1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
11 @c man begin COPYRIGHT
12 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
13 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
15 Permission is granted to copy, distribute and/or modify this document
16 under the terms of the GNU Free Documentation License, Version 1.2 or
17 any later version published by the Free Software Foundation; with the
18 Invariant Sections being ``GNU General Public License'' and ``Funding
19 Free Software'', the Front-Cover texts being (a) (see below), and with
20 the Back-Cover Texts being (b) (see below). A copy of the license is
21 included in the gfdl(7) man page.
23 (a) The FSF's Front-Cover Text is:
27 (b) The FSF's Back-Cover Text is:
29 You have freedom to copy and modify this GNU Manual, like GNU
30 software. Copies published by the Free Software Foundation raise
31 funds for GNU development.
33 @c Set file name and title for the man page.
35 @settitle GNU project C and C++ compiler
37 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
38 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
39 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
40 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
41 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
42 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
43 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
45 Only the most useful options are listed here; see below for the
46 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
49 gpl(7), gfdl(7), fsf-funding(7),
50 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
51 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
52 @file{ld}, @file{binutils} and @file{gdb}.
55 For instructions on reporting bugs, see
56 @w{@uref{http://gcc.gnu.org/bugs.html}}.
59 See the Info entry for @command{gcc}, or
60 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
61 for contributors to GCC@.
66 @chapter GCC Command Options
67 @cindex GCC command options
68 @cindex command options
69 @cindex options, GCC command
71 @c man begin DESCRIPTION
72 When you invoke GCC, it normally does preprocessing, compilation,
73 assembly and linking. The ``overall options'' allow you to stop this
74 process at an intermediate stage. For example, the @option{-c} option
75 says not to run the linker. Then the output consists of object files
76 output by the assembler.
78 Other options are passed on to one stage of processing. Some options
79 control the preprocessor and others the compiler itself. Yet other
80 options control the assembler and linker; most of these are not
81 documented here, since you rarely need to use any of them.
83 @cindex C compilation options
84 Most of the command line options that you can use with GCC are useful
85 for C programs; when an option is only useful with another language
86 (usually C++), the explanation says so explicitly. If the description
87 for a particular option does not mention a source language, you can use
88 that option with all supported languages.
90 @cindex C++ compilation options
91 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
92 options for compiling C++ programs.
94 @cindex grouping options
95 @cindex options, grouping
96 The @command{gcc} program accepts options and file names as operands. Many
97 options have multi-letter names; therefore multiple single-letter options
98 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
101 @cindex order of options
102 @cindex options, order
103 You can mix options and other arguments. For the most part, the order
104 you use doesn't matter. Order does matter when you use several options
105 of the same kind; for example, if you specify @option{-L} more than once,
106 the directories are searched in the order specified.
108 Many options have long names starting with @samp{-f} or with
109 @samp{-W}---for example,
110 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
111 these have both positive and negative forms; the negative form of
112 @option{-ffoo} would be @option{-fno-foo}. This manual documents
113 only one of these two forms, whichever one is not the default.
117 @xref{Option Index}, for an index to GCC's options.
120 * Option Summary:: Brief list of all options, without explanations.
121 * Overall Options:: Controlling the kind of output:
122 an executable, object files, assembler files,
123 or preprocessed source.
124 * Invoking G++:: Compiling C++ programs.
125 * C Dialect Options:: Controlling the variant of C language compiled.
126 * C++ Dialect Options:: Variations on C++.
127 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
129 * Language Independent Options:: Controlling how diagnostics should be
131 * Warning Options:: How picky should the compiler be?
132 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
133 * Optimize Options:: How much optimization?
134 * Preprocessor Options:: Controlling header files and macro definitions.
135 Also, getting dependency information for Make.
136 * Assembler Options:: Passing options to the assembler.
137 * Link Options:: Specifying libraries and so on.
138 * Directory Options:: Where to find header files and libraries.
139 Where to find the compiler executable files.
140 * Spec Files:: How to pass switches to sub-processes.
141 * Target Options:: Running a cross-compiler, or an old version of GCC.
142 * Submodel Options:: Specifying minor hardware or convention variations,
143 such as 68010 vs 68020.
144 * Code Gen Options:: Specifying conventions for function calls, data layout
146 * Environment Variables:: Env vars that affect GCC.
147 * Precompiled Headers:: Compiling a header once, and using it many times.
148 * Running Protoize:: Automatically adding or removing function prototypes.
154 @section Option Summary
156 Here is a summary of all the options, grouped by type. Explanations are
157 in the following sections.
160 @item Overall Options
161 @xref{Overall Options,,Options Controlling the Kind of Output}.
162 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
163 -x @var{language} -v -### --help --target-help --version @@@var{file}}
165 @item C Language Options
166 @xref{C Dialect Options,,Options Controlling C Dialect}.
167 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
168 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
169 -fhosted -ffreestanding -fms-extensions @gol
170 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
171 -fallow-single-precision -fcond-mismatch @gol
172 -fsigned-bitfields -fsigned-char @gol
173 -funsigned-bitfields -funsigned-char}
175 @item C++ Language Options
176 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
177 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
178 -fconserve-space -ffriend-injection @gol
179 -fno-elide-constructors @gol
180 -fno-enforce-eh-specs @gol
181 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
182 -fno-implicit-templates @gol
183 -fno-implicit-inline-templates @gol
184 -fno-implement-inlines -fms-extensions @gol
185 -fno-nonansi-builtins -fno-operator-names @gol
186 -fno-optional-diags -fpermissive @gol
187 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
188 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
189 -fno-default-inline -fvisibility-inlines-hidden @gol
190 -Wabi -Wctor-dtor-privacy @gol
191 -Wnon-virtual-dtor -Wreorder @gol
192 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
193 -Wno-non-template-friend -Wold-style-cast @gol
194 -Woverloaded-virtual -Wno-pmf-conversions @gol
197 @item Objective-C and Objective-C++ Language Options
198 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
199 Objective-C and Objective-C++ Dialects}.
201 -fconstant-string-class=@var{class-name} @gol
202 -fgnu-runtime -fnext-runtime @gol
203 -fno-nil-receivers @gol
204 -fobjc-call-cxx-cdtors @gol
205 -fobjc-direct-dispatch @gol
206 -fobjc-exceptions @gol
208 -freplace-objc-classes @gol
211 -Wassign-intercept @gol
212 -Wno-protocol -Wselector @gol
213 -Wstrict-selector-match @gol
214 -Wundeclared-selector}
216 @item Language Independent Options
217 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
218 @gccoptlist{-fmessage-length=@var{n} @gol
219 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}} @gol
220 -fdiagnostics-show-options
222 @item Warning Options
223 @xref{Warning Options,,Options to Request or Suppress Warnings}.
224 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
225 -w -Wextra -Wall -Waggregate-return -Walways-true -Wno-attributes @gol
226 -Wc++-compat -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
227 -Wconversion -Wno-deprecated-declarations @gol
228 -Wdisabled-optimization -Wno-div-by-zero -Wno-endif-labels @gol
229 -Werror -Werror-* -Werror-implicit-function-declaration @gol
230 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
231 -Wno-format-extra-args -Wformat-nonliteral @gol
232 -Wformat-security -Wformat-y2k @gol
233 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
234 -Wimport -Wno-import -Winit-self -Winline @gol
235 -Wno-int-to-pointer-cast @gol
236 -Wno-invalid-offsetof -Winvalid-pch @gol
237 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
238 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
239 -Wmissing-format-attribute -Wmissing-include-dirs @gol
240 -Wmissing-noreturn @gol
241 -Wno-multichar -Wnonnull -Wno-overflow @gol
242 -Woverlength-strings -Wpacked -Wpadded @gol
243 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
244 -Wredundant-decls @gol
245 -Wreturn-type -Wsequence-point -Wshadow @gol
246 -Wsign-compare -Wstack-protector @gol
247 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
248 -Wstring-literal-comparison @gol
249 -Wswitch -Wswitch-default -Wswitch-enum @gol
250 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
251 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
252 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
253 -Wunused-value -Wunused-variable -Wvariadic-macros @gol
254 -Wvolatile-register-var -Wwrite-strings}
256 @item C-only Warning Options
257 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
258 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
259 -Wstrict-prototypes -Wtraditional @gol
260 -Wdeclaration-after-statement -Wpointer-sign}
262 @item Debugging Options
263 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
264 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
265 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
266 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
267 -fdump-ipa-all -fdump-ipa-cgraph @gol
269 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
270 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
271 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
272 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
274 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
275 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
276 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
277 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
278 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
279 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-nrv -fdump-tree-vect @gol
283 -fdump-tree-sink @gol
284 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-salias @gol
286 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
288 -ftree-vectorizer-verbose=@var{n} @gol
289 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
290 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
291 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
292 -fmem-report -fprofile-arcs @gol
293 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
294 -ftest-coverage -ftime-report -fvar-tracking @gol
295 -g -g@var{level} -gcoff -gdwarf-2 @gol
296 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
297 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
298 -print-multi-directory -print-multi-lib @gol
299 -print-prog-name=@var{program} -print-search-dirs -Q @gol
302 @item Optimization Options
303 @xref{Optimize Options,,Options that Control Optimization}.
304 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
305 -falign-labels=@var{n} -falign-loops=@var{n} @gol
306 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
307 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
308 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
309 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
310 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
311 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
312 -fexpensive-optimizations -ffast-math -ffloat-store @gol
313 -fforce-addr -ffunction-sections @gol
314 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
315 -fcrossjumping -fif-conversion -fif-conversion2 @gol
316 -finline-functions -finline-functions-called-once @gol
317 -finline-limit=@var{n} -fkeep-inline-functions @gol
318 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
319 -fmodulo-sched -fno-branch-count-reg @gol
320 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
321 -fno-function-cse -fno-guess-branch-probability @gol
322 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
323 -funsafe-math-optimizations -funsafe-loop-optimizations -ffinite-math-only @gol
324 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
325 -fomit-frame-pointer -foptimize-register-move @gol
326 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
327 -fprofile-generate -fprofile-use @gol
328 -fregmove -frename-registers @gol
329 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
330 -frerun-cse-after-loop @gol
331 -frounding-math -frtl-abstract-sequences @gol
332 -fschedule-insns -fschedule-insns2 @gol
333 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
334 -fsched-spec-load-dangerous @gol
335 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
336 -fsched2-use-superblocks @gol
337 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
338 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
339 -fstack-protector -fstack-protector-all @gol
340 -fstrict-aliasing -ftracer -fthread-jumps @gol
341 -funroll-all-loops -funroll-loops -fpeel-loops @gol
342 -fsplit-ivs-in-unroller -funswitch-loops @gol
343 -fvariable-expansion-in-unroller @gol
344 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
345 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
346 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
347 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
348 -ftree-vect-loop-version -ftree-salias -fipa-pta -fweb @gol
349 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
350 --param @var{name}=@var{value}
351 -O -O0 -O1 -O2 -O3 -Os}
353 @item Preprocessor Options
354 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
355 @gccoptlist{-A@var{question}=@var{answer} @gol
356 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
357 -C -dD -dI -dM -dN @gol
358 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
359 -idirafter @var{dir} @gol
360 -include @var{file} -imacros @var{file} @gol
361 -iprefix @var{file} -iwithprefix @var{dir} @gol
362 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
363 -imultilib @var{dir} -isysroot @var{dir} @gol
364 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
365 -P -fworking-directory -remap @gol
366 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
367 -Xpreprocessor @var{option}}
369 @item Assembler Option
370 @xref{Assembler Options,,Passing Options to the Assembler}.
371 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
374 @xref{Link Options,,Options for Linking}.
375 @gccoptlist{@var{object-file-name} -l@var{library} @gol
376 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
377 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
378 -Wl,@var{option} -Xlinker @var{option} @gol
381 @item Directory Options
382 @xref{Directory Options,,Options for Directory Search}.
383 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
384 -specs=@var{file} -I- --sysroot=@var{dir}}
387 @c I wrote this xref this way to avoid overfull hbox. -- rms
388 @xref{Target Options}.
389 @gccoptlist{-V @var{version} -b @var{machine}}
391 @item Machine Dependent Options
392 @xref{Submodel Options,,Hardware Models and Configurations}.
393 @c This list is ordered alphanumerically by subsection name.
394 @c Try and put the significant identifier (CPU or system) first,
395 @c so users have a clue at guessing where the ones they want will be.
