7725 As a special kludge, if the path provided by @option{-B} is
7726 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
7727 9, then it will be replaced by @file{[dir/]include}. This is to help
7728 with boot-strapping the compiler.
7730 @item -specs=@var{file}
7732 Process @var{file} after the compiler reads in the standard @file{specs}
7733 file, in order to override the defaults that the @file{gcc} driver
7734 program uses when determining what switches to pass to @file{cc1},
7735 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
7736 @option{-specs=@var{file}} can be specified on the command line, and they
7737 are processed in order, from left to right.
7739 @item --sysroot=@var{dir}
7741 Use @var{dir} as the logical root directory for headers and libraries.
7742 For example, if the compiler would normally search for headers in
7743 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
7744 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
7746 If you use both this option and the @option{-isysroot} option, then
7747 the @option{--sysroot} option will apply to libraries, but the
7748 @option{-isysroot} option will apply to header files.
7750 The GNU linker (beginning with version 2.16) has the necessary support
7751 for this option. If your linker does not support this option, the
7752 header file aspect of @option{--sysroot} will still work, but the
7753 library aspect will not.
7757 This option has been deprecated. Please use @option{-iquote} instead for
7758 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
7759 Any directories you specify with @option{-I} options before the @option{-I-}
7760 option are searched only for the case of @samp{#include "@var{file}"};
7761 they are not searched for @samp{#include <@var{file}>}.
7763 If additional directories are specified with @option{-I} options after
7764 the @option{-I-}, these directories are searched for all @samp{#include}
7765 directives. (Ordinarily @emph{all} @option{-I} directories are used
7768 In addition, the @option{-I-} option inhibits the use of the current
7769 directory (where the current input file came from) as the first search
7770 directory for @samp{#include "@var{file}"}. There is no way to
7771 override this effect of @option{-I-}. With @option{-I.} you can specify
7772 searching the directory which was current when the compiler was
7773 invoked. That is not exactly the same as what the preprocessor does
7774 by default, but it is often satisfactory.
7776 @option{-I-} does not inhibit the use of the standard system directories
7777 for header files. Thus, @option{-I-} and @option{-nostdinc} are
7784 @section Specifying subprocesses and the switches to pass to them
7787 @command{gcc} is a driver program. It performs its job by invoking a
7788 sequence of other programs to do the work of compiling, assembling and
7789 linking. GCC interprets its command-line parameters and uses these to
7790 deduce which programs it should invoke, and which command-line options
7791 it ought to place on their command lines. This behavior is controlled
7792 by @dfn{spec strings}. In most cases there is one spec string for each
7793 program that GCC can invoke, but a few programs have multiple spec
7794 strings to control their behavior. The spec strings built into GCC can
7795 be overridden by using the @option{-specs=} command-line switch to specify
7798 @dfn{Spec files} are plaintext files that are used to construct spec
7799 strings. They consist of a sequence of directives separated by blank
7800 lines. The type of directive is determined by the first non-whitespace
7801 character on the line and it can be one of the following:
7804 @item %@var{command}
7805 Issues a @var{command} to the spec file processor. The commands that can
7809 @item %include <@var{file}>
7811 Search for @var{file} and insert its text at the current point in the
7814 @item %include_noerr <@var{file}>
7815 @cindex %include_noerr
7816 Just like @samp{%include}, but do not generate an error message if the include
7817 file cannot be found.
7819 @item %rename @var{old_name} @var{new_name}
7821 Rename the spec string @var{old_name} to @var{new_name}.
7825 @item *[@var{spec_name}]:
7826 This tells the compiler to create, override or delete the named spec
7827 string. All lines after this directive up to the next directive or
7828 blank line are considered to be the text for the spec string. If this
7829 results in an empty string then the spec will be deleted. (Or, if the
7830 spec did not exist, then nothing will happened.) Otherwise, if the spec
7831 does not currently exist a new spec will be created. If the spec does
7832 exist then its contents will be overridden by the text of this
7833 directive, unless the first character of that text is the @samp{+}
7834 character, in which case the text will be appended to the spec.
7836 @item [@var{suffix}]:
7837 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
7838 and up to the next directive or blank line are considered to make up the
7839 spec string for the indicated suffix. When the compiler encounters an
7840 input file with the named suffix, it will processes the spec string in
7841 order to work out how to compile that file. For example:
7848 This says that any input file whose name ends in @samp{.ZZ} should be
7849 passed to the program @samp{z-compile}, which should be invoked with the
7850 command-line switch @option{-input} and with the result of performing the
7851 @samp{%i} substitution. (See below.)
7853 As an alternative to providing a spec string, the text that follows a
7854 suffix directive can be one of the following:
7857 @item @@@var{language}
7858 This says that the suffix is an alias for a known @var{language}. This is
7859 similar to using the @option{-x} command-line switch to GCC to specify a
7860 language explicitly. For example:
7867 Says that .ZZ files are, in fact, C++ source files.
7870 This causes an error messages saying:
7873 @var{name} compiler not installed on this system.
7877 GCC already has an extensive list of suffixes built into it.
7878 This directive will add an entry to the end of the list of suffixes, but
7879 since the list is searched from the end backwards, it is effectively
7880 possible to override earlier entries using this technique.
7884 GCC has the following spec strings built into it. Spec files can
7885 override these strings or create their own. Note that individual
7886 targets can also add their own spec strings to this list.
7889 asm Options to pass to the assembler
7890 asm_final Options to pass to the assembler post-processor
7891 cpp Options to pass to the C preprocessor
7892 cc1 Options to pass to the C compiler
7893 cc1plus Options to pass to the C++ compiler
7894 endfile Object files to include at the end of the link
7895 link Options to pass to the linker
7896 lib Libraries to include on the command line to the linker
7897 libgcc Decides which GCC support library to pass to the linker
7898 linker Sets the name of the linker
7899 predefines Defines to be passed to the C preprocessor
7900 signed_char Defines to pass to CPP to say whether @code{char} is signed
7902 startfile Object files to include at the start of the link
7905 Here is a small example of a spec file:
7911 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
7914 This example renames the spec called @samp{lib} to @samp{old_lib} and
7915 then overrides the previous definition of @samp{lib} with a new one.
7916 The new definition adds in some extra command-line options before
7917 including the text of the old definition.
7919 @dfn{Spec strings} are a list of command-line options to be passed to their
7920 corresponding program. In addition, the spec strings can contain
7921 @samp{%}-prefixed sequences to substitute variable text or to
7922 conditionally insert text into the command line. Using these constructs
7923 it is possible to generate quite complex command lines.
7925 Here is a table of all defined @samp{%}-sequences for spec
7926 strings. Note that spaces are not generated automatically around the
7927 results of expanding these sequences. Therefore you can concatenate them
7928 together or combine them with constant text in a single argument.
7932 Substitute one @samp{%} into the program name or argument.
7935 Substitute the name of the input file being processed.
7938 Substitute the basename of the input file being processed.
7939 This is the substring up to (and not including) the last period
7940 and not including the directory.
7943 This is the same as @samp{%b}, but include the file suffix (text after
7947 Marks the argument containing or following the @samp{%d} as a
7948 temporary file name, so that that file will be deleted if GCC exits
7949 successfully. Unlike @samp{%g}, this contributes no text to the
7952 @item %g@var{suffix}
7953 Substitute a file name that has suffix @var{suffix} and is chosen
7954 once per compilation, and mark the argument in the same way as
7955 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
7956 name is now chosen in a way that is hard to predict even when previously
7957 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
7958 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
7959 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
7960 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
7961 was simply substituted with a file name chosen once per compilation,
7962 without regard to any appended suffix (which was therefore treated
7963 just like ordinary text), making such attacks more likely to succeed.
7965 @item %u@var{suffix}
7966 Like @samp{%g}, but generates a new temporary file name even if
7967 @samp{%u@var{suffix}} was already seen.
7969 @item %U@var{suffix}
7970 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
7971 new one if there is no such last file name. In the absence of any
7972 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
7973 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
7974 would involve the generation of two distinct file names, one
7975 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
7976 simply substituted with a file name chosen for the previous @samp{%u},
7977 without regard to any appended suffix.
7979 @item %j@var{suffix}
7980 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
7981 writable, and if save-temps is off; otherwise, substitute the name
7982 of a temporary file, just like @samp{%u}. This temporary file is not
7983 meant for communication between processes, but rather as a junk
7986 @item %|@var{suffix}
7987 @itemx %m@var{suffix}
7988 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
7989 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
7990 all. These are the two most common ways to instruct a program that it
7991 should read from standard input or write to standard output. If you
7992 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
7993 construct: see for example @file{f/lang-specs.h}.
7995 @item %.@var{SUFFIX}
7996 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
7997 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
7998 terminated by the next space or %.
8001 Marks the argument containing or following the @samp{%w} as the
8002 designated output file of this compilation. This puts the argument
8003 into the sequence of arguments that @samp{%o} will substitute later.
8006 Substitutes the names of all the output files, with spaces
8007 automatically placed around them. You should write spaces
8008 around the @samp{%o} as well or the results are undefined.
8009 @samp{%o} is for use in the specs for running the linker.
8010 Input files whose names have no recognized suffix are not compiled
8011 at all, but they are included among the output files, so they will
8015 Substitutes the suffix for object files. Note that this is
8016 handled specially when it immediately follows @samp{%g, %u, or %U},
8017 because of the need for those to form complete file names. The
8018 handling is such that @samp{%O} is treated exactly as if it had already
8019 been substituted, except that @samp{%g, %u, and %U} do not currently
8020 support additional @var{suffix} characters following @samp{%O} as they would
8021 following, for example, @samp{.o}.
8024 Substitutes the standard macro predefinitions for the
8025 current target machine. Use this when running @code{cpp}.
8028 Like @samp{%p}, but puts @samp{__} before and after the name of each
8029 predefined macro, except for macros that start with @samp{__} or with
8030 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
8034 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
8035 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
8036 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
8037 and @option{-imultilib} as necessary.
8040 Current argument is the name of a library or startup file of some sort.
8041 Search for that file in a standard list of directories and substitute
8042 the full name found.
8045 Print @var{str} as an error message. @var{str} is terminated by a newline.
8046 Use this when inconsistent options are detected.
8049 Substitute the contents of spec string @var{name} at this point.
8052 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
8054 @item %x@{@var{option}@}
8055 Accumulate an option for @samp{%X}.
8058 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
8062 Output the accumulated assembler options specified by @option{-Wa}.
8065 Output the accumulated preprocessor options specified by @option{-Wp}.
8068 Process the @code{asm} spec. This is used to compute the
8069 switches to be passed to the assembler.
8072 Process the @code{asm_final} spec. This is a spec string for
8073 passing switches to an assembler post-processor, if such a program is
8077 Process the @code{link} spec. This is the spec for computing the
8078 command line passed to the linker. Typically it will make use of the
8079 @samp{%L %G %S %D and %E} sequences.
8082 Dump out a @option{-L} option for each directory that GCC believes might
8083 contain startup files. If the target supports multilibs then the
8084 current multilib directory will be prepended to each of these paths.
8087 Process the @code{lib} spec. This is a spec string for deciding which
8088 libraries should be included on the command line to the linker.
8091 Process the @code{libgcc} spec. This is a spec string for deciding
8092 which GCC support library should be included on the command line to the linker.
8095 Process the @code{startfile} spec. This is a spec for deciding which
8096 object files should be the first ones passed to the linker. Typically
8097 this might be a file named @file{crt0.o}.
8100 Process the @code{endfile} spec. This is a spec string that specifies
8101 the last object files that will be passed to the linker.
8104 Process the @code{cpp} spec. This is used to construct the arguments
8105 to be passed to the C preprocessor.
8108 Process the @code{cc1} spec. This is used to construct the options to be
8109 passed to the actual C compiler (@samp{cc1}).
8112 Process the @code{cc1plus} spec. This is used to construct the options to be
8113 passed to the actual C++ compiler (@samp{cc1plus}).
8116 Substitute the variable part of a matched option. See below.
8117 Note that each comma in the substituted string is replaced by
8121 Remove all occurrences of @code{-S} from the command line. Note---this
8122 command is position dependent. @samp{%} commands in the spec string
8123 before this one will see @code{-S}, @samp{%} commands in the spec string
8124 after this one will not.
8126 @item %:@var{function}(@var{args})
8127 Call the named function @var{function}, passing it @var{args}.
8128 @var{args} is first processed as a nested spec string, then split
8129 into an argument vector in the usual fashion. The function returns
8130 a string which is processed as if it had appeared literally as part
8131 of the current spec.
8133 The following built-in spec functions are provided:
8137 The @code{getenv} spec function takes two arguments: an environment
8138 variable name and a string. If the environment variable is not
8139 defined, a fatal error is issued. Otherwise, the return value is the
8140 value of the environment variable concatenated with the string. For
8141 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8144 %:getenv(TOPDIR /include)
8147 expands to @file{/path/to/top/include}.
8149 @item @code{if-exists}
8150 The @code{if-exists} spec function takes one argument, an absolute
8151 pathname to a file. If the file exists, @code{if-exists} returns the
8152 pathname. Here is a small example of its usage:
8156 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
8159 @item @code{if-exists-else}
8160 The @code{if-exists-else} spec function is similar to the @code{if-exists}
8161 spec function, except that it takes two arguments. The first argument is
8162 an absolute pathname to a file. If the file exists, @code{if-exists-else}
8163 returns the pathname. If it does not exist, it returns the second argument.
8164 This way, @code{if-exists-else} can be used to select one file or another,
8165 based on the existence of the first. Here is a small example of its usage:
8169 crt0%O%s %:if-exists(crti%O%s) \
8170 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
8173 @item @code{replace-outfile}
8174 The @code{replace-outfile} spec function takes two arguments. It looks for the
8175 first argument in the outfiles array and replaces it with the second argument. Here
8176 is a small example of its usage:
8179 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
8182 @item @code{print-asm-header}
8183 The @code{print-asm-header} function takes no arguments and simply
8184 prints a banner like:
8190 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
8193 It is used to separate compiler options from assembler options
8194 in the @option{--target-help} output.
8198 Substitutes the @code{-S} switch, if that switch was given to GCC@.
8199 If that switch was not specified, this substitutes nothing. Note that
8200 the leading dash is omitted when specifying this option, and it is
8201 automatically inserted if the substitution is performed. Thus the spec
8202 string @samp{%@{foo@}} would match the command-line option @option{-foo}
8203 and would output the command line option @option{-foo}.
8205 @item %W@{@code{S}@}
8206 Like %@{@code{S}@} but mark last argument supplied within as a file to be
8209 @item %@{@code{S}*@}
8210 Substitutes all the switches specified to GCC whose names start
8211 with @code{-S}, but which also take an argument. This is used for
8212 switches like @option{-o}, @option{-D}, @option{-I}, etc.
8213 GCC considers @option{-o foo} as being
8214 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
8215 text, including the space. Thus two arguments would be generated.
8217 @item %@{@code{S}*&@code{T}*@}
8218 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
8219 (the order of @code{S} and @code{T} in the spec is not significant).
8220 There can be any number of ampersand-separated variables; for each the
8221 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
8223 @item %@{@code{S}:@code{X}@}
8224 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
8226 @item %@{!@code{S}:@code{X}@}
8227 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
8229 @item %@{@code{S}*:@code{X}@}
8230 Substitutes @code{X} if one or more switches whose names start with
8231 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
8232 once, no matter how many such switches appeared. However, if @code{%*}
8233 appears somewhere in @code{X}, then @code{X} will be substituted once
8234 for each matching switch, with the @code{%*} replaced by the part of
8235 that switch that matched the @code{*}.
8237 @item %@{.@code{S}:@code{X}@}
8238 Substitutes @code{X}, if processing a file with suffix @code{S}.
8240 @item %@{!.@code{S}:@code{X}@}
8241 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
8243 @item %@{,@code{S}:@code{X}@}
8244 Substitutes @code{X}, if processing a file for language @code{S}.
8246 @item %@{!,@code{S}:@code{X}@}
8247 Substitutes @code{X}, if not processing a file for language @code{S}.
8249 @item %@{@code{S}|@code{P}:@code{X}@}
8250 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
8251 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
8252 @code{*} sequences as well, although they have a stronger binding than
8253 the @samp{|}. If @code{%*} appears in @code{X}, all of the
8254 alternatives must be starred, and only the first matching alternative
8257 For example, a spec string like this:
8260 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
8263 will output the following command-line options from the following input
8264 command-line options:
8269 -d fred.c -foo -baz -boggle
8270 -d jim.d -bar -baz -boggle
8273 @item %@{S:X; T:Y; :D@}
8275 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
8276 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
8277 be as many clauses as you need. This may be combined with @code{.},
8278 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
8283 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
8284 construct may contain other nested @samp{%} constructs or spaces, or
8285 even newlines. They are processed as usual, as described above.
8286 Trailing white space in @code{X} is ignored. White space may also
8287 appear anywhere on the left side of the colon in these constructs,
8288 except between @code{.} or @code{*} and the corresponding word.
8290 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
8291 handled specifically in these constructs. If another value of
8292 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
8293 @option{-W} switch is found later in the command line, the earlier
8294 switch value is ignored, except with @{@code{S}*@} where @code{S} is
8295 just one letter, which passes all matching options.
8297 The character @samp{|} at the beginning of the predicate text is used to
8298 indicate that a command should be piped to the following command, but
8299 only if @option{-pipe} is specified.
8301 It is built into GCC which switches take arguments and which do not.
8302 (You might think it would be useful to generalize this to allow each
8303 compiler's spec to say which switches take arguments. But this cannot
8304 be done in a consistent fashion. GCC cannot even decide which input
8305 files have been specified without knowing which switches take arguments,
8306 and it must know which input files to compile in order to tell which
8309 GCC also knows implicitly that arguments starting in @option{-l} are to be
8310 treated as compiler output files, and passed to the linker in their
8311 proper position among the other output files.
8313 @c man begin OPTIONS
8315 @node Target Options
8316 @section Specifying Target Machine and Compiler Version
8317 @cindex target options
8318 @cindex cross compiling
8319 @cindex specifying machine version
8320 @cindex specifying compiler version and target machine
8321 @cindex compiler version, specifying
8322 @cindex target machine, specifying
8324 The usual way to run GCC is to run the executable called @file{gcc}, or
8325 @file{<machine>-gcc} when cross-compiling, or
8326 @file{<machine>-gcc-<version>} to run a version other than the one that
8327 was installed last. Sometimes this is inconvenient, so GCC provides
8328 options that will switch to another cross-compiler or version.
8331 @item -b @var{machine}
8333 The argument @var{machine} specifies the target machine for compilation.
8335 The value to use for @var{machine} is the same as was specified as the
8336 machine type when configuring GCC as a cross-compiler. For
8337 example, if a cross-compiler was configured with @samp{configure
8338 arm-elf}, meaning to compile for an arm processor with elf binaries,
8339 then you would specify @option{-b arm-elf} to run that cross compiler.
8340 Because there are other options beginning with @option{-b}, the
8341 configuration must contain a hyphen.
8343 @item -V @var{version}
8345 The argument @var{version} specifies which version of GCC to run.
8346 This is useful when multiple versions are installed. For example,
8347 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
8350 The @option{-V} and @option{-b} options work by running the
8351 @file{<machine>-gcc-<version>} executable, so there's no real reason to
8352 use them if you can just run that directly.
8354 @node Submodel Options
8355 @section Hardware Models and Configurations
8356 @cindex submodel options
8357 @cindex specifying hardware config
8358 @cindex hardware models and configurations, specifying
8359 @cindex machine dependent options
8361 Earlier we discussed the standard option @option{-b} which chooses among
8362 different installed compilers for completely different target
8363 machines, such as VAX vs.@: 68000 vs.@: 80386.
8365 In addition, each of these target machine types can have its own
8366 special options, starting with @samp{-m}, to choose among various
8367 hardware models or configurations---for example, 68010 vs 68020,
8368 floating coprocessor or none. A single installed version of the
8369 compiler can compile for any model or configuration, according to the
8372 Some configurations of the compiler also support additional special
8373 options, usually for compatibility with other compilers on the same
8376 @c This list is ordered alphanumerically by subsection name.
8377 @c It should be the same order and spelling as these options are listed
8378 @c in Machine Dependent Options
8384 * Blackfin Options::
8388 * DEC Alpha Options::
8389 * DEC Alpha/VMS Options::
8391 * GNU/Linux Options::
8394 * i386 and x86-64 Options::
8406 * RS/6000 and PowerPC Options::
8407 * S/390 and zSeries Options::
8412 * System V Options::
8417 * Xstormy16 Options::
8423 @subsection ARC Options
8426 These options are defined for ARC implementations:
8431 Compile code for little endian mode. This is the default.
8435 Compile code for big endian mode.
8438 @opindex mmangle-cpu
8439 Prepend the name of the cpu to all public symbol names.
8440 In multiple-processor systems, there are many ARC variants with different
8441 instruction and register set characteristics. This flag prevents code
8442 compiled for one cpu to be linked with code compiled for another.
8443 No facility exists for handling variants that are ``almost identical''.
8444 This is an all or nothing option.
8446 @item -mcpu=@var{cpu}
8448 Compile code for ARC variant @var{cpu}.
8449 Which variants are supported depend on the configuration.
8450 All variants support @option{-mcpu=base}, this is the default.
8452 @item -mtext=@var{text-section}
8453 @itemx -mdata=@var{data-section}
8454 @itemx -mrodata=@var{readonly-data-section}
8458 Put functions, data, and readonly data in @var{text-section},
8459 @var{data-section}, and @var{readonly-data-section} respectively
8460 by default. This can be overridden with the @code{section} attribute.
8461 @xref{Variable Attributes}.
8466 @subsection ARM Options
8469 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
8473 @item -mabi=@var{name}
8475 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
8476 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
8479 @opindex mapcs-frame
8480 Generate a stack frame that is compliant with the ARM Procedure Call
8481 Standard for all functions, even if this is not strictly necessary for
8482 correct execution of the code. Specifying @option{-fomit-frame-pointer}
8483 with this option will cause the stack frames not to be generated for
8484 leaf functions. The default is @option{-mno-apcs-frame}.
8488 This is a synonym for @option{-mapcs-frame}.
8491 @c not currently implemented
8492 @item -mapcs-stack-check
8493 @opindex mapcs-stack-check
8494 Generate code to check the amount of stack space available upon entry to
8495 every function (that actually uses some stack space). If there is
8496 insufficient space available then either the function
8497 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
8498 called, depending upon the amount of stack space required. The run time
8499 system is required to provide these functions. The default is
8500 @option{-mno-apcs-stack-check}, since this produces smaller code.
8502 @c not currently implemented
8504 @opindex mapcs-float
8505 Pass floating point arguments using the float point registers. This is
8506 one of the variants of the APCS@. This option is recommended if the
8507 target hardware has a floating point unit or if a lot of floating point
8508 arithmetic is going to be performed by the code. The default is
8509 @option{-mno-apcs-float}, since integer only code is slightly increased in
8510 size if @option{-mapcs-float} is used.
8512 @c not currently implemented
8513 @item -mapcs-reentrant
8514 @opindex mapcs-reentrant
8515 Generate reentrant, position independent code. The default is
8516 @option{-mno-apcs-reentrant}.
8519 @item -mthumb-interwork
8520 @opindex mthumb-interwork
8521 Generate code which supports calling between the ARM and Thumb
8522 instruction sets. Without this option the two instruction sets cannot
8523 be reliably used inside one program. The default is
8524 @option{-mno-thumb-interwork}, since slightly larger code is generated
8525 when @option{-mthumb-interwork} is specified.
8527 @item -mno-sched-prolog
8528 @opindex mno-sched-prolog
8529 Prevent the reordering of instructions in the function prolog, or the
8530 merging of those instruction with the instructions in the function's
8531 body. This means that all functions will start with a recognizable set
8532 of instructions (or in fact one of a choice from a small set of
8533 different function prologues), and this information can be used to
8534 locate the start if functions inside an executable piece of code. The
8535 default is @option{-msched-prolog}.
8538 @opindex mhard-float
8539 Generate output containing floating point instructions. This is the
8543 @opindex msoft-float
8544 Generate output containing library calls for floating point.
8545 @strong{Warning:} the requisite libraries are not available for all ARM
8546 targets. Normally the facilities of the machine's usual C compiler are
8547 used, but this cannot be done directly in cross-compilation. You must make
8548 your own arrangements to provide suitable library functions for
8551 @option{-msoft-float} changes the calling convention in the output file;
8552 therefore, it is only useful if you compile @emph{all} of a program with
8553 this option. In particular, you need to compile @file{libgcc.a}, the
8554 library that comes with GCC, with @option{-msoft-float} in order for
8557 @item -mfloat-abi=@var{name}
8559 Specifies which ABI to use for floating point values. Permissible values
8560 are: @samp{soft}, @samp{softfp} and @samp{hard}.
8562 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
8563 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
8564 of floating point instructions, but still uses the soft-float calling
8567 @item -mlittle-endian
8568 @opindex mlittle-endian
8569 Generate code for a processor running in little-endian mode. This is
8570 the default for all standard configurations.
8573 @opindex mbig-endian
8574 Generate code for a processor running in big-endian mode; the default is
8575 to compile code for a little-endian processor.
8577 @item -mwords-little-endian
8578 @opindex mwords-little-endian
8579 This option only applies when generating code for big-endian processors.
8580 Generate code for a little-endian word order but a big-endian byte
8581 order. That is, a byte order of the form @samp{32107654}. Note: this
8582 option should only be used if you require compatibility with code for
8583 big-endian ARM processors generated by versions of the compiler prior to
8586 @item -mcpu=@var{name}
8588 This specifies the name of the target ARM processor. GCC uses this name
8589 to determine what kind of instructions it can emit when generating
8590 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
8591 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
8592 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
8593 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
8594 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
8595 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
8596 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
8597 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
8598 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
8599 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
8600 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
8601 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
8602 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
8603 @samp{arm1156t2-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
8604 @samp{cortex-a8}, @samp{cortex-r4}, @samp{cortex-m3}, @samp{cortex-m1},
8605 @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
8607 @item -mtune=@var{name}
8609 This option is very similar to the @option{-mcpu=} option, except that
8610 instead of specifying the actual target processor type, and hence
8611 restricting which instructions can be used, it specifies that GCC should
8612 tune the performance of the code as if the target were of the type
8613 specified in this option, but still choosing the instructions that it
8614 will generate based on the cpu specified by a @option{-mcpu=} option.
8615 For some ARM implementations better performance can be obtained by using
8618 @item -march=@var{name}
8620 This specifies the name of the target ARM architecture. GCC uses this
8621 name to determine what kind of instructions it can emit when generating
8622 assembly code. This option can be used in conjunction with or instead
8623 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
8624 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
8625 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
8626 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
8627 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
8628 @samp{iwmmxt}, @samp{ep9312}.
8630 @item -mfpu=@var{name}
8631 @itemx -mfpe=@var{number}
8632 @itemx -mfp=@var{number}
8636 This specifies what floating point hardware (or hardware emulation) is
8637 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
8638 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
8639 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
8640 with older versions of GCC@.
8642 If @option{-msoft-float} is specified this specifies the format of
8643 floating point values.
8645 @item -mstructure-size-boundary=@var{n}
8646 @opindex mstructure-size-boundary
8647 The size of all structures and unions will be rounded up to a multiple
8648 of the number of bits set by this option. Permissible values are 8, 32
8649 and 64. The default value varies for different toolchains. For the COFF
8650 targeted toolchain the default value is 8. A value of 64 is only allowed
8651 if the underlying ABI supports it.
8653 Specifying the larger number can produce faster, more efficient code, but
8654 can also increase the size of the program. Different values are potentially
8655 incompatible. Code compiled with one value cannot necessarily expect to
8656 work with code or libraries compiled with another value, if they exchange
8657 information using structures or unions.
8659 @item -mabort-on-noreturn
8660 @opindex mabort-on-noreturn
8661 Generate a call to the function @code{abort} at the end of a
8662 @code{noreturn} function. It will be executed if the function tries to
8666 @itemx -mno-long-calls
8667 @opindex mlong-calls
8668 @opindex mno-long-calls
8669 Tells the compiler to perform function calls by first loading the
8670 address of the function into a register and then performing a subroutine
8671 call on this register. This switch is needed if the target function
8672 will lie outside of the 64 megabyte addressing range of the offset based
8673 version of subroutine call instruction.
8675 Even if this switch is enabled, not all function calls will be turned
8676 into long calls. The heuristic is that static functions, functions
8677 which have the @samp{short-call} attribute, functions that are inside
8678 the scope of a @samp{#pragma no_long_calls} directive and functions whose
8679 definitions have already been compiled within the current compilation
8680 unit, will not be turned into long calls. The exception to this rule is
8681 that weak function definitions, functions with the @samp{long-call}
8682 attribute or the @samp{section} attribute, and functions that are within
8683 the scope of a @samp{#pragma long_calls} directive, will always be
8684 turned into long calls.
8686 This feature is not enabled by default. Specifying
8687 @option{-mno-long-calls} will restore the default behavior, as will
8688 placing the function calls within the scope of a @samp{#pragma
8689 long_calls_off} directive. Note these switches have no effect on how
8690 the compiler generates code to handle function calls via function
8693 @item -mnop-fun-dllimport
8694 @opindex mnop-fun-dllimport
8695 Disable support for the @code{dllimport} attribute.
8697 @item -msingle-pic-base
8698 @opindex msingle-pic-base
8699 Treat the register used for PIC addressing as read-only, rather than
8700 loading it in the prologue for each function. The run-time system is
8701 responsible for initializing this register with an appropriate value
8702 before execution begins.
8704 @item -mpic-register=@var{reg}
8705 @opindex mpic-register
8706 Specify the register to be used for PIC addressing. The default is R10
8707 unless stack-checking is enabled, when R9 is used.
8709 @item -mcirrus-fix-invalid-insns
8710 @opindex mcirrus-fix-invalid-insns
8711 @opindex mno-cirrus-fix-invalid-insns
8712 Insert NOPs into the instruction stream to in order to work around
8713 problems with invalid Maverick instruction combinations. This option
8714 is only valid if the @option{-mcpu=ep9312} option has been used to
8715 enable generation of instructions for the Cirrus Maverick floating
8716 point co-processor. This option is not enabled by default, since the
8717 problem is only present in older Maverick implementations. The default
8718 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
8721 @item -mpoke-function-name
8722 @opindex mpoke-function-name
8723 Write the name of each function into the text section, directly
8724 preceding the function prologue. The generated code is similar to this:
8728 .ascii "arm_poke_function_name", 0
8731 .word 0xff000000 + (t1 - t0)
8732 arm_poke_function_name
8734 stmfd sp!, @{fp, ip, lr, pc@}
8738 When performing a stack backtrace, code can inspect the value of
8739 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
8740 location @code{pc - 12} and the top 8 bits are set, then we know that
8741 there is a function name embedded immediately preceding this location
8742 and has length @code{((pc[-3]) & 0xff000000)}.