398 @gccoptlist{-EB -EL @gol
399 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
400 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
403 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
404 -mabi=@var{name} @gol
405 -mapcs-stack-check -mno-apcs-stack-check @gol
406 -mapcs-float -mno-apcs-float @gol
407 -mapcs-reentrant -mno-apcs-reentrant @gol
408 -msched-prolog -mno-sched-prolog @gol
409 -mlittle-endian -mbig-endian -mwords-little-endian @gol
410 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
411 -mthumb-interwork -mno-thumb-interwork @gol
412 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
413 -mstructure-size-boundary=@var{n} @gol
414 -mabort-on-noreturn @gol
415 -mlong-calls -mno-long-calls @gol
416 -msingle-pic-base -mno-single-pic-base @gol
417 -mpic-register=@var{reg} @gol
418 -mnop-fun-dllimport @gol
419 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
420 -mpoke-function-name @gol
422 -mtpcs-frame -mtpcs-leaf-frame @gol
423 -mcaller-super-interworking -mcallee-super-interworking @gol
427 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
428 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
430 @emph{Blackfin Options}
431 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
432 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
433 -mlow-64k -mno-low64k -mid-shared-library @gol
434 -mno-id-shared-library -mshared-library-id=@var{n} @gol
435 -mlong-calls -mno-long-calls}
438 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
439 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
440 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
441 -mstack-align -mdata-align -mconst-align @gol
442 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
443 -melf -maout -melinux -mlinux -sim -sim2 @gol
444 -mmul-bug-workaround -mno-mul-bug-workaround}
447 @gccoptlist{-mmac -mpush-args}
449 @emph{Darwin Options}
450 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
451 -arch_only -bind_at_load -bundle -bundle_loader @gol
452 -client_name -compatibility_version -current_version @gol
454 -dependency-file -dylib_file -dylinker_install_name @gol
455 -dynamic -dynamiclib -exported_symbols_list @gol
456 -filelist -flat_namespace -force_cpusubtype_ALL @gol
457 -force_flat_namespace -headerpad_max_install_names @gol
458 -image_base -init -install_name -keep_private_externs @gol
459 -multi_module -multiply_defined -multiply_defined_unused @gol
460 -noall_load -no_dead_strip_inits_and_terms @gol
461 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
462 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
463 -private_bundle -read_only_relocs -sectalign @gol
464 -sectobjectsymbols -whyload -seg1addr @gol
465 -sectcreate -sectobjectsymbols -sectorder @gol
466 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
467 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
468 -segprot -segs_read_only_addr -segs_read_write_addr @gol
469 -single_module -static -sub_library -sub_umbrella @gol
470 -twolevel_namespace -umbrella -undefined @gol
471 -unexported_symbols_list -weak_reference_mismatches @gol
472 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
475 @emph{DEC Alpha Options}
476 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
477 -mieee -mieee-with-inexact -mieee-conformant @gol
478 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
479 -mtrap-precision=@var{mode} -mbuild-constants @gol
480 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
481 -mbwx -mmax -mfix -mcix @gol
482 -mfloat-vax -mfloat-ieee @gol
483 -mexplicit-relocs -msmall-data -mlarge-data @gol
484 -msmall-text -mlarge-text @gol
485 -mmemory-latency=@var{time}}
487 @emph{DEC Alpha/VMS Options}
488 @gccoptlist{-mvms-return-codes}
491 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
492 -mhard-float -msoft-float @gol
493 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
494 -mdouble -mno-double @gol
495 -mmedia -mno-media -mmuladd -mno-muladd @gol
496 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
497 -mlinked-fp -mlong-calls -malign-labels @gol
498 -mlibrary-pic -macc-4 -macc-8 @gol
499 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
500 -moptimize-membar -mno-optimize-membar @gol
501 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
502 -mvliw-branch -mno-vliw-branch @gol
503 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
504 -mno-nested-cond-exec -mtomcat-stats @gol
508 @emph{GNU/Linux Options}
509 @gccoptlist{-muclibc}
511 @emph{H8/300 Options}
512 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
515 @gccoptlist{-march=@var{architecture-type} @gol
516 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
517 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
518 -mfixed-range=@var{register-range} @gol
519 -mjump-in-delay -mlinker-opt -mlong-calls @gol
520 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
521 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
522 -mno-jump-in-delay -mno-long-load-store @gol
523 -mno-portable-runtime -mno-soft-float @gol
524 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
525 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
526 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
527 -munix=@var{unix-std} -nolibdld -static -threads}
529 @emph{i386 and x86-64 Options}
530 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
531 -mfpmath=@var{unit} @gol
532 -masm=@var{dialect} -mno-fancy-math-387 @gol
533 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
534 -mno-wide-multiply -mrtd -malign-double @gol
535 -mpreferred-stack-boundary=@var{num} @gol
536 -mmmx -msse -msse2 -msse3 -m3dnow -msselibm @gol
537 -mthreads -mno-align-stringops -minline-all-stringops @gol
538 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
539 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
540 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
541 -mcmodel=@var{code-model} @gol
542 -m32 -m64 -mlarge-data-threshold=@var{num}}
545 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
546 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
547 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
548 -minline-float-divide-max-throughput @gol
549 -minline-int-divide-min-latency @gol
550 -minline-int-divide-max-throughput @gol
551 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
552 -mno-dwarf2-asm -mearly-stop-bits @gol
553 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
554 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
555 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
556 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
557 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
558 -mno-sched-prefer-non-data-spec-insns @gol
559 -mno-sched-prefer-non-control-spec-insns @gol
560 -mno-sched-count-spec-in-critical-path}
562 @emph{M32R/D Options}
563 @gccoptlist{-m32r2 -m32rx -m32r @gol
565 -malign-loops -mno-align-loops @gol
566 -missue-rate=@var{number} @gol
567 -mbranch-cost=@var{number} @gol
568 -mmodel=@var{code-size-model-type} @gol
569 -msdata=@var{sdata-type} @gol
570 -mno-flush-func -mflush-func=@var{name} @gol
571 -mno-flush-trap -mflush-trap=@var{number} @gol
575 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
577 @emph{M680x0 Options}
578 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
579 -m68060 -mcpu32 -m5200 -mcfv4e -m68881 -mbitfield @gol
580 -mc68000 -mc68020 @gol
581 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
582 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
583 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
585 @emph{M68hc1x Options}
586 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
587 -mauto-incdec -minmax -mlong-calls -mshort @gol
588 -msoft-reg-count=@var{count}}
591 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
592 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
593 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
594 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
595 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
598 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
599 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
600 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
601 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
602 -mfp32 -mfp64 -mhard-float -msoft-float @gol
603 -msingle-float -mdouble-float -mdsp -mpaired-single -mips3d @gol
604 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
605 -G@var{num} -membedded-data -mno-embedded-data @gol
606 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
607 -msplit-addresses -mno-split-addresses @gol
608 -mexplicit-relocs -mno-explicit-relocs @gol
609 -mcheck-zero-division -mno-check-zero-division @gol
610 -mdivide-traps -mdivide-breaks @gol
611 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
612 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
613 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
614 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
615 -mfix-sb1 -mno-fix-sb1 @gol
616 -mflush-func=@var{func} -mno-flush-func @gol
617 -mbranch-likely -mno-branch-likely @gol
618 -mfp-exceptions -mno-fp-exceptions @gol
619 -mvr4130-align -mno-vr4130-align}
622 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
623 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
624 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
625 -mno-base-addresses -msingle-exit -mno-single-exit}
627 @emph{MN10300 Options}
628 @gccoptlist{-mmult-bug -mno-mult-bug @gol
629 -mam33 -mno-am33 @gol
630 -mam33-2 -mno-am33-2 @gol
631 -mreturn-pointer-on-d0 @gol
635 @gccoptlist{-mno-crt0 -mbacc -msim @gol
636 -march=@var{cpu-type} }
638 @emph{PDP-11 Options}
639 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
640 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
641 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
642 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
643 -mbranch-expensive -mbranch-cheap @gol
644 -msplit -mno-split -munix-asm -mdec-asm}
646 @emph{PowerPC Options}
647 See RS/6000 and PowerPC Options.
649 @emph{RS/6000 and PowerPC Options}
650 @gccoptlist{-mcpu=@var{cpu-type} @gol
651 -mtune=@var{cpu-type} @gol
652 -mpower -mno-power -mpower2 -mno-power2 @gol
653 -mpowerpc -mpowerpc64 -mno-powerpc @gol
654 -maltivec -mno-altivec @gol
655 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
656 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
657 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
658 -mnew-mnemonics -mold-mnemonics @gol
659 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
660 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
661 -malign-power -malign-natural @gol
662 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
663 -mstring -mno-string -mupdate -mno-update @gol
664 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
665 -mstrict-align -mno-strict-align -mrelocatable @gol
666 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
667 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
668 -mdynamic-no-pic -maltivec -mswdiv @gol
669 -mprioritize-restricted-insns=@var{priority} @gol
670 -msched-costly-dep=@var{dependence_type} @gol
671 -minsert-sched-nops=@var{scheme} @gol
672 -mcall-sysv -mcall-netbsd @gol
673 -maix-struct-return -msvr4-struct-return @gol
674 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
675 -misel -mno-isel @gol
676 -misel=yes -misel=no @gol
678 -mspe=yes -mspe=no @gol
679 -mvrsave -mno-vrsave @gol
680 -mmulhw -mno-mulhw @gol
681 -mdlmzb -mno-dlmzb @gol
682 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
683 -mprototype -mno-prototype @gol
684 -msim -mmvme -mads -myellowknife -memb -msdata @gol
685 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
687 @emph{S/390 and zSeries Options}
688 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
689 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
690 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
691 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
692 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
693 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
694 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
697 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
698 -m4-nofpu -m4-single-only -m4-single -m4 @gol
699 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
700 -m5-64media -m5-64media-nofpu @gol
701 -m5-32media -m5-32media-nofpu @gol
702 -m5-compact -m5-compact-nofpu @gol
703 -mb -ml -mdalign -mrelax @gol
704 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
705 -mieee -misize -mpadstruct -mspace @gol
706 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
707 -mdivsi3_libfunc=@var{name} @gol
708 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
712 @gccoptlist{-mcpu=@var{cpu-type} @gol
713 -mtune=@var{cpu-type} @gol
714 -mcmodel=@var{code-model} @gol
715 -m32 -m64 -mapp-regs -mno-app-regs @gol
716 -mfaster-structs -mno-faster-structs @gol
717 -mfpu -mno-fpu -mhard-float -msoft-float @gol
718 -mhard-quad-float -msoft-quad-float @gol
719 -mimpure-text -mno-impure-text -mlittle-endian @gol
720 -mstack-bias -mno-stack-bias @gol
721 -munaligned-doubles -mno-unaligned-doubles @gol
722 -mv8plus -mno-v8plus -mvis -mno-vis
723 -threads -pthreads -pthread}
725 @emph{System V Options}
726 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
728 @emph{TMS320C3x/C4x Options}
729 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
730 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
731 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
732 -mparallel-insns -mparallel-mpy -mpreserve-float}
735 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
736 -mprolog-function -mno-prolog-function -mspace @gol
737 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
738 -mapp-regs -mno-app-regs @gol
739 -mdisable-callt -mno-disable-callt @gol
745 @gccoptlist{-mg -mgnu -munix}
747 @emph{x86-64 Options}
748 See i386 and x86-64 Options.
750 @emph{Xstormy16 Options}
753 @emph{Xtensa Options}
754 @gccoptlist{-mconst16 -mno-const16 @gol
755 -mfused-madd -mno-fused-madd @gol
756 -mtext-section-literals -mno-text-section-literals @gol
757 -mtarget-align -mno-target-align @gol
758 -mlongcalls -mno-longcalls}
760 @emph{zSeries Options}
761 See S/390 and zSeries Options.
763 @item Code Generation Options
764 @xref{Code Gen Options,,Options for Code Generation Conventions}.
765 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
766 -ffixed-@var{reg} -fexceptions @gol
767 -fnon-call-exceptions -funwind-tables @gol
768 -fasynchronous-unwind-tables @gol
769 -finhibit-size-directive -finstrument-functions @gol
770 -fno-common -fno-ident @gol
771 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
772 -fno-jump-tables @gol
773 -freg-struct-return -fshort-enums @gol
774 -fshort-double -fshort-wchar @gol
775 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
776 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
777 -fargument-alias -fargument-noalias @gol
778 -fargument-noalias-global -fargument-noalias-anything
779 -fleading-underscore -ftls-model=@var{model} @gol
780 -ftrapv -fwrapv -fbounds-check @gol
781 -fvisibility -fopenmp}
785 * Overall Options:: Controlling the kind of output:
786 an executable, object files, assembler files,
787 or preprocessed source.
788 * C Dialect Options:: Controlling the variant of C language compiled.
789 * C++ Dialect Options:: Variations on C++.
790 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
792 * Language Independent Options:: Controlling how diagnostics should be
794 * Warning Options:: How picky should the compiler be?
795 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
796 * Optimize Options:: How much optimization?
797 * Preprocessor Options:: Controlling header files and macro definitions.