8746 Generate code for the Thumb instruction set. The default is to
8747 use the 32-bit ARM instruction set.
8748 This option automatically enables either 16-bit Thumb-1 or
8749 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
8750 and @option{-march=@var{name}} options.
8753 @opindex mtpcs-frame
8754 Generate a stack frame that is compliant with the Thumb Procedure Call
8755 Standard for all non-leaf functions. (A leaf function is one that does
8756 not call any other functions.) The default is @option{-mno-tpcs-frame}.
8758 @item -mtpcs-leaf-frame
8759 @opindex mtpcs-leaf-frame
8760 Generate a stack frame that is compliant with the Thumb Procedure Call
8761 Standard for all leaf functions. (A leaf function is one that does
8762 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
8764 @item -mcallee-super-interworking
8765 @opindex mcallee-super-interworking
8766 Gives all externally visible functions in the file being compiled an ARM
8767 instruction set header which switches to Thumb mode before executing the
8768 rest of the function. This allows these functions to be called from
8769 non-interworking code.
8771 @item -mcaller-super-interworking
8772 @opindex mcaller-super-interworking
8773 Allows calls via function pointers (including virtual functions) to
8774 execute correctly regardless of whether the target code has been
8775 compiled for interworking or not. There is a small overhead in the cost
8776 of executing a function pointer if this option is enabled.
8778 @item -mtp=@var{name}
8780 Specify the access model for the thread local storage pointer. The valid
8781 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
8782 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
8783 (supported in the arm6k architecture), and @option{auto}, which uses the
8784 best available method for the selected processor. The default setting is
8790 @subsection AVR Options
8793 These options are defined for AVR implementations:
8796 @item -mmcu=@var{mcu}
8798 Specify ATMEL AVR instruction set or MCU type.
8800 Instruction set avr1 is for the minimal AVR core, not supported by the C
8801 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8802 attiny11, attiny12, attiny15, attiny28).
8804 Instruction set avr2 (default) is for the classic AVR core with up to
8805 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8806 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8807 at90c8534, at90s8535).
8809 Instruction set avr3 is for the classic AVR core with up to 128K program
8810 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
8812 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8813 memory space (MCU types: atmega8, atmega83, atmega85).
8815 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8816 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
8817 atmega64, atmega128, at43usb355, at94k).
8821 Output instruction sizes to the asm file.
8823 @item -minit-stack=@var{N}
8824 @opindex minit-stack
8825 Specify the initial stack address, which may be a symbol or numeric value,
8826 @samp{__stack} is the default.
8828 @item -mno-interrupts
8829 @opindex mno-interrupts
8830 Generated code is not compatible with hardware interrupts.
8831 Code size will be smaller.
8833 @item -mcall-prologues
8834 @opindex mcall-prologues
8835 Functions prologues/epilogues expanded as call to appropriate
8836 subroutines. Code size will be smaller.
8838 @item -mno-tablejump
8839 @opindex mno-tablejump
8840 Do not generate tablejump insns which sometimes increase code size.
8843 @opindex mtiny-stack
8844 Change only the low 8 bits of the stack pointer.
8848 Assume int to be 8 bit integer. This affects the sizes of all types: A
8849 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
8850 and long long will be 4 bytes. Please note that this option does not
8851 comply to the C standards, but it will provide you with smaller code
8855 @node Blackfin Options
8856 @subsection Blackfin Options
8857 @cindex Blackfin Options
8860 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
8862 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
8863 can be one of @samp{bf522}, @samp{bf523}, @samp{bf524},
8864 @samp{bf525}, @samp{bf526}, @samp{bf527},
8865 @samp{bf531}, @samp{bf532}, @samp{bf533}, @samp{bf534},
8866 @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
8867 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
8869 The optional @var{sirevision} specifies the silicon revision of the target
8870 Blackfin processor. Any workarounds available for the targeted silicon revision
8871 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
8872 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
8873 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
8874 hexadecimal digits representing the major and minor numbers in the silicon
8875 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
8876 is not defined. If @var{sirevision} is @samp{any}, the
8877 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
8878 If this optional @var{sirevision} is not used, GCC assumes the latest known
8879 silicon revision of the targeted Blackfin processor.
8881 Support for @samp{bf561} is incomplete. For @samp{bf561},
8882 Only the processor macro is defined.
8883 Without this option, @samp{bf532} is used as the processor by default.
8884 The corresponding predefined processor macros for @var{cpu} is to
8885 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
8886 provided by libgloss to be linked in if @option{-msim} is not given.
8890 Specifies that the program will be run on the simulator. This causes
8891 the simulator BSP provided by libgloss to be linked in. This option
8892 has effect only for @samp{bfin-elf} toolchain.
8893 Certain other options, such as @option{-mid-shared-library} and
8894 @option{-mfdpic}, imply @option{-msim}.
8896 @item -momit-leaf-frame-pointer
8897 @opindex momit-leaf-frame-pointer
8898 Don't keep the frame pointer in a register for leaf functions. This
8899 avoids the instructions to save, set up and restore frame pointers and
8900 makes an extra register available in leaf functions. The option
8901 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8902 which might make debugging harder.
8904 @item -mspecld-anomaly
8905 @opindex mspecld-anomaly
8906 When enabled, the compiler will ensure that the generated code does not
8907 contain speculative loads after jump instructions. If this option is used,
8908 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
8910 @item -mno-specld-anomaly
8911 @opindex mno-specld-anomaly
8912 Don't generate extra code to prevent speculative loads from occurring.
8914 @item -mcsync-anomaly
8915 @opindex mcsync-anomaly
8916 When enabled, the compiler will ensure that the generated code does not
8917 contain CSYNC or SSYNC instructions too soon after conditional branches.
8918 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
8920 @item -mno-csync-anomaly
8921 @opindex mno-csync-anomaly
8922 Don't generate extra code to prevent CSYNC or SSYNC instructions from
8923 occurring too soon after a conditional branch.
8927 When enabled, the compiler is free to take advantage of the knowledge that
8928 the entire program fits into the low 64k of memory.
8931 @opindex mno-low-64k
8932 Assume that the program is arbitrarily large. This is the default.
8934 @item -mstack-check-l1
8935 @opindex mstack-check-l1
8936 Do stack checking using information placed into L1 scratchpad memory by the
8939 @item -mid-shared-library
8940 @opindex mid-shared-library
8941 Generate code that supports shared libraries via the library ID method.
8942 This allows for execute in place and shared libraries in an environment
8943 without virtual memory management. This option implies @option{-fPIC}.
8944 With a @samp{bfin-elf} target, this option implies @option{-msim}.
8946 @item -mno-id-shared-library
8947 @opindex mno-id-shared-library
8948 Generate code that doesn't assume ID based shared libraries are being used.
8949 This is the default.
8951 @item -mleaf-id-shared-library
8952 @opindex mleaf-id-shared-library
8953 Generate code that supports shared libraries via the library ID method,
8954 but assumes that this library or executable won't link against any other
8955 ID shared libraries. That allows the compiler to use faster code for jumps
8958 @item -mno-leaf-id-shared-library
8959 @opindex mno-leaf-id-shared-library
8960 Do not assume that the code being compiled won't link against any ID shared
8961 libraries. Slower code will be generated for jump and call insns.
8963 @item -mshared-library-id=n
8964 @opindex mshared-library-id
8965 Specified the identification number of the ID based shared library being
8966 compiled. Specifying a value of 0 will generate more compact code, specifying
8967 other values will force the allocation of that number to the current
8968 library but is no more space or time efficient than omitting this option.
8972 Generate code that allows the data segment to be located in a different
8973 area of memory from the text segment. This allows for execute in place in
8974 an environment without virtual memory management by eliminating relocations
8975 against the text section.
8978 @opindex mno-sep-data
8979 Generate code that assumes that the data segment follows the text segment.
8980 This is the default.
8983 @itemx -mno-long-calls
8984 @opindex mlong-calls
8985 @opindex mno-long-calls
8986 Tells the compiler to perform function calls by first loading the
8987 address of the function into a register and then performing a subroutine
8988 call on this register. This switch is needed if the target function
8989 will lie outside of the 24 bit addressing range of the offset based
8990 version of subroutine call instruction.
8992 This feature is not enabled by default. Specifying
8993 @option{-mno-long-calls} will restore the default behavior. Note these
8994 switches have no effect on how the compiler generates code to handle
8995 function calls via function pointers.
8999 Link with the fast floating-point library. This library relaxes some of
9000 the IEEE floating-point standard's rules for checking inputs against
9001 Not-a-Number (NAN), in the interest of performance.
9004 @opindex minline-plt
9005 Enable inlining of PLT entries in function calls to functions that are
9006 not known to bind locally. It has no effect without @option{-mfdpic}.
9010 Build standalone application for multicore Blackfin processor. Proper
9011 start files and link scripts will be used to support multicore.
9012 This option defines @code{__BFIN_MULTICORE}. It can only be used with
9013 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
9014 @option{-mcorea} or @option{-mcoreb}. If it's used without
9015 @option{-mcorea} or @option{-mcoreb}, single application/dual core
9016 programming model is used. In this model, the main function of Core B
9017 should be named as coreb_main. If it's used with @option{-mcorea} or
9018 @option{-mcoreb}, one application per core programming model is used.
9019 If this option is not used, single core application programming
9024 Build standalone application for Core A of BF561 when using
9025 one application per core programming model. Proper start files
9026 and link scripts will be used to support Core A. This option
9027 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
9031 Build standalone application for Core B of BF561 when using
9032 one application per core programming model. Proper start files
9033 and link scripts will be used to support Core B. This option
9034 defines @code{__BFIN_COREB}. When this option is used, coreb_main
9035 should be used instead of main. It must be used with
9036 @option{-mmulticore}.
9040 Build standalone application for SDRAM. Proper start files and
9041 link scripts will be used to put the application into SDRAM.
9042 Loader should initialize SDRAM before loading the application
9043 into SDRAM. This option defines @code{__BFIN_SDRAM}.
9047 @subsection CRIS Options
9048 @cindex CRIS Options
9050 These options are defined specifically for the CRIS ports.
9053 @item -march=@var{architecture-type}
9054 @itemx -mcpu=@var{architecture-type}
9057 Generate code for the specified architecture. The choices for
9058 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9059 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
9060 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9063 @item -mtune=@var{architecture-type}
9065 Tune to @var{architecture-type} everything applicable about the generated
9066 code, except for the ABI and the set of available instructions. The
9067 choices for @var{architecture-type} are the same as for
9068 @option{-march=@var{architecture-type}}.
9070 @item -mmax-stack-frame=@var{n}
9071 @opindex mmax-stack-frame
9072 Warn when the stack frame of a function exceeds @var{n} bytes.
9078 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9079 @option{-march=v3} and @option{-march=v8} respectively.
9081 @item -mmul-bug-workaround
9082 @itemx -mno-mul-bug-workaround
9083 @opindex mmul-bug-workaround
9084 @opindex mno-mul-bug-workaround
9085 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
9086 models where it applies. This option is active by default.
9090 Enable CRIS-specific verbose debug-related information in the assembly
9091 code. This option also has the effect to turn off the @samp{#NO_APP}
9092 formatted-code indicator to the assembler at the beginning of the
9097 Do not use condition-code results from previous instruction; always emit
9098 compare and test instructions before use of condition codes.
9100 @item -mno-side-effects
9101 @opindex mno-side-effects
9102 Do not emit instructions with side-effects in addressing modes other than
9106 @itemx -mno-stack-align
9108 @itemx -mno-data-align
9109 @itemx -mconst-align
9110 @itemx -mno-const-align
9111 @opindex mstack-align
9112 @opindex mno-stack-align
9113 @opindex mdata-align
9114 @opindex mno-data-align
9115 @opindex mconst-align
9116 @opindex mno-const-align
9117 These options (no-options) arranges (eliminate arrangements) for the
9118 stack-frame, individual data and constants to be aligned for the maximum
9119 single data access size for the chosen CPU model. The default is to
9120 arrange for 32-bit alignment. ABI details such as structure layout are
9121 not affected by these options.
9129 Similar to the stack- data- and const-align options above, these options
9130 arrange for stack-frame, writable data and constants to all be 32-bit,
9131 16-bit or 8-bit aligned. The default is 32-bit alignment.
9133 @item -mno-prologue-epilogue
9134 @itemx -mprologue-epilogue
9135 @opindex mno-prologue-epilogue
9136 @opindex mprologue-epilogue
9137 With @option{-mno-prologue-epilogue}, the normal function prologue and
9138 epilogue that sets up the stack-frame are omitted and no return
9139 instructions or return sequences are generated in the code. Use this
9140 option only together with visual inspection of the compiled code: no
9141 warnings or errors are generated when call-saved registers must be saved,
9142 or storage for local variable needs to be allocated.
9148 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9149 instruction sequences that load addresses for functions from the PLT part
9150 of the GOT rather than (traditional on other architectures) calls to the
9151 PLT@. The default is @option{-mgotplt}.
9155 Legacy no-op option only recognized with the cris-axis-elf and
9156 cris-axis-linux-gnu targets.
9160 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9164 This option, recognized for the cris-axis-elf arranges
9165 to link with input-output functions from a simulator library. Code,
9166 initialized data and zero-initialized data are allocated consecutively.
9170 Like @option{-sim}, but pass linker options to locate initialized data at
9171 0x40000000 and zero-initialized data at 0x80000000.
9175 @subsection CRX Options
9178 These options are defined specifically for the CRX ports.
9184 Enable the use of multiply-accumulate instructions. Disabled by default.
9188 Push instructions will be used to pass outgoing arguments when functions
9189 are called. Enabled by default.
9192 @node Darwin Options
9193 @subsection Darwin Options
9194 @cindex Darwin options
9196 These options are defined for all architectures running the Darwin operating
9199 FSF GCC on Darwin does not create ``fat'' object files; it will create
9200 an object file for the single architecture that it was built to
9201 target. Apple's GCC on Darwin does create ``fat'' files if multiple
9202 @option{-arch} options are used; it does so by running the compiler or
9203 linker multiple times and joining the results together with
9206 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
9207 @samp{i686}) is determined by the flags that specify the ISA
9208 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
9209 @option{-force_cpusubtype_ALL} option can be used to override this.
9211 The Darwin tools vary in their behavior when presented with an ISA
9212 mismatch. The assembler, @file{as}, will only permit instructions to
9213 be used that are valid for the subtype of the file it is generating,
9214 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
9215 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
9216 and print an error if asked to create a shared library with a less
9217 restrictive subtype than its input files (for instance, trying to put
9218 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
9219 for executables, @file{ld}, will quietly give the executable the most
9220 restrictive subtype of any of its input files.
9225 Add the framework directory @var{dir} to the head of the list of
9226 directories to be searched for header files. These directories are
9227 interleaved with those specified by @option{-I} options and are
9228 scanned in a left-to-right order.
9230 A framework directory is a directory with frameworks in it. A
9231 framework is a directory with a @samp{"Headers"} and/or
9232 @samp{"PrivateHeaders"} directory contained directly in it that ends
9233 in @samp{".framework"}. The name of a framework is the name of this
9234 directory excluding the @samp{".framework"}. Headers associated with
9235 the framework are found in one of those two directories, with
9236 @samp{"Headers"} being searched first. A subframework is a framework
9237 directory that is in a framework's @samp{"Frameworks"} directory.
9238 Includes of subframework headers can only appear in a header of a
9239 framework that contains the subframework, or in a sibling subframework
9240 header. Two subframeworks are siblings if they occur in the same
9241 framework. A subframework should not have the same name as a
9242 framework, a warning will be issued if this is violated. Currently a
9243 subframework cannot have subframeworks, in the future, the mechanism
9244 may be extended to support this. The standard frameworks can be found
9245 in @samp{"/System/Library/Frameworks"} and
9246 @samp{"/Library/Frameworks"}. An example include looks like
9247 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
9248 the name of the framework and header.h is found in the
9249 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
9251 @item -iframework@var{dir}
9253 Like @option{-F} except the directory is a treated as a system
9254 directory. The main difference between this @option{-iframework} and
9255 @option{-F} is that with @option{-iframework} the compiler does not
9256 warn about constructs contained within header files found via
9257 @var{dir}. This option is valid only for the C family of languages.
9261 Emit debugging information for symbols that are used. For STABS
9262 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
9263 This is by default ON@.
9267 Emit debugging information for all symbols and types.
9269 @item -mmacosx-version-min=@var{version}
9270 The earliest version of MacOS X that this executable will run on
9271 is @var{version}. Typical values of @var{version} include @code{10.1},
9272 @code{10.2}, and @code{10.3.9}.
9274 If the compiler was built to use the system's headers by default,
9275 then the default for this option is the system version on which the
9276 compiler is running, otherwise the default is to make choices which
9277 are compatible with as many systems and code bases as possible.
9281 Enable kernel development mode. The @option{-mkernel} option sets
9282 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
9283 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
9284 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
9285 applicable. This mode also sets @option{-mno-altivec},
9286 @option{-msoft-float}, @option{-fno-builtin} and
9287 @option{-mlong-branch} for PowerPC targets.
9289 @item -mone-byte-bool
9290 @opindex mone-byte-bool
9291 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
9292 By default @samp{sizeof(bool)} is @samp{4} when compiling for
9293 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
9294 option has no effect on x86.
9296 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
9297 to generate code that is not binary compatible with code generated
9298 without that switch. Using this switch may require recompiling all
9299 other modules in a program, including system libraries. Use this
9300 switch to conform to a non-default data model.
9302 @item -mfix-and-continue
9303 @itemx -ffix-and-continue
9304 @itemx -findirect-data
9305 @opindex mfix-and-continue
9306 @opindex ffix-and-continue
9307 @opindex findirect-data
9308 Generate code suitable for fast turn around development. Needed to
9309 enable gdb to dynamically load @code{.o} files into already running
9310 programs. @option{-findirect-data} and @option{-ffix-and-continue}
9311 are provided for backwards compatibility.
9315 Loads all members of static archive libraries.
9316 See man ld(1) for more information.
9318 @item -arch_errors_fatal
9319 @opindex arch_errors_fatal
9320 Cause the errors having to do with files that have the wrong architecture
9324 @opindex bind_at_load
9325 Causes the output file to be marked such that the dynamic linker will
9326 bind all undefined references when the file is loaded or launched.
9330 Produce a Mach-o bundle format file.
9331 See man ld(1) for more information.
9333 @item -bundle_loader @var{executable}
9334 @opindex bundle_loader
9335 This option specifies the @var{executable} that will be loading the build
9336 output file being linked. See man ld(1) for more information.
9340 When passed this option, GCC will produce a dynamic library instead of
9341 an executable when linking, using the Darwin @file{libtool} command.
9343 @item -force_cpusubtype_ALL
9344 @opindex force_cpusubtype_ALL
9345 This causes GCC's output file to have the @var{ALL} subtype, instead of
9346 one controlled by the @option{-mcpu} or @option{-march} option.
9348 @item -allowable_client @var{client_name}
9350 @itemx -compatibility_version
9351 @itemx -current_version
9353 @itemx -dependency-file
9355 @itemx -dylinker_install_name
9357 @itemx -exported_symbols_list
9359 @itemx -flat_namespace
9360 @itemx -force_flat_namespace
9361 @itemx -headerpad_max_install_names
9364 @itemx -install_name
9365 @itemx -keep_private_externs
9366 @itemx -multi_module
9367 @itemx -multiply_defined
9368 @itemx -multiply_defined_unused
9370 @itemx -no_dead_strip_inits_and_terms
9371 @itemx -nofixprebinding
9374 @itemx -noseglinkedit
9375 @itemx -pagezero_size
9377 @itemx -prebind_all_twolevel_modules
9378 @itemx -private_bundle
9379 @itemx -read_only_relocs
9381 @itemx -sectobjectsymbols
9385 @itemx -sectobjectsymbols
9388 @itemx -segs_read_only_addr
9389 @itemx -segs_read_write_addr
9390 @itemx -seg_addr_table
9391 @itemx -seg_addr_table_filename
9394 @itemx -segs_read_only_addr
9395 @itemx -segs_read_write_addr
9396 @itemx -single_module
9399 @itemx -sub_umbrella
9400 @itemx -twolevel_namespace
9403 @itemx -unexported_symbols_list
9404 @itemx -weak_reference_mismatches
9406 @opindex allowable_client
9407 @opindex client_name
9408 @opindex compatibility_version
9409 @opindex current_version
9411 @opindex dependency-file
9413 @opindex dylinker_install_name
9415 @opindex exported_symbols_list
9417 @opindex flat_namespace
9418 @opindex force_flat_namespace
9419 @opindex headerpad_max_install_names
9422 @opindex install_name
9423 @opindex keep_private_externs
9424 @opindex multi_module
9425 @opindex multiply_defined
9426 @opindex multiply_defined_unused
9428 @opindex no_dead_strip_inits_and_terms
9429 @opindex nofixprebinding
9430 @opindex nomultidefs
9432 @opindex noseglinkedit
9433 @opindex pagezero_size
9435 @opindex prebind_all_twolevel_modules
9436 @opindex private_bundle
9437 @opindex read_only_relocs
9439 @opindex sectobjectsymbols
9443 @opindex sectobjectsymbols
9446 @opindex segs_read_only_addr
9447 @opindex segs_read_write_addr
9448 @opindex seg_addr_table
9449 @opindex seg_addr_table_filename
9450 @opindex seglinkedit
9452 @opindex segs_read_only_addr
9453 @opindex segs_read_write_addr
9454 @opindex single_module
9456 @opindex sub_library
9457 @opindex sub_umbrella
9458 @opindex twolevel_namespace
9461 @opindex unexported_symbols_list
9462 @opindex weak_reference_mismatches
9463 @opindex whatsloaded
9464 These options are passed to the Darwin linker. The Darwin linker man page
9465 describes them in detail.
9468 @node DEC Alpha Options
9469 @subsection DEC Alpha Options
9471 These @samp{-m} options are defined for the DEC Alpha implementations:
9474 @item -mno-soft-float
9476 @opindex mno-soft-float
9477 @opindex msoft-float
9478 Use (do not use) the hardware floating-point instructions for
9479 floating-point operations. When @option{-msoft-float} is specified,
9480 functions in @file{libgcc.a} will be used to perform floating-point
9481 operations. Unless they are replaced by routines that emulate the
9482 floating-point operations, or compiled in such a way as to call such
9483 emulations routines, these routines will issue floating-point
9484 operations. If you are compiling for an Alpha without floating-point
9485 operations, you must ensure that the library is built so as not to call
9488 Note that Alpha implementations without floating-point operations are
9489 required to have floating-point registers.
9494 @opindex mno-fp-regs
9495 Generate code that uses (does not use) the floating-point register set.
9496 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9497 register set is not used, floating point operands are passed in integer
9498 registers as if they were integers and floating-point results are passed
9499 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9500 so any function with a floating-point argument or return value called by code
9501 compiled with @option{-mno-fp-regs} must also be compiled with that
9504 A typical use of this option is building a kernel that does not use,
9505 and hence need not save and restore, any floating-point registers.
9509 The Alpha architecture implements floating-point hardware optimized for
9510 maximum performance. It is mostly compliant with the IEEE floating
9511 point standard. However, for full compliance, software assistance is
9512 required. This option generates code fully IEEE compliant code
9513 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9514 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9515 defined during compilation. The resulting code is less efficient but is
9516 able to correctly support denormalized numbers and exceptional IEEE
9517 values such as not-a-number and plus/minus infinity. Other Alpha
9518 compilers call this option @option{-ieee_with_no_inexact}.
9520 @item -mieee-with-inexact
9521 @opindex mieee-with-inexact
9522 This is like @option{-mieee} except the generated code also maintains
9523 the IEEE @var{inexact-flag}. Turning on this option causes the
9524 generated code to implement fully-compliant IEEE math. In addition to
9525 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9526 macro. On some Alpha implementations the resulting code may execute
9527 significantly slower than the code generated by default. Since there is
9528 very little code that depends on the @var{inexact-flag}, you should
9529 normally not specify this option. Other Alpha compilers call this
9530 option @option{-ieee_with_inexact}.
9532 @item -mfp-trap-mode=@var{trap-mode}
9533 @opindex mfp-trap-mode
9534 This option controls what floating-point related traps are enabled.
9535 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9536 The trap mode can be set to one of four values:
9540 This is the default (normal) setting. The only traps that are enabled
9541 are the ones that cannot be disabled in software (e.g., division by zero
9545 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9549 Like @samp{u}, but the instructions are marked to be safe for software
9550 completion (see Alpha architecture manual for details).
9553 Like @samp{su}, but inexact traps are enabled as well.
9556 @item -mfp-rounding-mode=@var{rounding-mode}
9557 @opindex mfp-rounding-mode
9558 Selects the IEEE rounding mode. Other Alpha compilers call this option
9559 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9564 Normal IEEE rounding mode. Floating point numbers are rounded towards
9565 the nearest machine number or towards the even machine number in case
9569 Round towards minus infinity.
9572 Chopped rounding mode. Floating point numbers are rounded towards zero.
9575 Dynamic rounding mode. A field in the floating point control register
9576 (@var{fpcr}, see Alpha architecture reference manual) controls the
9577 rounding mode in effect. The C library initializes this register for
9578 rounding towards plus infinity. Thus, unless your program modifies the
9579 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9582 @item -mtrap-precision=@var{trap-precision}
9583 @opindex mtrap-precision
9584 In the Alpha architecture, floating point traps are imprecise. This
9585 means without software assistance it is impossible to recover from a
9586 floating trap and program execution normally needs to be terminated.
9587 GCC can generate code that can assist operating system trap handlers
9588 in determining the exact location that caused a floating point trap.
9589 Depending on the requirements of an application, different levels of
9590 precisions can be selected:
9594 Program precision. This option is the default and means a trap handler
9595 can only identify which program caused a floating point exception.
9598 Function precision. The trap handler can determine the function that
9599 caused a floating point exception.
9602 Instruction precision. The trap handler can determine the exact
9603 instruction that caused a floating point exception.
9606 Other Alpha compilers provide the equivalent options called
9607 @option{-scope_safe} and @option{-resumption_safe}.
9609 @item -mieee-conformant
9610 @opindex mieee-conformant
9611 This option marks the generated code as IEEE conformant. You must not
9612 use this option unless you also specify @option{-mtrap-precision=i} and either
9613 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9614 is to emit the line @samp{.eflag 48} in the function prologue of the
9615 generated assembly file. Under DEC Unix, this has the effect that
9616 IEEE-conformant math library routines will be linked in.
9618 @item -mbuild-constants
9619 @opindex mbuild-constants
9620 Normally GCC examines a 32- or 64-bit integer constant to
9621 see if it can construct it from smaller constants in two or three
9622 instructions. If it cannot, it will output the constant as a literal and
9623 generate code to load it from the data segment at runtime.
9625 Use this option to require GCC to construct @emph{all} integer constants
9626 using code, even if it takes more instructions (the maximum is six).
9628 You would typically use this option to build a shared library dynamic
9629 loader. Itself a shared library, it must relocate itself in memory
9630 before it can find the variables and constants in its own data segment.
9636 Select whether to generate code to be assembled by the vendor-supplied
9637 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9655 Indicate whether GCC should generate code to use the optional BWX,
9656 CIX, FIX and MAX instruction sets. The default is to use the instruction
9657 sets supported by the CPU type specified via @option{-mcpu=} option or that
9658 of the CPU on which GCC was built if none was specified.
9663 @opindex mfloat-ieee
9664 Generate code that uses (does not use) VAX F and G floating point
9665 arithmetic instead of IEEE single and double precision.
9667 @item -mexplicit-relocs
9668 @itemx -mno-explicit-relocs
9669 @opindex mexplicit-relocs
9670 @opindex mno-explicit-relocs
9671 Older Alpha assemblers provided no way to generate symbol relocations
9672 except via assembler macros. Use of these macros does not allow
9673 optimal instruction scheduling. GNU binutils as of version 2.12
9674 supports a new syntax that allows the compiler to explicitly mark
9675 which relocations should apply to which instructions. This option
9676 is mostly useful for debugging, as GCC detects the capabilities of
9677 the assembler when it is built and sets the default accordingly.
9681 @opindex msmall-data
9682 @opindex mlarge-data
9683 When @option{-mexplicit-relocs} is in effect, static data is
9684 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9685 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9686 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9687 16-bit relocations off of the @code{$gp} register. This limits the
9688 size of the small data area to 64KB, but allows the variables to be
9689 directly accessed via a single instruction.
9691 The default is @option{-mlarge-data}. With this option the data area
9692 is limited to just below 2GB@. Programs that require more than 2GB of
9693 data must use @code{malloc} or @code{mmap} to allocate the data in the
9694 heap instead of in the program's data segment.
9696 When generating code for shared libraries, @option{-fpic} implies
9697 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9701 @opindex msmall-text
9702 @opindex mlarge-text
9703 When @option{-msmall-text} is used, the compiler assumes that the
9704 code of the entire program (or shared library) fits in 4MB, and is
9705 thus reachable with a branch instruction. When @option{-msmall-data}
9706 is used, the compiler can assume that all local symbols share the
9707 same @code{$gp} value, and thus reduce the number of instructions
9708 required for a function call from 4 to 1.
9710 The default is @option{-mlarge-text}.
9712 @item -mcpu=@var{cpu_type}
9714 Set the instruction set and instruction scheduling parameters for
9715 machine type @var{cpu_type}. You can specify either the @samp{EV}
9716 style name or the corresponding chip number. GCC supports scheduling
9717 parameters for the EV4, EV5 and EV6 family of processors and will
9718 choose the default values for the instruction set from the processor
9719 you specify. If you do not specify a processor type, GCC will default
9720 to the processor on which the compiler was built.
9722 Supported values for @var{cpu_type} are
9728 Schedules as an EV4 and has no instruction set extensions.
9732 Schedules as an EV5 and has no instruction set extensions.
9736 Schedules as an EV5 and supports the BWX extension.
9741 Schedules as an EV5 and supports the BWX and MAX extensions.
9745 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9749 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9752 @item -mtune=@var{cpu_type}
9754 Set only the instruction scheduling parameters for machine type
9755 @var{cpu_type}. The instruction set is not changed.
9757 @item -mmemory-latency=@var{time}
9758 @opindex mmemory-latency
9759 Sets the latency the scheduler should assume for typical memory
9760 references as seen by the application. This number is highly
9761 dependent on the memory access patterns used by the application
9762 and the size of the external cache on the machine.