798 Also, getting dependency information for Make.
799 * Assembler Options:: Passing options to the assembler.
800 * Link Options:: Specifying libraries and so on.
801 * Directory Options:: Where to find header files and libraries.
802 Where to find the compiler executable files.
803 * Spec Files:: How to pass switches to sub-processes.
804 * Target Options:: Running a cross-compiler, or an old version of GCC.
807 @node Overall Options
808 @section Options Controlling the Kind of Output
810 Compilation can involve up to four stages: preprocessing, compilation
811 proper, assembly and linking, always in that order. GCC is capable of
812 preprocessing and compiling several files either into several
813 assembler input files, or into one assembler input file; then each
814 assembler input file produces an object file, and linking combines all
815 the object files (those newly compiled, and those specified as input)
816 into an executable file.
818 @cindex file name suffix
819 For any given input file, the file name suffix determines what kind of
824 C source code which must be preprocessed.
827 C source code which should not be preprocessed.
830 C++ source code which should not be preprocessed.
833 Objective-C source code. Note that you must link with the @file{libobjc}
834 library to make an Objective-C program work.
837 Objective-C source code which should not be preprocessed.
841 Objective-C++ source code. Note that you must link with the @file{libobjc}
842 library to make an Objective-C++ program work. Note that @samp{.M} refers
843 to a literal capital M@.
846 Objective-C++ source code which should not be preprocessed.
849 C, C++, Objective-C or Objective-C++ header file to be turned into a
854 @itemx @var{file}.cxx
855 @itemx @var{file}.cpp
856 @itemx @var{file}.CPP
857 @itemx @var{file}.c++
859 C++ source code which must be preprocessed. Note that in @samp{.cxx},
860 the last two letters must both be literally @samp{x}. Likewise,
861 @samp{.C} refers to a literal capital C@.
865 Objective-C++ source code which must be preprocessed.
868 Objective-C++ source code which should not be preprocessed.
872 C++ header file to be turned into a precompiled header.
875 @itemx @var{file}.for
876 @itemx @var{file}.FOR
877 Fixed form Fortran source code which should not be preprocessed.
880 @itemx @var{file}.fpp
881 @itemx @var{file}.FPP
882 Fixed form Fortran source code which must be preprocessed (with the traditional
886 @itemx @var{file}.f95
887 Free form Fortran source code which should not be preprocessed.
890 @itemx @var{file}.F95
891 Free form Fortran source code which must be preprocessed (with the
892 traditional preprocessor).
894 @c FIXME: Descriptions of Java file types.
901 Ada source code file which contains a library unit declaration (a
902 declaration of a package, subprogram, or generic, or a generic
903 instantiation), or a library unit renaming declaration (a package,
904 generic, or subprogram renaming declaration). Such files are also
907 @itemx @var{file}.adb
908 Ada source code file containing a library unit body (a subprogram or
909 package body). Such files are also called @dfn{bodies}.
911 @c GCC also knows about some suffixes for languages not yet included:
922 Assembler code which must be preprocessed.
925 An object file to be fed straight into linking.
926 Any file name with no recognized suffix is treated this way.
930 You can specify the input language explicitly with the @option{-x} option:
933 @item -x @var{language}
934 Specify explicitly the @var{language} for the following input files
935 (rather than letting the compiler choose a default based on the file
936 name suffix). This option applies to all following input files until
937 the next @option{-x} option. Possible values for @var{language} are:
939 c c-header c-cpp-output
940 c++ c++-header c++-cpp-output
941 objective-c objective-c-header objective-c-cpp-output
942 objective-c++ objective-c++-header objective-c++-cpp-output
943 assembler assembler-with-cpp
952 Turn off any specification of a language, so that subsequent files are
953 handled according to their file name suffixes (as they are if @option{-x}
954 has not been used at all).
956 @item -pass-exit-codes
957 @opindex pass-exit-codes
958 Normally the @command{gcc} program will exit with the code of 1 if any
959 phase of the compiler returns a non-success return code. If you specify
960 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
961 numerically highest error produced by any phase that returned an error
962 indication. The C, C++, and Fortran frontends return 4, if an internal
963 compiler error is encountered.
966 If you only want some of the stages of compilation, you can use
967 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
968 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
969 @command{gcc} is to stop. Note that some combinations (for example,
970 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
975 Compile or assemble the source files, but do not link. The linking
976 stage simply is not done. The ultimate output is in the form of an
977 object file for each source file.
979 By default, the object file name for a source file is made by replacing
980 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
982 Unrecognized input files, not requiring compilation or assembly, are
987 Stop after the stage of compilation proper; do not assemble. The output
988 is in the form of an assembler code file for each non-assembler input
991 By default, the assembler file name for a source file is made by
992 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
994 Input files that don't require compilation are ignored.
998 Stop after the preprocessing stage; do not run the compiler proper. The
999 output is in the form of preprocessed source code, which is sent to the
1002 Input files which don't require preprocessing are ignored.
1004 @cindex output file option
1007 Place output in file @var{file}. This applies regardless to whatever
1008 sort of output is being produced, whether it be an executable file,
1009 an object file, an assembler file or preprocessed C code.
1011 If @option{-o} is not specified, the default is to put an executable
1012 file in @file{a.out}, the object file for
1013 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1014 assembler file in @file{@var{source}.s}, a precompiled header file in
1015 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1020 Print (on standard error output) the commands executed to run the stages
1021 of compilation. Also print the version number of the compiler driver
1022 program and of the preprocessor and the compiler proper.
1026 Like @option{-v} except the commands are not executed and all command
1027 arguments are quoted. This is useful for shell scripts to capture the
1028 driver-generated command lines.
1032 Use pipes rather than temporary files for communication between the
1033 various stages of compilation. This fails to work on some systems where
1034 the assembler is unable to read from a pipe; but the GNU assembler has
1039 If you are compiling multiple source files, this option tells the driver
1040 to pass all the source files to the compiler at once (for those
1041 languages for which the compiler can handle this). This will allow
1042 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1043 language for which this is supported is C@. If you pass source files for
1044 multiple languages to the driver, using this option, the driver will invoke
1045 the compiler(s) that support IMA once each, passing each compiler all the
1046 source files appropriate for it. For those languages that do not support
1047 IMA this option will be ignored, and the compiler will be invoked once for
1048 each source file in that language. If you use this option in conjunction
1049 with @option{-save-temps}, the compiler will generate multiple
1051 (one for each source file), but only one (combined) @file{.o} or
1056 Print (on the standard output) a description of the command line options
1057 understood by @command{gcc}. If the @option{-v} option is also specified
1058 then @option{--help} will also be passed on to the various processes
1059 invoked by @command{gcc}, so that they can display the command line options
1060 they accept. If the @option{-Wextra} option is also specified then command
1061 line options which have no documentation associated with them will also
1065 @opindex target-help
1066 Print (on the standard output) a description of target specific command
1067 line options for each tool.
1071 Display the version number and copyrights of the invoked GCC@.
1073 @include @value{srcdir}/../libiberty/at-file.texi
1077 @section Compiling C++ Programs
1079 @cindex suffixes for C++ source
1080 @cindex C++ source file suffixes
1081 C++ source files conventionally use one of the suffixes @samp{.C},
1082 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1083 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1084 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1085 files with these names and compiles them as C++ programs even if you
1086 call the compiler the same way as for compiling C programs (usually
1087 with the name @command{gcc}).
1091 However, C++ programs often require class libraries as well as a
1092 compiler that understands the C++ language---and under some
1093 circumstances, you might want to compile programs or header files from
1094 standard input, or otherwise without a suffix that flags them as C++
1095 programs. You might also like to precompile a C header file with a
1096 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1097 program that calls GCC with the default language set to C++, and
1098 automatically specifies linking against the C++ library. On many
1099 systems, @command{g++} is also installed with the name @command{c++}.
1101 @cindex invoking @command{g++}
1102 When you compile C++ programs, you may specify many of the same
1103 command-line options that you use for compiling programs in any
1104 language; or command-line options meaningful for C and related
1105 languages; or options that are meaningful only for C++ programs.
1106 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1107 explanations of options for languages related to C@.
1108 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1109 explanations of options that are meaningful only for C++ programs.
1111 @node C Dialect Options
1112 @section Options Controlling C Dialect
1113 @cindex dialect options
1114 @cindex language dialect options
1115 @cindex options, dialect
1117 The following options control the dialect of C (or languages derived
1118 from C, such as C++, Objective-C and Objective-C++) that the compiler
1122 @cindex ANSI support
1126 In C mode, support all ISO C90 programs. In C++ mode,
1127 remove GNU extensions that conflict with ISO C++.
1129 This turns off certain features of GCC that are incompatible with ISO
1130 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1131 such as the @code{asm} and @code{typeof} keywords, and
1132 predefined macros such as @code{unix} and @code{vax} that identify the
1133 type of system you are using. It also enables the undesirable and
1134 rarely used ISO trigraph feature. For the C compiler,
1135 it disables recognition of C++ style @samp{//} comments as well as
1136 the @code{inline} keyword.
1138 The alternate keywords @code{__asm__}, @code{__extension__},
1139 @code{__inline__} and @code{__typeof__} continue to work despite
1140 @option{-ansi}. You would not want to use them in an ISO C program, of
1141 course, but it is useful to put them in header files that might be included
1142 in compilations done with @option{-ansi}. Alternate predefined macros
1143 such as @code{__unix__} and @code{__vax__} are also available, with or
1144 without @option{-ansi}.
1146 The @option{-ansi} option does not cause non-ISO programs to be
1147 rejected gratuitously. For that, @option{-pedantic} is required in
1148 addition to @option{-ansi}. @xref{Warning Options}.
1150 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1151 option is used. Some header files may notice this macro and refrain
1152 from declaring certain functions or defining certain macros that the
1153 ISO standard doesn't call for; this is to avoid interfering with any
1154 programs that might use these names for other things.
1156 Functions which would normally be built in but do not have semantics
1157 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1158 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1159 built-in functions provided by GCC}, for details of the functions
1164 Determine the language standard. This option is currently only
1165 supported when compiling C or C++. A value for this option must be
1166 provided; possible values are
1171 ISO C90 (same as @option{-ansi}).
1173 @item iso9899:199409
1174 ISO C90 as modified in amendment 1.
1180 ISO C99. Note that this standard is not yet fully supported; see
1181 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1182 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1185 Default, ISO C90 plus GNU extensions (including some C99 features).
1189 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1190 this will become the default. The name @samp{gnu9x} is deprecated.
1193 The 1998 ISO C++ standard plus amendments.
1196 The same as @option{-std=c++98} plus GNU extensions. This is the
1197 default for C++ code.
1200 Even when this option is not specified, you can still use some of the
1201 features of newer standards in so far as they do not conflict with
1202 previous C standards. For example, you may use @code{__restrict__} even
1203 when @option{-std=c99} is not specified.
1205 The @option{-std} options specifying some version of ISO C have the same
1206 effects as @option{-ansi}, except that features that were not in ISO C90
1207 but are in the specified version (for example, @samp{//} comments and
1208 the @code{inline} keyword in ISO C99) are not disabled.
1210 @xref{Standards,,Language Standards Supported by GCC}, for details of
1211 these standard versions.
1213 @item -aux-info @var{filename}
1215 Output to the given filename prototyped declarations for all functions
1216 declared and/or defined in a translation unit, including those in header
1217 files. This option is silently ignored in any language other than C@.
1219 Besides declarations, the file indicates, in comments, the origin of
1220 each declaration (source file and line), whether the declaration was
1221 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1222 @samp{O} for old, respectively, in the first character after the line
1223 number and the colon), and whether it came from a declaration or a
1224 definition (@samp{C} or @samp{F}, respectively, in the following
1225 character). In the case of function definitions, a K&R-style list of
1226 arguments followed by their declarations is also provided, inside
1227 comments, after the declaration.
1231 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1232 keyword, so that code can use these words as identifiers. You can use
1233 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1234 instead. @option{-ansi} implies @option{-fno-asm}.
1236 In C++, this switch only affects the @code{typeof} keyword, since
1237 @code{asm} and @code{inline} are standard keywords. You may want to
1238 use the @option{-fno-gnu-keywords} flag instead, which has the same
1239 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1240 switch only affects the @code{asm} and @code{typeof} keywords, since
1241 @code{inline} is a standard keyword in ISO C99.
1244 @itemx -fno-builtin-@var{function}
1245 @opindex fno-builtin
1246 @cindex built-in functions
1247 Don't recognize built-in functions that do not begin with
1248 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1249 functions provided by GCC}, for details of the functions affected,
1250 including those which are not built-in functions when @option{-ansi} or
1251 @option{-std} options for strict ISO C conformance are used because they
1252 do not have an ISO standard meaning.