9764 Valid options for @var{time} are
9768 A decimal number representing clock cycles.
9774 The compiler contains estimates of the number of clock cycles for
9775 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9776 (also called Dcache, Scache, and Bcache), as well as to main memory.
9777 Note that L3 is only valid for EV5.
9782 @node DEC Alpha/VMS Options
9783 @subsection DEC Alpha/VMS Options
9785 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9788 @item -mvms-return-codes
9789 @opindex mvms-return-codes
9790 Return VMS condition codes from main. The default is to return POSIX
9791 style condition (e.g.@: error) codes.
9795 @subsection FRV Options
9802 Only use the first 32 general purpose registers.
9807 Use all 64 general purpose registers.
9812 Use only the first 32 floating point registers.
9817 Use all 64 floating point registers
9820 @opindex mhard-float
9822 Use hardware instructions for floating point operations.
9825 @opindex msoft-float
9827 Use library routines for floating point operations.
9832 Dynamically allocate condition code registers.
9837 Do not try to dynamically allocate condition code registers, only
9838 use @code{icc0} and @code{fcc0}.
9843 Change ABI to use double word insns.
9848 Do not use double word instructions.
9853 Use floating point double instructions.
9858 Do not use floating point double instructions.
9863 Use media instructions.
9868 Do not use media instructions.
9873 Use multiply and add/subtract instructions.
9878 Do not use multiply and add/subtract instructions.
9883 Select the FDPIC ABI, that uses function descriptors to represent
9884 pointers to functions. Without any PIC/PIE-related options, it
9885 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9886 assumes GOT entries and small data are within a 12-bit range from the
9887 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9888 are computed with 32 bits.
9889 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9892 @opindex minline-plt
9894 Enable inlining of PLT entries in function calls to functions that are
9895 not known to bind locally. It has no effect without @option{-mfdpic}.
9896 It's enabled by default if optimizing for speed and compiling for
9897 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9898 optimization option such as @option{-O3} or above is present in the
9904 Assume a large TLS segment when generating thread-local code.
9909 Do not assume a large TLS segment when generating thread-local code.
9914 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9915 that is known to be in read-only sections. It's enabled by default,
9916 except for @option{-fpic} or @option{-fpie}: even though it may help
9917 make the global offset table smaller, it trades 1 instruction for 4.
9918 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9919 one of which may be shared by multiple symbols, and it avoids the need
9920 for a GOT entry for the referenced symbol, so it's more likely to be a
9921 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9923 @item -multilib-library-pic
9924 @opindex multilib-library-pic
9926 Link with the (library, not FD) pic libraries. It's implied by
9927 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9928 @option{-fpic} without @option{-mfdpic}. You should never have to use
9934 Follow the EABI requirement of always creating a frame pointer whenever
9935 a stack frame is allocated. This option is enabled by default and can
9936 be disabled with @option{-mno-linked-fp}.
9939 @opindex mlong-calls
9941 Use indirect addressing to call functions outside the current
9942 compilation unit. This allows the functions to be placed anywhere
9943 within the 32-bit address space.
9945 @item -malign-labels
9946 @opindex malign-labels
9948 Try to align labels to an 8-byte boundary by inserting nops into the
9949 previous packet. This option only has an effect when VLIW packing
9950 is enabled. It doesn't create new packets; it merely adds nops to
9954 @opindex mlibrary-pic
9956 Generate position-independent EABI code.
9961 Use only the first four media accumulator registers.
9966 Use all eight media accumulator registers.
9971 Pack VLIW instructions.
9976 Do not pack VLIW instructions.
9981 Do not mark ABI switches in e_flags.
9986 Enable the use of conditional-move instructions (default).
9988 This switch is mainly for debugging the compiler and will likely be removed
9989 in a future version.
9991 @item -mno-cond-move
9992 @opindex mno-cond-move
9994 Disable the use of conditional-move instructions.
9996 This switch is mainly for debugging the compiler and will likely be removed
9997 in a future version.
10002 Enable the use of conditional set instructions (default).
10004 This switch is mainly for debugging the compiler and will likely be removed
10005 in a future version.
10010 Disable the use of conditional set instructions.
10012 This switch is mainly for debugging the compiler and will likely be removed
10013 in a future version.
10016 @opindex mcond-exec
10018 Enable the use of conditional execution (default).
10020 This switch is mainly for debugging the compiler and will likely be removed
10021 in a future version.
10023 @item -mno-cond-exec
10024 @opindex mno-cond-exec
10026 Disable the use of conditional execution.
10028 This switch is mainly for debugging the compiler and will likely be removed
10029 in a future version.
10031 @item -mvliw-branch
10032 @opindex mvliw-branch
10034 Run a pass to pack branches into VLIW instructions (default).
10036 This switch is mainly for debugging the compiler and will likely be removed
10037 in a future version.
10039 @item -mno-vliw-branch
10040 @opindex mno-vliw-branch
10042 Do not run a pass to pack branches into VLIW instructions.
10044 This switch is mainly for debugging the compiler and will likely be removed
10045 in a future version.
10047 @item -mmulti-cond-exec
10048 @opindex mmulti-cond-exec
10050 Enable optimization of @code{&&} and @code{||} in conditional execution
10053 This switch is mainly for debugging the compiler and will likely be removed
10054 in a future version.
10056 @item -mno-multi-cond-exec
10057 @opindex mno-multi-cond-exec
10059 Disable optimization of @code{&&} and @code{||} in conditional execution.
10061 This switch is mainly for debugging the compiler and will likely be removed
10062 in a future version.
10064 @item -mnested-cond-exec
10065 @opindex mnested-cond-exec
10067 Enable nested conditional execution optimizations (default).
10069 This switch is mainly for debugging the compiler and will likely be removed
10070 in a future version.
10072 @item -mno-nested-cond-exec
10073 @opindex mno-nested-cond-exec
10075 Disable nested conditional execution optimizations.
10077 This switch is mainly for debugging the compiler and will likely be removed
10078 in a future version.
10080 @item -moptimize-membar
10081 @opindex moptimize-membar
10083 This switch removes redundant @code{membar} instructions from the
10084 compiler generated code. It is enabled by default.
10086 @item -mno-optimize-membar
10087 @opindex mno-optimize-membar
10089 This switch disables the automatic removal of redundant @code{membar}
10090 instructions from the generated code.
10092 @item -mtomcat-stats
10093 @opindex mtomcat-stats
10095 Cause gas to print out tomcat statistics.
10097 @item -mcpu=@var{cpu}
10100 Select the processor type for which to generate code. Possible values are
10101 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
10102 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
10106 @node GNU/Linux Options
10107 @subsection GNU/Linux Options
10109 These @samp{-m} options are defined for GNU/Linux targets:
10114 Use the GNU C library instead of uClibc. This is the default except
10115 on @samp{*-*-linux-*uclibc*} targets.
10119 Use uClibc instead of the GNU C library. This is the default on
10120 @samp{*-*-linux-*uclibc*} targets.
10123 @node H8/300 Options
10124 @subsection H8/300 Options
10126 These @samp{-m} options are defined for the H8/300 implementations:
10131 Shorten some address references at link time, when possible; uses the
10132 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
10133 ld, Using ld}, for a fuller description.
10137 Generate code for the H8/300H@.
10141 Generate code for the H8S@.
10145 Generate code for the H8S and H8/300H in the normal mode. This switch
10146 must be used either with @option{-mh} or @option{-ms}.
10150 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
10154 Make @code{int} data 32 bits by default.
10157 @opindex malign-300
10158 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
10159 The default for the H8/300H and H8S is to align longs and floats on 4
10161 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
10162 This option has no effect on the H8/300.
10166 @subsection HPPA Options
10167 @cindex HPPA Options
10169 These @samp{-m} options are defined for the HPPA family of computers:
10172 @item -march=@var{architecture-type}
10174 Generate code for the specified architecture. The choices for
10175 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
10176 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
10177 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
10178 architecture option for your machine. Code compiled for lower numbered
10179 architectures will run on higher numbered architectures, but not the
10182 @item -mpa-risc-1-0
10183 @itemx -mpa-risc-1-1
10184 @itemx -mpa-risc-2-0
10185 @opindex mpa-risc-1-0
10186 @opindex mpa-risc-1-1
10187 @opindex mpa-risc-2-0
10188 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
10191 @opindex mbig-switch
10192 Generate code suitable for big switch tables. Use this option only if
10193 the assembler/linker complain about out of range branches within a switch
10196 @item -mjump-in-delay
10197 @opindex mjump-in-delay
10198 Fill delay slots of function calls with unconditional jump instructions
10199 by modifying the return pointer for the function call to be the target
10200 of the conditional jump.
10202 @item -mdisable-fpregs
10203 @opindex mdisable-fpregs
10204 Prevent floating point registers from being used in any manner. This is
10205 necessary for compiling kernels which perform lazy context switching of
10206 floating point registers. If you use this option and attempt to perform
10207 floating point operations, the compiler will abort.
10209 @item -mdisable-indexing
10210 @opindex mdisable-indexing
10211 Prevent the compiler from using indexing address modes. This avoids some
10212 rather obscure problems when compiling MIG generated code under MACH@.
10214 @item -mno-space-regs
10215 @opindex mno-space-regs
10216 Generate code that assumes the target has no space registers. This allows
10217 GCC to generate faster indirect calls and use unscaled index address modes.
10219 Such code is suitable for level 0 PA systems and kernels.
10221 @item -mfast-indirect-calls
10222 @opindex mfast-indirect-calls
10223 Generate code that assumes calls never cross space boundaries. This
10224 allows GCC to emit code which performs faster indirect calls.
10226 This option will not work in the presence of shared libraries or nested
10229 @item -mfixed-range=@var{register-range}
10230 @opindex mfixed-range
10231 Generate code treating the given register range as fixed registers.
10232 A fixed register is one that the register allocator can not use. This is
10233 useful when compiling kernel code. A register range is specified as
10234 two registers separated by a dash. Multiple register ranges can be
10235 specified separated by a comma.
10237 @item -mlong-load-store
10238 @opindex mlong-load-store
10239 Generate 3-instruction load and store sequences as sometimes required by
10240 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
10243 @item -mportable-runtime
10244 @opindex mportable-runtime
10245 Use the portable calling conventions proposed by HP for ELF systems.
10249 Enable the use of assembler directives only GAS understands.
10251 @item -mschedule=@var{cpu-type}
10253 Schedule code according to the constraints for the machine type
10254 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
10255 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
10256 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
10257 proper scheduling option for your machine. The default scheduling is
10261 @opindex mlinker-opt
10262 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
10263 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
10264 linkers in which they give bogus error messages when linking some programs.
10267 @opindex msoft-float
10268 Generate output containing library calls for floating point.
10269 @strong{Warning:} the requisite libraries are not available for all HPPA
10270 targets. Normally the facilities of the machine's usual C compiler are
10271 used, but this cannot be done directly in cross-compilation. You must make
10272 your own arrangements to provide suitable library functions for
10275 @option{-msoft-float} changes the calling convention in the output file;
10276 therefore, it is only useful if you compile @emph{all} of a program with
10277 this option. In particular, you need to compile @file{libgcc.a}, the
10278 library that comes with GCC, with @option{-msoft-float} in order for
10283 Generate the predefine, @code{_SIO}, for server IO@. The default is
10284 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
10285 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
10286 options are available under HP-UX and HI-UX@.
10290 Use GNU ld specific options. This passes @option{-shared} to ld when
10291 building a shared library. It is the default when GCC is configured,
10292 explicitly or implicitly, with the GNU linker. This option does not
10293 have any affect on which ld is called, it only changes what parameters
10294 are passed to that ld. The ld that is called is determined by the
10295 @option{--with-ld} configure option, GCC's program search path, and
10296 finally by the user's @env{PATH}. The linker used by GCC can be printed
10297 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
10298 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10302 Use HP ld specific options. This passes @option{-b} to ld when building
10303 a shared library and passes @option{+Accept TypeMismatch} to ld on all
10304 links. It is the default when GCC is configured, explicitly or
10305 implicitly, with the HP linker. This option does not have any affect on
10306 which ld is called, it only changes what parameters are passed to that
10307 ld. The ld that is called is determined by the @option{--with-ld}
10308 configure option, GCC's program search path, and finally by the user's
10309 @env{PATH}. The linker used by GCC can be printed using @samp{which
10310 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
10311 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
10314 @opindex mno-long-calls
10315 Generate code that uses long call sequences. This ensures that a call
10316 is always able to reach linker generated stubs. The default is to generate
10317 long calls only when the distance from the call site to the beginning
10318 of the function or translation unit, as the case may be, exceeds a
10319 predefined limit set by the branch type being used. The limits for
10320 normal calls are 7,600,000 and 240,000 bytes, respectively for the
10321 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
10324 Distances are measured from the beginning of functions when using the
10325 @option{-ffunction-sections} option, or when using the @option{-mgas}
10326 and @option{-mno-portable-runtime} options together under HP-UX with
10329 It is normally not desirable to use this option as it will degrade
10330 performance. However, it may be useful in large applications,
10331 particularly when partial linking is used to build the application.
10333 The types of long calls used depends on the capabilities of the
10334 assembler and linker, and the type of code being generated. The
10335 impact on systems that support long absolute calls, and long pic
10336 symbol-difference or pc-relative calls should be relatively small.
10337 However, an indirect call is used on 32-bit ELF systems in pic code
10338 and it is quite long.
10340 @item -munix=@var{unix-std}
10342 Generate compiler predefines and select a startfile for the specified
10343 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
10344 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
10345 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
10346 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
10347 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
10350 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
10351 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
10352 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
10353 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
10354 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
10355 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
10357 It is @emph{important} to note that this option changes the interfaces
10358 for various library routines. It also affects the operational behavior
10359 of the C library. Thus, @emph{extreme} care is needed in using this
10362 Library code that is intended to operate with more than one UNIX
10363 standard must test, set and restore the variable @var{__xpg4_extended_mask}
10364 as appropriate. Most GNU software doesn't provide this capability.
10368 Suppress the generation of link options to search libdld.sl when the
10369 @option{-static} option is specified on HP-UX 10 and later.
10373 The HP-UX implementation of setlocale in libc has a dependency on
10374 libdld.sl. There isn't an archive version of libdld.sl. Thus,
10375 when the @option{-static} option is specified, special link options
10376 are needed to resolve this dependency.
10378 On HP-UX 10 and later, the GCC driver adds the necessary options to
10379 link with libdld.sl when the @option{-static} option is specified.
10380 This causes the resulting binary to be dynamic. On the 64-bit port,
10381 the linkers generate dynamic binaries by default in any case. The
10382 @option{-nolibdld} option can be used to prevent the GCC driver from
10383 adding these link options.
10387 Add support for multithreading with the @dfn{dce thread} library
10388 under HP-UX@. This option sets flags for both the preprocessor and
10392 @node i386 and x86-64 Options
10393 @subsection Intel 386 and AMD x86-64 Options
10394 @cindex i386 Options
10395 @cindex x86-64 Options
10396 @cindex Intel 386 Options
10397 @cindex AMD x86-64 Options
10399 These @samp{-m} options are defined for the i386 and x86-64 family of
10403 @item -mtune=@var{cpu-type}
10405 Tune to @var{cpu-type} everything applicable about the generated code, except
10406 for the ABI and the set of available instructions. The choices for
10407 @var{cpu-type} are:
10410 Produce code optimized for the most common IA32/AMD64/EM64T processors.
10411 If you know the CPU on which your code will run, then you should use
10412 the corresponding @option{-mtune} option instead of
10413 @option{-mtune=generic}. But, if you do not know exactly what CPU users
10414 of your application will have, then you should use this option.
10416 As new processors are deployed in the marketplace, the behavior of this
10417 option will change. Therefore, if you upgrade to a newer version of
10418 GCC, the code generated option will change to reflect the processors
10419 that were most common when that version of GCC was released.
10421 There is no @option{-march=generic} option because @option{-march}
10422 indicates the instruction set the compiler can use, and there is no
10423 generic instruction set applicable to all processors. In contrast,
10424 @option{-mtune} indicates the processor (or, in this case, collection of
10425 processors) for which the code is optimized.
10427 This selects the CPU to tune for at compilation time by determining
10428 the processor type of the compiling machine. Using @option{-mtune=native}
10429 will produce code optimized for the local machine under the constraints
10430 of the selected instruction set. Using @option{-march=native} will
10431 enable all instruction subsets supported by the local machine (hence
10432 the result might not run on different machines).
10434 Original Intel's i386 CPU@.
10436 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
10437 @item i586, pentium
10438 Intel Pentium CPU with no MMX support.
10440 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
10442 Intel PentiumPro CPU@.
10444 Same as @code{generic}, but when used as @code{march} option, PentiumPro
10445 instruction set will be used, so the code will run on all i686 family chips.
10447 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
10448 @item pentium3, pentium3m
10449 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
10452 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
10453 support. Used by Centrino notebooks.
10454 @item pentium4, pentium4m
10455 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
10457 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
10460 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
10461 SSE2 and SSE3 instruction set support.
10463 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
10464 instruction set support.
10466 AMD K6 CPU with MMX instruction set support.
10468 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
10469 @item athlon, athlon-tbird
10470 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
10472 @item athlon-4, athlon-xp, athlon-mp
10473 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
10474 instruction set support.
10475 @item k8, opteron, athlon64, athlon-fx
10476 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
10477 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
10478 @item k8-sse3, opteron-sse3, athlon64-sse3
10479 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
10480 @item amdfam10, barcelona
10481 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
10482 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
10483 instruction set extensions.)
10485 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
10488 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
10489 instruction set support.
10491 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
10492 implemented for this chip.)
10494 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
10495 implemented for this chip.)
10497 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
10500 While picking a specific @var{cpu-type} will schedule things appropriately
10501 for that particular chip, the compiler will not generate any code that
10502 does not run on the i386 without the @option{-march=@var{cpu-type}} option
10505 @item -march=@var{cpu-type}
10507 Generate instructions for the machine type @var{cpu-type}. The choices
10508 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
10509 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
10511 @item -mcpu=@var{cpu-type}
10513 A deprecated synonym for @option{-mtune}.
10515 @item -mfpmath=@var{unit}
10517 Generate floating point arithmetics for selected unit @var{unit}. The choices
10518 for @var{unit} are:
10522 Use the standard 387 floating point coprocessor present majority of chips and
10523 emulated otherwise. Code compiled with this option will run almost everywhere.
10524 The temporary results are computed in 80bit precision instead of precision
10525 specified by the type resulting in slightly different results compared to most
10526 of other chips. See @option{-ffloat-store} for more detailed description.
10528 This is the default choice for i386 compiler.
10531 Use scalar floating point instructions present in the SSE instruction set.
10532 This instruction set is supported by Pentium3 and newer chips, in the AMD line
10533 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
10534 instruction set supports only single precision arithmetics, thus the double and
10535 extended precision arithmetics is still done using 387. Later version, present
10536 only in Pentium4 and the future AMD x86-64 chips supports double precision
10539 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
10540 or @option{-msse2} switches to enable SSE extensions and make this option
10541 effective. For the x86-64 compiler, these extensions are enabled by default.
10543 The resulting code should be considerably faster in the majority of cases and avoid
10544 the numerical instability problems of 387 code, but may break some existing
10545 code that expects temporaries to be 80bit.
10547 This is the default choice for the x86-64 compiler.
10552 Attempt to utilize both instruction sets at once. This effectively double the
10553 amount of available registers and on chips with separate execution units for
10554 387 and SSE the execution resources too. Use this option with care, as it is
10555 still experimental, because the GCC register allocator does not model separate
10556 functional units well resulting in instable performance.
10559 @item -masm=@var{dialect}
10560 @opindex masm=@var{dialect}
10561 Output asm instructions using selected @var{dialect}. Supported
10562 choices are @samp{intel} or @samp{att} (the default one). Darwin does
10563 not support @samp{intel}.
10566 @itemx -mno-ieee-fp
10568 @opindex mno-ieee-fp
10569 Control whether or not the compiler uses IEEE floating point
10570 comparisons. These handle correctly the case where the result of a
10571 comparison is unordered.
10574 @opindex msoft-float
10575 Generate output containing library calls for floating point.
10576 @strong{Warning:} the requisite libraries are not part of GCC@.
10577 Normally the facilities of the machine's usual C compiler are used, but
10578 this can't be done directly in cross-compilation. You must make your
10579 own arrangements to provide suitable library functions for
10582 On machines where a function returns floating point results in the 80387
10583 register stack, some floating point opcodes may be emitted even if
10584 @option{-msoft-float} is used.
10586 @item -mno-fp-ret-in-387
10587 @opindex mno-fp-ret-in-387
10588 Do not use the FPU registers for return values of functions.
10590 The usual calling convention has functions return values of types
10591 @code{float} and @code{double} in an FPU register, even if there
10592 is no FPU@. The idea is that the operating system should emulate
10595 The option @option{-mno-fp-ret-in-387} causes such values to be returned
10596 in ordinary CPU registers instead.
10598 @item -mno-fancy-math-387
10599 @opindex mno-fancy-math-387
10600 Some 387 emulators do not support the @code{sin}, @code{cos} and
10601 @code{sqrt} instructions for the 387. Specify this option to avoid
10602 generating those instructions. This option is the default on FreeBSD,
10603 OpenBSD and NetBSD@. This option is overridden when @option{-march}
10604 indicates that the target cpu will always have an FPU and so the
10605 instruction will not need emulation. As of revision 2.6.1, these
10606 instructions are not generated unless you also use the
10607 @option{-funsafe-math-optimizations} switch.
10609 @item -malign-double
10610 @itemx -mno-align-double
10611 @opindex malign-double
10612 @opindex mno-align-double
10613 Control whether GCC aligns @code{double}, @code{long double}, and
10614 @code{long long} variables on a two word boundary or a one word
10615 boundary. Aligning @code{double} variables on a two word boundary will
10616 produce code that runs somewhat faster on a @samp{Pentium} at the
10617 expense of more memory.
10619 On x86-64, @option{-malign-double} is enabled by default.
10621 @strong{Warning:} if you use the @option{-malign-double} switch,
10622 structures containing the above types will be aligned differently than
10623 the published application binary interface specifications for the 386
10624 and will not be binary compatible with structures in code compiled
10625 without that switch.
10627 @item -m96bit-long-double
10628 @itemx -m128bit-long-double
10629 @opindex m96bit-long-double
10630 @opindex m128bit-long-double
10631 These switches control the size of @code{long double} type. The i386
10632 application binary interface specifies the size to be 96 bits,
10633 so @option{-m96bit-long-double} is the default in 32 bit mode.
10635 Modern architectures (Pentium and newer) would prefer @code{long double}
10636 to be aligned to an 8 or 16 byte boundary. In arrays or structures
10637 conforming to the ABI, this would not be possible. So specifying a
10638 @option{-m128bit-long-double} will align @code{long double}
10639 to a 16 byte boundary by padding the @code{long double} with an additional
10642 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
10643 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
10645 Notice that neither of these options enable any extra precision over the x87
10646 standard of 80 bits for a @code{long double}.
10648 @strong{Warning:} if you override the default value for your target ABI, the
10649 structures and arrays containing @code{long double} variables will change
10650 their size as well as function calling convention for function taking
10651 @code{long double} will be modified. Hence they will not be binary
10652 compatible with arrays or structures in code compiled without that switch.
10654 @item -mmlarge-data-threshold=@var{number}
10655 @opindex mlarge-data-threshold=@var{number}
10656 When @option{-mcmodel=medium} is specified, the data greater than
10657 @var{threshold} are placed in large data section. This value must be the
10658 same across all object linked into the binary and defaults to 65535.
10662 Use a different function-calling convention, in which functions that
10663 take a fixed number of arguments return with the @code{ret} @var{num}
10664 instruction, which pops their arguments while returning. This saves one
10665 instruction in the caller since there is no need to pop the arguments
10668 You can specify that an individual function is called with this calling
10669 sequence with the function attribute @samp{stdcall}. You can also
10670 override the @option{-mrtd} option by using the function attribute
10671 @samp{cdecl}. @xref{Function Attributes}.
10673 @strong{Warning:} this calling convention is incompatible with the one
10674 normally used on Unix, so you cannot use it if you need to call
10675 libraries compiled with the Unix compiler.
10677 Also, you must provide function prototypes for all functions that
10678 take variable numbers of arguments (including @code{printf});
10679 otherwise incorrect code will be generated for calls to those
10682 In addition, seriously incorrect code will result if you call a
10683 function with too many arguments. (Normally, extra arguments are
10684 harmlessly ignored.)
10686 @item -mregparm=@var{num}
10688 Control how many registers are used to pass integer arguments. By
10689 default, no registers are used to pass arguments, and at most 3
10690 registers can be used. You can control this behavior for a specific
10691 function by using the function attribute @samp{regparm}.
10692 @xref{Function Attributes}.
10694 @strong{Warning:} if you use this switch, and
10695 @var{num} is nonzero, then you must build all modules with the same
10696 value, including any libraries. This includes the system libraries and
10700 @opindex msseregparm
10701 Use SSE register passing conventions for float and double arguments
10702 and return values. You can control this behavior for a specific
10703 function by using the function attribute @samp{sseregparm}.
10704 @xref{Function Attributes}.
10706 @strong{Warning:} if you use this switch then you must build all
10707 modules with the same value, including any libraries. This includes
10708 the system libraries and startup modules.
10717 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
10718 is specified, the significands of results of floating-point operations are
10719 rounded to 24 bits (single precision); @option{-mpc64} rounds the the
10720 significands of results of floating-point operations to 53 bits (double
10721 precision) and @option{-mpc80} rounds the significands of results of
10722 floating-point operations to 64 bits (extended double precision), which is
10723 the default. When this option is used, floating-point operations in higher
10724 precisions are not available to the programmer without setting the FPU
10725 control word explicitly.
10727 Setting the rounding of floating-point operations to less than the default
10728 80 bits can speed some programs by 2% or more. Note that some mathematical
10729 libraries assume that extended precision (80 bit) floating-point operations
10730 are enabled by default; routines in such libraries could suffer significant
10731 loss of accuracy, typically through so-called "catastrophic cancellation",
10732 when this option is used to set the precision to less than extended precision.
10734 @item -mstackrealign
10735 @opindex mstackrealign
10736 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
10737 option will generate an alternate prologue and epilogue that realigns the
10738 runtime stack if necessary. This supports mixing legacy codes that keep
10739 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
10740 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
10741 applicable to individual functions.
10743 @item -mpreferred-stack-boundary=@var{num}
10744 @opindex mpreferred-stack-boundary
10745 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
10746 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
10747 the default is 4 (16 bytes or 128 bits).
10749 @item -mincoming-stack-boundary=@var{num}
10750 @opindex mincoming-stack-boundary
10751 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
10752 boundary. If @option{-mincoming-stack-boundary} is not specified,
10753 the one specified by @option{-mpreferred-stack-boundary} will be used.
10755 On Pentium and PentiumPro, @code{double} and @code{long double} values
10756 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
10757 suffer significant run time performance penalties. On Pentium III, the
10758 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
10759 properly if it is not 16 byte aligned.
10761 To ensure proper alignment of this values on the stack, the stack boundary
10762 must be as aligned as that required by any value stored on the stack.
10763 Further, every function must be generated such that it keeps the stack
10764 aligned. Thus calling a function compiled with a higher preferred
10765 stack boundary from a function compiled with a lower preferred stack
10766 boundary will most likely misalign the stack. It is recommended that
10767 libraries that use callbacks always use the default setting.
10769 This extra alignment does consume extra stack space, and generally
10770 increases code size. Code that is sensitive to stack space usage, such
10771 as embedded systems and operating system kernels, may want to reduce the
10772 preferred alignment to @option{-mpreferred-stack-boundary=2}.
10810 These switches enable or disable the use of instructions in the MMX,
10811 SSE, SSE2, SSE3, SSSE3, SSE4.1, AES, PCLMUL, SSE4A, SSE5, ABM or
10812 3DNow!@: extended instruction sets.
10813 These extensions are also available as built-in functions: see
10814 @ref{X86 Built-in Functions}, for details of the functions enabled and
10815 disabled by these switches.
10817 To have SSE/SSE2 instructions generated automatically from floating-point
10818 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
10820 These options will enable GCC to use these extended instructions in
10821 generated code, even without @option{-mfpmath=sse}. Applications which
10822 perform runtime CPU detection must compile separate files for each
10823 supported architecture, using the appropriate flags. In particular,
10824 the file containing the CPU detection code should be compiled without
10829 This option instructs GCC to emit a @code{cld} instruction in the prologue
10830 of functions that use string instructions. String instructions depend on
10831 the DF flag to select between autoincrement or autodecrement mode. While the
10832 ABI specifies the DF flag to be cleared on function entry, some operating
10833 systems violate this specification by not clearing the DF flag in their
10834 exception dispatchers. The exception handler can be invoked with the DF flag
10835 set which leads to wrong direction mode, when string instructions are used.
10836 This option can be enabled by default on 32-bit x86 targets by configuring
10837 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
10838 instructions can be suppressed with the @option{-mno-cld} compiler option
10843 This option will enable GCC to use CMPXCHG16B instruction in generated code.
10844 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
10845 data types. This is useful for high resolution counters that could be updated
10846 by multiple processors (or cores). This instruction is generated as part of
10847 atomic built-in functions: see @ref{Atomic Builtins} for details.
10851 This option will enable GCC to use SAHF instruction in generated 64-bit code.
10852 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
10853 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
10854 SAHF are load and store instructions, respectively, for certain status flags.
10855 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
10856 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
10860 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
10861 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Rhapson step
10862 to increase precision instead of DIVSS and SQRTSS (and their vectorized
10863 variants) for single precision floating point arguments. These instructions
10864 are generated only when @option{-funsafe-math-optimizations} is enabled
10865 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
10866 Note that while the throughput of the sequence is higher than the throughput
10867 of the non-reciprocal instruction, the precision of the sequence can be
10868 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
10870 @item -mveclibabi=@var{type}
10871 @opindex mveclibabi
10872 Specifies the ABI type to use for vectorizing intrinsics using an
10873 external library. Supported types are @code{svml} for the Intel short
10874 vector math library and @code{acml} for the AMD math core library style
10875 of interfacing. GCC will currently emit calls to @code{vmldExp2},
10876 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
10877 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
10878 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
10879 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
10880 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
10881 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
10882 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
10883 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
10884 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
10885 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
10886 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
10887 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
10888 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
10889 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
10890 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
10891 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
10892 compatible library will have to be specified at link time.
10895 @itemx -mno-push-args
10896 @opindex mpush-args
10897 @opindex mno-push-args
10898 Use PUSH operations to store outgoing parameters. This method is shorter
10899 and usually equally fast as method using SUB/MOV operations and is enabled
10900 by default. In some cases disabling it may improve performance because of
10901 improved scheduling and reduced dependencies.
10903 @item -maccumulate-outgoing-args
10904 @opindex maccumulate-outgoing-args
10905 If enabled, the maximum amount of space required for outgoing arguments will be
10906 computed in the function prologue. This is faster on most modern CPUs
10907 because of reduced dependencies, improved scheduling and reduced stack usage
10908 when preferred stack boundary is not equal to 2. The drawback is a notable
10909 increase in code size. This switch implies @option{-mno-push-args}.
10913 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
10914 on thread-safe exception handling must compile and link all code with the
10915 @option{-mthreads} option. When compiling, @option{-mthreads} defines
10916 @option{-D_MT}; when linking, it links in a special thread helper library
10917 @option{-lmingwthrd} which cleans up per thread exception handling data.
10919 @item -mno-align-stringops
10920 @opindex mno-align-stringops
10921 Do not align destination of inlined string operations. This switch reduces
10922 code size and improves performance in case the destination is already aligned,
10923 but GCC doesn't know about it.
10925 @item -minline-all-stringops
10926 @opindex minline-all-stringops
10927 By default GCC inlines string operations only when destination is known to be
10928 aligned at least to 4 byte boundary. This enables more inlining, increase code
10929 size, but may improve performance of code that depends on fast memcpy, strlen
10930 and memset for short lengths.
10932 @item -minline-stringops-dynamically
10933 @opindex minline-stringops-dynamically
10934 For string operation of unknown size, inline runtime checks so for small
10935 blocks inline code is used, while for large blocks library call is used.
10937 @item -mstringop-strategy=@var{alg}
10938 @opindex mstringop-strategy=@var{alg}
10939 Overwrite internal decision heuristic about particular algorithm to inline
10940 string operation with. The allowed values are @code{rep_byte},
10941 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
10942 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
10943 expanding inline loop, @code{libcall} for always expanding library call.
10945 @item -momit-leaf-frame-pointer
10946 @opindex momit-leaf-frame-pointer
10947 Don't keep the frame pointer in a register for leaf functions. This
10948 avoids the instructions to save, set up and restore frame pointers and
10949 makes an extra register available in leaf functions. The option
10950 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10951 which might make debugging harder.
10953 @item -mtls-direct-seg-refs
10954 @itemx -mno-tls-direct-seg-refs
10955 @opindex mtls-direct-seg-refs
10956 Controls whether TLS variables may be accessed with offsets from the
10957 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
10958 or whether the thread base pointer must be added. Whether or not this
10959 is legal depends on the operating system, and whether it maps the
10960 segment to cover the entire TLS area.
10962 For systems that use GNU libc, the default is on.
10965 @itemx -mno-fused-madd
10966 @opindex mfused-madd
10967 Enable automatic generation of fused floating point multiply-add instructions
10968 if the ISA supports such instructions. The -mfused-madd option is on by
10969 default. The fused multiply-add instructions have a different
10970 rounding behavior compared to executing a multiply followed by an add.
10973 These @samp{-m} switches are supported in addition to the above
10974 on AMD x86-64 processors in 64-bit environments.
10981 Generate code for a 32-bit or 64-bit environment.
10982 The 32-bit environment sets int, long and pointer to 32 bits and
10983 generates code that runs on any i386 system.
10984 The 64-bit environment sets int to 32 bits and long and pointer
10985 to 64 bits and generates code for AMD's x86-64 architecture. For
10986 darwin only the -m64 option turns off the @option{-fno-pic} and
10987 @option{-mdynamic-no-pic} options.
10989 @item -mno-red-zone
10990 @opindex no-red-zone
10991 Do not use a so called red zone for x86-64 code. The red zone is mandated
10992 by the x86-64 ABI, it is a 128-byte area beyond the location of the
10993 stack pointer that will not be modified by signal or interrupt handlers
10994 and therefore can be used for temporary data without adjusting the stack
10995 pointer. The flag @option{-mno-red-zone} disables this red zone.
10997 @item -mcmodel=small
10998 @opindex mcmodel=small
10999 Generate code for the small code model: the program and its symbols must
11000 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
11001 Programs can be statically or dynamically linked. This is the default
11004 @item -mcmodel=kernel
11005 @opindex mcmodel=kernel
11006 Generate code for the kernel code model. The kernel runs in the
11007 negative 2 GB of the address space.
11008 This model has to be used for Linux kernel code.
11010 @item -mcmodel=medium
11011 @opindex mcmodel=medium
11012 Generate code for the medium model: The program is linked in the lower 2
11013 GB of the address space but symbols can be located anywhere in the
11014 address space. Programs can be statically or dynamically linked, but
11015 building of shared libraries are not supported with the medium model.
11017 @item -mcmodel=large
11018 @opindex mcmodel=large
11019 Generate code for the large model: This model makes no assumptions
11020 about addresses and sizes of sections.
11023 @node IA-64 Options
11024 @subsection IA-64 Options
11025 @cindex IA-64 Options
11027 These are the @samp{-m} options defined for the Intel IA-64 architecture.
11031 @opindex mbig-endian
11032 Generate code for a big endian target. This is the default for HP-UX@.
11034 @item -mlittle-endian
11035 @opindex mlittle-endian
11036 Generate code for a little endian target. This is the default for AIX5
11042 @opindex mno-gnu-as
11043 Generate (or don't) code for the GNU assembler. This is the default.
11044 @c Also, this is the default if the configure option @option{--with-gnu-as}
11050 @opindex mno-gnu-ld
11051 Generate (or don't) code for the GNU linker. This is the default.
11052 @c Also, this is the default if the configure option @option{--with-gnu-ld}
11057 Generate code that does not use a global pointer register. The result
11058 is not position independent code, and violates the IA-64 ABI@.
11060 @item -mvolatile-asm-stop
11061 @itemx -mno-volatile-asm-stop
11062 @opindex mvolatile-asm-stop
11063 @opindex mno-volatile-asm-stop
11064 Generate (or don't) a stop bit immediately before and after volatile asm
11067 @item -mregister-names
11068 @itemx -mno-register-names
11069 @opindex mregister-names
11070 @opindex mno-register-names
11071 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
11072 the stacked registers. This may make assembler output more readable.
11078 Disable (or enable) optimizations that use the small data section. This may
11079 be useful for working around optimizer bugs.
11081 @item -mconstant-gp
11082 @opindex mconstant-gp
11083 Generate code that uses a single constant global pointer value. This is
11084 useful when compiling kernel code.
11088 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
11089 This is useful when compiling firmware code.
11091 @item -minline-float-divide-min-latency
11092 @opindex minline-float-divide-min-latency
11093 Generate code for inline divides of floating point values
11094 using the minimum latency algorithm.
11096 @item -minline-float-divide-max-throughput
11097 @opindex minline-float-divide-max-throughput
11098 Generate code for inline divides of floating point values
11099 using the maximum throughput algorithm.
11101 @item -minline-int-divide-min-latency
11102 @opindex minline-int-divide-min-latency
11103 Generate code for inline divides of integer values
11104 using the minimum latency algorithm.
11106 @item -minline-int-divide-max-throughput
11107 @opindex minline-int-divide-max-throughput
11108 Generate code for inline divides of integer values
11109 using the maximum throughput algorithm.
11111 @item -minline-sqrt-min-latency
11112 @opindex minline-sqrt-min-latency
11113 Generate code for inline square roots
11114 using the minimum latency algorithm.
11116 @item -minline-sqrt-max-throughput
11117 @opindex minline-sqrt-max-throughput
11118 Generate code for inline square roots
11119 using the maximum throughput algorithm.
11121 @item -mno-dwarf2-asm
11122 @itemx -mdwarf2-asm
11123 @opindex mno-dwarf2-asm
11124 @opindex mdwarf2-asm
11125 Don't (or do) generate assembler code for the DWARF2 line number debugging
11126 info. This may be useful when not using the GNU assembler.
11128 @item -mearly-stop-bits
11129 @itemx -mno-early-stop-bits
11130 @opindex mearly-stop-bits
11131 @opindex mno-early-stop-bits
11132 Allow stop bits to be placed earlier than immediately preceding the
11133 instruction that triggered the stop bit. This can improve instruction
11134 scheduling, but does not always do so.
11136 @item -mfixed-range=@var{register-range}
11137 @opindex mfixed-range
11138 Generate code treating the given register range as fixed registers.
11139 A fixed register is one that the register allocator can not use. This is
11140 useful when compiling kernel code. A register range is specified as
11141 two registers separated by a dash. Multiple register ranges can be
11142 specified separated by a comma.
11144 @item -mtls-size=@var{tls-size}
11146 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
11149 @item -mtune=@var{cpu-type}
11151 Tune the instruction scheduling for a particular CPU, Valid values are
11152 itanium, itanium1, merced, itanium2, and mckinley.
11158 Add support for multithreading using the POSIX threads library. This
11159 option sets flags for both the preprocessor and linker. It does
11160 not affect the thread safety of object code produced by the compiler or
11161 that of libraries supplied with it. These are HP-UX specific flags.
11167 Generate code for a 32-bit or 64-bit environment.
11168 The 32-bit environment sets int, long and pointer to 32 bits.
11169 The 64-bit environment sets int to 32 bits and long and pointer
11170 to 64 bits. These are HP-UX specific flags.
11172 @item -mno-sched-br-data-spec
11173 @itemx -msched-br-data-spec
11174 @opindex mno-sched-br-data-spec
11175 @opindex msched-br-data-spec
11176 (Dis/En)able data speculative scheduling before reload.
11177 This will result in generation of the ld.a instructions and
11178 the corresponding check instructions (ld.c / chk.a).
11179 The default is 'disable'.
11181 @item -msched-ar-data-spec
11182 @itemx -mno-sched-ar-data-spec
11183 @opindex msched-ar-data-spec
11184 @opindex mno-sched-ar-data-spec
11185 (En/Dis)able data speculative scheduling after reload.
11186 This will result in generation of the ld.a instructions and
11187 the corresponding check instructions (ld.c / chk.a).
11188 The default is 'enable'.
11190 @item -mno-sched-control-spec
11191 @itemx -msched-control-spec
11192 @opindex mno-sched-control-spec
11193 @opindex msched-control-spec
11194 (Dis/En)able control speculative scheduling. This feature is
11195 available only during region scheduling (i.e.@: before reload).
11196 This will result in generation of the ld.s instructions and
11197 the corresponding check instructions chk.s .
11198 The default is 'disable'.
11200 @item -msched-br-in-data-spec
11201 @itemx -mno-sched-br-in-data-spec
11202 @opindex msched-br-in-data-spec
11203 @opindex mno-sched-br-in-data-spec
11204 (En/Dis)able speculative scheduling of the instructions that
11205 are dependent on the data speculative loads before reload.
11206 This is effective only with @option{-msched-br-data-spec} enabled.
11207 The default is 'enable'.
11209 @item -msched-ar-in-data-spec
11210 @itemx -mno-sched-ar-in-data-spec
11211 @opindex msched-ar-in-data-spec
11212 @opindex mno-sched-ar-in-data-spec
11213 (En/Dis)able speculative scheduling of the instructions that
11214 are dependent on the data speculative loads after reload.
11215 This is effective only with @option{-msched-ar-data-spec} enabled.
11216 The default is 'enable'.
11218 @item -msched-in-control-spec
11219 @itemx -mno-sched-in-control-spec
11220 @opindex msched-in-control-spec
11221 @opindex mno-sched-in-control-spec
11222 (En/Dis)able speculative scheduling of the instructions that
11223 are dependent on the control speculative loads.
11224 This is effective only with @option{-msched-control-spec} enabled.
11225 The default is 'enable'.
11228 @itemx -mno-sched-ldc
11229 @opindex msched-ldc
11230 @opindex mno-sched-ldc
11231 (En/Dis)able use of simple data speculation checks ld.c .
11232 If disabled, only chk.a instructions will be emitted to check
11233 data speculative loads.
11234 The default is 'enable'.
11236 @item -mno-sched-control-ldc
11237 @itemx -msched-control-ldc
11238 @opindex mno-sched-control-ldc
11239 @opindex msched-control-ldc
11240 (Dis/En)able use of ld.c instructions to check control speculative loads.
11241 If enabled, in case of control speculative load with no speculatively
11242 scheduled dependent instructions this load will be emitted as ld.sa and
11243 ld.c will be used to check it.
11244 The default is 'disable'.
11246 @item -mno-sched-spec-verbose
11247 @itemx -msched-spec-verbose
11248 @opindex mno-sched-spec-verbose
11249 @opindex msched-spec-verbose
11250 (Dis/En)able printing of the information about speculative motions.
11252 @item -mno-sched-prefer-non-data-spec-insns
11253 @itemx -msched-prefer-non-data-spec-insns
11254 @opindex mno-sched-prefer-non-data-spec-insns
11255 @opindex msched-prefer-non-data-spec-insns
11256 If enabled, data speculative instructions will be chosen for schedule
11257 only if there are no other choices at the moment. This will make
11258 the use of the data speculation much more conservative.
11259 The default is 'disable'.
11261 @item -mno-sched-prefer-non-control-spec-insns
11262 @itemx -msched-prefer-non-control-spec-insns
11263 @opindex mno-sched-prefer-non-control-spec-insns
11264 @opindex msched-prefer-non-control-spec-insns
11265 If enabled, control speculative instructions will be chosen for schedule
11266 only if there are no other choices at the moment. This will make
11267 the use of the control speculation much more conservative.
11268 The default is 'disable'.
11270 @item -mno-sched-count-spec-in-critical-path
11271 @itemx -msched-count-spec-in-critical-path
11272 @opindex mno-sched-count-spec-in-critical-path
11273 @opindex msched-count-spec-in-critical-path
11274 If enabled, speculative dependencies will be considered during
11275 computation of the instructions priorities. This will make the use of the
11276 speculation a bit more conservative.
11277 The default is 'disable'.
11282 @subsection M32C Options
11283 @cindex M32C options
11286 @item -mcpu=@var{name}
11288 Select the CPU for which code is generated. @var{name} may be one of
11289 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
11290 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
11291 the M32C/80 series.
11295 Specifies that the program will be run on the simulator. This causes
11296 an alternate runtime library to be linked in which supports, for
11297 example, file I/O@. You must not use this option when generating
11298 programs that will run on real hardware; you must provide your own
11299 runtime library for whatever I/O functions are needed.
11301 @item -memregs=@var{number}
11303 Specifies the number of memory-based pseudo-registers GCC will use
11304 during code generation. These pseudo-registers will be used like real
11305 registers, so there is a tradeoff between GCC's ability to fit the
11306 code into available registers, and the performance penalty of using
11307 memory instead of registers. Note that all modules in a program must
11308 be compiled with the same value for this option. Because of that, you
11309 must not use this option with the default runtime libraries gcc
11314 @node M32R/D Options
11315 @subsection M32R/D Options
11316 @cindex M32R/D options
11318 These @option{-m} options are defined for Renesas M32R/D architectures:
11323 Generate code for the M32R/2@.
11327 Generate code for the M32R/X@.
11331 Generate code for the M32R@. This is the default.
11333 @item -mmodel=small
11334 @opindex mmodel=small
11335 Assume all objects live in the lower 16MB of memory (so that their addresses
11336 can be loaded with the @code{ld24} instruction), and assume all subroutines
11337 are reachable with the @code{bl} instruction.
11338 This is the default.
11340 The addressability of a particular object can be set with the
11341 @code{model} attribute.
11343 @item -mmodel=medium
11344 @opindex mmodel=medium
11345 Assume objects may be anywhere in the 32-bit address space (the compiler
11346 will generate @code{seth/add3} instructions to load their addresses), and
11347 assume all subroutines are reachable with the @code{bl} instruction.
11349 @item -mmodel=large
11350 @opindex mmodel=large
11351 Assume objects may be anywhere in the 32-bit address space (the compiler
11352 will generate @code{seth/add3} instructions to load their addresses), and
11353 assume subroutines may not be reachable with the @code{bl} instruction
11354 (the compiler will generate the much slower @code{seth/add3/jl}
11355 instruction sequence).
11358 @opindex msdata=none
11359 Disable use of the small data area. Variables will be put into
11360 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
11361 @code{section} attribute has been specified).
11362 This is the default.
11364 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
11365 Objects may be explicitly put in the small data area with the
11366 @code{section} attribute using one of these sections.
11368 @item -msdata=sdata
11369 @opindex msdata=sdata
11370 Put small global and static data in the small data area, but do not
11371 generate special code to reference them.
11374 @opindex msdata=use
11375 Put small global and static data in the small data area, and generate
11376 special instructions to reference them.
11380 @cindex smaller data references
11381 Put global and static objects less than or equal to @var{num} bytes
11382 into the small data or bss sections instead of the normal data or bss
11383 sections. The default value of @var{num} is 8.
11384 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
11385 for this option to have any effect.
11387 All modules should be compiled with the same @option{-G @var{num}} value.
11388 Compiling with different values of @var{num} may or may not work; if it
11389 doesn't the linker will give an error message---incorrect code will not be
11394 Makes the M32R specific code in the compiler display some statistics
11395 that might help in debugging programs.
11397 @item -malign-loops
11398 @opindex malign-loops
11399 Align all loops to a 32-byte boundary.
11401 @item -mno-align-loops
11402 @opindex mno-align-loops
11403 Do not enforce a 32-byte alignment for loops. This is the default.
11405 @item -missue-rate=@var{number}
11406 @opindex missue-rate=@var{number}
11407 Issue @var{number} instructions per cycle. @var{number} can only be 1
11410 @item -mbranch-cost=@var{number}
11411 @opindex mbranch-cost=@var{number}
11412 @var{number} can only be 1 or 2. If it is 1 then branches will be
11413 preferred over conditional code, if it is 2, then the opposite will
11416 @item -mflush-trap=@var{number}
11417 @opindex mflush-trap=@var{number}
11418 Specifies the trap number to use to flush the cache. The default is
11419 12. Valid numbers are between 0 and 15 inclusive.
11421 @item -mno-flush-trap
11422 @opindex mno-flush-trap
11423 Specifies that the cache cannot be flushed by using a trap.
11425 @item -mflush-func=@var{name}
11426 @opindex mflush-func=@var{name}
11427 Specifies the name of the operating system function to call to flush
11428 the cache. The default is @emph{_flush_cache}, but a function call
11429 will only be used if a trap is not available.
11431 @item -mno-flush-func
11432 @opindex mno-flush-func
11433 Indicates that there is no OS function for flushing the cache.
11437 @node M680x0 Options
11438 @subsection M680x0 Options
11439 @cindex M680x0 options
11441 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
11442 The default settings depend on which architecture was selected when
11443 the compiler was configured; the defaults for the most common choices
11447 @item -march=@var{arch}
11449 Generate code for a specific M680x0 or ColdFire instruction set
11450 architecture. Permissible values of @var{arch} for M680x0
11451 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
11452 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
11453 architectures are selected according to Freescale's ISA classification
11454 and the permissible values are: @samp{isaa}, @samp{isaaplus},
11455 @samp{isab} and @samp{isac}.
11457 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
11458 code for a ColdFire target. The @var{arch} in this macro is one of the
11459 @option{-march} arguments given above.
11461 When used together, @option{-march} and @option{-mtune} select code
11462 that runs on a family of similar processors but that is optimized
11463 for a particular microarchitecture.
11465 @item -mcpu=@var{cpu}
11467 Generate code for a specific M680x0 or ColdFire processor.
11468 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
11469 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
11470 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
11471 below, which also classifies the CPUs into families:
11473 @multitable @columnfractions 0.20 0.80
11474 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
11475 @item @samp{51qe} @tab @samp{51qe}
11476 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
11477 @item @samp{5206e} @tab @samp{5206e}
11478 @item @samp{5208} @tab @samp{5207} @samp{5208}
11479 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
11480 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
11481 @item @samp{5216} @tab @samp{5214} @samp{5216}
11482 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
11483 @item @samp{5225} @tab @samp{5224} @samp{5225}
11484 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
11485 @item @samp{5249} @tab @samp{5249}
11486 @item @samp{5250} @tab @samp{5250}
11487 @item @samp{5271} @tab @samp{5270} @samp{5271}
11488 @item @samp{5272} @tab @samp{5272}
11489 @item @samp{5275} @tab @samp{5274} @samp{5275}
11490 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
11491 @item @samp{5307} @tab @samp{5307}
11492 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
11493 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
11494 @item @samp{5407} @tab @samp{5407}
11495 @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}
11498 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
11499 @var{arch} is compatible with @var{cpu}. Other combinations of
11500 @option{-mcpu} and @option{-march} are rejected.
11502 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
11503 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
11504 where the value of @var{family} is given by the table above.
11506 @item -mtune=@var{tune}
11508 Tune the code for a particular microarchitecture, within the
11509 constraints set by @option{-march} and @option{-mcpu}.
11510 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
11511 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
11512 and @samp{cpu32}. The ColdFire microarchitectures
11513 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
11515 You can also use @option{-mtune=68020-40} for code that needs
11516 to run relatively well on 68020, 68030 and 68040 targets.
11517 @option{-mtune=68020-60} is similar but includes 68060 targets
11518 as well. These two options select the same tuning decisions as
11519 @option{-m68020-40} and @option{-m68020-60} respectively.
11521 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
11522 when tuning for 680x0 architecture @var{arch}. It also defines
11523 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
11524 option is used. If gcc is tuning for a range of architectures,
11525 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
11526 it defines the macros for every architecture in the range.
11528 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
11529 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
11530 of the arguments given above.
11536 Generate output for a 68000. This is the default
11537 when the compiler is configured for 68000-based systems.
11538 It is equivalent to @option{-march=68000}.
11540 Use this option for microcontrollers with a 68000 or EC000 core,
11541 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
11545 Generate output for a 68010. This is the default
11546 when the compiler is configured for 68010-based systems.
11547 It is equivalent to @option{-march=68010}.
11553 Generate output for a 68020. This is the default
11554 when the compiler is configured for 68020-based systems.
11555 It is equivalent to @option{-march=68020}.
11559 Generate output for a 68030. This is the default when the compiler is
11560 configured for 68030-based systems. It is equivalent to
11561 @option{-march=68030}.
11565 Generate output for a 68040. This is the default when the compiler is
11566 configured for 68040-based systems. It is equivalent to
11567 @option{-march=68040}.
11569 This option inhibits the use of 68881/68882 instructions that have to be
11570 emulated by software on the 68040. Use this option if your 68040 does not
11571 have code to emulate those instructions.
11575 Generate output for a 68060. This is the default when the compiler is
11576 configured for 68060-based systems. It is equivalent to
11577 @option{-march=68060}.
11579 This option inhibits the use of 68020 and 68881/68882 instructions that
11580 have to be emulated by software on the 68060. Use this option if your 68060
11581 does not have code to emulate those instructions.
11585 Generate output for a CPU32. This is the default
11586 when the compiler is configured for CPU32-based systems.
11587 It is equivalent to @option{-march=cpu32}.
11589 Use this option for microcontrollers with a
11590 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
11591 68336, 68340, 68341, 68349 and 68360.
11595 Generate output for a 520X ColdFire CPU@. This is the default
11596 when the compiler is configured for 520X-based systems.
11597 It is equivalent to @option{-mcpu=5206}, and is now deprecated
11598 in favor of that option.
11600 Use this option for microcontroller with a 5200 core, including
11601 the MCF5202, MCF5203, MCF5204 and MCF5206.
11605 Generate output for a 5206e ColdFire CPU@. The option is now
11606 deprecated in favor of the equivalent @option{-mcpu=5206e}.
11610 Generate output for a member of the ColdFire 528X family.
11611 The option is now deprecated in favor of the equivalent
11612 @option{-mcpu=528x}.
11616 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
11617 in favor of the equivalent @option{-mcpu=5307}.
11621 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
11622 in favor of the equivalent @option{-mcpu=5407}.
11626 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
11627 This includes use of hardware floating point instructions.
11628 The option is equivalent to @option{-mcpu=547x}, and is now
11629 deprecated in favor of that option.
11633 Generate output for a 68040, without using any of the new instructions.
11634 This results in code which can run relatively efficiently on either a
11635 68020/68881 or a 68030 or a 68040. The generated code does use the
11636 68881 instructions that are emulated on the 68040.
11638 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
11642 Generate output for a 68060, without using any of the new instructions.
11643 This results in code which can run relatively efficiently on either a
11644 68020/68881 or a 68030 or a 68040. The generated code does use the
11645 68881 instructions that are emulated on the 68060.
11647 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
11651 @opindex mhard-float
11653 Generate floating-point instructions. This is the default for 68020
11654 and above, and for ColdFire devices that have an FPU@. It defines the
11655 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
11656 on ColdFire targets.
11659 @opindex msoft-float
11660 Do not generate floating-point instructions; use library calls instead.
11661 This is the default for 68000, 68010, and 68832 targets. It is also
11662 the default for ColdFire devices that have no FPU.
11668 Generate (do not generate) ColdFire hardware divide and remainder
11669 instructions. If @option{-march} is used without @option{-mcpu},
11670 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
11671 architectures. Otherwise, the default is taken from the target CPU
11672 (either the default CPU, or the one specified by @option{-mcpu}). For
11673 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
11674 @option{-mcpu=5206e}.
11676 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
11680 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11681 Additionally, parameters passed on the stack are also aligned to a
11682 16-bit boundary even on targets whose API mandates promotion to 32-bit.
11686 Do not consider type @code{int} to be 16 bits wide. This is the default.
11689 @itemx -mno-bitfield
11690 @opindex mnobitfield
11691 @opindex mno-bitfield
11692 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
11693 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
11697 Do use the bit-field instructions. The @option{-m68020} option implies
11698 @option{-mbitfield}. This is the default if you use a configuration
11699 designed for a 68020.
11703 Use a different function-calling convention, in which functions
11704 that take a fixed number of arguments return with the @code{rtd}
11705 instruction, which pops their arguments while returning. This
11706 saves one instruction in the caller since there is no need to pop
11707 the arguments there.
11709 This calling convention is incompatible with the one normally
11710 used on Unix, so you cannot use it if you need to call libraries
11711 compiled with the Unix compiler.
11713 Also, you must provide function prototypes for all functions that
11714 take variable numbers of arguments (including @code{printf});
11715 otherwise incorrect code will be generated for calls to those
11718 In addition, seriously incorrect code will result if you call a
11719 function with too many arguments. (Normally, extra arguments are
11720 harmlessly ignored.)
11722 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
11723 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
11727 Do not use the calling conventions selected by @option{-mrtd}.
11728 This is the default.
11731 @itemx -mno-align-int
11732 @opindex malign-int
11733 @opindex mno-align-int
11734 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
11735 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
11736 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
11737 Aligning variables on 32-bit boundaries produces code that runs somewhat
11738 faster on processors with 32-bit busses at the expense of more memory.
11740 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
11741 align structures containing the above types differently than
11742 most published application binary interface specifications for the m68k.
11746 Use the pc-relative addressing mode of the 68000 directly, instead of
11747 using a global offset table. At present, this option implies @option{-fpic},
11748 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
11749 not presently supported with @option{-mpcrel}, though this could be supported for
11750 68020 and higher processors.
11752 @item -mno-strict-align
11753 @itemx -mstrict-align
11754 @opindex mno-strict-align
11755 @opindex mstrict-align
11756 Do not (do) assume that unaligned memory references will be handled by
11760 Generate code that allows the data segment to be located in a different
11761 area of memory from the text segment. This allows for execute in place in
11762 an environment without virtual memory management. This option implies
11765 @item -mno-sep-data
11766 Generate code that assumes that the data segment follows the text segment.
11767 This is the default.
11769 @item -mid-shared-library
11770 Generate code that supports shared libraries via the library ID method.
11771 This allows for execute in place and shared libraries in an environment
11772 without virtual memory management. This option implies @option{-fPIC}.
11774 @item -mno-id-shared-library
11775 Generate code that doesn't assume ID based shared libraries are being used.
11776 This is the default.
11778 @item -mshared-library-id=n
11779 Specified the identification number of the ID based shared library being
11780 compiled. Specifying a value of 0 will generate more compact code, specifying
11781 other values will force the allocation of that number to the current
11782 library but is no more space or time efficient than omitting this option.
11788 When generating position-independent code for ColdFire, generate code
11789 that works if the GOT has more than 8192 entries. This code is
11790 larger and slower than code generated without this option. On M680x0
11791 processors, this option is not needed; @option{-fPIC} suffices.
11793 GCC normally uses a single instruction to load values from the GOT@.
11794 While this is relatively efficient, it only works if the GOT
11795 is smaller than about 64k. Anything larger causes the linker
11796 to report an error such as:
11798 @cindex relocation truncated to fit (ColdFire)
11800 relocation truncated to fit: R_68K_GOT16O foobar
11803 If this happens, you should recompile your code with @option{-mxgot}.
11804 It should then work with very large GOTs. However, code generated with
11805 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
11806 the value of a global symbol.
11808 Note that some linkers, including newer versions of the GNU linker,
11809 can create multiple GOTs and sort GOT entries. If you have such a linker,
11810 you should only need to use @option{-mxgot} when compiling a single
11811 object file that accesses more than 8192 GOT entries. Very few do.
11813 These options have no effect unless GCC is generating
11814 position-independent code.
11818 @node M68hc1x Options
11819 @subsection M68hc1x Options
11820 @cindex M68hc1x options
11822 These are the @samp{-m} options defined for the 68hc11 and 68hc12
11823 microcontrollers. The default values for these options depends on
11824 which style of microcontroller was selected when the compiler was configured;
11825 the defaults for the most common choices are given below.
11832 Generate output for a 68HC11. This is the default
11833 when the compiler is configured for 68HC11-based systems.
11839 Generate output for a 68HC12. This is the default
11840 when the compiler is configured for 68HC12-based systems.
11846 Generate output for a 68HCS12.
11848 @item -mauto-incdec
11849 @opindex mauto-incdec
11850 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
11857 Enable the use of 68HC12 min and max instructions.
11860 @itemx -mno-long-calls
11861 @opindex mlong-calls
11862 @opindex mno-long-calls
11863 Treat all calls as being far away (near). If calls are assumed to be
11864 far away, the compiler will use the @code{call} instruction to
11865 call a function and the @code{rtc} instruction for returning.
11869 Consider type @code{int} to be 16 bits wide, like @code{short int}.
11871 @item -msoft-reg-count=@var{count}
11872 @opindex msoft-reg-count
11873 Specify the number of pseudo-soft registers which are used for the
11874 code generation. The maximum number is 32. Using more pseudo-soft
11875 register may or may not result in better code depending on the program.