1254 GCC normally generates special code to handle certain built-in functions
1255 more efficiently; for instance, calls to @code{alloca} may become single
1256 instructions that adjust the stack directly, and calls to @code{memcpy}
1257 may become inline copy loops. The resulting code is often both smaller
1258 and faster, but since the function calls no longer appear as such, you
1259 cannot set a breakpoint on those calls, nor can you change the behavior
1260 of the functions by linking with a different library. In addition,
1261 when a function is recognized as a built-in function, GCC may use
1262 information about that function to warn about problems with calls to
1263 that function, or to generate more efficient code, even if the
1264 resulting code still contains calls to that function. For example,
1265 warnings are given with @option{-Wformat} for bad calls to
1266 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1267 known not to modify global memory.
1269 With the @option{-fno-builtin-@var{function}} option
1270 only the built-in function @var{function} is
1271 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1272 function is named this is not built-in in this version of GCC, this
1273 option is ignored. There is no corresponding
1274 @option{-fbuiltin-@var{function}} option; if you wish to enable
1275 built-in functions selectively when using @option{-fno-builtin} or
1276 @option{-ffreestanding}, you may define macros such as:
1279 #define abs(n) __builtin_abs ((n))
1280 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1285 @cindex hosted environment
1287 Assert that compilation takes place in a hosted environment. This implies
1288 @option{-fbuiltin}. A hosted environment is one in which the
1289 entire standard library is available, and in which @code{main} has a return
1290 type of @code{int}. Examples are nearly everything except a kernel.
1291 This is equivalent to @option{-fno-freestanding}.
1293 @item -ffreestanding
1294 @opindex ffreestanding
1295 @cindex hosted environment
1297 Assert that compilation takes place in a freestanding environment. This
1298 implies @option{-fno-builtin}. A freestanding environment
1299 is one in which the standard library may not exist, and program startup may
1300 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1301 This is equivalent to @option{-fno-hosted}.
1303 @xref{Standards,,Language Standards Supported by GCC}, for details of
1304 freestanding and hosted environments.
1306 @item -fms-extensions
1307 @opindex fms-extensions
1308 Accept some non-standard constructs used in Microsoft header files.
1310 Some cases of unnamed fields in structures and unions are only
1311 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1312 fields within structs/unions}, for details.
1316 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1317 options for strict ISO C conformance) implies @option{-trigraphs}.
1319 @item -no-integrated-cpp
1320 @opindex no-integrated-cpp
1321 Performs a compilation in two passes: preprocessing and compiling. This
1322 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1323 @option{-B} option. The user supplied compilation step can then add in
1324 an additional preprocessing step after normal preprocessing but before
1325 compiling. The default is to use the integrated cpp (internal cpp)
1327 The semantics of this option will change if "cc1", "cc1plus", and
1328 "cc1obj" are merged.
1330 @cindex traditional C language
1331 @cindex C language, traditional
1333 @itemx -traditional-cpp
1334 @opindex traditional-cpp
1335 @opindex traditional
1336 Formerly, these options caused GCC to attempt to emulate a pre-standard
1337 C compiler. They are now only supported with the @option{-E} switch.
1338 The preprocessor continues to support a pre-standard mode. See the GNU
1339 CPP manual for details.
1341 @item -fcond-mismatch
1342 @opindex fcond-mismatch
1343 Allow conditional expressions with mismatched types in the second and
1344 third arguments. The value of such an expression is void. This option
1345 is not supported for C++.
1347 @item -funsigned-char
1348 @opindex funsigned-char
1349 Let the type @code{char} be unsigned, like @code{unsigned char}.
1351 Each kind of machine has a default for what @code{char} should
1352 be. It is either like @code{unsigned char} by default or like
1353 @code{signed char} by default.
1355 Ideally, a portable program should always use @code{signed char} or
1356 @code{unsigned char} when it depends on the signedness of an object.
1357 But many programs have been written to use plain @code{char} and
1358 expect it to be signed, or expect it to be unsigned, depending on the
1359 machines they were written for. This option, and its inverse, let you
1360 make such a program work with the opposite default.
1362 The type @code{char} is always a distinct type from each of
1363 @code{signed char} or @code{unsigned char}, even though its behavior
1364 is always just like one of those two.
1367 @opindex fsigned-char
1368 Let the type @code{char} be signed, like @code{signed char}.
1370 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1371 the negative form of @option{-funsigned-char}. Likewise, the option
1372 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1374 @item -fsigned-bitfields
1375 @itemx -funsigned-bitfields
1376 @itemx -fno-signed-bitfields
1377 @itemx -fno-unsigned-bitfields
1378 @opindex fsigned-bitfields
1379 @opindex funsigned-bitfields
1380 @opindex fno-signed-bitfields
1381 @opindex fno-unsigned-bitfields
1382 These options control whether a bit-field is signed or unsigned, when the
1383 declaration does not use either @code{signed} or @code{unsigned}. By
1384 default, such a bit-field is signed, because this is consistent: the
1385 basic integer types such as @code{int} are signed types.
1388 @node C++ Dialect Options
1389 @section Options Controlling C++ Dialect
1391 @cindex compiler options, C++
1392 @cindex C++ options, command line
1393 @cindex options, C++
1394 This section describes the command-line options that are only meaningful
1395 for C++ programs; but you can also use most of the GNU compiler options
1396 regardless of what language your program is in. For example, you
1397 might compile a file @code{firstClass.C} like this:
1400 g++ -g -frepo -O -c firstClass.C
1404 In this example, only @option{-frepo} is an option meant
1405 only for C++ programs; you can use the other options with any
1406 language supported by GCC@.
1408 Here is a list of options that are @emph{only} for compiling C++ programs:
1412 @item -fabi-version=@var{n}
1413 @opindex fabi-version
1414 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1415 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1416 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1417 the version that conforms most closely to the C++ ABI specification.
1418 Therefore, the ABI obtained using version 0 will change as ABI bugs
1421 The default is version 2.
1423 @item -fno-access-control
1424 @opindex fno-access-control
1425 Turn off all access checking. This switch is mainly useful for working
1426 around bugs in the access control code.
1430 Check that the pointer returned by @code{operator new} is non-null
1431 before attempting to modify the storage allocated. This check is
1432 normally unnecessary because the C++ standard specifies that
1433 @code{operator new} will only return @code{0} if it is declared
1434 @samp{throw()}, in which case the compiler will always check the
1435 return value even without this option. In all other cases, when
1436 @code{operator new} has a non-empty exception specification, memory
1437 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1438 @samp{new (nothrow)}.
1440 @item -fconserve-space
1441 @opindex fconserve-space
1442 Put uninitialized or runtime-initialized global variables into the
1443 common segment, as C does. This saves space in the executable at the
1444 cost of not diagnosing duplicate definitions. If you compile with this
1445 flag and your program mysteriously crashes after @code{main()} has
1446 completed, you may have an object that is being destroyed twice because
1447 two definitions were merged.
1449 This option is no longer useful on most targets, now that support has
1450 been added for putting variables into BSS without making them common.
1452 @item -ffriend-injection
1453 @opindex ffriend-injection
1454 Inject friend functions into the enclosing namespace, so that they are
1455 visible outside the scope of the class in which they are declared.
1456 Friend functions were documented to work this way in the old Annotated
1457 C++ Reference Manual, and versions of G++ before 4.1 always worked
1458 that way. However, in ISO C++ a friend function which is not declared
1459 in an enclosing scope can only be found using argument dependent
1460 lookup. This option causes friends to be injected as they were in
1463 This option is for compatibility, and may be removed in a future
1466 @item -fno-elide-constructors
1467 @opindex fno-elide-constructors
1468 The C++ standard allows an implementation to omit creating a temporary
1469 which is only used to initialize another object of the same type.
1470 Specifying this option disables that optimization, and forces G++ to
1471 call the copy constructor in all cases.
1473 @item -fno-enforce-eh-specs
1474 @opindex fno-enforce-eh-specs
1475 Don't generate code to check for violation of exception specifications
1476 at runtime. This option violates the C++ standard, but may be useful
1477 for reducing code size in production builds, much like defining
1478 @samp{NDEBUG}. This does not give user code permission to throw
1479 exceptions in violation of the exception specifications; the compiler
1480 will still optimize based on the specifications, so throwing an
1481 unexpected exception will result in undefined behavior.
1484 @itemx -fno-for-scope
1486 @opindex fno-for-scope
1487 If @option{-ffor-scope} is specified, the scope of variables declared in
1488 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1489 as specified by the C++ standard.
1490 If @option{-fno-for-scope} is specified, the scope of variables declared in
1491 a @i{for-init-statement} extends to the end of the enclosing scope,
1492 as was the case in old versions of G++, and other (traditional)
1493 implementations of C++.
1495 The default if neither flag is given to follow the standard,
1496 but to allow and give a warning for old-style code that would
1497 otherwise be invalid, or have different behavior.
1499 @item -fno-gnu-keywords
1500 @opindex fno-gnu-keywords
1501 Do not recognize @code{typeof} as a keyword, so that code can use this
1502 word as an identifier. You can use the keyword @code{__typeof__} instead.
1503 @option{-ansi} implies @option{-fno-gnu-keywords}.
1505 @item -fno-implicit-templates
1506 @opindex fno-implicit-templates
1507 Never emit code for non-inline templates which are instantiated
1508 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1509 @xref{Template Instantiation}, for more information.
1511 @item -fno-implicit-inline-templates
1512 @opindex fno-implicit-inline-templates
1513 Don't emit code for implicit instantiations of inline templates, either.
1514 The default is to handle inlines differently so that compiles with and
1515 without optimization will need the same set of explicit instantiations.
1517 @item -fno-implement-inlines
1518 @opindex fno-implement-inlines
1519 To save space, do not emit out-of-line copies of inline functions
1520 controlled by @samp{#pragma implementation}. This will cause linker
1521 errors if these functions are not inlined everywhere they are called.
1523 @item -fms-extensions
1524 @opindex fms-extensions
1525 Disable pedantic warnings about constructs used in MFC, such as implicit
1526 int and getting a pointer to member function via non-standard syntax.
1528 @item -fno-nonansi-builtins
1529 @opindex fno-nonansi-builtins
1530 Disable built-in declarations of functions that are not mandated by
1531 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1532 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1534 @item -fno-operator-names
1535 @opindex fno-operator-names
1536 Do not treat the operator name keywords @code{and}, @code{bitand},
1537 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1538 synonyms as keywords.
1540 @item -fno-optional-diags
1541 @opindex fno-optional-diags
1542 Disable diagnostics that the standard says a compiler does not need to
1543 issue. Currently, the only such diagnostic issued by G++ is the one for
1544 a name having multiple meanings within a class.
1547 @opindex fpermissive
1548 Downgrade some diagnostics about nonconformant code from errors to
1549 warnings. Thus, using @option{-fpermissive} will allow some
1550 nonconforming code to compile.
1554 Enable automatic template instantiation at link time. This option also
1555 implies @option{-fno-implicit-templates}. @xref{Template
1556 Instantiation}, for more information.
1560 Disable generation of information about every class with virtual
1561 functions for use by the C++ runtime type identification features
1562 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1563 of the language, you can save some space by using this flag. Note that
1564 exception handling uses the same information, but it will generate it as
1569 Emit statistics about front-end processing at the end of the compilation.
1570 This information is generally only useful to the G++ development team.
1572 @item -ftemplate-depth-@var{n}
1573 @opindex ftemplate-depth
1574 Set the maximum instantiation depth for template classes to @var{n}.
1575 A limit on the template instantiation depth is needed to detect
1576 endless recursions during template class instantiation. ANSI/ISO C++
1577 conforming programs must not rely on a maximum depth greater than 17.
1579 @item -fno-threadsafe-statics
1580 @opindex fno-threadsafe-statics
1581 Do not emit the extra code to use the routines specified in the C++
1582 ABI for thread-safe initialization of local statics. You can use this
1583 option to reduce code size slightly in code that doesn't need to be
1586 @item -fuse-cxa-atexit
1587 @opindex fuse-cxa-atexit
1588 Register destructors for objects with static storage duration with the
1589 @code{__cxa_atexit} function rather than the @code{atexit} function.
1590 This option is required for fully standards-compliant handling of static
1591 destructors, but will only work if your C library supports
1592 @code{__cxa_atexit}.
1594 @item -fno-use-cxa-get-exception-ptr
1595 @opindex fno-use-cxa-get-exception-ptr
1596 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1597 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1598 if the runtime routine is not available.
1600 @item -fvisibility-inlines-hidden
1601 @opindex fvisibility-inlines-hidden
1602 This switch declares that the user does not attempt to compare
1603 pointers to inline methods where the addresses of the two functions
1604 were taken in different shared objects.