11876 The default is 4 for 68HC11 and 2 for 68HC12.
11880 @node MCore Options
11881 @subsection MCore Options
11882 @cindex MCore options
11884 These are the @samp{-m} options defined for the Motorola M*Core
11890 @itemx -mno-hardlit
11892 @opindex mno-hardlit
11893 Inline constants into the code stream if it can be done in two
11894 instructions or less.
11900 Use the divide instruction. (Enabled by default).
11902 @item -mrelax-immediate
11903 @itemx -mno-relax-immediate
11904 @opindex mrelax-immediate
11905 @opindex mno-relax-immediate
11906 Allow arbitrary sized immediates in bit operations.
11908 @item -mwide-bitfields
11909 @itemx -mno-wide-bitfields
11910 @opindex mwide-bitfields
11911 @opindex mno-wide-bitfields
11912 Always treat bit-fields as int-sized.
11914 @item -m4byte-functions
11915 @itemx -mno-4byte-functions
11916 @opindex m4byte-functions
11917 @opindex mno-4byte-functions
11918 Force all functions to be aligned to a four byte boundary.
11920 @item -mcallgraph-data
11921 @itemx -mno-callgraph-data
11922 @opindex mcallgraph-data
11923 @opindex mno-callgraph-data
11924 Emit callgraph information.
11927 @itemx -mno-slow-bytes
11928 @opindex mslow-bytes
11929 @opindex mno-slow-bytes
11930 Prefer word access when reading byte quantities.
11932 @item -mlittle-endian
11933 @itemx -mbig-endian
11934 @opindex mlittle-endian
11935 @opindex mbig-endian
11936 Generate code for a little endian target.
11942 Generate code for the 210 processor.
11946 @subsection MIPS Options
11947 @cindex MIPS options
11953 Generate big-endian code.
11957 Generate little-endian code. This is the default for @samp{mips*el-*-*}
11960 @item -march=@var{arch}
11962 Generate code that will run on @var{arch}, which can be the name of a
11963 generic MIPS ISA, or the name of a particular processor.
11965 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
11966 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
11967 The processor names are:
11968 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
11969 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
11970 @samp{5kc}, @samp{5kf},
11972 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
11973 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
11974 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
11975 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
11976 @samp{loongson2e}, @samp{loongson2f},
11979 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
11980 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
11981 @samp{rm7000}, @samp{rm9000},
11984 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
11985 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
11987 The special value @samp{from-abi} selects the
11988 most compatible architecture for the selected ABI (that is,
11989 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
11991 Native Linux/GNU toolchains also support the value @samp{native},
11992 which selects the best architecture option for the host processor.
11993 @option{-march=native} has no effect if GCC does not recognize
11996 In processor names, a final @samp{000} can be abbreviated as @samp{k}
11997 (for example, @samp{-march=r2k}). Prefixes are optional, and
11998 @samp{vr} may be written @samp{r}.
12000 Names of the form @samp{@var{n}f2_1} refer to processors with
12001 FPUs clocked at half the rate of the core, names of the form
12002 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
12003 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
12004 processors with FPUs clocked a ratio of 3:2 with respect to the core.
12005 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
12006 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
12007 accepted as synonyms for @samp{@var{n}f1_1}.
12009 GCC defines two macros based on the value of this option. The first
12010 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
12011 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
12012 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
12013 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
12014 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
12016 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
12017 above. In other words, it will have the full prefix and will not
12018 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
12019 the macro names the resolved architecture (either @samp{"mips1"} or
12020 @samp{"mips3"}). It names the default architecture when no
12021 @option{-march} option is given.
12023 @item -mtune=@var{arch}
12025 Optimize for @var{arch}. Among other things, this option controls
12026 the way instructions are scheduled, and the perceived cost of arithmetic
12027 operations. The list of @var{arch} values is the same as for
12030 When this option is not used, GCC will optimize for the processor
12031 specified by @option{-march}. By using @option{-march} and
12032 @option{-mtune} together, it is possible to generate code that will
12033 run on a family of processors, but optimize the code for one
12034 particular member of that family.
12036 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
12037 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
12038 @samp{-march} ones described above.
12042 Equivalent to @samp{-march=mips1}.
12046 Equivalent to @samp{-march=mips2}.
12050 Equivalent to @samp{-march=mips3}.
12054 Equivalent to @samp{-march=mips4}.
12058 Equivalent to @samp{-march=mips32}.
12062 Equivalent to @samp{-march=mips32r2}.
12066 Equivalent to @samp{-march=mips64}.
12071 @opindex mno-mips16
12072 Generate (do not generate) MIPS16 code. If GCC is targetting a
12073 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
12075 MIPS16 code generation can also be controlled on a per-function basis
12076 by means of @code{mips16} and @code{nomips16} attributes.
12077 @xref{Function Attributes}, for more information.
12079 @item -mflip-mips16
12080 @opindex mflip-mips16
12081 Generate MIPS16 code on alternating functions. This option is provided
12082 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
12083 not intended for ordinary use in compiling user code.
12085 @item -minterlink-mips16
12086 @itemx -mno-interlink-mips16
12087 @opindex minterlink-mips16
12088 @opindex mno-interlink-mips16
12089 Require (do not require) that non-MIPS16 code be link-compatible with
12092 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
12093 it must either use a call or an indirect jump. @option{-minterlink-mips16}
12094 therefore disables direct jumps unless GCC knows that the target of the
12095 jump is not MIPS16.
12107 Generate code for the given ABI@.
12109 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
12110 generates 64-bit code when you select a 64-bit architecture, but you
12111 can use @option{-mgp32} to get 32-bit code instead.
12113 For information about the O64 ABI, see
12114 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
12116 GCC supports a variant of the o32 ABI in which floating-point registers
12117 are 64 rather than 32 bits wide. You can select this combination with
12118 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
12119 and @samp{mfhc1} instructions and is therefore only supported for
12120 MIPS32R2 processors.
12122 The register assignments for arguments and return values remain the
12123 same, but each scalar value is passed in a single 64-bit register
12124 rather than a pair of 32-bit registers. For example, scalar
12125 floating-point values are returned in @samp{$f0} only, not a
12126 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
12127 remains the same, but all 64 bits are saved.
12130 @itemx -mno-abicalls
12132 @opindex mno-abicalls
12133 Generate (do not generate) code that is suitable for SVR4-style
12134 dynamic objects. @option{-mabicalls} is the default for SVR4-based
12139 Generate (do not generate) code that is fully position-independent,
12140 and that can therefore be linked into shared libraries. This option
12141 only affects @option{-mabicalls}.
12143 All @option{-mabicalls} code has traditionally been position-independent,
12144 regardless of options like @option{-fPIC} and @option{-fpic}. However,
12145 as an extension, the GNU toolchain allows executables to use absolute
12146 accesses for locally-binding symbols. It can also use shorter GP
12147 initialization sequences and generate direct calls to locally-defined
12148 functions. This mode is selected by @option{-mno-shared}.
12150 @option{-mno-shared} depends on binutils 2.16 or higher and generates
12151 objects that can only be linked by the GNU linker. However, the option
12152 does not affect the ABI of the final executable; it only affects the ABI
12153 of relocatable objects. Using @option{-mno-shared} will generally make
12154 executables both smaller and quicker.
12156 @option{-mshared} is the default.
12162 Lift (do not lift) the usual restrictions on the size of the global
12165 GCC normally uses a single instruction to load values from the GOT@.
12166 While this is relatively efficient, it will only work if the GOT
12167 is smaller than about 64k. Anything larger will cause the linker
12168 to report an error such as:
12170 @cindex relocation truncated to fit (MIPS)
12172 relocation truncated to fit: R_MIPS_GOT16 foobar
12175 If this happens, you should recompile your code with @option{-mxgot}.
12176 It should then work with very large GOTs, although it will also be
12177 less efficient, since it will take three instructions to fetch the
12178 value of a global symbol.
12180 Note that some linkers can create multiple GOTs. If you have such a
12181 linker, you should only need to use @option{-mxgot} when a single object
12182 file accesses more than 64k's worth of GOT entries. Very few do.
12184 These options have no effect unless GCC is generating position
12189 Assume that general-purpose registers are 32 bits wide.
12193 Assume that general-purpose registers are 64 bits wide.
12197 Assume that floating-point registers are 32 bits wide.
12201 Assume that floating-point registers are 64 bits wide.
12204 @opindex mhard-float
12205 Use floating-point coprocessor instructions.
12208 @opindex msoft-float
12209 Do not use floating-point coprocessor instructions. Implement
12210 floating-point calculations using library calls instead.
12212 @item -msingle-float
12213 @opindex msingle-float
12214 Assume that the floating-point coprocessor only supports single-precision
12217 @item -mdouble-float
12218 @opindex mdouble-float
12219 Assume that the floating-point coprocessor supports double-precision
12220 operations. This is the default.
12226 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
12227 implement atomic memory built-in functions. When neither option is
12228 specified, GCC will use the instructions if the target architecture
12231 @option{-mllsc} is useful if the runtime environment can emulate the
12232 instructions and @option{-mno-llsc} can be useful when compiling for
12233 nonstandard ISAs. You can make either option the default by
12234 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
12235 respectively. @option{--with-llsc} is the default for some
12236 configurations; see the installation documentation for details.
12242 Use (do not use) revision 1 of the MIPS DSP ASE@.
12243 @xref{MIPS DSP Built-in Functions}. This option defines the
12244 preprocessor macro @samp{__mips_dsp}. It also defines
12245 @samp{__mips_dsp_rev} to 1.
12251 Use (do not use) revision 2 of the MIPS DSP ASE@.
12252 @xref{MIPS DSP Built-in Functions}. This option defines the
12253 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
12254 It also defines @samp{__mips_dsp_rev} to 2.
12257 @itemx -mno-smartmips
12258 @opindex msmartmips
12259 @opindex mno-smartmips
12260 Use (do not use) the MIPS SmartMIPS ASE.
12262 @item -mpaired-single
12263 @itemx -mno-paired-single
12264 @opindex mpaired-single
12265 @opindex mno-paired-single
12266 Use (do not use) paired-single floating-point instructions.
12267 @xref{MIPS Paired-Single Support}. This option requires
12268 hardware floating-point support to be enabled.
12274 Use (do not use) MIPS Digital Media Extension instructions.
12275 This option can only be used when generating 64-bit code and requires
12276 hardware floating-point support to be enabled.
12281 @opindex mno-mips3d
12282 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
12283 The option @option{-mips3d} implies @option{-mpaired-single}.
12289 Use (do not use) MT Multithreading instructions.
12293 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
12294 an explanation of the default and the way that the pointer size is
12299 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
12301 The default size of @code{int}s, @code{long}s and pointers depends on
12302 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
12303 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
12304 32-bit @code{long}s. Pointers are the same size as @code{long}s,
12305 or the same size as integer registers, whichever is smaller.
12311 Assume (do not assume) that all symbols have 32-bit values, regardless
12312 of the selected ABI@. This option is useful in combination with
12313 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
12314 to generate shorter and faster references to symbolic addresses.
12318 Put definitions of externally-visible data in a small data section
12319 if that data is no bigger than @var{num} bytes. GCC can then access
12320 the data more efficiently; see @option{-mgpopt} for details.
12322 The default @option{-G} option depends on the configuration.
12324 @item -mlocal-sdata
12325 @itemx -mno-local-sdata
12326 @opindex mlocal-sdata
12327 @opindex mno-local-sdata
12328 Extend (do not extend) the @option{-G} behavior to local data too,
12329 such as to static variables in C@. @option{-mlocal-sdata} is the
12330 default for all configurations.
12332 If the linker complains that an application is using too much small data,
12333 you might want to try rebuilding the less performance-critical parts with
12334 @option{-mno-local-sdata}. You might also want to build large
12335 libraries with @option{-mno-local-sdata}, so that the libraries leave
12336 more room for the main program.
12338 @item -mextern-sdata
12339 @itemx -mno-extern-sdata
12340 @opindex mextern-sdata
12341 @opindex mno-extern-sdata
12342 Assume (do not assume) that externally-defined data will be in
12343 a small data section if that data is within the @option{-G} limit.
12344 @option{-mextern-sdata} is the default for all configurations.
12346 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
12347 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
12348 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
12349 is placed in a small data section. If @var{Var} is defined by another
12350 module, you must either compile that module with a high-enough
12351 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
12352 definition. If @var{Var} is common, you must link the application
12353 with a high-enough @option{-G} setting.
12355 The easiest way of satisfying these restrictions is to compile
12356 and link every module with the same @option{-G} option. However,
12357 you may wish to build a library that supports several different
12358 small data limits. You can do this by compiling the library with
12359 the highest supported @option{-G} setting and additionally using
12360 @option{-mno-extern-sdata} to stop the library from making assumptions
12361 about externally-defined data.
12367 Use (do not use) GP-relative accesses for symbols that are known to be
12368 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
12369 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
12372 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
12373 might not hold the value of @code{_gp}. For example, if the code is
12374 part of a library that might be used in a boot monitor, programs that
12375 call boot monitor routines will pass an unknown value in @code{$gp}.
12376 (In such situations, the boot monitor itself would usually be compiled
12377 with @option{-G0}.)
12379 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
12380 @option{-mno-extern-sdata}.
12382 @item -membedded-data
12383 @itemx -mno-embedded-data
12384 @opindex membedded-data
12385 @opindex mno-embedded-data
12386 Allocate variables to the read-only data section first if possible, then
12387 next in the small data section if possible, otherwise in data. This gives
12388 slightly slower code than the default, but reduces the amount of RAM required
12389 when executing, and thus may be preferred for some embedded systems.
12391 @item -muninit-const-in-rodata
12392 @itemx -mno-uninit-const-in-rodata
12393 @opindex muninit-const-in-rodata
12394 @opindex mno-uninit-const-in-rodata
12395 Put uninitialized @code{const} variables in the read-only data section.
12396 This option is only meaningful in conjunction with @option{-membedded-data}.
12398 @item -mcode-readable=@var{setting}
12399 @opindex mcode-readable
12400 Specify whether GCC may generate code that reads from executable sections.
12401 There are three possible settings:
12404 @item -mcode-readable=yes
12405 Instructions may freely access executable sections. This is the
12408 @item -mcode-readable=pcrel
12409 MIPS16 PC-relative load instructions can access executable sections,
12410 but other instructions must not do so. This option is useful on 4KSc
12411 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
12412 It is also useful on processors that can be configured to have a dual
12413 instruction/data SRAM interface and that, like the M4K, automatically
12414 redirect PC-relative loads to the instruction RAM.
12416 @item -mcode-readable=no
12417 Instructions must not access executable sections. This option can be
12418 useful on targets that are configured to have a dual instruction/data
12419 SRAM interface but that (unlike the M4K) do not automatically redirect
12420 PC-relative loads to the instruction RAM.
12423 @item -msplit-addresses
12424 @itemx -mno-split-addresses
12425 @opindex msplit-addresses
12426 @opindex mno-split-addresses
12427 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
12428 relocation operators. This option has been superseded by
12429 @option{-mexplicit-relocs} but is retained for backwards compatibility.
12431 @item -mexplicit-relocs
12432 @itemx -mno-explicit-relocs
12433 @opindex mexplicit-relocs
12434 @opindex mno-explicit-relocs
12435 Use (do not use) assembler relocation operators when dealing with symbolic
12436 addresses. The alternative, selected by @option{-mno-explicit-relocs},
12437 is to use assembler macros instead.
12439 @option{-mexplicit-relocs} is the default if GCC was configured
12440 to use an assembler that supports relocation operators.
12442 @item -mcheck-zero-division
12443 @itemx -mno-check-zero-division
12444 @opindex mcheck-zero-division
12445 @opindex mno-check-zero-division
12446 Trap (do not trap) on integer division by zero.
12448 The default is @option{-mcheck-zero-division}.
12450 @item -mdivide-traps
12451 @itemx -mdivide-breaks
12452 @opindex mdivide-traps
12453 @opindex mdivide-breaks
12454 MIPS systems check for division by zero by generating either a
12455 conditional trap or a break instruction. Using traps results in
12456 smaller code, but is only supported on MIPS II and later. Also, some
12457 versions of the Linux kernel have a bug that prevents trap from
12458 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
12459 allow conditional traps on architectures that support them and
12460 @option{-mdivide-breaks} to force the use of breaks.
12462 The default is usually @option{-mdivide-traps}, but this can be
12463 overridden at configure time using @option{--with-divide=breaks}.
12464 Divide-by-zero checks can be completely disabled using
12465 @option{-mno-check-zero-division}.
12470 @opindex mno-memcpy
12471 Force (do not force) the use of @code{memcpy()} for non-trivial block
12472 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
12473 most constant-sized copies.
12476 @itemx -mno-long-calls
12477 @opindex mlong-calls
12478 @opindex mno-long-calls
12479 Disable (do not disable) use of the @code{jal} instruction. Calling
12480 functions using @code{jal} is more efficient but requires the caller
12481 and callee to be in the same 256 megabyte segment.
12483 This option has no effect on abicalls code. The default is
12484 @option{-mno-long-calls}.
12490 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
12491 instructions, as provided by the R4650 ISA@.
12494 @itemx -mno-fused-madd
12495 @opindex mfused-madd
12496 @opindex mno-fused-madd
12497 Enable (disable) use of the floating point multiply-accumulate
12498 instructions, when they are available. The default is
12499 @option{-mfused-madd}.
12501 When multiply-accumulate instructions are used, the intermediate
12502 product is calculated to infinite precision and is not subject to
12503 the FCSR Flush to Zero bit. This may be undesirable in some
12508 Tell the MIPS assembler to not run its preprocessor over user
12509 assembler files (with a @samp{.s} suffix) when assembling them.
12512 @itemx -mno-fix-r4000
12513 @opindex mfix-r4000
12514 @opindex mno-fix-r4000
12515 Work around certain R4000 CPU errata:
12518 A double-word or a variable shift may give an incorrect result if executed
12519 immediately after starting an integer division.
12521 A double-word or a variable shift may give an incorrect result if executed
12522 while an integer multiplication is in progress.
12524 An integer division may give an incorrect result if started in a delay slot
12525 of a taken branch or a jump.
12529 @itemx -mno-fix-r4400
12530 @opindex mfix-r4400
12531 @opindex mno-fix-r4400
12532 Work around certain R4400 CPU errata:
12535 A double-word or a variable shift may give an incorrect result if executed
12536 immediately after starting an integer division.
12540 @itemx -mno-fix-vr4120
12541 @opindex mfix-vr4120
12542 Work around certain VR4120 errata:
12545 @code{dmultu} does not always produce the correct result.
12547 @code{div} and @code{ddiv} do not always produce the correct result if one
12548 of the operands is negative.
12550 The workarounds for the division errata rely on special functions in
12551 @file{libgcc.a}. At present, these functions are only provided by
12552 the @code{mips64vr*-elf} configurations.
12554 Other VR4120 errata require a nop to be inserted between certain pairs of
12555 instructions. These errata are handled by the assembler, not by GCC itself.
12558 @opindex mfix-vr4130
12559 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
12560 workarounds are implemented by the assembler rather than by GCC,
12561 although GCC will avoid using @code{mflo} and @code{mfhi} if the
12562 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
12563 instructions are available instead.
12566 @itemx -mno-fix-sb1
12568 Work around certain SB-1 CPU core errata.
12569 (This flag currently works around the SB-1 revision 2
12570 ``F1'' and ``F2'' floating point errata.)
12572 @item -mflush-func=@var{func}
12573 @itemx -mno-flush-func
12574 @opindex mflush-func
12575 Specifies the function to call to flush the I and D caches, or to not
12576 call any such function. If called, the function must take the same
12577 arguments as the common @code{_flush_func()}, that is, the address of the
12578 memory range for which the cache is being flushed, the size of the
12579 memory range, and the number 3 (to flush both caches). The default
12580 depends on the target GCC was configured for, but commonly is either
12581 @samp{_flush_func} or @samp{__cpu_flush}.
12583 @item mbranch-cost=@var{num}
12584 @opindex mbranch-cost
12585 Set the cost of branches to roughly @var{num} ``simple'' instructions.
12586 This cost is only a heuristic and is not guaranteed to produce
12587 consistent results across releases. A zero cost redundantly selects
12588 the default, which is based on the @option{-mtune} setting.
12590 @item -mbranch-likely
12591 @itemx -mno-branch-likely
12592 @opindex mbranch-likely
12593 @opindex mno-branch-likely
12594 Enable or disable use of Branch Likely instructions, regardless of the
12595 default for the selected architecture. By default, Branch Likely
12596 instructions may be generated if they are supported by the selected
12597 architecture. An exception is for the MIPS32 and MIPS64 architectures
12598 and processors which implement those architectures; for those, Branch
12599 Likely instructions will not be generated by default because the MIPS32
12600 and MIPS64 architectures specifically deprecate their use.
12602 @item -mfp-exceptions
12603 @itemx -mno-fp-exceptions
12604 @opindex mfp-exceptions
12605 Specifies whether FP exceptions are enabled. This affects how we schedule
12606 FP instructions for some processors. The default is that FP exceptions are
12609 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
12610 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
12613 @item -mvr4130-align
12614 @itemx -mno-vr4130-align
12615 @opindex mvr4130-align
12616 The VR4130 pipeline is two-way superscalar, but can only issue two
12617 instructions together if the first one is 8-byte aligned. When this
12618 option is enabled, GCC will align pairs of instructions that it
12619 thinks should execute in parallel.
12621 This option only has an effect when optimizing for the VR4130.
12622 It normally makes code faster, but at the expense of making it bigger.
12623 It is enabled by default at optimization level @option{-O3}.
12627 @subsection MMIX Options
12628 @cindex MMIX Options
12630 These options are defined for the MMIX:
12634 @itemx -mno-libfuncs
12636 @opindex mno-libfuncs
12637 Specify that intrinsic library functions are being compiled, passing all
12638 values in registers, no matter the size.
12641 @itemx -mno-epsilon
12643 @opindex mno-epsilon
12644 Generate floating-point comparison instructions that compare with respect
12645 to the @code{rE} epsilon register.
12647 @item -mabi=mmixware
12649 @opindex mabi-mmixware
12651 Generate code that passes function parameters and return values that (in
12652 the called function) are seen as registers @code{$0} and up, as opposed to
12653 the GNU ABI which uses global registers @code{$231} and up.
12655 @item -mzero-extend
12656 @itemx -mno-zero-extend
12657 @opindex mzero-extend
12658 @opindex mno-zero-extend
12659 When reading data from memory in sizes shorter than 64 bits, use (do not
12660 use) zero-extending load instructions by default, rather than
12661 sign-extending ones.
12664 @itemx -mno-knuthdiv
12666 @opindex mno-knuthdiv
12667 Make the result of a division yielding a remainder have the same sign as
12668 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
12669 remainder follows the sign of the dividend. Both methods are
12670 arithmetically valid, the latter being almost exclusively used.
12672 @item -mtoplevel-symbols
12673 @itemx -mno-toplevel-symbols
12674 @opindex mtoplevel-symbols
12675 @opindex mno-toplevel-symbols
12676 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
12677 code can be used with the @code{PREFIX} assembly directive.
12681 Generate an executable in the ELF format, rather than the default
12682 @samp{mmo} format used by the @command{mmix} simulator.
12684 @item -mbranch-predict
12685 @itemx -mno-branch-predict
12686 @opindex mbranch-predict
12687 @opindex mno-branch-predict
12688 Use (do not use) the probable-branch instructions, when static branch
12689 prediction indicates a probable branch.
12691 @item -mbase-addresses
12692 @itemx -mno-base-addresses
12693 @opindex mbase-addresses
12694 @opindex mno-base-addresses
12695 Generate (do not generate) code that uses @emph{base addresses}. Using a
12696 base address automatically generates a request (handled by the assembler
12697 and the linker) for a constant to be set up in a global register. The
12698 register is used for one or more base address requests within the range 0
12699 to 255 from the value held in the register. The generally leads to short
12700 and fast code, but the number of different data items that can be
12701 addressed is limited. This means that a program that uses lots of static
12702 data may require @option{-mno-base-addresses}.
12704 @item -msingle-exit
12705 @itemx -mno-single-exit
12706 @opindex msingle-exit
12707 @opindex mno-single-exit
12708 Force (do not force) generated code to have a single exit point in each
12712 @node MN10300 Options
12713 @subsection MN10300 Options
12714 @cindex MN10300 options
12716 These @option{-m} options are defined for Matsushita MN10300 architectures:
12721 Generate code to avoid bugs in the multiply instructions for the MN10300
12722 processors. This is the default.
12724 @item -mno-mult-bug
12725 @opindex mno-mult-bug
12726 Do not generate code to avoid bugs in the multiply instructions for the
12727 MN10300 processors.
12731 Generate code which uses features specific to the AM33 processor.
12735 Do not generate code which uses features specific to the AM33 processor. This
12738 @item -mreturn-pointer-on-d0
12739 @opindex mreturn-pointer-on-d0
12740 When generating a function which returns a pointer, return the pointer
12741 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
12742 only in a0, and attempts to call such functions without a prototype
12743 would result in errors. Note that this option is on by default; use
12744 @option{-mno-return-pointer-on-d0} to disable it.
12748 Do not link in the C run-time initialization object file.
12752 Indicate to the linker that it should perform a relaxation optimization pass
12753 to shorten branches, calls and absolute memory addresses. This option only
12754 has an effect when used on the command line for the final link step.
12756 This option makes symbolic debugging impossible.
12759 @node PDP-11 Options
12760 @subsection PDP-11 Options
12761 @cindex PDP-11 Options
12763 These options are defined for the PDP-11:
12768 Use hardware FPP floating point. This is the default. (FIS floating
12769 point on the PDP-11/40 is not supported.)
12772 @opindex msoft-float
12773 Do not use hardware floating point.
12777 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
12781 Return floating-point results in memory. This is the default.
12785 Generate code for a PDP-11/40.
12789 Generate code for a PDP-11/45. This is the default.
12793 Generate code for a PDP-11/10.
12795 @item -mbcopy-builtin
12796 @opindex bcopy-builtin
12797 Use inline @code{movmemhi} patterns for copying memory. This is the
12802 Do not use inline @code{movmemhi} patterns for copying memory.
12808 Use 16-bit @code{int}. This is the default.
12814 Use 32-bit @code{int}.
12817 @itemx -mno-float32
12819 @opindex mno-float32
12820 Use 64-bit @code{float}. This is the default.
12823 @itemx -mno-float64
12825 @opindex mno-float64
12826 Use 32-bit @code{float}.
12830 Use @code{abshi2} pattern. This is the default.
12834 Do not use @code{abshi2} pattern.
12836 @item -mbranch-expensive
12837 @opindex mbranch-expensive
12838 Pretend that branches are expensive. This is for experimenting with
12839 code generation only.
12841 @item -mbranch-cheap
12842 @opindex mbranch-cheap
12843 Do not pretend that branches are expensive. This is the default.
12847 Generate code for a system with split I&D@.
12851 Generate code for a system without split I&D@. This is the default.
12855 Use Unix assembler syntax. This is the default when configured for
12856 @samp{pdp11-*-bsd}.
12860 Use DEC assembler syntax. This is the default when configured for any
12861 PDP-11 target other than @samp{pdp11-*-bsd}.
12864 @node PowerPC Options
12865 @subsection PowerPC Options
12866 @cindex PowerPC options
12868 These are listed under @xref{RS/6000 and PowerPC Options}.
12870 @node RS/6000 and PowerPC Options
12871 @subsection IBM RS/6000 and PowerPC Options
12872 @cindex RS/6000 and PowerPC Options
12873 @cindex IBM RS/6000 and PowerPC Options
12875 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
12882 @itemx -mno-powerpc
12883 @itemx -mpowerpc-gpopt
12884 @itemx -mno-powerpc-gpopt
12885 @itemx -mpowerpc-gfxopt
12886 @itemx -mno-powerpc-gfxopt
12888 @itemx -mno-powerpc64
12892 @itemx -mno-popcntb
12900 @itemx -mno-hard-dfp
12904 @opindex mno-power2
12906 @opindex mno-powerpc
12907 @opindex mpowerpc-gpopt
12908 @opindex mno-powerpc-gpopt
12909 @opindex mpowerpc-gfxopt
12910 @opindex mno-powerpc-gfxopt
12911 @opindex mpowerpc64
12912 @opindex mno-powerpc64
12916 @opindex mno-popcntb
12922 @opindex mno-mfpgpr
12924 @opindex mno-hard-dfp
12925 GCC supports two related instruction set architectures for the
12926 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
12927 instructions supported by the @samp{rios} chip set used in the original
12928 RS/6000 systems and the @dfn{PowerPC} instruction set is the
12929 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
12930 the IBM 4xx, 6xx, and follow-on microprocessors.
12932 Neither architecture is a subset of the other. However there is a
12933 large common subset of instructions supported by both. An MQ
12934 register is included in processors supporting the POWER architecture.
12936 You use these options to specify which instructions are available on the
12937 processor you are using. The default value of these options is
12938 determined when configuring GCC@. Specifying the
12939 @option{-mcpu=@var{cpu_type}} overrides the specification of these
12940 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
12941 rather than the options listed above.
12943 The @option{-mpower} option allows GCC to generate instructions that
12944 are found only in the POWER architecture and to use the MQ register.
12945 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
12946 to generate instructions that are present in the POWER2 architecture but
12947 not the original POWER architecture.
12949 The @option{-mpowerpc} option allows GCC to generate instructions that
12950 are found only in the 32-bit subset of the PowerPC architecture.
12951 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
12952 GCC to use the optional PowerPC architecture instructions in the
12953 General Purpose group, including floating-point square root. Specifying
12954 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
12955 use the optional PowerPC architecture instructions in the Graphics
12956 group, including floating-point select.
12958 The @option{-mmfcrf} option allows GCC to generate the move from
12959 condition register field instruction implemented on the POWER4
12960 processor and other processors that support the PowerPC V2.01
12962 The @option{-mpopcntb} option allows GCC to generate the popcount and
12963 double precision FP reciprocal estimate instruction implemented on the
12964 POWER5 processor and other processors that support the PowerPC V2.02
12966 The @option{-mfprnd} option allows GCC to generate the FP round to
12967 integer instructions implemented on the POWER5+ processor and other
12968 processors that support the PowerPC V2.03 architecture.
12969 The @option{-mcmpb} option allows GCC to generate the compare bytes
12970 instruction implemented on the POWER6 processor and other processors
12971 that support the PowerPC V2.05 architecture.