1606 The effect of this is that GCC may, effectively, mark inline methods with
1607 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1608 appear in the export table of a DSO and do not require a PLT indirection
1609 when used within the DSO@. Enabling this option can have a dramatic effect
1610 on load and link times of a DSO as it massively reduces the size of the
1611 dynamic export table when the library makes heavy use of templates.
1613 The behaviour of this switch is not quite the same as marking the
1614 methods as hidden directly. Normally if there is a class with default
1615 visibility which has a hidden method, the effect of this is that the
1616 method must be defined in only one shared object. This switch does
1617 not have this restriction.
1619 You may mark a method as having a visibility explicitly to negate the
1620 effect of the switch for that method. For example, if you do want to
1621 compare pointers to a particular inline method, you might mark it as
1622 having default visibility.
1626 Do not use weak symbol support, even if it is provided by the linker.
1627 By default, G++ will use weak symbols if they are available. This
1628 option exists only for testing, and should not be used by end-users;
1629 it will result in inferior code and has no benefits. This option may
1630 be removed in a future release of G++.
1634 Do not search for header files in the standard directories specific to
1635 C++, but do still search the other standard directories. (This option
1636 is used when building the C++ library.)
1639 In addition, these optimization, warning, and code generation options
1640 have meanings only for C++ programs:
1643 @item -fno-default-inline
1644 @opindex fno-default-inline
1645 Do not assume @samp{inline} for functions defined inside a class scope.
1646 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1647 functions will have linkage like inline functions; they just won't be
1650 @item -Wabi @r{(C++ only)}
1652 Warn when G++ generates code that is probably not compatible with the
1653 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1654 all such cases, there are probably some cases that are not warned about,
1655 even though G++ is generating incompatible code. There may also be
1656 cases where warnings are emitted even though the code that is generated
1659 You should rewrite your code to avoid these warnings if you are
1660 concerned about the fact that code generated by G++ may not be binary
1661 compatible with code generated by other compilers.
1663 The known incompatibilities at this point include:
1668 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1669 pack data into the same byte as a base class. For example:
1672 struct A @{ virtual void f(); int f1 : 1; @};
1673 struct B : public A @{ int f2 : 1; @};
1677 In this case, G++ will place @code{B::f2} into the same byte
1678 as@code{A::f1}; other compilers will not. You can avoid this problem
1679 by explicitly padding @code{A} so that its size is a multiple of the
1680 byte size on your platform; that will cause G++ and other compilers to
1681 layout @code{B} identically.
1684 Incorrect handling of tail-padding for virtual bases. G++ does not use
1685 tail padding when laying out virtual bases. For example:
1688 struct A @{ virtual void f(); char c1; @};
1689 struct B @{ B(); char c2; @};
1690 struct C : public A, public virtual B @{@};
1694 In this case, G++ will not place @code{B} into the tail-padding for
1695 @code{A}; other compilers will. You can avoid this problem by
1696 explicitly padding @code{A} so that its size is a multiple of its
1697 alignment (ignoring virtual base classes); that will cause G++ and other
1698 compilers to layout @code{C} identically.
1701 Incorrect handling of bit-fields with declared widths greater than that
1702 of their underlying types, when the bit-fields appear in a union. For
1706 union U @{ int i : 4096; @};
1710 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1711 union too small by the number of bits in an @code{int}.
1714 Empty classes can be placed at incorrect offsets. For example:
1724 struct C : public B, public A @{@};
1728 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1729 it should be placed at offset zero. G++ mistakenly believes that the
1730 @code{A} data member of @code{B} is already at offset zero.
1733 Names of template functions whose types involve @code{typename} or
1734 template template parameters can be mangled incorrectly.
1737 template <typename Q>
1738 void f(typename Q::X) @{@}
1740 template <template <typename> class Q>
1741 void f(typename Q<int>::X) @{@}
1745 Instantiations of these templates may be mangled incorrectly.
1749 @item -Wctor-dtor-privacy @r{(C++ only)}
1750 @opindex Wctor-dtor-privacy
1751 Warn when a class seems unusable because all the constructors or
1752 destructors in that class are private, and it has neither friends nor
1753 public static member functions.
1755 @item -Wnon-virtual-dtor @r{(C++ only)}
1756 @opindex Wnon-virtual-dtor
1757 Warn when a class appears to be polymorphic, thereby requiring a virtual
1758 destructor, yet it declares a non-virtual one. This warning is also
1759 enabled if -Weffc++ is specified.
1761 @item -Wreorder @r{(C++ only)}
1763 @cindex reordering, warning
1764 @cindex warning for reordering of member initializers
1765 Warn when the order of member initializers given in the code does not
1766 match the order in which they must be executed. For instance:
1772 A(): j (0), i (1) @{ @}
1776 The compiler will rearrange the member initializers for @samp{i}
1777 and @samp{j} to match the declaration order of the members, emitting
1778 a warning to that effect. This warning is enabled by @option{-Wall}.
1781 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1784 @item -Weffc++ @r{(C++ only)}
1786 Warn about violations of the following style guidelines from Scott Meyers'
1787 @cite{Effective C++} book:
1791 Item 11: Define a copy constructor and an assignment operator for classes
1792 with dynamically allocated memory.
1795 Item 12: Prefer initialization to assignment in constructors.
1798 Item 14: Make destructors virtual in base classes.
1801 Item 15: Have @code{operator=} return a reference to @code{*this}.
1804 Item 23: Don't try to return a reference when you must return an object.
1808 Also warn about violations of the following style guidelines from
1809 Scott Meyers' @cite{More Effective C++} book:
1813 Item 6: Distinguish between prefix and postfix forms of increment and
1814 decrement operators.
1817 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1821 When selecting this option, be aware that the standard library
1822 headers do not obey all of these guidelines; use @samp{grep -v}
1823 to filter out those warnings.
1825 @item -Wno-deprecated @r{(C++ only)}
1826 @opindex Wno-deprecated
1827 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1829 @item -Wstrict-null-sentinel @r{(C++ only)}
1830 @opindex Wstrict-null-sentinel
1831 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1832 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1833 to @code{__null}. Although it is a null pointer constant not a null pointer,
1834 it is guaranteed to of the same size as a pointer. But this use is
1835 not portable across different compilers.
1837 @item -Wno-non-template-friend @r{(C++ only)}
1838 @opindex Wno-non-template-friend
1839 Disable warnings when non-templatized friend functions are declared
1840 within a template. Since the advent of explicit template specification
1841 support in G++, if the name of the friend is an unqualified-id (i.e.,
1842 @samp{friend foo(int)}), the C++ language specification demands that the
1843 friend declare or define an ordinary, nontemplate function. (Section
1844 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1845 could be interpreted as a particular specialization of a templatized
1846 function. Because this non-conforming behavior is no longer the default
1847 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1848 check existing code for potential trouble spots and is on by default.
1849 This new compiler behavior can be turned off with
1850 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1851 but disables the helpful warning.
1853 @item -Wold-style-cast @r{(C++ only)}
1854 @opindex Wold-style-cast
1855 Warn if an old-style (C-style) cast to a non-void type is used within
1856 a C++ program. The new-style casts (@samp{dynamic_cast},
1857 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1858 less vulnerable to unintended effects and much easier to search for.
1860 @item -Woverloaded-virtual @r{(C++ only)}
1861 @opindex Woverloaded-virtual
1862 @cindex overloaded virtual fn, warning
1863 @cindex warning for overloaded virtual fn
1864 Warn when a function declaration hides virtual functions from a
1865 base class. For example, in:
1872 struct B: public A @{
1877 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1885 will fail to compile.
1887 @item -Wno-pmf-conversions @r{(C++ only)}
1888 @opindex Wno-pmf-conversions
1889 Disable the diagnostic for converting a bound pointer to member function
1892 @item -Wsign-promo @r{(C++ only)}
1893 @opindex Wsign-promo
1894 Warn when overload resolution chooses a promotion from unsigned or
1895 enumerated type to a signed type, over a conversion to an unsigned type of
1896 the same size. Previous versions of G++ would try to preserve
1897 unsignedness, but the standard mandates the current behavior.
1902 A& operator = (int);
1912 In this example, G++ will synthesize a default @samp{A& operator =
1913 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1916 @node Objective-C and Objective-C++ Dialect Options
1917 @section Options Controlling Objective-C and Objective-C++ Dialects
1919 @cindex compiler options, Objective-C and Objective-C++
1920 @cindex Objective-C and Objective-C++ options, command line
1921 @cindex options, Objective-C and Objective-C++
1922 (NOTE: This manual does not describe the Objective-C and Objective-C++
1923 languages themselves. See @xref{Standards,,Language Standards
1924 Supported by GCC}, for references.)
1926 This section describes the command-line options that are only meaningful
1927 for Objective-C and Objective-C++ programs, but you can also use most of
1928 the language-independent GNU compiler options.
1929 For example, you might compile a file @code{some_class.m} like this:
1932 gcc -g -fgnu-runtime -O -c some_class.m
1936 In this example, @option{-fgnu-runtime} is an option meant only for
1937 Objective-C and Objective-C++ programs; you can use the other options with
1938 any language supported by GCC@.
1940 Note that since Objective-C is an extension of the C language, Objective-C
1941 compilations may also use options specific to the C front-end (e.g.,
1942 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1943 C++-specific options (e.g., @option{-Wabi}).
1945 Here is a list of options that are @emph{only} for compiling Objective-C
1946 and Objective-C++ programs:
1949 @item -fconstant-string-class=@var{class-name}
1950 @opindex fconstant-string-class
1951 Use @var{class-name} as the name of the class to instantiate for each
1952 literal string specified with the syntax @code{@@"@dots{}"}. The default
1953 class name is @code{NXConstantString} if the GNU runtime is being used, and
1954 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1955 @option{-fconstant-cfstrings} option, if also present, will override the
1956 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1957 to be laid out as constant CoreFoundation strings.
1960 @opindex fgnu-runtime
1961 Generate object code compatible with the standard GNU Objective-C
1962 runtime. This is the default for most types of systems.
1964 @item -fnext-runtime
1965 @opindex fnext-runtime
1966 Generate output compatible with the NeXT runtime. This is the default
1967 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1968 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1971 @item -fno-nil-receivers
1972 @opindex fno-nil-receivers
1973 Assume that all Objective-C message dispatches (e.g.,
1974 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1975 is not @code{nil}. This allows for more efficient entry points in the runtime
1976 to be used. Currently, this option is only available in conjunction with
1977 the NeXT runtime on Mac OS X 10.3 and later.
1979 @item -fobjc-call-cxx-cdtors
1980 @opindex fobjc-call-cxx-cdtors
1981 For each Objective-C class, check if any of its instance variables is a
1982 C++ object with a non-trivial default constructor. If so, synthesize a
1983 special @code{- (id) .cxx_construct} instance method that will run
1984 non-trivial default constructors on any such instance variables, in order,
1985 and then return @code{self}. Similarly, check if any instance variable
1986 is a C++ object with a non-trivial destructor, and if so, synthesize a
1987 special @code{- (void) .cxx_destruct} method that will run
1988 all such default destructors, in reverse order.
1990 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
1991 thusly generated will only operate on instance variables declared in the
1992 current Objective-C class, and not those inherited from superclasses. It
1993 is the responsibility of the Objective-C runtime to invoke all such methods
1994 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
1995 will be invoked by the runtime immediately after a new object
1996 instance is allocated; the @code{- (void) .cxx_destruct} methods will
1997 be invoked immediately before the runtime deallocates an object instance.
1999 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2000 support for invoking the @code{- (id) .cxx_construct} and
2001 @code{- (void) .cxx_destruct} methods.
2003 @item -fobjc-direct-dispatch
2004 @opindex fobjc-direct-dispatch
2005 Allow fast jumps to the message dispatcher. On Darwin this is
2006 accomplished via the comm page.
2008 @item -fobjc-exceptions
2009 @opindex fobjc-exceptions
2010 Enable syntactic support for structured exception handling in Objective-C,
2011 similar to what is offered by C++ and Java. This option is
2012 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2021 @@catch (AnObjCClass *exc) @{
2028 @@catch (AnotherClass *exc) @{
2031 @@catch (id allOthers) @{
2041 The @code{@@throw} statement may appear anywhere in an Objective-C or
2042 Objective-C++ program; when used inside of a @code{@@catch} block, the
2043 @code{@@throw} may appear without an argument (as shown above), in which case
2044 the object caught by the @code{@@catch} will be rethrown.
2046 Note that only (pointers to) Objective-C objects may be thrown and
2047 caught using this scheme. When an object is thrown, it will be caught
2048 by the nearest @code{@@catch} clause capable of handling objects of that type,
2049 analogously to how @code{catch} blocks work in C++ and Java. A
2050 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2051 any and all Objective-C exceptions not caught by previous @code{@@catch}
2054 The @code{@@finally} clause, if present, will be executed upon exit from the
2055 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2056 regardless of whether any exceptions are thrown, caught or rethrown
2057 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2058 of the @code{finally} clause in Java.
2060 There are several caveats to using the new exception mechanism:
2064 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2065 idioms provided by the @code{NSException} class, the new
2066 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2067 systems, due to additional functionality needed in the (NeXT) Objective-C
2071 As mentioned above, the new exceptions do not support handling
2072 types other than Objective-C objects. Furthermore, when used from
2073 Objective-C++, the Objective-C exception model does not interoperate with C++
2074 exceptions at this time. This means you cannot @code{@@throw} an exception
2075 from Objective-C and @code{catch} it in C++, or vice versa
2076 (i.e., @code{throw @dots{} @@catch}).
2079 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2080 blocks for thread-safe execution:
2083 @@synchronized (ObjCClass *guard) @{
2088 Upon entering the @code{@@synchronized} block, a thread of execution shall
2089 first check whether a lock has been placed on the corresponding @code{guard}
2090 object by another thread. If it has, the current thread shall wait until
2091 the other thread relinquishes its lock. Once @code{guard} becomes available,
2092 the current thread will place its own lock on it, execute the code contained in
2093 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2094 making @code{guard} available to other threads).
2096 Unlike Java, Objective-C does not allow for entire methods to be marked
2097 @code{@@synchronized}. Note that throwing exceptions out of
2098 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2099 to be unlocked properly.
2103 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2105 @item -freplace-objc-classes
2106 @opindex freplace-objc-classes
2107 Emit a special marker instructing @command{ld(1)} not to statically link in
2108 the resulting object file, and allow @command{dyld(1)} to load it in at
2109 run time instead. This is used in conjunction with the Fix-and-Continue
2110 debugging mode, where the object file in question may be recompiled and
2111 dynamically reloaded in the course of program execution, without the need
2112 to restart the program itself. Currently, Fix-and-Continue functionality
2113 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2118 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2119 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2120 compile time) with static class references that get initialized at load time,
2121 which improves run-time performance. Specifying the @option{-fzero-link} flag
2122 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2123 to be retained. This is useful in Zero-Link debugging mode, since it allows
2124 for individual class implementations to be modified during program execution.
2128 Dump interface declarations for all classes seen in the source file to a
2129 file named @file{@var{sourcename}.decl}.
2131 @item -Wassign-intercept
2132 @opindex Wassign-intercept
2133 Warn whenever an Objective-C assignment is being intercepted by the
2137 @opindex Wno-protocol
2138 If a class is declared to implement a protocol, a warning is issued for
2139 every method in the protocol that is not implemented by the class. The
2140 default behavior is to issue a warning for every method not explicitly
2141 implemented in the class, even if a method implementation is inherited
2142 from the superclass. If you use the @option{-Wno-protocol} option, then
2143 methods inherited from the superclass are considered to be implemented,
2144 and no warning is issued for them.
2148 Warn if multiple methods of different types for the same selector are
2149 found during compilation. The check is performed on the list of methods
2150 in the final stage of compilation. Additionally, a check is performed
2151 for each selector appearing in a @code{@@selector(@dots{})}
2152 expression, and a corresponding method for that selector has been found
2153 during compilation. Because these checks scan the method table only at
2154 the end of compilation, these warnings are not produced if the final
2155 stage of compilation is not reached, for example because an error is
2156 found during compilation, or because the @option{-fsyntax-only} option is
2159 @item -Wstrict-selector-match
2160 @opindex Wstrict-selector-match
2161 Warn if multiple methods with differing argument and/or return types are
2162 found for a given selector when attempting to send a message using this
2163 selector to a receiver of type @code{id} or @code{Class}. When this flag
2164 is off (which is the default behavior), the compiler will omit such warnings
2165 if any differences found are confined to types which share the same size
2168 @item -Wundeclared-selector
2169 @opindex Wundeclared-selector
2170 Warn if a @code{@@selector(@dots{})} expression referring to an
2171 undeclared selector is found. A selector is considered undeclared if no
2172 method with that name has been declared before the
2173 @code{@@selector(@dots{})} expression, either explicitly in an
2174 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2175 an @code{@@implementation} section. This option always performs its
2176 checks as soon as a @code{@@selector(@dots{})} expression is found,
2177 while @option{-Wselector} only performs its checks in the final stage of
2178 compilation. This also enforces the coding style convention
2179 that methods and selectors must be declared before being used.
2181 @item -print-objc-runtime-info
2182 @opindex print-objc-runtime-info
2183 Generate C header describing the largest structure that is passed by
2188 @node Language Independent Options
2189 @section Options to Control Diagnostic Messages Formatting
2190 @cindex options to control diagnostics formatting
2191 @cindex diagnostic messages
2192 @cindex message formatting
2194 Traditionally, diagnostic messages have been formatted irrespective of
2195 the output device's aspect (e.g.@: its width, @dots{}). The options described
2196 below can be used to control the diagnostic messages formatting
2197 algorithm, e.g.@: how many characters per line, how often source location
2198 information should be reported. Right now, only the C++ front end can
2199 honor these options. However it is expected, in the near future, that
2200 the remaining front ends would be able to digest them correctly.
2203 @item -fmessage-length=@var{n}
2204 @opindex fmessage-length
2205 Try to format error messages so that they fit on lines of about @var{n}
2206 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2207 the front ends supported by GCC@. If @var{n} is zero, then no
2208 line-wrapping will be done; each error message will appear on a single
2211 @opindex fdiagnostics-show-location
2212 @item -fdiagnostics-show-location=once
2213 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2214 reporter to emit @emph{once} source location information; that is, in
2215 case the message is too long to fit on a single physical line and has to
2216 be wrapped, the source location won't be emitted (as prefix) again,
2217 over and over, in subsequent continuation lines. This is the default
2220 @item -fdiagnostics-show-location=every-line
2221 Only meaningful in line-wrapping mode. Instructs the diagnostic
2222 messages reporter to emit the same source location information (as
2223 prefix) for physical lines that result from the process of breaking
2224 a message which is too long to fit on a single line.
2226 @item -fdiagnostics-show-options
2227 @opindex fdiagnostics-show-options
2228 This option instructs the diagnostic machinery to add text to each
2229 diagnostic emitted, which indicates which command line option directly
2230 controls that diagnostic, when such an option is known to the
2231 diagnostic machinery.
2235 @node Warning Options
2236 @section Options to Request or Suppress Warnings
2237 @cindex options to control warnings
2238 @cindex warning messages
2239 @cindex messages, warning
2240 @cindex suppressing warnings
2242 Warnings are diagnostic messages that report constructions which
2243 are not inherently erroneous but which are risky or suggest there
2244 may have been an error.
2246 You can request many specific warnings with options beginning @samp{-W},
2247 for example @option{-Wimplicit} to request warnings on implicit
2248 declarations. Each of these specific warning options also has a
2249 negative form beginning @samp{-Wno-} to turn off warnings;
2250 for example, @option{-Wno-implicit}. This manual lists only one of the
2251 two forms, whichever is not the default.
2253 The following options control the amount and kinds of warnings produced
2254 by GCC; for further, language-specific options also refer to
2255 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2259 @cindex syntax checking
2261 @opindex fsyntax-only
2262 Check the code for syntax errors, but don't do anything beyond that.
2266 Issue all the warnings demanded by strict ISO C and ISO C++;
2267 reject all programs that use forbidden extensions, and some other
2268 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2269 version of the ISO C standard specified by any @option{-std} option used.
2271 Valid ISO C and ISO C++ programs should compile properly with or without
2272 this option (though a rare few will require @option{-ansi} or a
2273 @option{-std} option specifying the required version of ISO C)@. However,
2274 without this option, certain GNU extensions and traditional C and C++
2275 features are supported as well. With this option, they are rejected.
2277 @option{-pedantic} does not cause warning messages for use of the
2278 alternate keywords whose names begin and end with @samp{__}. Pedantic
2279 warnings are also disabled in the expression that follows
2280 @code{__extension__}. However, only system header files should use
2281 these escape routes; application programs should avoid them.
2282 @xref{Alternate Keywords}.
2284 Some users try to use @option{-pedantic} to check programs for strict ISO
2285 C conformance. They soon find that it does not do quite what they want:
2286 it finds some non-ISO practices, but not all---only those for which
2287 ISO C @emph{requires} a diagnostic, and some others for which
2288 diagnostics have been added.
2290 A feature to report any failure to conform to ISO C might be useful in
2291 some instances, but would require considerable additional work and would
2292 be quite different from @option{-pedantic}. We don't have plans to
2293 support such a feature in the near future.
2295 Where the standard specified with @option{-std} represents a GNU
2296 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2297 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2298 extended dialect is based. Warnings from @option{-pedantic} are given
2299 where they are required by the base standard. (It would not make sense
2300 for such warnings to be given only for features not in the specified GNU
2301 C dialect, since by definition the GNU dialects of C include all
2302 features the compiler supports with the given option, and there would be
2303 nothing to warn about.)
2305 @item -pedantic-errors
2306 @opindex pedantic-errors
2307 Like @option{-pedantic}, except that errors are produced rather than
2312 Inhibit all warning messages.
2316 Inhibit warning messages about the use of @samp{#import}.
2318 @item -Wchar-subscripts
2319 @opindex Wchar-subscripts
2320 Warn if an array subscript has type @code{char}. This is a common cause
2321 of error, as programmers often forget that this type is signed on some
2323 This warning is enabled by @option{-Wall}.
2327 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2328 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2329 This warning is enabled by @option{-Wall}.
2331 @item -Wfatal-errors
2332 @opindex Wfatal-errors
2333 This option causes the compiler to abort compilation on the first error
2334 occurred rather than trying to keep going and printing further error
2339 @opindex ffreestanding
2340 @opindex fno-builtin
2341 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2342 the arguments supplied have types appropriate to the format string
2343 specified, and that the conversions specified in the format string make
2344 sense. This includes standard functions, and others specified by format
2345 attributes (@pxref{Function Attributes}), in the @code{printf},
2346 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2347 not in the C standard) families (or other target-specific families).
2348 Which functions are checked without format attributes having been
2349 specified depends on the standard version selected, and such checks of
2350 functions without the attribute specified are disabled by
2351 @option{-ffreestanding} or @option{-fno-builtin}.
2353 The formats are checked against the format features supported by GNU
2354 libc version 2.2. These include all ISO C90 and C99 features, as well
2355 as features from the Single Unix Specification and some BSD and GNU
2356 extensions. Other library implementations may not support all these
2357 features; GCC does not support warning about features that go beyond a
2358 particular library's limitations. However, if @option{-pedantic} is used
2359 with @option{-Wformat}, warnings will be given about format features not
2360 in the selected standard version (but not for @code{strfmon} formats,
2361 since those are not in any version of the C standard). @xref{C Dialect
2362 Options,,Options Controlling C Dialect}.
2364 Since @option{-Wformat} also checks for null format arguments for
2365 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2367 @option{-Wformat} is included in @option{-Wall}. For more control over some
2368 aspects of format checking, the options @option{-Wformat-y2k},
2369 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2370 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2371 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2374 @opindex Wformat-y2k
2375 If @option{-Wformat} is specified, also warn about @code{strftime}
2376 formats which may yield only a two-digit year.
2378 @item -Wno-format-extra-args
2379 @opindex Wno-format-extra-args
2380 If @option{-Wformat} is specified, do not warn about excess arguments to a
2381 @code{printf} or @code{scanf} format function. The C standard specifies
2382 that such arguments are ignored.
2384 Where the unused arguments lie between used arguments that are
2385 specified with @samp{$} operand number specifications, normally
2386 warnings are still given, since the implementation could not know what
2387 type to pass to @code{va_arg} to skip the unused arguments. However,
2388 in the case of @code{scanf} formats, this option will suppress the
2389 warning if the unused arguments are all pointers, since the Single
2390 Unix Specification says that such unused arguments are allowed.
2392 @item -Wno-format-zero-length
2393 @opindex Wno-format-zero-length
2394 If @option{-Wformat} is specified, do not warn about zero-length formats.
2395 The C standard specifies that zero-length formats are allowed.
2397 @item -Wformat-nonliteral
2398 @opindex Wformat-nonliteral
2399 If @option{-Wformat} is specified, also warn if the format string is not a
2400 string literal and so cannot be checked, unless the format function
2401 takes its format arguments as a @code{va_list}.
2403 @item -Wformat-security
2404 @opindex Wformat-security
2405 If @option{-Wformat} is specified, also warn about uses of format
2406 functions that represent possible security problems. At present, this
2407 warns about calls to @code{printf} and @code{scanf} functions where the
2408 format string is not a string literal and there are no format arguments,
2409 as in @code{printf (foo);}. This may be a security hole if the format
2410 string came from untrusted input and contains @samp{%n}. (This is
2411 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2412 in future warnings may be added to @option{-Wformat-security} that are not
2413 included in @option{-Wformat-nonliteral}.)