12972 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
12973 general purpose register instructions implemented on the POWER6X
12974 processor and other processors that support the extended PowerPC V2.05
12976 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
12977 point instructions implemented on some POWER processors.
12979 The @option{-mpowerpc64} option allows GCC to generate the additional
12980 64-bit instructions that are found in the full PowerPC64 architecture
12981 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
12982 @option{-mno-powerpc64}.
12984 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
12985 will use only the instructions in the common subset of both
12986 architectures plus some special AIX common-mode calls, and will not use
12987 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
12988 permits GCC to use any instruction from either architecture and to
12989 allow use of the MQ register; specify this for the Motorola MPC601.
12991 @item -mnew-mnemonics
12992 @itemx -mold-mnemonics
12993 @opindex mnew-mnemonics
12994 @opindex mold-mnemonics
12995 Select which mnemonics to use in the generated assembler code. With
12996 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
12997 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
12998 assembler mnemonics defined for the POWER architecture. Instructions
12999 defined in only one architecture have only one mnemonic; GCC uses that
13000 mnemonic irrespective of which of these options is specified.
13002 GCC defaults to the mnemonics appropriate for the architecture in
13003 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
13004 value of these option. Unless you are building a cross-compiler, you
13005 should normally not specify either @option{-mnew-mnemonics} or
13006 @option{-mold-mnemonics}, but should instead accept the default.
13008 @item -mcpu=@var{cpu_type}
13010 Set architecture type, register usage, choice of mnemonics, and
13011 instruction scheduling parameters for machine type @var{cpu_type}.
13012 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
13013 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
13014 @samp{505}, @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
13015 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
13016 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
13017 @samp{860}, @samp{970}, @samp{8540}, @samp{e300c2}, @samp{e300c3},
13018 @samp{e500mc}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
13019 @samp{power}, @samp{power2}, @samp{power3}, @samp{power4},
13020 @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x},
13021 @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
13022 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
13024 @option{-mcpu=common} selects a completely generic processor. Code
13025 generated under this option will run on any POWER or PowerPC processor.
13026 GCC will use only the instructions in the common subset of both
13027 architectures, and will not use the MQ register. GCC assumes a generic
13028 processor model for scheduling purposes.
13030 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
13031 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
13032 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
13033 types, with an appropriate, generic processor model assumed for
13034 scheduling purposes.
13036 The other options specify a specific processor. Code generated under
13037 those options will run best on that processor, and may not run at all on
13040 The @option{-mcpu} options automatically enable or disable the
13043 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
13044 -mnew-mnemonics -mpopcntb -mpower -mpower2 -mpowerpc64 @gol
13045 -mpowerpc-gpopt -mpowerpc-gfxopt -mstring -mmulhw -mdlmzb -mmfpgpr}
13047 The particular options set for any particular CPU will vary between
13048 compiler versions, depending on what setting seems to produce optimal
13049 code for that CPU; it doesn't necessarily reflect the actual hardware's
13050 capabilities. If you wish to set an individual option to a particular
13051 value, you may specify it after the @option{-mcpu} option, like
13052 @samp{-mcpu=970 -mno-altivec}.
13054 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
13055 not enabled or disabled by the @option{-mcpu} option at present because
13056 AIX does not have full support for these options. You may still
13057 enable or disable them individually if you're sure it'll work in your
13060 @item -mtune=@var{cpu_type}
13062 Set the instruction scheduling parameters for machine type
13063 @var{cpu_type}, but do not set the architecture type, register usage, or
13064 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
13065 values for @var{cpu_type} are used for @option{-mtune} as for
13066 @option{-mcpu}. If both are specified, the code generated will use the
13067 architecture, registers, and mnemonics set by @option{-mcpu}, but the
13068 scheduling parameters set by @option{-mtune}.
13074 Generate code to compute division as reciprocal estimate and iterative
13075 refinement, creating opportunities for increased throughput. This
13076 feature requires: optional PowerPC Graphics instruction set for single
13077 precision and FRE instruction for double precision, assuming divides
13078 cannot generate user-visible traps, and the domain values not include
13079 Infinities, denormals or zero denominator.
13082 @itemx -mno-altivec
13084 @opindex mno-altivec
13085 Generate code that uses (does not use) AltiVec instructions, and also
13086 enable the use of built-in functions that allow more direct access to
13087 the AltiVec instruction set. You may also need to set
13088 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
13094 @opindex mno-vrsave
13095 Generate VRSAVE instructions when generating AltiVec code.
13098 @opindex msecure-plt
13099 Generate code that allows ld and ld.so to build executables and shared
13100 libraries with non-exec .plt and .got sections. This is a PowerPC
13101 32-bit SYSV ABI option.
13105 Generate code that uses a BSS .plt section that ld.so fills in, and
13106 requires .plt and .got sections that are both writable and executable.
13107 This is a PowerPC 32-bit SYSV ABI option.
13113 This switch enables or disables the generation of ISEL instructions.
13115 @item -misel=@var{yes/no}
13116 This switch has been deprecated. Use @option{-misel} and
13117 @option{-mno-isel} instead.
13123 This switch enables or disables the generation of SPE simd
13129 @opindex mno-paired
13130 This switch enables or disables the generation of PAIRED simd
13133 @item -mspe=@var{yes/no}
13134 This option has been deprecated. Use @option{-mspe} and
13135 @option{-mno-spe} instead.
13137 @item -mfloat-gprs=@var{yes/single/double/no}
13138 @itemx -mfloat-gprs
13139 @opindex mfloat-gprs
13140 This switch enables or disables the generation of floating point
13141 operations on the general purpose registers for architectures that
13144 The argument @var{yes} or @var{single} enables the use of
13145 single-precision floating point operations.
13147 The argument @var{double} enables the use of single and
13148 double-precision floating point operations.
13150 The argument @var{no} disables floating point operations on the
13151 general purpose registers.
13153 This option is currently only available on the MPC854x.
13159 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
13160 targets (including GNU/Linux). The 32-bit environment sets int, long
13161 and pointer to 32 bits and generates code that runs on any PowerPC
13162 variant. The 64-bit environment sets int to 32 bits and long and
13163 pointer to 64 bits, and generates code for PowerPC64, as for
13164 @option{-mpowerpc64}.
13167 @itemx -mno-fp-in-toc
13168 @itemx -mno-sum-in-toc
13169 @itemx -mminimal-toc
13171 @opindex mno-fp-in-toc
13172 @opindex mno-sum-in-toc
13173 @opindex mminimal-toc
13174 Modify generation of the TOC (Table Of Contents), which is created for
13175 every executable file. The @option{-mfull-toc} option is selected by
13176 default. In that case, GCC will allocate at least one TOC entry for
13177 each unique non-automatic variable reference in your program. GCC
13178 will also place floating-point constants in the TOC@. However, only
13179 16,384 entries are available in the TOC@.
13181 If you receive a linker error message that saying you have overflowed
13182 the available TOC space, you can reduce the amount of TOC space used
13183 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
13184 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
13185 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
13186 generate code to calculate the sum of an address and a constant at
13187 run-time instead of putting that sum into the TOC@. You may specify one
13188 or both of these options. Each causes GCC to produce very slightly
13189 slower and larger code at the expense of conserving TOC space.
13191 If you still run out of space in the TOC even when you specify both of
13192 these options, specify @option{-mminimal-toc} instead. This option causes
13193 GCC to make only one TOC entry for every file. When you specify this
13194 option, GCC will produce code that is slower and larger but which
13195 uses extremely little TOC space. You may wish to use this option
13196 only on files that contain less frequently executed code.
13202 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
13203 @code{long} type, and the infrastructure needed to support them.
13204 Specifying @option{-maix64} implies @option{-mpowerpc64} and
13205 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
13206 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
13209 @itemx -mno-xl-compat
13210 @opindex mxl-compat
13211 @opindex mno-xl-compat
13212 Produce code that conforms more closely to IBM XL compiler semantics
13213 when using AIX-compatible ABI@. Pass floating-point arguments to
13214 prototyped functions beyond the register save area (RSA) on the stack
13215 in addition to argument FPRs. Do not assume that most significant
13216 double in 128-bit long double value is properly rounded when comparing
13217 values and converting to double. Use XL symbol names for long double
13220 The AIX calling convention was extended but not initially documented to
13221 handle an obscure K&R C case of calling a function that takes the
13222 address of its arguments with fewer arguments than declared. IBM XL
13223 compilers access floating point arguments which do not fit in the
13224 RSA from the stack when a subroutine is compiled without
13225 optimization. Because always storing floating-point arguments on the
13226 stack is inefficient and rarely needed, this option is not enabled by
13227 default and only is necessary when calling subroutines compiled by IBM
13228 XL compilers without optimization.
13232 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
13233 application written to use message passing with special startup code to
13234 enable the application to run. The system must have PE installed in the
13235 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
13236 must be overridden with the @option{-specs=} option to specify the
13237 appropriate directory location. The Parallel Environment does not
13238 support threads, so the @option{-mpe} option and the @option{-pthread}
13239 option are incompatible.
13241 @item -malign-natural
13242 @itemx -malign-power
13243 @opindex malign-natural
13244 @opindex malign-power
13245 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
13246 @option{-malign-natural} overrides the ABI-defined alignment of larger
13247 types, such as floating-point doubles, on their natural size-based boundary.
13248 The option @option{-malign-power} instructs GCC to follow the ABI-specified
13249 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
13251 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
13255 @itemx -mhard-float
13256 @opindex msoft-float
13257 @opindex mhard-float
13258 Generate code that does not use (uses) the floating-point register set.
13259 Software floating point emulation is provided if you use the
13260 @option{-msoft-float} option, and pass the option to GCC when linking.
13263 @itemx -mno-multiple
13265 @opindex mno-multiple
13266 Generate code that uses (does not use) the load multiple word
13267 instructions and the store multiple word instructions. These
13268 instructions are generated by default on POWER systems, and not
13269 generated on PowerPC systems. Do not use @option{-mmultiple} on little
13270 endian PowerPC systems, since those instructions do not work when the
13271 processor is in little endian mode. The exceptions are PPC740 and
13272 PPC750 which permit the instructions usage in little endian mode.
13277 @opindex mno-string
13278 Generate code that uses (does not use) the load string instructions
13279 and the store string word instructions to save multiple registers and
13280 do small block moves. These instructions are generated by default on
13281 POWER systems, and not generated on PowerPC systems. Do not use
13282 @option{-mstring} on little endian PowerPC systems, since those
13283 instructions do not work when the processor is in little endian mode.
13284 The exceptions are PPC740 and PPC750 which permit the instructions
13285 usage in little endian mode.
13290 @opindex mno-update
13291 Generate code that uses (does not use) the load or store instructions
13292 that update the base register to the address of the calculated memory
13293 location. These instructions are generated by default. If you use
13294 @option{-mno-update}, there is a small window between the time that the
13295 stack pointer is updated and the address of the previous frame is
13296 stored, which means code that walks the stack frame across interrupts or
13297 signals may get corrupted data.
13300 @itemx -mno-fused-madd
13301 @opindex mfused-madd
13302 @opindex mno-fused-madd
13303 Generate code that uses (does not use) the floating point multiply and
13304 accumulate instructions. These instructions are generated by default if
13305 hardware floating is used.
13311 Generate code that uses (does not use) the half-word multiply and
13312 multiply-accumulate instructions on the IBM 405, 440 and 464 processors.
13313 These instructions are generated by default when targetting those
13320 Generate code that uses (does not use) the string-search @samp{dlmzb}
13321 instruction on the IBM 405, 440 and 464 processors. This instruction is
13322 generated by default when targetting those processors.
13324 @item -mno-bit-align
13326 @opindex mno-bit-align
13327 @opindex mbit-align
13328 On System V.4 and embedded PowerPC systems do not (do) force structures
13329 and unions that contain bit-fields to be aligned to the base type of the
13332 For example, by default a structure containing nothing but 8
13333 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
13334 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
13335 the structure would be aligned to a 1 byte boundary and be one byte in
13338 @item -mno-strict-align
13339 @itemx -mstrict-align
13340 @opindex mno-strict-align
13341 @opindex mstrict-align
13342 On System V.4 and embedded PowerPC systems do not (do) assume that
13343 unaligned memory references will be handled by the system.
13345 @item -mrelocatable
13346 @itemx -mno-relocatable
13347 @opindex mrelocatable
13348 @opindex mno-relocatable
13349 On embedded PowerPC systems generate code that allows (does not allow)
13350 the program to be relocated to a different address at runtime. If you
13351 use @option{-mrelocatable} on any module, all objects linked together must
13352 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
13354 @item -mrelocatable-lib
13355 @itemx -mno-relocatable-lib
13356 @opindex mrelocatable-lib
13357 @opindex mno-relocatable-lib
13358 On embedded PowerPC systems generate code that allows (does not allow)
13359 the program to be relocated to a different address at runtime. Modules
13360 compiled with @option{-mrelocatable-lib} can be linked with either modules
13361 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
13362 with modules compiled with the @option{-mrelocatable} options.
13368 On System V.4 and embedded PowerPC systems do not (do) assume that
13369 register 2 contains a pointer to a global area pointing to the addresses
13370 used in the program.
13373 @itemx -mlittle-endian
13375 @opindex mlittle-endian
13376 On System V.4 and embedded PowerPC systems compile code for the
13377 processor in little endian mode. The @option{-mlittle-endian} option is
13378 the same as @option{-mlittle}.
13381 @itemx -mbig-endian
13383 @opindex mbig-endian
13384 On System V.4 and embedded PowerPC systems compile code for the
13385 processor in big endian mode. The @option{-mbig-endian} option is
13386 the same as @option{-mbig}.
13388 @item -mdynamic-no-pic
13389 @opindex mdynamic-no-pic
13390 On Darwin and Mac OS X systems, compile code so that it is not
13391 relocatable, but that its external references are relocatable. The
13392 resulting code is suitable for applications, but not shared
13395 @item -mprioritize-restricted-insns=@var{priority}
13396 @opindex mprioritize-restricted-insns
13397 This option controls the priority that is assigned to
13398 dispatch-slot restricted instructions during the second scheduling
13399 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
13400 @var{no/highest/second-highest} priority to dispatch slot restricted
13403 @item -msched-costly-dep=@var{dependence_type}
13404 @opindex msched-costly-dep
13405 This option controls which dependences are considered costly
13406 by the target during instruction scheduling. The argument
13407 @var{dependence_type} takes one of the following values:
13408 @var{no}: no dependence is costly,
13409 @var{all}: all dependences are costly,
13410 @var{true_store_to_load}: a true dependence from store to load is costly,
13411 @var{store_to_load}: any dependence from store to load is costly,
13412 @var{number}: any dependence which latency >= @var{number} is costly.
13414 @item -minsert-sched-nops=@var{scheme}
13415 @opindex minsert-sched-nops
13416 This option controls which nop insertion scheme will be used during
13417 the second scheduling pass. The argument @var{scheme} takes one of the
13419 @var{no}: Don't insert nops.
13420 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
13421 according to the scheduler's grouping.
13422 @var{regroup_exact}: Insert nops to force costly dependent insns into
13423 separate groups. Insert exactly as many nops as needed to force an insn
13424 to a new group, according to the estimated processor grouping.
13425 @var{number}: Insert nops to force costly dependent insns into
13426 separate groups. Insert @var{number} nops to force an insn to a new group.
13429 @opindex mcall-sysv
13430 On System V.4 and embedded PowerPC systems compile code using calling
13431 conventions that adheres to the March 1995 draft of the System V
13432 Application Binary Interface, PowerPC processor supplement. This is the
13433 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
13435 @item -mcall-sysv-eabi
13436 @opindex mcall-sysv-eabi
13437 Specify both @option{-mcall-sysv} and @option{-meabi} options.
13439 @item -mcall-sysv-noeabi
13440 @opindex mcall-sysv-noeabi
13441 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
13443 @item -mcall-solaris
13444 @opindex mcall-solaris
13445 On System V.4 and embedded PowerPC systems compile code for the Solaris
13449 @opindex mcall-linux
13450 On System V.4 and embedded PowerPC systems compile code for the
13451 Linux-based GNU system.
13455 On System V.4 and embedded PowerPC systems compile code for the
13456 Hurd-based GNU system.
13458 @item -mcall-netbsd
13459 @opindex mcall-netbsd
13460 On System V.4 and embedded PowerPC systems compile code for the
13461 NetBSD operating system.
13463 @item -maix-struct-return
13464 @opindex maix-struct-return
13465 Return all structures in memory (as specified by the AIX ABI)@.
13467 @item -msvr4-struct-return
13468 @opindex msvr4-struct-return
13469 Return structures smaller than 8 bytes in registers (as specified by the
13472 @item -mabi=@var{abi-type}
13474 Extend the current ABI with a particular extension, or remove such extension.
13475 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
13476 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
13480 Extend the current ABI with SPE ABI extensions. This does not change
13481 the default ABI, instead it adds the SPE ABI extensions to the current
13485 @opindex mabi=no-spe
13486 Disable Booke SPE ABI extensions for the current ABI@.
13488 @item -mabi=ibmlongdouble
13489 @opindex mabi=ibmlongdouble
13490 Change the current ABI to use IBM extended precision long double.
13491 This is a PowerPC 32-bit SYSV ABI option.
13493 @item -mabi=ieeelongdouble
13494 @opindex mabi=ieeelongdouble
13495 Change the current ABI to use IEEE extended precision long double.
13496 This is a PowerPC 32-bit Linux ABI option.
13499 @itemx -mno-prototype
13500 @opindex mprototype
13501 @opindex mno-prototype
13502 On System V.4 and embedded PowerPC systems assume that all calls to
13503 variable argument functions are properly prototyped. Otherwise, the
13504 compiler must insert an instruction before every non prototyped call to
13505 set or clear bit 6 of the condition code register (@var{CR}) to
13506 indicate whether floating point values were passed in the floating point
13507 registers in case the function takes a variable arguments. With
13508 @option{-mprototype}, only calls to prototyped variable argument functions
13509 will set or clear the bit.
13513 On embedded PowerPC systems, assume that the startup module is called
13514 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
13515 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
13520 On embedded PowerPC systems, assume that the startup module is called
13521 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
13526 On embedded PowerPC systems, assume that the startup module is called
13527 @file{crt0.o} and the standard C libraries are @file{libads.a} and
13530 @item -myellowknife
13531 @opindex myellowknife
13532 On embedded PowerPC systems, assume that the startup module is called
13533 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
13538 On System V.4 and embedded PowerPC systems, specify that you are
13539 compiling for a VxWorks system.
13543 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
13544 header to indicate that @samp{eabi} extended relocations are used.
13550 On System V.4 and embedded PowerPC systems do (do not) adhere to the
13551 Embedded Applications Binary Interface (eabi) which is a set of
13552 modifications to the System V.4 specifications. Selecting @option{-meabi}
13553 means that the stack is aligned to an 8 byte boundary, a function
13554 @code{__eabi} is called to from @code{main} to set up the eabi
13555 environment, and the @option{-msdata} option can use both @code{r2} and
13556 @code{r13} to point to two separate small data areas. Selecting
13557 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
13558 do not call an initialization function from @code{main}, and the
13559 @option{-msdata} option will only use @code{r13} to point to a single
13560 small data area. The @option{-meabi} option is on by default if you
13561 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
13564 @opindex msdata=eabi
13565 On System V.4 and embedded PowerPC systems, put small initialized
13566 @code{const} global and static data in the @samp{.sdata2} section, which
13567 is pointed to by register @code{r2}. Put small initialized
13568 non-@code{const} global and static data in the @samp{.sdata} section,
13569 which is pointed to by register @code{r13}. Put small uninitialized
13570 global and static data in the @samp{.sbss} section, which is adjacent to
13571 the @samp{.sdata} section. The @option{-msdata=eabi} option is
13572 incompatible with the @option{-mrelocatable} option. The
13573 @option{-msdata=eabi} option also sets the @option{-memb} option.
13576 @opindex msdata=sysv
13577 On System V.4 and embedded PowerPC systems, put small global and static
13578 data in the @samp{.sdata} section, which is pointed to by register
13579 @code{r13}. Put small uninitialized global and static data in the
13580 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
13581 The @option{-msdata=sysv} option is incompatible with the
13582 @option{-mrelocatable} option.
13584 @item -msdata=default
13586 @opindex msdata=default
13588 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
13589 compile code the same as @option{-msdata=eabi}, otherwise compile code the
13590 same as @option{-msdata=sysv}.
13593 @opindex msdata-data
13594 On System V.4 and embedded PowerPC systems, put small global
13595 data in the @samp{.sdata} section. Put small uninitialized global
13596 data in the @samp{.sbss} section. Do not use register @code{r13}
13597 to address small data however. This is the default behavior unless
13598 other @option{-msdata} options are used.
13602 @opindex msdata=none
13604 On embedded PowerPC systems, put all initialized global and static data
13605 in the @samp{.data} section, and all uninitialized data in the
13606 @samp{.bss} section.
13610 @cindex smaller data references (PowerPC)
13611 @cindex .sdata/.sdata2 references (PowerPC)
13612 On embedded PowerPC systems, put global and static items less than or
13613 equal to @var{num} bytes into the small data or bss sections instead of
13614 the normal data or bss section. By default, @var{num} is 8. The
13615 @option{-G @var{num}} switch is also passed to the linker.
13616 All modules should be compiled with the same @option{-G @var{num}} value.
13619 @itemx -mno-regnames
13621 @opindex mno-regnames
13622 On System V.4 and embedded PowerPC systems do (do not) emit register
13623 names in the assembly language output using symbolic forms.
13626 @itemx -mno-longcall
13628 @opindex mno-longcall
13629 By default assume that all calls are far away so that a longer more
13630 expensive calling sequence is required. This is required for calls
13631 further than 32 megabytes (33,554,432 bytes) from the current location.
13632 A short call will be generated if the compiler knows
13633 the call cannot be that far away. This setting can be overridden by
13634 the @code{shortcall} function attribute, or by @code{#pragma
13637 Some linkers are capable of detecting out-of-range calls and generating
13638 glue code on the fly. On these systems, long calls are unnecessary and
13639 generate slower code. As of this writing, the AIX linker can do this,
13640 as can the GNU linker for PowerPC/64. It is planned to add this feature
13641 to the GNU linker for 32-bit PowerPC systems as well.
13643 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
13644 callee, L42'', plus a ``branch island'' (glue code). The two target
13645 addresses represent the callee and the ``branch island''. The
13646 Darwin/PPC linker will prefer the first address and generate a ``bl
13647 callee'' if the PPC ``bl'' instruction will reach the callee directly;
13648 otherwise, the linker will generate ``bl L42'' to call the ``branch
13649 island''. The ``branch island'' is appended to the body of the
13650 calling function; it computes the full 32-bit address of the callee
13653 On Mach-O (Darwin) systems, this option directs the compiler emit to
13654 the glue for every direct call, and the Darwin linker decides whether
13655 to use or discard it.
13657 In the future, we may cause GCC to ignore all longcall specifications
13658 when the linker is known to generate glue.
13662 Adds support for multithreading with the @dfn{pthreads} library.
13663 This option sets flags for both the preprocessor and linker.
13667 @node S/390 and zSeries Options
13668 @subsection S/390 and zSeries Options
13669 @cindex S/390 and zSeries Options
13671 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
13675 @itemx -msoft-float
13676 @opindex mhard-float
13677 @opindex msoft-float
13678 Use (do not use) the hardware floating-point instructions and registers
13679 for floating-point operations. When @option{-msoft-float} is specified,
13680 functions in @file{libgcc.a} will be used to perform floating-point
13681 operations. When @option{-mhard-float} is specified, the compiler
13682 generates IEEE floating-point instructions. This is the default.
13684 @item -mlong-double-64
13685 @itemx -mlong-double-128
13686 @opindex mlong-double-64
13687 @opindex mlong-double-128
13688 These switches control the size of @code{long double} type. A size
13689 of 64bit makes the @code{long double} type equivalent to the @code{double}
13690 type. This is the default.
13693 @itemx -mno-backchain
13694 @opindex mbackchain
13695 @opindex mno-backchain
13696 Store (do not store) the address of the caller's frame as backchain pointer
13697 into the callee's stack frame.
13698 A backchain may be needed to allow debugging using tools that do not understand
13699 DWARF-2 call frame information.
13700 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
13701 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
13702 the backchain is placed into the topmost word of the 96/160 byte register
13705 In general, code compiled with @option{-mbackchain} is call-compatible with
13706 code compiled with @option{-mmo-backchain}; however, use of the backchain
13707 for debugging purposes usually requires that the whole binary is built with
13708 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
13709 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13710 to build a linux kernel use @option{-msoft-float}.
13712 The default is to not maintain the backchain.
13714 @item -mpacked-stack
13715 @itemx -mno-packed-stack
13716 @opindex mpacked-stack
13717 @opindex mno-packed-stack
13718 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
13719 specified, the compiler uses the all fields of the 96/160 byte register save
13720 area only for their default purpose; unused fields still take up stack space.
13721 When @option{-mpacked-stack} is specified, register save slots are densely
13722 packed at the top of the register save area; unused space is reused for other
13723 purposes, allowing for more efficient use of the available stack space.
13724 However, when @option{-mbackchain} is also in effect, the topmost word of
13725 the save area is always used to store the backchain, and the return address
13726 register is always saved two words below the backchain.
13728 As long as the stack frame backchain is not used, code generated with
13729 @option{-mpacked-stack} is call-compatible with code generated with
13730 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
13731 S/390 or zSeries generated code that uses the stack frame backchain at run
13732 time, not just for debugging purposes. Such code is not call-compatible
13733 with code compiled with @option{-mpacked-stack}. Also, note that the
13734 combination of @option{-mbackchain},
13735 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
13736 to build a linux kernel use @option{-msoft-float}.
13738 The default is to not use the packed stack layout.
13741 @itemx -mno-small-exec
13742 @opindex msmall-exec
13743 @opindex mno-small-exec
13744 Generate (or do not generate) code using the @code{bras} instruction
13745 to do subroutine calls.
13746 This only works reliably if the total executable size does not
13747 exceed 64k. The default is to use the @code{basr} instruction instead,
13748 which does not have this limitation.
13754 When @option{-m31} is specified, generate code compliant to the
13755 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
13756 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
13757 particular to generate 64-bit instructions. For the @samp{s390}
13758 targets, the default is @option{-m31}, while the @samp{s390x}
13759 targets default to @option{-m64}.
13765 When @option{-mzarch} is specified, generate code using the
13766 instructions available on z/Architecture.
13767 When @option{-mesa} is specified, generate code using the
13768 instructions available on ESA/390. Note that @option{-mesa} is
13769 not possible with @option{-m64}.
13770 When generating code compliant to the GNU/Linux for S/390 ABI,
13771 the default is @option{-mesa}. When generating code compliant
13772 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
13778 Generate (or do not generate) code using the @code{mvcle} instruction
13779 to perform block moves. When @option{-mno-mvcle} is specified,
13780 use a @code{mvc} loop instead. This is the default unless optimizing for
13787 Print (or do not print) additional debug information when compiling.
13788 The default is to not print debug information.
13790 @item -march=@var{cpu-type}
13792 Generate code that will run on @var{cpu-type}, which is the name of a system
13793 representing a certain processor type. Possible values for
13794 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
13795 When generating code using the instructions available on z/Architecture,
13796 the default is @option{-march=z900}. Otherwise, the default is
13797 @option{-march=g5}.
13799 @item -mtune=@var{cpu-type}
13801 Tune to @var{cpu-type} everything applicable about the generated code,
13802 except for the ABI and the set of available instructions.
13803 The list of @var{cpu-type} values is the same as for @option{-march}.
13804 The default is the value used for @option{-march}.
13807 @itemx -mno-tpf-trace
13808 @opindex mtpf-trace
13809 @opindex mno-tpf-trace
13810 Generate code that adds (does not add) in TPF OS specific branches to trace
13811 routines in the operating system. This option is off by default, even
13812 when compiling for the TPF OS@.
13815 @itemx -mno-fused-madd
13816 @opindex mfused-madd
13817 @opindex mno-fused-madd
13818 Generate code that uses (does not use) the floating point multiply and
13819 accumulate instructions. These instructions are generated by default if
13820 hardware floating point is used.
13822 @item -mwarn-framesize=@var{framesize}
13823 @opindex mwarn-framesize
13824 Emit a warning if the current function exceeds the given frame size. Because
13825 this is a compile time check it doesn't need to be a real problem when the program
13826 runs. It is intended to identify functions which most probably cause
13827 a stack overflow. It is useful to be used in an environment with limited stack
13828 size e.g.@: the linux kernel.
13830 @item -mwarn-dynamicstack
13831 @opindex mwarn-dynamicstack
13832 Emit a warning if the function calls alloca or uses dynamically
13833 sized arrays. This is generally a bad idea with a limited stack size.
13835 @item -mstack-guard=@var{stack-guard}
13836 @itemx -mstack-size=@var{stack-size}
13837 @opindex mstack-guard
13838 @opindex mstack-size
13839 If these options are provided the s390 back end emits additional instructions in
13840 the function prologue which trigger a trap if the stack size is @var{stack-guard}
13841 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
13842 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
13843 the frame size of the compiled function is chosen.
13844 These options are intended to be used to help debugging stack overflow problems.
13845 The additionally emitted code causes only little overhead and hence can also be
13846 used in production like systems without greater performance degradation. The given
13847 values have to be exact powers of 2 and @var{stack-size} has to be greater than
13848 @var{stack-guard} without exceeding 64k.
13849 In order to be efficient the extra code makes the assumption that the stack starts
13850 at an address aligned to the value given by @var{stack-size}.
13851 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
13854 @node Score Options
13855 @subsection Score Options
13856 @cindex Score Options
13858 These options are defined for Score implementations:
13863 Compile code for big endian mode. This is the default.
13867 Compile code for little endian mode.
13871 Disable generate bcnz instruction.
13875 Enable generate unaligned load and store instruction.
13879 Enable the use of multiply-accumulate instructions. Disabled by default.
13883 Specify the SCORE5 as the target architecture.
13887 Specify the SCORE5U of the target architecture.
13891 Specify the SCORE7 as the target architecture. This is the default.
13895 Specify the SCORE7D as the target architecture.
13899 @subsection SH Options
13901 These @samp{-m} options are defined for the SH implementations:
13906 Generate code for the SH1.
13910 Generate code for the SH2.
13913 Generate code for the SH2e.
13917 Generate code for the SH3.
13921 Generate code for the SH3e.
13925 Generate code for the SH4 without a floating-point unit.
13927 @item -m4-single-only
13928 @opindex m4-single-only
13929 Generate code for the SH4 with a floating-point unit that only
13930 supports single-precision arithmetic.