2417 Enable @option{-Wformat} plus format checks not included in
2418 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2419 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2423 Warn about passing a null pointer for arguments marked as
2424 requiring a non-null value by the @code{nonnull} function attribute.
2426 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2427 can be disabled with the @option{-Wno-nonnull} option.
2429 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2431 Warn about uninitialized variables which are initialized with themselves.
2432 Note this option can only be used with the @option{-Wuninitialized} option,
2433 which in turn only works with @option{-O1} and above.
2435 For example, GCC will warn about @code{i} being uninitialized in the
2436 following snippet only when @option{-Winit-self} has been specified:
2447 @item -Wimplicit-int
2448 @opindex Wimplicit-int
2449 Warn when a declaration does not specify a type.
2450 This warning is enabled by @option{-Wall}.
2452 @item -Wimplicit-function-declaration
2453 @itemx -Werror-implicit-function-declaration
2454 @opindex Wimplicit-function-declaration
2455 @opindex Werror-implicit-function-declaration
2456 Give a warning (or error) whenever a function is used before being
2457 declared. The form @option{-Wno-error-implicit-function-declaration}
2459 This warning is enabled by @option{-Wall} (as a warning, not an error).
2463 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2464 This warning is enabled by @option{-Wall}.
2468 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2469 function with external linkage, returning int, taking either zero
2470 arguments, two, or three arguments of appropriate types.
2471 This warning is enabled by @option{-Wall}.
2473 @item -Wmissing-braces
2474 @opindex Wmissing-braces
2475 Warn if an aggregate or union initializer is not fully bracketed. In
2476 the following example, the initializer for @samp{a} is not fully
2477 bracketed, but that for @samp{b} is fully bracketed.
2480 int a[2][2] = @{ 0, 1, 2, 3 @};
2481 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2484 This warning is enabled by @option{-Wall}.
2486 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2487 @opindex Wmissing-include-dirs
2488 Warn if a user-supplied include directory does not exist.
2491 @opindex Wparentheses
2492 Warn if parentheses are omitted in certain contexts, such
2493 as when there is an assignment in a context where a truth value
2494 is expected, or when operators are nested whose precedence people
2495 often get confused about. Only the warning for an assignment used as
2496 a truth value is supported when compiling C++; the other warnings are
2497 only supported when compiling C@.
2499 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2500 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2501 interpretation from that of ordinary mathematical notation.
2503 Also warn about constructions where there may be confusion to which
2504 @code{if} statement an @code{else} branch belongs. Here is an example of
2519 In C, every @code{else} branch belongs to the innermost possible @code{if}
2520 statement, which in this example is @code{if (b)}. This is often not
2521 what the programmer expected, as illustrated in the above example by
2522 indentation the programmer chose. When there is the potential for this
2523 confusion, GCC will issue a warning when this flag is specified.
2524 To eliminate the warning, add explicit braces around the innermost
2525 @code{if} statement so there is no way the @code{else} could belong to
2526 the enclosing @code{if}. The resulting code would look like this:
2542 This warning is enabled by @option{-Wall}.
2544 @item -Wsequence-point
2545 @opindex Wsequence-point
2546 Warn about code that may have undefined semantics because of violations
2547 of sequence point rules in the C and C++ standards.
2549 The C and C++ standards defines the order in which expressions in a C/C++
2550 program are evaluated in terms of @dfn{sequence points}, which represent
2551 a partial ordering between the execution of parts of the program: those
2552 executed before the sequence point, and those executed after it. These
2553 occur after the evaluation of a full expression (one which is not part
2554 of a larger expression), after the evaluation of the first operand of a
2555 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2556 function is called (but after the evaluation of its arguments and the
2557 expression denoting the called function), and in certain other places.
2558 Other than as expressed by the sequence point rules, the order of
2559 evaluation of subexpressions of an expression is not specified. All
2560 these rules describe only a partial order rather than a total order,
2561 since, for example, if two functions are called within one expression
2562 with no sequence point between them, the order in which the functions
2563 are called is not specified. However, the standards committee have
2564 ruled that function calls do not overlap.
2566 It is not specified when between sequence points modifications to the
2567 values of objects take effect. Programs whose behavior depends on this
2568 have undefined behavior; the C and C++ standards specify that ``Between
2569 the previous and next sequence point an object shall have its stored
2570 value modified at most once by the evaluation of an expression.
2571 Furthermore, the prior value shall be read only to determine the value
2572 to be stored.''. If a program breaks these rules, the results on any
2573 particular implementation are entirely unpredictable.
2575 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2576 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2577 diagnosed by this option, and it may give an occasional false positive
2578 result, but in general it has been found fairly effective at detecting
2579 this sort of problem in programs.
2581 The standard is worded confusingly, therefore there is some debate
2582 over the precise meaning of the sequence point rules in subtle cases.
2583 Links to discussions of the problem, including proposed formal
2584 definitions, may be found on the GCC readings page, at
2585 @w{@uref{http://gcc.gnu.org/readings.html}}.
2587 This warning is enabled by @option{-Wall} for C and C++.
2590 @opindex Wreturn-type
2591 Warn whenever a function is defined with a return-type that defaults to
2592 @code{int}. Also warn about any @code{return} statement with no
2593 return-value in a function whose return-type is not @code{void}.
2595 For C, also warn if the return type of a function has a type qualifier
2596 such as @code{const}. Such a type qualifier has no effect, since the
2597 value returned by a function is not an lvalue. ISO C prohibits
2598 qualified @code{void} return types on function definitions, so such
2599 return types always receive a warning even without this option.
2601 For C++, a function without return type always produces a diagnostic
2602 message, even when @option{-Wno-return-type} is specified. The only
2603 exceptions are @samp{main} and functions defined in system headers.
2605 This warning is enabled by @option{-Wall}.
2609 Warn whenever a @code{switch} statement has an index of enumerated type
2610 and lacks a @code{case} for one or more of the named codes of that
2611 enumeration. (The presence of a @code{default} label prevents this
2612 warning.) @code{case} labels outside the enumeration range also
2613 provoke warnings when this option is used.
2614 This warning is enabled by @option{-Wall}.
2616 @item -Wswitch-default
2617 @opindex Wswitch-switch
2618 Warn whenever a @code{switch} statement does not have a @code{default}
2622 @opindex Wswitch-enum
2623 Warn whenever a @code{switch} statement has an index of enumerated type
2624 and lacks a @code{case} for one or more of the named codes of that
2625 enumeration. @code{case} labels outside the enumeration range also
2626 provoke warnings when this option is used.
2630 Warn if any trigraphs are encountered that might change the meaning of
2631 the program (trigraphs within comments are not warned about).
2632 This warning is enabled by @option{-Wall}.
2634 @item -Wunused-function
2635 @opindex Wunused-function
2636 Warn whenever a static function is declared but not defined or a
2637 non-inline static function is unused.
2638 This warning is enabled by @option{-Wall}.
2640 @item -Wunused-label
2641 @opindex Wunused-label
2642 Warn whenever a label is declared but not used.
2643 This warning is enabled by @option{-Wall}.
2645 To suppress this warning use the @samp{unused} attribute
2646 (@pxref{Variable Attributes}).
2648 @item -Wunused-parameter
2649 @opindex Wunused-parameter
2650 Warn whenever a function parameter is unused aside from its declaration.
2652 To suppress this warning use the @samp{unused} attribute
2653 (@pxref{Variable Attributes}).
2655 @item -Wunused-variable
2656 @opindex Wunused-variable
2657 Warn whenever a local variable or non-constant static variable is unused
2658 aside from its declaration
2659 This warning is enabled by @option{-Wall}.
2661 To suppress this warning use the @samp{unused} attribute
2662 (@pxref{Variable Attributes}).
2664 @item -Wunused-value
2665 @opindex Wunused-value
2666 Warn whenever a statement computes a result that is explicitly not used.
2667 This warning is enabled by @option{-Wall}.
2669 To suppress this warning cast the expression to @samp{void}.
2673 All the above @option{-Wunused} options combined.
2675 In order to get a warning about an unused function parameter, you must
2676 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2677 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2679 @item -Wuninitialized
2680 @opindex Wuninitialized
2681 Warn if an automatic variable is used without first being initialized or
2682 if a variable may be clobbered by a @code{setjmp} call.
2684 These warnings are possible only in optimizing compilation,
2685 because they require data flow information that is computed only
2686 when optimizing. If you do not specify @option{-O}, you will not get
2687 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2688 requiring @option{-O}.
2690 If you want to warn about code which uses the uninitialized value of the
2691 variable in its own initializer, use the @option{-Winit-self} option.
2693 These warnings occur for individual uninitialized or clobbered
2694 elements of structure, union or array variables as well as for
2695 variables which are uninitialized or clobbered as a whole. They do
2696 not occur for variables or elements declared @code{volatile}. Because
2697 these warnings depend on optimization, the exact variables or elements
2698 for which there are warnings will depend on the precise optimization
2699 options and version of GCC used.
2701 Note that there may be no warning about a variable that is used only
2702 to compute a value that itself is never used, because such
2703 computations may be deleted by data flow analysis before the warnings
2706 These warnings are made optional because GCC is not smart
2707 enough to see all the reasons why the code might be correct
2708 despite appearing to have an error. Here is one example of how
2729 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2730 always initialized, but GCC doesn't know this. Here is
2731 another common case:
2736 if (change_y) save_y = y, y = new_y;
2738 if (change_y) y = save_y;
2743 This has no bug because @code{save_y} is used only if it is set.
2745 @cindex @code{longjmp} warnings
2746 This option also warns when a non-volatile automatic variable might be
2747 changed by a call to @code{longjmp}. These warnings as well are possible
2748 only in optimizing compilation.
2750 The compiler sees only the calls to @code{setjmp}. It cannot know
2751 where @code{longjmp} will be called; in fact, a signal handler could
2752 call it at any point in the code. As a result, you may get a warning
2753 even when there is in fact no problem because @code{longjmp} cannot
2754 in fact be called at the place which would cause a problem.
2756 Some spurious warnings can be avoided if you declare all the functions
2757 you use that never return as @code{noreturn}. @xref{Function
2760 This warning is enabled by @option{-Wall}.
2762 @item -Wunknown-pragmas
2763 @opindex Wunknown-pragmas
2764 @cindex warning for unknown pragmas
2765 @cindex unknown pragmas, warning
2766 @cindex pragmas, warning of unknown
2767 Warn when a #pragma directive is encountered which is not understood by
2768 GCC@. If this command line option is used, warnings will even be issued
2769 for unknown pragmas in system header files. This is not the case if
2770 the warnings were only enabled by the @option{-Wall} command line option.
2773 @opindex Wno-pragmas
2775 Do not warn about misuses of pragmas, such as incorrect parameters,
2776 invalid syntax, or conflicts between pragmas. See also
2777 @samp{-Wunknown-pragmas}.
2779 @item -Wstrict-aliasing
2780 @opindex Wstrict-aliasing
2781 This option is only active when @option{-fstrict-aliasing} is active.
2782 It warns about code which might break the strict aliasing rules that the
2783 compiler is using for optimization. The warning does not catch all
2784 cases, but does attempt to catch the more common pitfalls. It is
2785 included in @option{-Wall}.
2787 @item -Wstrict-aliasing=2
2788 @opindex Wstrict-aliasing=2
2789 This option is only active when @option{-fstrict-aliasing} is active.
2790 It warns about code which might break the strict aliasing rules that the
2791 compiler is using for optimization. This warning catches more cases than
2792 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2793 cases that are safe.
2797 All of the above @samp{-W} options combined. This enables all the
2798 warnings about constructions that some users consider questionable, and
2799 that are easy to avoid (or modify to prevent the warning), even in
2800 conjunction with macros. This also enables some language-specific
2801 warnings described in @ref{C++ Dialect Options} and
2802 @ref{Objective-C and Objective-C++ Dialect Options}.
2805 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2806 Some of them warn about constructions that users generally do not
2807 consider questionable, but which occasionally you might wish to check
2808 for; others warn about constructions that are necessary or hard to avoid
2809 in some cases, and there is no simple way to modify the code to suppress
2816 (This option used to be called @option{-W}. The older name is still
2817 supported, but the newer name is more descriptive.) Print extra warning
2818 messages for these events:
2822 A function can return either with or without a value. (Falling
2823 off the end of the function body is considered returning without
2824 a value.) For example, this function would evoke such a
2838 An expression-statement or the left-hand side of a comma expression
2839 contains no side effects.