13934 Generate code for the SH4 assuming the floating-point unit is in
13935 single-precision mode by default.
13939 Generate code for the SH4.
13943 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
13944 floating-point unit is not used.
13946 @item -m4a-single-only
13947 @opindex m4a-single-only
13948 Generate code for the SH4a, in such a way that no double-precision
13949 floating point operations are used.
13952 @opindex m4a-single
13953 Generate code for the SH4a assuming the floating-point unit is in
13954 single-precision mode by default.
13958 Generate code for the SH4a.
13962 Same as @option{-m4a-nofpu}, except that it implicitly passes
13963 @option{-dsp} to the assembler. GCC doesn't generate any DSP
13964 instructions at the moment.
13968 Compile code for the processor in big endian mode.
13972 Compile code for the processor in little endian mode.
13976 Align doubles at 64-bit boundaries. Note that this changes the calling
13977 conventions, and thus some functions from the standard C library will
13978 not work unless you recompile it first with @option{-mdalign}.
13982 Shorten some address references at link time, when possible; uses the
13983 linker option @option{-relax}.
13987 Use 32-bit offsets in @code{switch} tables. The default is to use
13992 Enable the use of bit manipulation instructions on SH2A.
13996 Enable the use of the instruction @code{fmovd}.
14000 Comply with the calling conventions defined by Renesas.
14004 Comply with the calling conventions defined by Renesas.
14008 Comply with the calling conventions defined for GCC before the Renesas
14009 conventions were available. This option is the default for all
14010 targets of the SH toolchain except for @samp{sh-symbianelf}.
14013 @opindex mnomacsave
14014 Mark the @code{MAC} register as call-clobbered, even if
14015 @option{-mhitachi} is given.
14019 Increase IEEE-compliance of floating-point code.
14020 At the moment, this is equivalent to @option{-fno-finite-math-only}.
14021 When generating 16 bit SH opcodes, getting IEEE-conforming results for
14022 comparisons of NANs / infinities incurs extra overhead in every
14023 floating point comparison, therefore the default is set to
14024 @option{-ffinite-math-only}.
14026 @item -minline-ic_invalidate
14027 @opindex minline-ic_invalidate
14028 Inline code to invalidate instruction cache entries after setting up
14029 nested function trampolines.
14030 This option has no effect if -musermode is in effect and the selected
14031 code generation option (e.g. -m4) does not allow the use of the icbi
14033 If the selected code generation option does not allow the use of the icbi
14034 instruction, and -musermode is not in effect, the inlined code will
14035 manipulate the instruction cache address array directly with an associative
14036 write. This not only requires privileged mode, but it will also
14037 fail if the cache line had been mapped via the TLB and has become unmapped.
14041 Dump instruction size and location in the assembly code.
14044 @opindex mpadstruct
14045 This option is deprecated. It pads structures to multiple of 4 bytes,
14046 which is incompatible with the SH ABI@.
14050 Optimize for space instead of speed. Implied by @option{-Os}.
14053 @opindex mprefergot
14054 When generating position-independent code, emit function calls using
14055 the Global Offset Table instead of the Procedure Linkage Table.
14059 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
14060 if the inlined code would not work in user mode.
14061 This is the default when the target is @code{sh-*-linux*}.
14063 @item -multcost=@var{number}
14064 @opindex multcost=@var{number}
14065 Set the cost to assume for a multiply insn.
14067 @item -mdiv=@var{strategy}
14068 @opindex mdiv=@var{strategy}
14069 Set the division strategy to use for SHmedia code. @var{strategy} must be
14070 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
14071 inv:call2, inv:fp .
14072 "fp" performs the operation in floating point. This has a very high latency,
14073 but needs only a few instructions, so it might be a good choice if
14074 your code has enough easily exploitable ILP to allow the compiler to
14075 schedule the floating point instructions together with other instructions.
14076 Division by zero causes a floating point exception.
14077 "inv" uses integer operations to calculate the inverse of the divisor,
14078 and then multiplies the dividend with the inverse. This strategy allows
14079 cse and hoisting of the inverse calculation. Division by zero calculates
14080 an unspecified result, but does not trap.
14081 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
14082 have been found, or if the entire operation has been hoisted to the same
14083 place, the last stages of the inverse calculation are intertwined with the
14084 final multiply to reduce the overall latency, at the expense of using a few
14085 more instructions, and thus offering fewer scheduling opportunities with
14087 "call" calls a library function that usually implements the inv:minlat
14089 This gives high code density for m5-*media-nofpu compilations.
14090 "call2" uses a different entry point of the same library function, where it
14091 assumes that a pointer to a lookup table has already been set up, which
14092 exposes the pointer load to cse / code hoisting optimizations.
14093 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
14094 code generation, but if the code stays unoptimized, revert to the "call",
14095 "call2", or "fp" strategies, respectively. Note that the
14096 potentially-trapping side effect of division by zero is carried by a
14097 separate instruction, so it is possible that all the integer instructions
14098 are hoisted out, but the marker for the side effect stays where it is.
14099 A recombination to fp operations or a call is not possible in that case.
14100 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
14101 that the inverse calculation was nor separated from the multiply, they speed
14102 up division where the dividend fits into 20 bits (plus sign where applicable),
14103 by inserting a test to skip a number of operations in this case; this test
14104 slows down the case of larger dividends. inv20u assumes the case of a such
14105 a small dividend to be unlikely, and inv20l assumes it to be likely.
14107 @item -mdivsi3_libfunc=@var{name}
14108 @opindex mdivsi3_libfunc=@var{name}
14109 Set the name of the library function used for 32 bit signed division to
14110 @var{name}. This only affect the name used in the call and inv:call
14111 division strategies, and the compiler will still expect the same
14112 sets of input/output/clobbered registers as if this option was not present.
14114 @item -mfixed-range=@var{register-range}
14115 @opindex mfixed-range
14116 Generate code treating the given register range as fixed registers.
14117 A fixed register is one that the register allocator can not use. This is
14118 useful when compiling kernel code. A register range is specified as
14119 two registers separated by a dash. Multiple register ranges can be
14120 specified separated by a comma.
14122 @item -madjust-unroll
14123 @opindex madjust-unroll
14124 Throttle unrolling to avoid thrashing target registers.
14125 This option only has an effect if the gcc code base supports the
14126 TARGET_ADJUST_UNROLL_MAX target hook.
14128 @item -mindexed-addressing
14129 @opindex mindexed-addressing
14130 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
14131 This is only safe if the hardware and/or OS implement 32 bit wrap-around
14132 semantics for the indexed addressing mode. The architecture allows the
14133 implementation of processors with 64 bit MMU, which the OS could use to
14134 get 32 bit addressing, but since no current hardware implementation supports
14135 this or any other way to make the indexed addressing mode safe to use in
14136 the 32 bit ABI, the default is -mno-indexed-addressing.
14138 @item -mgettrcost=@var{number}
14139 @opindex mgettrcost=@var{number}
14140 Set the cost assumed for the gettr instruction to @var{number}.
14141 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
14145 Assume pt* instructions won't trap. This will generally generate better
14146 scheduled code, but is unsafe on current hardware. The current architecture
14147 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
14148 This has the unintentional effect of making it unsafe to schedule ptabs /
14149 ptrel before a branch, or hoist it out of a loop. For example,
14150 __do_global_ctors, a part of libgcc that runs constructors at program
14151 startup, calls functions in a list which is delimited by @minus{}1. With the
14152 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
14153 That means that all the constructors will be run a bit quicker, but when
14154 the loop comes to the end of the list, the program crashes because ptabs
14155 loads @minus{}1 into a target register. Since this option is unsafe for any
14156 hardware implementing the current architecture specification, the default
14157 is -mno-pt-fixed. Unless the user specifies a specific cost with
14158 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
14159 this deters register allocation using target registers for storing
14162 @item -minvalid-symbols
14163 @opindex minvalid-symbols
14164 Assume symbols might be invalid. Ordinary function symbols generated by
14165 the compiler will always be valid to load with movi/shori/ptabs or
14166 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
14167 to generate symbols that will cause ptabs / ptrel to trap.
14168 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
14169 It will then prevent cross-basic-block cse, hoisting and most scheduling
14170 of symbol loads. The default is @option{-mno-invalid-symbols}.
14173 @node SPARC Options
14174 @subsection SPARC Options
14175 @cindex SPARC options
14177 These @samp{-m} options are supported on the SPARC:
14180 @item -mno-app-regs
14182 @opindex mno-app-regs
14184 Specify @option{-mapp-regs} to generate output using the global registers
14185 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
14188 To be fully SVR4 ABI compliant at the cost of some performance loss,
14189 specify @option{-mno-app-regs}. You should compile libraries and system
14190 software with this option.
14193 @itemx -mhard-float
14195 @opindex mhard-float
14196 Generate output containing floating point instructions. This is the
14200 @itemx -msoft-float
14202 @opindex msoft-float
14203 Generate output containing library calls for floating point.
14204 @strong{Warning:} the requisite libraries are not available for all SPARC
14205 targets. Normally the facilities of the machine's usual C compiler are
14206 used, but this cannot be done directly in cross-compilation. You must make
14207 your own arrangements to provide suitable library functions for
14208 cross-compilation. The embedded targets @samp{sparc-*-aout} and
14209 @samp{sparclite-*-*} do provide software floating point support.
14211 @option{-msoft-float} changes the calling convention in the output file;
14212 therefore, it is only useful if you compile @emph{all} of a program with
14213 this option. In particular, you need to compile @file{libgcc.a}, the
14214 library that comes with GCC, with @option{-msoft-float} in order for
14217 @item -mhard-quad-float
14218 @opindex mhard-quad-float
14219 Generate output containing quad-word (long double) floating point
14222 @item -msoft-quad-float
14223 @opindex msoft-quad-float
14224 Generate output containing library calls for quad-word (long double)
14225 floating point instructions. The functions called are those specified
14226 in the SPARC ABI@. This is the default.
14228 As of this writing, there are no SPARC implementations that have hardware
14229 support for the quad-word floating point instructions. They all invoke
14230 a trap handler for one of these instructions, and then the trap handler
14231 emulates the effect of the instruction. Because of the trap handler overhead,
14232 this is much slower than calling the ABI library routines. Thus the
14233 @option{-msoft-quad-float} option is the default.
14235 @item -mno-unaligned-doubles
14236 @itemx -munaligned-doubles
14237 @opindex mno-unaligned-doubles
14238 @opindex munaligned-doubles
14239 Assume that doubles have 8 byte alignment. This is the default.
14241 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
14242 alignment only if they are contained in another type, or if they have an
14243 absolute address. Otherwise, it assumes they have 4 byte alignment.
14244 Specifying this option avoids some rare compatibility problems with code
14245 generated by other compilers. It is not the default because it results
14246 in a performance loss, especially for floating point code.
14248 @item -mno-faster-structs
14249 @itemx -mfaster-structs
14250 @opindex mno-faster-structs
14251 @opindex mfaster-structs
14252 With @option{-mfaster-structs}, the compiler assumes that structures
14253 should have 8 byte alignment. This enables the use of pairs of
14254 @code{ldd} and @code{std} instructions for copies in structure
14255 assignment, in place of twice as many @code{ld} and @code{st} pairs.
14256 However, the use of this changed alignment directly violates the SPARC
14257 ABI@. Thus, it's intended only for use on targets where the developer
14258 acknowledges that their resulting code will not be directly in line with
14259 the rules of the ABI@.
14261 @item -mimpure-text
14262 @opindex mimpure-text
14263 @option{-mimpure-text}, used in addition to @option{-shared}, tells
14264 the compiler to not pass @option{-z text} to the linker when linking a
14265 shared object. Using this option, you can link position-dependent
14266 code into a shared object.
14268 @option{-mimpure-text} suppresses the ``relocations remain against
14269 allocatable but non-writable sections'' linker error message.
14270 However, the necessary relocations will trigger copy-on-write, and the
14271 shared object is not actually shared across processes. Instead of
14272 using @option{-mimpure-text}, you should compile all source code with
14273 @option{-fpic} or @option{-fPIC}.
14275 This option is only available on SunOS and Solaris.
14277 @item -mcpu=@var{cpu_type}
14279 Set the instruction set, register set, and instruction scheduling parameters
14280 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
14281 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
14282 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
14283 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
14284 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
14286 Default instruction scheduling parameters are used for values that select
14287 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
14288 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
14290 Here is a list of each supported architecture and their supported
14295 v8: supersparc, hypersparc
14296 sparclite: f930, f934, sparclite86x
14298 v9: ultrasparc, ultrasparc3, niagara, niagara2
14301 By default (unless configured otherwise), GCC generates code for the V7
14302 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
14303 additionally optimizes it for the Cypress CY7C602 chip, as used in the
14304 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
14305 SPARCStation 1, 2, IPX etc.
14307 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
14308 architecture. The only difference from V7 code is that the compiler emits
14309 the integer multiply and integer divide instructions which exist in SPARC-V8
14310 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
14311 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
14314 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
14315 the SPARC architecture. This adds the integer multiply, integer divide step
14316 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
14317 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
14318 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
14319 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
14320 MB86934 chip, which is the more recent SPARClite with FPU@.
14322 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
14323 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
14324 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
14325 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
14326 optimizes it for the TEMIC SPARClet chip.
14328 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
14329 architecture. This adds 64-bit integer and floating-point move instructions,
14330 3 additional floating-point condition code registers and conditional move
14331 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
14332 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
14333 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
14334 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
14335 @option{-mcpu=niagara}, the compiler additionally optimizes it for
14336 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
14337 additionally optimizes it for Sun UltraSPARC T2 chips.
14339 @item -mtune=@var{cpu_type}
14341 Set the instruction scheduling parameters for machine type
14342 @var{cpu_type}, but do not set the instruction set or register set that the
14343 option @option{-mcpu=@var{cpu_type}} would.
14345 The same values for @option{-mcpu=@var{cpu_type}} can be used for
14346 @option{-mtune=@var{cpu_type}}, but the only useful values are those
14347 that select a particular cpu implementation. Those are @samp{cypress},
14348 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
14349 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
14350 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
14355 @opindex mno-v8plus
14356 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
14357 difference from the V8 ABI is that the global and out registers are
14358 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
14359 mode for all SPARC-V9 processors.
14365 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
14366 Visual Instruction Set extensions. The default is @option{-mno-vis}.
14369 These @samp{-m} options are supported in addition to the above
14370 on SPARC-V9 processors in 64-bit environments:
14373 @item -mlittle-endian
14374 @opindex mlittle-endian
14375 Generate code for a processor running in little-endian mode. It is only
14376 available for a few configurations and most notably not on Solaris and Linux.
14382 Generate code for a 32-bit or 64-bit environment.
14383 The 32-bit environment sets int, long and pointer to 32 bits.
14384 The 64-bit environment sets int to 32 bits and long and pointer
14387 @item -mcmodel=medlow
14388 @opindex mcmodel=medlow
14389 Generate code for the Medium/Low code model: 64-bit addresses, programs
14390 must be linked in the low 32 bits of memory. Programs can be statically
14391 or dynamically linked.
14393 @item -mcmodel=medmid
14394 @opindex mcmodel=medmid
14395 Generate code for the Medium/Middle code model: 64-bit addresses, programs
14396 must be linked in the low 44 bits of memory, the text and data segments must
14397 be less than 2GB in size and the data segment must be located within 2GB of
14400 @item -mcmodel=medany
14401 @opindex mcmodel=medany
14402 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
14403 may be linked anywhere in memory, the text and data segments must be less
14404 than 2GB in size and the data segment must be located within 2GB of the
14407 @item -mcmodel=embmedany
14408 @opindex mcmodel=embmedany
14409 Generate code for the Medium/Anywhere code model for embedded systems:
14410 64-bit addresses, the text and data segments must be less than 2GB in
14411 size, both starting anywhere in memory (determined at link time). The
14412 global register %g4 points to the base of the data segment. Programs
14413 are statically linked and PIC is not supported.
14416 @itemx -mno-stack-bias
14417 @opindex mstack-bias
14418 @opindex mno-stack-bias
14419 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
14420 frame pointer if present, are offset by @minus{}2047 which must be added back
14421 when making stack frame references. This is the default in 64-bit mode.
14422 Otherwise, assume no such offset is present.
14425 These switches are supported in addition to the above on Solaris:
14430 Add support for multithreading using the Solaris threads library. This
14431 option sets flags for both the preprocessor and linker. This option does
14432 not affect the thread safety of object code produced by the compiler or
14433 that of libraries supplied with it.
14437 Add support for multithreading using the POSIX 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 This is a synonym for @option{-pthreads}.
14448 @subsection SPU Options
14449 @cindex SPU options
14451 These @samp{-m} options are supported on the SPU:
14455 @itemx -merror-reloc
14456 @opindex mwarn-reloc
14457 @opindex merror-reloc
14459 The loader for SPU does not handle dynamic relocations. By default, GCC
14460 will give an error when it generates code that requires a dynamic
14461 relocation. @option{-mno-error-reloc} disables the error,
14462 @option{-mwarn-reloc} will generate a warning instead.
14465 @itemx -munsafe-dma
14467 @opindex munsafe-dma
14469 Instructions which initiate or test completion of DMA must not be
14470 reordered with respect to loads and stores of the memory which is being
14471 accessed. Users typically address this problem using the volatile
14472 keyword, but that can lead to inefficient code in places where the
14473 memory is known to not change. Rather than mark the memory as volatile
14474 we treat the DMA instructions as potentially effecting all memory. With
14475 @option{-munsafe-dma} users must use the volatile keyword to protect
14478 @item -mbranch-hints
14479 @opindex mbranch-hints
14481 By default, GCC will generate a branch hint instruction to avoid
14482 pipeline stalls for always taken or probably taken branches. A hint
14483 will not be generated closer than 8 instructions away from its branch.
14484 There is little reason to disable them, except for debugging purposes,
14485 or to make an object a little bit smaller.
14489 @opindex msmall-mem
14490 @opindex mlarge-mem
14492 By default, GCC generates code assuming that addresses are never larger
14493 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
14494 a full 32 bit address.
14499 By default, GCC links against startup code that assumes the SPU-style
14500 main function interface (which has an unconventional parameter list).
14501 With @option{-mstdmain}, GCC will link your program against startup
14502 code that assumes a C99-style interface to @code{main}, including a
14503 local copy of @code{argv} strings.
14505 @item -mfixed-range=@var{register-range}
14506 @opindex mfixed-range
14507 Generate code treating the given register range as fixed registers.
14508 A fixed register is one that the register allocator can not use. This is
14509 useful when compiling kernel code. A register range is specified as
14510 two registers separated by a dash. Multiple register ranges can be
14511 specified separated by a comma.
14515 @node System V Options
14516 @subsection Options for System V
14518 These additional options are available on System V Release 4 for
14519 compatibility with other compilers on those systems:
14524 Create a shared object.
14525 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
14529 Identify the versions of each tool used by the compiler, in a
14530 @code{.ident} assembler directive in the output.
14534 Refrain from adding @code{.ident} directives to the output file (this is
14537 @item -YP,@var{dirs}
14539 Search the directories @var{dirs}, and no others, for libraries
14540 specified with @option{-l}.
14542 @item -Ym,@var{dir}
14544 Look in the directory @var{dir} to find the M4 preprocessor.
14545 The assembler uses this option.
14546 @c This is supposed to go with a -Yd for predefined M4 macro files, but
14547 @c the generic assembler that comes with Solaris takes just -Ym.
14551 @subsection V850 Options
14552 @cindex V850 Options
14554 These @samp{-m} options are defined for V850 implementations:
14558 @itemx -mno-long-calls
14559 @opindex mlong-calls
14560 @opindex mno-long-calls
14561 Treat all calls as being far away (near). If calls are assumed to be
14562 far away, the compiler will always load the functions address up into a
14563 register, and call indirect through the pointer.
14569 Do not optimize (do optimize) basic blocks that use the same index
14570 pointer 4 or more times to copy pointer into the @code{ep} register, and
14571 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
14572 option is on by default if you optimize.
14574 @item -mno-prolog-function
14575 @itemx -mprolog-function
14576 @opindex mno-prolog-function
14577 @opindex mprolog-function
14578 Do not use (do use) external functions to save and restore registers
14579 at the prologue and epilogue of a function. The external functions
14580 are slower, but use less code space if more than one function saves
14581 the same number of registers. The @option{-mprolog-function} option
14582 is on by default if you optimize.
14586 Try to make the code as small as possible. At present, this just turns
14587 on the @option{-mep} and @option{-mprolog-function} options.
14589 @item -mtda=@var{n}
14591 Put static or global variables whose size is @var{n} bytes or less into
14592 the tiny data area that register @code{ep} points to. The tiny data
14593 area can hold up to 256 bytes in total (128 bytes for byte references).
14595 @item -msda=@var{n}
14597 Put static or global variables whose size is @var{n} bytes or less into
14598 the small data area that register @code{gp} points to. The small data
14599 area can hold up to 64 kilobytes.
14601 @item -mzda=@var{n}
14603 Put static or global variables whose size is @var{n} bytes or less into
14604 the first 32 kilobytes of memory.
14608 Specify that the target processor is the V850.
14611 @opindex mbig-switch
14612 Generate code suitable for big switch tables. Use this option only if
14613 the assembler/linker complain about out of range branches within a switch
14618 This option will cause r2 and r5 to be used in the code generated by
14619 the compiler. This setting is the default.
14621 @item -mno-app-regs
14622 @opindex mno-app-regs
14623 This option will cause r2 and r5 to be treated as fixed registers.
14627 Specify that the target processor is the V850E1. The preprocessor
14628 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
14629 this option is used.
14633 Specify that the target processor is the V850E@. The preprocessor
14634 constant @samp{__v850e__} will be defined if this option is used.
14636 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
14637 are defined then a default target processor will be chosen and the
14638 relevant @samp{__v850*__} preprocessor constant will be defined.
14640 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
14641 defined, regardless of which processor variant is the target.
14643 @item -mdisable-callt
14644 @opindex mdisable-callt
14645 This option will suppress generation of the CALLT instruction for the
14646 v850e and v850e1 flavors of the v850 architecture. The default is
14647 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
14652 @subsection VAX Options
14653 @cindex VAX options
14655 These @samp{-m} options are defined for the VAX:
14660 Do not output certain jump instructions (@code{aobleq} and so on)
14661 that the Unix assembler for the VAX cannot handle across long
14666 Do output those jump instructions, on the assumption that you
14667 will assemble with the GNU assembler.
14671 Output code for g-format floating point numbers instead of d-format.
14674 @node VxWorks Options
14675 @subsection VxWorks Options
14676 @cindex VxWorks Options
14678 The options in this section are defined for all VxWorks targets.
14679 Options specific to the target hardware are listed with the other
14680 options for that target.
14685 GCC can generate code for both VxWorks kernels and real time processes
14686 (RTPs). This option switches from the former to the latter. It also
14687 defines the preprocessor macro @code{__RTP__}.
14690 @opindex non-static
14691 Link an RTP executable against shared libraries rather than static
14692 libraries. The options @option{-static} and @option{-shared} can
14693 also be used for RTPs (@pxref{Link Options}); @option{-static}
14700 These options are passed down to the linker. They are defined for
14701 compatibility with Diab.
14704 @opindex Xbind-lazy
14705 Enable lazy binding of function calls. This option is equivalent to
14706 @option{-Wl,-z,now} and is defined for compatibility with Diab.
14710 Disable lazy binding of function calls. This option is the default and
14711 is defined for compatibility with Diab.
14714 @node x86-64 Options
14715 @subsection x86-64 Options
14716 @cindex x86-64 options
14718 These are listed under @xref{i386 and x86-64 Options}.
14720 @node Xstormy16 Options
14721 @subsection Xstormy16 Options
14722 @cindex Xstormy16 Options
14724 These options are defined for Xstormy16:
14729 Choose startup files and linker script suitable for the simulator.
14732 @node Xtensa Options
14733 @subsection Xtensa Options
14734 @cindex Xtensa Options
14736 These options are supported for Xtensa targets:
14740 @itemx -mno-const16
14742 @opindex mno-const16
14743 Enable or disable use of @code{CONST16} instructions for loading
14744 constant values. The @code{CONST16} instruction is currently not a
14745 standard option from Tensilica. When enabled, @code{CONST16}
14746 instructions are always used in place of the standard @code{L32R}
14747 instructions. The use of @code{CONST16} is enabled by default only if
14748 the @code{L32R} instruction is not available.
14751 @itemx -mno-fused-madd
14752 @opindex mfused-madd
14753 @opindex mno-fused-madd
14754 Enable or disable use of fused multiply/add and multiply/subtract
14755 instructions in the floating-point option. This has no effect if the
14756 floating-point option is not also enabled. Disabling fused multiply/add
14757 and multiply/subtract instructions forces the compiler to use separate
14758 instructions for the multiply and add/subtract operations. This may be
14759 desirable in some cases where strict IEEE 754-compliant results are
14760 required: the fused multiply add/subtract instructions do not round the
14761 intermediate result, thereby producing results with @emph{more} bits of
14762 precision than specified by the IEEE standard. Disabling fused multiply
14763 add/subtract instructions also ensures that the program output is not
14764 sensitive to the compiler's ability to combine multiply and add/subtract
14767 @item -mserialize-volatile
14768 @itemx -mno-serialize-volatile
14769 @opindex mserialize-volatile
14770 @opindex mno-serialize-volatile
14771 When this option is enabled, GCC inserts @code{MEMW} instructions before
14772 @code{volatile} memory references to guarantee sequential consistency.
14773 The default is @option{-mserialize-volatile}. Use
14774 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
14776 @item -mtext-section-literals
14777 @itemx -mno-text-section-literals
14778 @opindex mtext-section-literals
14779 @opindex mno-text-section-literals
14780 Control the treatment of literal pools. The default is
14781 @option{-mno-text-section-literals}, which places literals in a separate
14782 section in the output file. This allows the literal pool to be placed
14783 in a data RAM/ROM, and it also allows the linker to combine literal
14784 pools from separate object files to remove redundant literals and
14785 improve code size. With @option{-mtext-section-literals}, the literals
14786 are interspersed in the text section in order to keep them as close as
14787 possible to their references. This may be necessary for large assembly
14790 @item -mtarget-align
14791 @itemx -mno-target-align
14792 @opindex mtarget-align
14793 @opindex mno-target-align
14794 When this option is enabled, GCC instructs the assembler to
14795 automatically align instructions to reduce branch penalties at the
14796 expense of some code density. The assembler attempts to widen density
14797 instructions to align branch targets and the instructions following call
14798 instructions. If there are not enough preceding safe density
14799 instructions to align a target, no widening will be performed. The
14800 default is @option{-mtarget-align}. These options do not affect the
14801 treatment of auto-aligned instructions like @code{LOOP}, which the
14802 assembler will always align, either by widening density instructions or
14803 by inserting no-op instructions.
14806 @itemx -mno-longcalls
14807 @opindex mlongcalls
14808 @opindex mno-longcalls
14809 When this option is enabled, GCC instructs the assembler to translate
14810 direct calls to indirect calls unless it can determine that the target
14811 of a direct call is in the range allowed by the call instruction. This
14812 translation typically occurs for calls to functions in other source
14813 files. Specifically, the assembler translates a direct @code{CALL}
14814 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
14815 The default is @option{-mno-longcalls}. This option should be used in
14816 programs where the call target can potentially be out of range. This
14817 option is implemented in the assembler, not the compiler, so the
14818 assembly code generated by GCC will still show direct call
14819 instructions---look at the disassembled object code to see the actual
14820 instructions. Note that the assembler will use an indirect call for
14821 every cross-file call, not just those that really will be out of range.
14824 @node zSeries Options
14825 @subsection zSeries Options
14826 @cindex zSeries options
14828 These are listed under @xref{S/390 and zSeries Options}.
14830 @node Code Gen Options
14831 @section Options for Code Generation Conventions
14832 @cindex code generation conventions
14833 @cindex options, code generation
14834 @cindex run-time options
14836 These machine-independent options control the interface conventions
14837 used in code generation.
14839 Most of them have both positive and negative forms; the negative form
14840 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
14841 one of the forms is listed---the one which is not the default. You
14842 can figure out the other form by either removing @samp{no-} or adding
14846 @item -fbounds-check
14847 @opindex fbounds-check
14848 For front-ends that support it, generate additional code to check that
14849 indices used to access arrays are within the declared range. This is
14850 currently only supported by the Java and Fortran front-ends, where
14851 this option defaults to true and false respectively.
14855 This option generates traps for signed overflow on addition, subtraction,
14856 multiplication operations.
14860 This option instructs the compiler to assume that signed arithmetic
14861 overflow of addition, subtraction and multiplication wraps around
14862 using twos-complement representation. This flag enables some optimizations
14863 and disables others. This option is enabled by default for the Java
14864 front-end, as required by the Java language specification.
14867 @opindex fexceptions
14868 Enable exception handling. Generates extra code needed to propagate
14869 exceptions. For some targets, this implies GCC will generate frame
14870 unwind information for all functions, which can produce significant data
14871 size overhead, although it does not affect execution. If you do not
14872 specify this option, GCC will enable it by default for languages like
14873 C++ which normally require exception handling, and disable it for
14874 languages like C that do not normally require it. However, you may need
14875 to enable this option when compiling C code that needs to interoperate
14876 properly with exception handlers written in C++. You may also wish to
14877 disable this option if you are compiling older C++ programs that don't
14878 use exception handling.
14880 @item -fnon-call-exceptions
14881 @opindex fnon-call-exceptions
14882 Generate code that allows trapping instructions to throw exceptions.
14883 Note that this requires platform-specific runtime support that does
14884 not exist everywhere. Moreover, it only allows @emph{trapping}
14885 instructions to throw exceptions, i.e.@: memory references or floating
14886 point instructions. It does not allow exceptions to be thrown from
14887 arbitrary signal handlers such as @code{SIGALRM}.
14889 @item -funwind-tables
14890 @opindex funwind-tables
14891 Similar to @option{-fexceptions}, except that it will just generate any needed
14892 static data, but will not affect the generated code in any other way.
14893 You will normally not enable this option; instead, a language processor
14894 that needs this handling would enable it on your behalf.
14896 @item -fasynchronous-unwind-tables
14897 @opindex fasynchronous-unwind-tables
14898 Generate unwind table in dwarf2 format, if supported by target machine. The
14899 table is exact at each instruction boundary, so it can be used for stack
14900 unwinding from asynchronous events (such as debugger or garbage collector).
14902 @item -fpcc-struct-return
14903 @opindex fpcc-struct-return
14904 Return ``short'' @code{struct} and @code{union} values in memory like
14905 longer ones, rather than in registers. This convention is less
14906 efficient, but it has the advantage of allowing intercallability between
14907 GCC-compiled files and files compiled with other compilers, particularly
14908 the Portable C Compiler (pcc).
14910 The precise convention for returning structures in memory depends
14911 on the target configuration macros.
14913 Short structures and unions are those whose size and alignment match
14914 that of some integer type.
14916 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
14917 switch is not binary compatible with code compiled with the
14918 @option{-freg-struct-return} switch.
14919 Use it to conform to a non-default application binary interface.
14921 @item -freg-struct-return
14922 @opindex freg-struct-return
14923 Return @code{struct} and @code{union} values in registers when possible.
14924 This is more efficient for small structures than
14925 @option{-fpcc-struct-return}.
14927 If you specify neither @option{-fpcc-struct-return} nor
14928 @option{-freg-struct-return}, GCC defaults to whichever convention is
14929 standard for the target. If there is no standard convention, GCC
14930 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
14931 the principal compiler. In those cases, we can choose the standard, and
14932 we chose the more efficient register return alternative.
14934 @strong{Warning:} code compiled with the @option{-freg-struct-return}
14935 switch is not binary compatible with code compiled with the
14936 @option{-fpcc-struct-return} switch.
14937 Use it to conform to a non-default application binary interface.
14939 @item -fshort-enums
14940 @opindex fshort-enums
14941 Allocate to an @code{enum} type only as many bytes as it needs for the
14942 declared range of possible values. Specifically, the @code{enum} type
14943 will be equivalent to the smallest integer type which has enough room.
14945 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
14946 code that is not binary compatible with code generated without that switch.
14947 Use it to conform to a non-default application binary interface.
14949 @item -fshort-double
14950 @opindex fshort-double
14951 Use the same size for @code{double} as for @code{float}.
14953 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
14954 code that is not binary compatible with code generated without that switch.
14955 Use it to conform to a non-default application binary interface.
14957 @item -fshort-wchar
14958 @opindex fshort-wchar
14959 Override the underlying type for @samp{wchar_t} to be @samp{short
14960 unsigned int} instead of the default for the target. This option is
14961 useful for building programs to run under WINE@.
14963 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
14964 code that is not binary compatible with code generated without that switch.
14965 Use it to conform to a non-default application binary interface.
14968 @opindex fno-common
14969 In C, allocate even uninitialized global variables in the data section of the
14970 object file, rather than generating them as common blocks. This has the
14971 effect that if the same variable is declared (without @code{extern}) in
14972 two different compilations, you will get an error when you link them.
14973 The only reason this might be useful is if you wish to verify that the
14974 program will work on other systems which always work this way.
14978 Ignore the @samp{#ident} directive.
14980 @item -finhibit-size-directive
14981 @opindex finhibit-size-directive
14982 Don't output a @code{.size} assembler directive, or anything else that
14983 would cause trouble if the function is split in the middle, and the
14984 two halves are placed at locations far apart in memory. This option is
14985 used when compiling @file{crtstuff.c}; you should not need to use it
14988 @item -fverbose-asm
14989 @opindex fverbose-asm
14990 Put extra commentary information in the generated assembly code to
14991 make it more readable. This option is generally only of use to those
14992 who actually need to read the generated assembly code (perhaps while
14993 debugging the compiler itself).
14995 @option{-fno-verbose-asm}, the default, causes the
14996 extra information to be omitted and is useful when comparing two assembler
14999 @item -frecord-gcc-switches
15000 @opindex frecord-gcc-switches
15001 This switch causes the command line that was used to invoke the
15002 compiler to be recorded into the object file that is being created.
15003 This switch is only implemented on some targets and the exact format
15004 of the recording is target and binary file format dependent, but it
15005 usually takes the form of a section containing ASCII text. This
15006 switch is related to the @option{-fverbose-asm} switch, but that
15007 switch only records information in the assembler output file as
15008 comments, so it never reaches the object file.
15012 @cindex global offset table
15014 Generate position-independent code (PIC) suitable for use in a shared
15015 library, if supported for the target machine. Such code accesses all
15016 constant addresses through a global offset table (GOT)@. The dynamic
15017 loader resolves the GOT entries when the program starts (the dynamic
15018 loader is not part of GCC; it is part of the operating system). If
15019 the GOT size for the linked executable exceeds a machine-specific
15020 maximum size, you get an error message from the linker indicating that
15021 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
15022 instead. (These maximums are 8k on the SPARC and 32k
15023 on the m68k and RS/6000. The 386 has no such limit.)
15025 Position-independent code requires special support, and therefore works
15026 only on certain machines. For the 386, GCC supports PIC for System V
15027 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
15028 position-independent.
15030 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15035 If supported for the target machine, emit position-independent code,
15036 suitable for dynamic linking and avoiding any limit on the size of the
15037 global offset table. This option makes a difference on the m68k,
15038 PowerPC and SPARC@.
15040 Position-independent code requires special support, and therefore works
15041 only on certain machines.
15043 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
15050 These options are similar to @option{-fpic} and @option{-fPIC}, but
15051 generated position independent code can be only linked into executables.
15052 Usually these options are used when @option{-pie} GCC option will be
15053 used during linking.
15055 @option{-fpie} and @option{-fPIE} both define the macros
15056 @code{__pie__} and @code{__PIE__}. The macros have the value 1
15057 for @option{-fpie} and 2 for @option{-fPIE}.
15059 @item -fno-jump-tables
15060 @opindex fno-jump-tables
15061 Do not use jump tables for switch statements even where it would be
15062 more efficient than other code generation strategies. This option is
15063 of use in conjunction with @option{-fpic} or @option{-fPIC} for
15064 building code which forms part of a dynamic linker and cannot
15065 reference the address of a jump table. On some targets, jump tables
15066 do not require a GOT and this option is not needed.
15068 @item -ffixed-@var{reg}
15070 Treat the register named @var{reg} as a fixed register; generated code
15071 should never refer to it (except perhaps as a stack pointer, frame
15072 pointer or in some other fixed role).
15074 @var{reg} must be the name of a register. The register names accepted
15075 are machine-specific and are defined in the @code{REGISTER_NAMES}
15076 macro in the machine description macro file.
15078 This flag does not have a negative form, because it specifies a
15081 @item -fcall-used-@var{reg}
15082 @opindex fcall-used
15083 Treat the register named @var{reg} as an allocable register that is
15084 clobbered by function calls. It may be allocated for temporaries or
15085 variables that do not live across a call. Functions compiled this way
15086 will not save and restore the register @var{reg}.
15088 It is an error to used this flag with the frame pointer or stack pointer.
15089 Use of this flag for other registers that have fixed pervasive roles in
15090 the machine's execution model will produce disastrous results.
15092 This flag does not have a negative form, because it specifies a
15095 @item -fcall-saved-@var{reg}
15096 @opindex fcall-saved
15097 Treat the register named @var{reg} as an allocable register saved by
15098 functions. It may be allocated even for temporaries or variables that
15099 live across a call. Functions compiled this way will save and restore
15100 the register @var{reg} if they use it.
15102 It is an error to used this flag with the frame pointer or stack pointer.
15103 Use of this flag for other registers that have fixed pervasive roles in
15104 the machine's execution model will produce disastrous results.
15106 A different sort of disaster will result from the use of this flag for
15107 a register in which function values may be returned.
15109 This flag does not have a negative form, because it specifies a
15112 @item -fpack-struct[=@var{n}]
15113 @opindex fpack-struct
15114 Without a value specified, pack all structure members together without
15115 holes. When a value is specified (which must be a small power of two), pack
15116 structure members according to this value, representing the maximum
15117 alignment (that is, objects with default alignment requirements larger than
15118 this will be output potentially unaligned at the next fitting location.
15120 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
15121 code that is not binary compatible with code generated without that switch.
15122 Additionally, it makes the code suboptimal.
15123 Use it to conform to a non-default application binary interface.
15125 @item -finstrument-functions
15126 @opindex finstrument-functions
15127 Generate instrumentation calls for entry and exit to functions. Just
15128 after function entry and just before function exit, the following
15129 profiling functions will be called with the address of the current
15130 function and its call site. (On some platforms,
15131 @code{__builtin_return_address} does not work beyond the current
15132 function, so the call site information may not be available to the
15133 profiling functions otherwise.)
15136 void __cyg_profile_func_enter (void *this_fn,
15138 void __cyg_profile_func_exit (void *this_fn,
15142 The first argument is the address of the start of the current function,
15143 which may be looked up exactly in the symbol table.
15145 This instrumentation is also done for functions expanded inline in other
15146 functions. The profiling calls will indicate where, conceptually, the
15147 inline function is entered and exited. This means that addressable
15148 versions of such functions must be available. If all your uses of a
15149 function are expanded inline, this may mean an additional expansion of
15150 code size. If you use @samp{extern inline} in your C code, an
15151 addressable version of such functions must be provided. (This is
15152 normally the case anyways, but if you get lucky and the optimizer always
15153 expands the functions inline, you might have gotten away without
15154 providing static copies.)
15156 A function may be given the attribute @code{no_instrument_function}, in
15157 which case this instrumentation will not be done. This can be used, for
15158 example, for the profiling functions listed above, high-priority
15159 interrupt routines, and any functions from which the profiling functions
15160 cannot safely be called (perhaps signal handlers, if the profiling
15161 routines generate output or allocate memory).
15163 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
15164 @opindex finstrument-functions-exclude-file-list
15166 Set the list of functions that are excluded from instrumentation (see
15167 the description of @code{-finstrument-functions}). If the file that
15168 contains a function definition matches with one of @var{file}, then
15169 that function is not instrumented. The match is done on substrings:
15170 if the @var{file} parameter is a substring of the file name, it is
15171 considered to be a match.
15174 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
15175 will exclude any inline function defined in files whose pathnames
15176 contain @code{/bits/stl} or @code{include/sys}.
15178 If, for some reason, you want to include letter @code{','} in one of
15179 @var{sym}, write @code{'\,'}. For example,
15180 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
15181 (note the single quote surrounding the option).
15183 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
15184 @opindex finstrument-functions-exclude-function-list
15186 This is similar to @code{-finstrument-functions-exclude-file-list},
15187 but this option sets the list of function names to be excluded from
15188 instrumentation. The function name to be matched is its user-visible
15189 name, such as @code{vector<int> blah(const vector<int> &)}, not the
15190 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
15191 match is done on substrings: if the @var{sym} parameter is a substring
15192 of the function name, it is considered to be a match.
15194 @item -fstack-check
15195 @opindex fstack-check
15196 Generate code to verify that you do not go beyond the boundary of the
15197 stack. You should specify this flag if you are running in an
15198 environment with multiple threads, but only rarely need to specify it in
15199 a single-threaded environment since stack overflow is automatically
15200 detected on nearly all systems if there is only one stack.
15202 Note that this switch does not actually cause checking to be done; the
15203 operating system must do that. The switch causes generation of code
15204 to ensure that the operating system sees the stack being extended.
15206 @item -fstack-limit-register=@var{reg}
15207 @itemx -fstack-limit-symbol=@var{sym}
15208 @itemx -fno-stack-limit
15209 @opindex fstack-limit-register
15210 @opindex fstack-limit-symbol
15211 @opindex fno-stack-limit
15212 Generate code to ensure that the stack does not grow beyond a certain value,
15213 either the value of a register or the address of a symbol. If the stack
15214 would grow beyond the value, a signal is raised. For most targets,
15215 the signal is raised before the stack overruns the boundary, so
15216 it is possible to catch the signal without taking special precautions.
15218 For instance, if the stack starts at absolute address @samp{0x80000000}
15219 and grows downwards, you can use the flags
15220 @option{-fstack-limit-symbol=__stack_limit} and
15221 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
15222 of 128KB@. Note that this may only work with the GNU linker.
15224 @cindex aliasing of parameters
15225 @cindex parameters, aliased
15226 @item -fargument-alias
15227 @itemx -fargument-noalias
15228 @itemx -fargument-noalias-global
15229 @itemx -fargument-noalias-anything
15230 @opindex fargument-alias
15231 @opindex fargument-noalias
15232 @opindex fargument-noalias-global
15233 @opindex fargument-noalias-anything
15234 Specify the possible relationships among parameters and between
15235 parameters and global data.
15237 @option{-fargument-alias} specifies that arguments (parameters) may
15238 alias each other and may alias global storage.@*
15239 @option{-fargument-noalias} specifies that arguments do not alias
15240 each other, but may alias global storage.@*
15241 @option{-fargument-noalias-global} specifies that arguments do not
15242 alias each other and do not alias global storage.
15243 @option{-fargument-noalias-anything} specifies that arguments do not
15244 alias any other storage.
15246 Each language will automatically use whatever option is required by
15247 the language standard. You should not need to use these options yourself.
15249 @item -fleading-underscore
15250 @opindex fleading-underscore
15251 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
15252 change the way C symbols are represented in the object file. One use
15253 is to help link with legacy assembly code.
15255 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
15256 generate code that is not binary compatible with code generated without that
15257 switch. Use it to conform to a non-default application binary interface.
15258 Not all targets provide complete support for this switch.
15260 @item -ftls-model=@var{model}
15261 @opindex ftls-model
15262 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
15263 The @var{model} argument should be one of @code{global-dynamic},
15264 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
15266 The default without @option{-fpic} is @code{initial-exec}; with
15267 @option{-fpic} the default is @code{global-dynamic}.
15269 @item -fvisibility=@var{default|internal|hidden|protected}
15270 @opindex fvisibility
15271 Set the default ELF image symbol visibility to the specified option---all
15272 symbols will be marked with this unless overridden within the code.
15273 Using this feature can very substantially improve linking and
15274 load times of shared object libraries, produce more optimized
15275 code, provide near-perfect API export and prevent symbol clashes.
15276 It is @strong{strongly} recommended that you use this in any shared objects
15279 Despite the nomenclature, @code{default} always means public ie;
15280 available to be linked against from outside the shared object.
15281 @code{protected} and @code{internal} are pretty useless in real-world
15282 usage so the only other commonly used option will be @code{hidden}.
15283 The default if @option{-fvisibility} isn't specified is
15284 @code{default}, i.e., make every
15285 symbol public---this causes the same behavior as previous versions of
15288 A good explanation of the benefits offered by ensuring ELF
15289 symbols have the correct visibility is given by ``How To Write
15290 Shared Libraries'' by Ulrich Drepper (which can be found at
15291 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
15292 solution made possible by this option to marking things hidden when
15293 the default is public is to make the default hidden and mark things
15294 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
15295 and @code{__attribute__ ((visibility("default")))} instead of
15296 @code{__declspec(dllexport)} you get almost identical semantics with
15297 identical syntax. This is a great boon to those working with
15298 cross-platform projects.
15300 For those adding visibility support to existing code, you may find
15301 @samp{#pragma GCC visibility} of use. This works by you enclosing
15302 the declarations you wish to set visibility for with (for example)
15303 @samp{#pragma GCC visibility push(hidden)} and
15304 @samp{#pragma GCC visibility pop}.
15305 Bear in mind that symbol visibility should be viewed @strong{as
15306 part of the API interface contract} and thus all new code should
15307 always specify visibility when it is not the default ie; declarations
15308 only for use within the local DSO should @strong{always} be marked explicitly
15309 as hidden as so to avoid PLT indirection overheads---making this
15310 abundantly clear also aids readability and self-documentation of the code.
15311 Note that due to ISO C++ specification requirements, operator new and
15312 operator delete must always be of default visibility.
15314 Be aware that headers from outside your project, in particular system
15315 headers and headers from any other library you use, may not be
15316 expecting to be compiled with visibility other than the default. You
15317 may need to explicitly say @samp{#pragma GCC visibility push(default)}
15318 before including any such headers.
15320 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
15321 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
15322 no modifications. However, this means that calls to @samp{extern}
15323 functions with no explicit visibility will use the PLT, so it is more
15324 effective to use @samp{__attribute ((visibility))} and/or
15325 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
15326 declarations should be treated as hidden.
15328 Note that @samp{-fvisibility} does affect C++ vague linkage
15329 entities. This means that, for instance, an exception class that will
15330 be thrown between DSOs must be explicitly marked with default
15331 visibility so that the @samp{type_info} nodes will be unified between
15334 An overview of these techniques, their benefits and how to use them
15335 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
15341 @node Environment Variables
15342 @section Environment Variables Affecting GCC
15343 @cindex environment variables
15345 @c man begin ENVIRONMENT
15346 This section describes several environment variables that affect how GCC
15347 operates. Some of them work by specifying directories or prefixes to use
15348 when searching for various kinds of files. Some are used to specify other
15349 aspects of the compilation environment.
15351 Note that you can also specify places to search using options such as
15352 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
15353 take precedence over places specified using environment variables, which
15354 in turn take precedence over those specified by the configuration of GCC@.
15355 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
15356 GNU Compiler Collection (GCC) Internals}.
15361 @c @itemx LC_COLLATE
15363 @c @itemx LC_MONETARY
15364 @c @itemx LC_NUMERIC
15369 @c @findex LC_COLLATE
15370 @findex LC_MESSAGES
15371 @c @findex LC_MONETARY
15372 @c @findex LC_NUMERIC
15376 These environment variables control the way that GCC uses
15377 localization information that allow GCC to work with different
15378 national conventions. GCC inspects the locale categories
15379 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
15380 so. These locale categories can be set to any value supported by your
15381 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
15382 Kingdom encoded in UTF-8.
15384 The @env{LC_CTYPE} environment variable specifies character
15385 classification. GCC uses it to determine the character boundaries in
15386 a string; this is needed for some multibyte encodings that contain quote
15387 and escape characters that would otherwise be interpreted as a string
15390 The @env{LC_MESSAGES} environment variable specifies the language to
15391 use in diagnostic messages.
15393 If the @env{LC_ALL} environment variable is set, it overrides the value
15394 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
15395 and @env{LC_MESSAGES} default to the value of the @env{LANG}
15396 environment variable. If none of these variables are set, GCC
15397 defaults to traditional C English behavior.
15401 If @env{TMPDIR} is set, it specifies the directory to use for temporary
15402 files. GCC uses temporary files to hold the output of one stage of
15403 compilation which is to be used as input to the next stage: for example,
15404 the output of the preprocessor, which is the input to the compiler
15407 @item GCC_EXEC_PREFIX
15408 @findex GCC_EXEC_PREFIX
15409 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
15410 names of the subprograms executed by the compiler. No slash is added
15411 when this prefix is combined with the name of a subprogram, but you can
15412 specify a prefix that ends with a slash if you wish.
15414 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
15415 an appropriate prefix to use based on the pathname it was invoked with.
15417 If GCC cannot find the subprogram using the specified prefix, it
15418 tries looking in the usual places for the subprogram.
15420 The default value of @env{GCC_EXEC_PREFIX} is
15421 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
15422 the installed compiler. In many cases @var{prefix} is the value
15423 of @code{prefix} when you ran the @file{configure} script.
15425 Other prefixes specified with @option{-B} take precedence over this prefix.
15427 This prefix is also used for finding files such as @file{crt0.o} that are
15430 In addition, the prefix is used in an unusual way in finding the
15431 directories to search for header files. For each of the standard
15432 directories whose name normally begins with @samp{/usr/local/lib/gcc}
15433 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
15434 replacing that beginning with the specified prefix to produce an
15435 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
15436 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
15437 These alternate directories are searched first; the standard directories
15438 come next. If a standard directory begins with the configured
15439 @var{prefix} then the value of @var{prefix} is replaced by
15440 @env{GCC_EXEC_PREFIX} when looking for header files.
15442 @item COMPILER_PATH
15443 @findex COMPILER_PATH
15444 The value of @env{COMPILER_PATH} is a colon-separated list of
15445 directories, much like @env{PATH}. GCC tries the directories thus
15446 specified when searching for subprograms, if it can't find the
15447 subprograms using @env{GCC_EXEC_PREFIX}.
15450 @findex LIBRARY_PATH
15451 The value of @env{LIBRARY_PATH} is a colon-separated list of
15452 directories, much like @env{PATH}. When configured as a native compiler,
15453 GCC tries the directories thus specified when searching for special
15454 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
15455 using GCC also uses these directories when searching for ordinary
15456 libraries for the @option{-l} option (but directories specified with
15457 @option{-L} come first).
15461 @cindex locale definition
15462 This variable is used to pass locale information to the compiler. One way in
15463 which this information is used is to determine the character set to be used
15464 when character literals, string literals and comments are parsed in C and C++.
15465 When the compiler is configured to allow multibyte characters,
15466 the following values for @env{LANG} are recognized:
15470 Recognize JIS characters.
15472 Recognize SJIS characters.
15474 Recognize EUCJP characters.
15477 If @env{LANG} is not defined, or if it has some other value, then the
15478 compiler will use mblen and mbtowc as defined by the default locale to
15479 recognize and translate multibyte characters.
15483 Some additional environments variables affect the behavior of the
15486 @include cppenv.texi
15490 @node Precompiled Headers
15491 @section Using Precompiled Headers
15492 @cindex precompiled headers
15493 @cindex speed of compilation
15495 Often large projects have many header files that are included in every
15496 source file. The time the compiler takes to process these header files
15497 over and over again can account for nearly all of the time required to
15498 build the project. To make builds faster, GCC allows users to
15499 `precompile' a header file; then, if builds can use the precompiled
15500 header file they will be much faster.
15502 To create a precompiled header file, simply compile it as you would any
15503 other file, if necessary using the @option{-x} option to make the driver
15504 treat it as a C or C++ header file. You will probably want to use a
15505 tool like @command{make} to keep the precompiled header up-to-date when
15506 the headers it contains change.
15508 A precompiled header file will be searched for when @code{#include} is
15509 seen in the compilation. As it searches for the included file
15510 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
15511 compiler looks for a precompiled header in each directory just before it
15512 looks for the include file in that directory. The name searched for is
15513 the name specified in the @code{#include} with @samp{.gch} appended. If
15514 the precompiled header file can't be used, it is ignored.
15516 For instance, if you have @code{#include "all.h"}, and you have
15517 @file{all.h.gch} in the same directory as @file{all.h}, then the
15518 precompiled header file will be used if possible, and the original
15519 header will be used otherwise.
15521 Alternatively, you might decide to put the precompiled header file in a
15522 directory and use @option{-I} to ensure that directory is searched
15523 before (or instead of) the directory containing the original header.
15524 Then, if you want to check that the precompiled header file is always
15525 used, you can put a file of the same name as the original header in this
15526 directory containing an @code{#error} command.
15528 This also works with @option{-include}. So yet another way to use
15529 precompiled headers, good for projects not designed with precompiled
15530 header files in mind, is to simply take most of the header files used by
15531 a project, include them from another header file, precompile that header
15532 file, and @option{-include} the precompiled header. If the header files
15533 have guards against multiple inclusion, they will be skipped because
15534 they've already been included (in the precompiled header).
15536 If you need to precompile the same header file for different
15537 languages, targets, or compiler options, you can instead make a
15538 @emph{directory} named like @file{all.h.gch}, and put each precompiled
15539 header in the directory, perhaps using @option{-o}. It doesn't matter
15540 what you call the files in the directory, every precompiled header in
15541 the directory will be considered. The first precompiled header
15542 encountered in the directory that is valid for this compilation will
15543 be used; they're searched in no particular order.
15545 There are many other possibilities, limited only by your imagination,
15546 good sense, and the constraints of your build system.
15548 A precompiled header file can be used only when these conditions apply:
15552 Only one precompiled header can be used in a particular compilation.
15555 A precompiled header can't be used once the first C token is seen. You
15556 can have preprocessor directives before a precompiled header; you can
15557 even include a precompiled header from inside another header, so long as
15558 there are no C tokens before the @code{#include}.
15561 The precompiled header file must be produced for the same language as
15562 the current compilation. You can't use a C precompiled header for a C++
15566 The precompiled header file must have been produced by the same compiler
15567 binary as the current compilation is using.
15570 Any macros defined before the precompiled header is included must
15571 either be defined in the same way as when the precompiled header was
15572 generated, or must not affect the precompiled header, which usually
15573 means that they don't appear in the precompiled header at all.
15575 The @option{-D} option is one way to define a macro before a
15576 precompiled header is included; using a @code{#define} can also do it.
15577 There are also some options that define macros implicitly, like
15578 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
15581 @item If debugging information is output when using the precompiled
15582 header, using @option{-g} or similar, the same kind of debugging information
15583 must have been output when building the precompiled header. However,
15584 a precompiled header built using @option{-g} can be used in a compilation
15585 when no debugging information is being output.
15587 @item The same @option{-m} options must generally be used when building
15588 and using the precompiled header. @xref{Submodel Options},
15589 for any cases where this rule is relaxed.
15591 @item Each of the following options must be the same when building and using
15592 the precompiled header:
15594 @gccoptlist{-fexceptions}
15597 Some other command-line options starting with @option{-f},
15598 @option{-p}, or @option{-O} must be defined in the same way as when
15599 the precompiled header was generated. At present, it's not clear
15600 which options are safe to change and which are not; the safest choice
15601 is to use exactly the same options when generating and using the
15602 precompiled header. The following are known to be safe:
15604 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
15605 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
15606 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
15611 For all of these except the last, the compiler will automatically
15612 ignore the precompiled header if the conditions aren't met. If you
15613 find an option combination that doesn't work and doesn't cause the
15614 precompiled header to be ignored, please consider filing a bug report,
15617 If you do use differing options when generating and using the
15618 precompiled header, the actual behavior will be a mixture of the
15619 behavior for the options. For instance, if you use @option{-g} to
15620 generate the precompiled header but not when using it, you may or may
15621 not get debugging information for routines in the precompiled header.
15623 @node Running Protoize
15624 @section Running Protoize
15626 The program @code{protoize} is an optional part of GCC@. You can use
15627 it to add prototypes to a program, thus converting the program to ISO
15628 C in one respect. The companion program @code{unprotoize} does the
15629 reverse: it removes argument types from any prototypes that are found.
15631 When you run these programs, you must specify a set of source files as
15632 command line arguments. The conversion programs start out by compiling
15633 these files to see what functions they define. The information gathered
15634 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
15636 After scanning comes actual conversion. The specified files are all
15637 eligible to be converted; any files they include (whether sources or
15638 just headers) are eligible as well.
15640 But not all the eligible files are converted. By default,
15641 @code{protoize} and @code{unprotoize} convert only source and header
15642 files in the current directory. You can specify additional directories
15643 whose files should be converted with the @option{-d @var{directory}}
15644 option. You can also specify particular files to exclude with the
15645 @option{-x @var{file}} option. A file is converted if it is eligible, its
15646 directory name matches one of the specified directory names, and its
15647 name within the directory has not been excluded.
15649 Basic conversion with @code{protoize} consists of rewriting most
15650 function definitions and function declarations to specify the types of
15651 the arguments. The only ones not rewritten are those for varargs
15654 @code{protoize} optionally inserts prototype declarations at the
15655 beginning of the source file, to make them available for any calls that
15656 precede the function's definition. Or it can insert prototype
15657 declarations with block scope in the blocks where undeclared functions
15660 Basic conversion with @code{unprotoize} consists of rewriting most
15661 function declarations to remove any argument types, and rewriting
15662 function definitions to the old-style pre-ISO form.
15664 Both conversion programs print a warning for any function declaration or
15665 definition that they can't convert. You can suppress these warnings
15668 The output from @code{protoize} or @code{unprotoize} replaces the
15669 original source file. The original file is renamed to a name ending
15670 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
15671 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
15672 for DOS) file already exists, then the source file is simply discarded.
15674 @code{protoize} and @code{unprotoize} both depend on GCC itself to
15675 scan the program and collect information about the functions it uses.
15676 So neither of these programs will work until GCC is installed.
15678 Here is a table of the options you can use with @code{protoize} and
15679 @code{unprotoize}. Each option works with both programs unless
15683 @item -B @var{directory}
15684 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
15685 usual directory (normally @file{/usr/local/lib}). This file contains
15686 prototype information about standard system functions. This option
15687 applies only to @code{protoize}.
15689 @item -c @var{compilation-options}
15690 Use @var{compilation-options} as the options when running @command{gcc} to
15691 produce the @samp{.X} files. The special option @option{-aux-info} is
15692 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
15694 Note that the compilation options must be given as a single argument to
15695 @code{protoize} or @code{unprotoize}. If you want to specify several
15696 @command{gcc} options, you must quote the entire set of compilation options
15697 to make them a single word in the shell.
15699 There are certain @command{gcc} arguments that you cannot use, because they
15700 would produce the wrong kind of output. These include @option{-g},
15701 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
15702 the @var{compilation-options}, they are ignored.
15705 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
15706 systems) instead of @samp{.c}. This is convenient if you are converting
15707 a C program to C++. This option applies only to @code{protoize}.
15710 Add explicit global declarations. This means inserting explicit
15711 declarations at the beginning of each source file for each function
15712 that is called in the file and was not declared. These declarations
15713 precede the first function definition that contains a call to an
15714 undeclared function. This option applies only to @code{protoize}.
15716 @item -i @var{string}
15717 Indent old-style parameter declarations with the string @var{string}.
15718 This option applies only to @code{protoize}.
15720 @code{unprotoize} converts prototyped function definitions to old-style
15721 function definitions, where the arguments are declared between the
15722 argument list and the initial @samp{@{}. By default, @code{unprotoize}
15723 uses five spaces as the indentation. If you want to indent with just
15724 one space instead, use @option{-i " "}.
15727 Keep the @samp{.X} files. Normally, they are deleted after conversion
15731 Add explicit local declarations. @code{protoize} with @option{-l} inserts
15732 a prototype declaration for each function in each block which calls the
15733 function without any declaration. This option applies only to
15737 Make no real changes. This mode just prints information about the conversions
15738 that would have been done without @option{-n}.
15741 Make no @samp{.save} files. The original files are simply deleted.
15742 Use this option with caution.
15744 @item -p @var{program}
15745 Use the program @var{program} as the compiler. Normally, the name
15746 @file{gcc} is used.
15749 Work quietly. Most warnings are suppressed.
15752 Print the version number, just like @option{-v} for @command{gcc}.
15755 If you need special compiler options to compile one of your program's
15756 source files, then you should generate that file's @samp{.X} file
15757 specially, by running @command{gcc} on that source file with the
15758 appropriate options and the option @option{-aux-info}. Then run
15759 @code{protoize} on the entire set of files. @code{protoize} will use
15760 the existing @samp{.X} file because it is newer than the source file.
15764 gcc -Dfoo=bar file1.c -aux-info file1.X
15769 You need to include the special files along with the rest in the
15770 @code{protoize} command, even though their @samp{.X} files already
15771 exist, because otherwise they won't get converted.
15773 @xref{Protoize Caveats}, for more information on how to use
15774 @code{protoize} successfully.