2840 To suppress the warning, cast the unused expression to void.
2841 For example, an expression such as @samp{x[i,j]} will cause a warning,
2842 but @samp{x[(void)i,j]} will not.
2845 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2848 Storage-class specifiers like @code{static} are not the first things in
2849 a declaration. According to the C Standard, this usage is obsolescent.
2852 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2856 A comparison between signed and unsigned values could produce an
2857 incorrect result when the signed value is converted to unsigned.
2858 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2861 An aggregate has an initializer which does not initialize all members.
2862 This warning can be independently controlled by
2863 @option{-Wmissing-field-initializers}.
2866 A function parameter is declared without a type specifier in K&R-style
2874 An empty body occurs in an @samp{if} or @samp{else} statement.
2877 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2878 @samp{>}, or @samp{>=}.
2881 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2884 Any of several floating-point events that often indicate errors, such as
2885 overflow, underflow, loss of precision, etc.
2887 @item @r{(C++ only)}
2888 An enumerator and a non-enumerator both appear in a conditional expression.
2890 @item @r{(C++ only)}
2891 A non-static reference or non-static @samp{const} member appears in a
2892 class without constructors.
2894 @item @r{(C++ only)}
2895 Ambiguous virtual bases.
2897 @item @r{(C++ only)}
2898 Subscripting an array which has been declared @samp{register}.
2900 @item @r{(C++ only)}
2901 Taking the address of a variable which has been declared @samp{register}.
2903 @item @r{(C++ only)}
2904 A base class is not initialized in a derived class' copy constructor.
2907 @item -Wno-div-by-zero
2908 @opindex Wno-div-by-zero
2909 @opindex Wdiv-by-zero
2910 Do not warn about compile-time integer division by zero. Floating point
2911 division by zero is not warned about, as it can be a legitimate way of
2912 obtaining infinities and NaNs.
2914 @item -Wsystem-headers
2915 @opindex Wsystem-headers
2916 @cindex warnings from system headers
2917 @cindex system headers, warnings from
2918 Print warning messages for constructs found in system header files.
2919 Warnings from system headers are normally suppressed, on the assumption
2920 that they usually do not indicate real problems and would only make the
2921 compiler output harder to read. Using this command line option tells
2922 GCC to emit warnings from system headers as if they occurred in user
2923 code. However, note that using @option{-Wall} in conjunction with this
2924 option will @emph{not} warn about unknown pragmas in system
2925 headers---for that, @option{-Wunknown-pragmas} must also be used.
2928 @opindex Wfloat-equal
2929 Warn if floating point values are used in equality comparisons.
2931 The idea behind this is that sometimes it is convenient (for the
2932 programmer) to consider floating-point values as approximations to
2933 infinitely precise real numbers. If you are doing this, then you need
2934 to compute (by analyzing the code, or in some other way) the maximum or
2935 likely maximum error that the computation introduces, and allow for it
2936 when performing comparisons (and when producing output, but that's a
2937 different problem). In particular, instead of testing for equality, you
2938 would check to see whether the two values have ranges that overlap; and
2939 this is done with the relational operators, so equality comparisons are
2942 @item -Wtraditional @r{(C only)}
2943 @opindex Wtraditional
2944 Warn about certain constructs that behave differently in traditional and
2945 ISO C@. Also warn about ISO C constructs that have no traditional C
2946 equivalent, and/or problematic constructs which should be avoided.
2950 Macro parameters that appear within string literals in the macro body.
2951 In traditional C macro replacement takes place within string literals,
2952 but does not in ISO C@.
2955 In traditional C, some preprocessor directives did not exist.
2956 Traditional preprocessors would only consider a line to be a directive
2957 if the @samp{#} appeared in column 1 on the line. Therefore
2958 @option{-Wtraditional} warns about directives that traditional C
2959 understands but would ignore because the @samp{#} does not appear as the
2960 first character on the line. It also suggests you hide directives like
2961 @samp{#pragma} not understood by traditional C by indenting them. Some
2962 traditional implementations would not recognize @samp{#elif}, so it
2963 suggests avoiding it altogether.
2966 A function-like macro that appears without arguments.
2969 The unary plus operator.
2972 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2973 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2974 constants.) Note, these suffixes appear in macros defined in the system
2975 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2976 Use of these macros in user code might normally lead to spurious
2977 warnings, however GCC's integrated preprocessor has enough context to
2978 avoid warning in these cases.
2981 A function declared external in one block and then used after the end of
2985 A @code{switch} statement has an operand of type @code{long}.
2988 A non-@code{static} function declaration follows a @code{static} one.
2989 This construct is not accepted by some traditional C compilers.
2992 The ISO type of an integer constant has a different width or
2993 signedness from its traditional type. This warning is only issued if
2994 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2995 typically represent bit patterns, are not warned about.
2998 Usage of ISO string concatenation is detected.
3001 Initialization of automatic aggregates.
3004 Identifier conflicts with labels. Traditional C lacks a separate
3005 namespace for labels.
3008 Initialization of unions. If the initializer is zero, the warning is
3009 omitted. This is done under the assumption that the zero initializer in
3010 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3011 initializer warnings and relies on default initialization to zero in the
3015 Conversions by prototypes between fixed/floating point values and vice
3016 versa. The absence of these prototypes when compiling with traditional
3017 C would cause serious problems. This is a subset of the possible
3018 conversion warnings, for the full set use @option{-Wconversion}.
3021 Use of ISO C style function definitions. This warning intentionally is
3022 @emph{not} issued for prototype declarations or variadic functions
3023 because these ISO C features will appear in your code when using
3024 libiberty's traditional C compatibility macros, @code{PARAMS} and
3025 @code{VPARAMS}. This warning is also bypassed for nested functions
3026 because that feature is already a GCC extension and thus not relevant to
3027 traditional C compatibility.
3030 @item -Wdeclaration-after-statement @r{(C only)}
3031 @opindex Wdeclaration-after-statement
3032 Warn when a declaration is found after a statement in a block. This
3033 construct, known from C++, was introduced with ISO C99 and is by default
3034 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3035 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3039 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3041 @item -Wno-endif-labels
3042 @opindex Wno-endif-labels
3043 @opindex Wendif-labels
3044 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3048 Warn whenever a local variable shadows another local variable, parameter or
3049 global variable or whenever a built-in function is shadowed.
3051 @item -Wlarger-than-@var{len}
3052 @opindex Wlarger-than
3053 Warn whenever an object of larger than @var{len} bytes is defined.
3055 @item -Wunsafe-loop-optimizations
3056 @opindex Wunsafe-loop-optimizations
3057 Warn if the loop cannot be optimized because the compiler could not
3058 assume anything on the bounds of the loop indices. With
3059 @option{-funsafe-loop-optimizations} warn if the compiler made
3062 @item -Wpointer-arith
3063 @opindex Wpointer-arith
3064 Warn about anything that depends on the ``size of'' a function type or
3065 of @code{void}. GNU C assigns these types a size of 1, for
3066 convenience in calculations with @code{void *} pointers and pointers
3069 @item -Wbad-function-cast @r{(C only)}
3070 @opindex Wbad-function-cast
3071 Warn whenever a function call is cast to a non-matching type.
3072 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3075 Warn about ISO C constructs that are outside of the common subset of
3076 ISO C and ISO C++, e.g.@: request for implicit conversion from
3077 @code{void *} to a pointer to non-@code{void} type.
3081 Warn whenever a pointer is cast so as to remove a type qualifier from
3082 the target type. For example, warn if a @code{const char *} is cast
3083 to an ordinary @code{char *}.
3086 @opindex Wcast-align
3087 Warn whenever a pointer is cast such that the required alignment of the
3088 target is increased. For example, warn if a @code{char *} is cast to
3089 an @code{int *} on machines where integers can only be accessed at
3090 two- or four-byte boundaries.
3092 @item -Wwrite-strings
3093 @opindex Wwrite-strings
3094 When compiling C, give string constants the type @code{const
3095 char[@var{length}]} so that
3096 copying the address of one into a non-@code{const} @code{char *}
3097 pointer will get a warning; when compiling C++, warn about the
3098 deprecated conversion from string literals to @code{char *}. This
3099 warning, by default, is enabled for C++ programs.
3100 These warnings will help you find at
3101 compile time code that can try to write into a string constant, but
3102 only if you have been very careful about using @code{const} in
3103 declarations and prototypes. Otherwise, it will just be a nuisance;
3104 this is why we did not make @option{-Wall} request these warnings.
3107 @opindex Wconversion
3108 Warn if a prototype causes a type conversion that is different from what
3109 would happen to the same argument in the absence of a prototype. This
3110 includes conversions of fixed point to floating and vice versa, and
3111 conversions changing the width or signedness of a fixed point argument
3112 except when the same as the default promotion.
3114 Also, warn if a negative integer constant expression is implicitly
3115 converted to an unsigned type. For example, warn about the assignment
3116 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3117 casts like @code{(unsigned) -1}.
3119 @item -Wsign-compare
3120 @opindex Wsign-compare
3121 @cindex warning for comparison of signed and unsigned values
3122 @cindex comparison of signed and unsigned values, warning
3123 @cindex signed and unsigned values, comparison warning
3124 Warn when a comparison between signed and unsigned values could produce
3125 an incorrect result when the signed value is converted to unsigned.
3126 This warning is also enabled by @option{-Wextra}; to get the other warnings
3127 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3129 @item -Waggregate-return
3130 @opindex Waggregate-return
3131 Warn if any functions that return structures or unions are defined or
3132 called. (In languages where you can return an array, this also elicits
3136 @opindex Walways-true
3137 Warn about comparisons which are always true such as testing if
3138 unsigned values are greater than or equal to zero. This warning is
3139 enabled by @option{-Wall}.
3141 @item -Wno-attributes
3142 @opindex Wno-attributes
3143 @opindex Wattributes
3144 Do not warn if an unexpected @code{__attribute__} is used, such as
3145 unrecognized attributes, function attributes applied to variables,
3146 etc. This will not stop errors for incorrect use of supported
3149 @item -Wstrict-prototypes @r{(C only)}
3150 @opindex Wstrict-prototypes
3151 Warn if a function is declared or defined without specifying the
3152 argument types. (An old-style function definition is permitted without
3153 a warning if preceded by a declaration which specifies the argument
3156 @item -Wold-style-definition @r{(C only)}
3157 @opindex Wold-style-definition
3158 Warn if an old-style function definition is used. A warning is given
3159 even if there is a previous prototype.
3161 @item -Wmissing-prototypes @r{(C only)}
3162 @opindex Wmissing-prototypes
3163 Warn if a global function is defined without a previous prototype
3164 declaration. This warning is issued even if the definition itself
3165 provides a prototype. The aim is to detect global functions that fail
3166 to be declared in header files.
3168 @item -Wmissing-declarations @r{(C only)}
3169 @opindex Wmissing-declarations
3170 Warn if a global function is defined without a previous declaration.
3171 Do so even if the definition itself provides a prototype.
3172 Use this option to detect global functions that are not declared in
3175 @item -Wmissing-field-initializers
3176 @opindex Wmissing-field-initializers
3179 Warn if a structure's initializer has some fields missing. For
3180 example, the following code would cause such a warning, because
3181 @code{x.h} is implicitly zero:
3184 struct s @{ int f, g, h; @};
3185 struct s x = @{ 3, 4 @};
3188 This option does not warn about designated initializers, so the following
3189 modification would not trigger a warning:
3192 struct s @{ int f, g, h; @};
3193 struct s x = @{ .f = 3, .g = 4 @};
3196 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3197 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3199 @item -Wmissing-noreturn
3200 @opindex Wmissing-noreturn
3201 Warn about functions which might be candidates for attribute @code{noreturn}.
3202 Note these are only possible candidates, not absolute ones. Care should
3203 be taken to manually verify functions actually do not ever return before
3204 adding the @code{noreturn} attribute, otherwise subtle code generation
3205 bugs could be introduced. You will not get a warning for @code{main} in
3206 hosted C environments.
3208 @item -Wmissing-format-attribute
3209 @opindex Wmissing-format-attribute
3211 Warn about function pointers which might be candidates for @code{format}
3212 attributes. Note these are only possible candidates, not absolute ones.
3213 GCC will guess that function pointers with @code{format} attributes that
3214 are used in assignment, initialization, parameter passing or return
3215 statements should have a corresponding @code{format} attribute in the
3216 resulting type. I.e.@: the left-hand side of the assignment or
3217 initialization, the type of the parameter variable, or the return type
3218 of the containing function respectively should also have a @code{format}
3219 attribute to avoid the warning.
3221 GCC will also warn about function definitions which might be
3222 candidates for @code{format} attributes. Again, these are only
3223 possible candidates. GCC will guess that @code{format} attributes
3224 might be appropriate for any function that calls a function like
3225 @code{vprintf} or @code{vscanf}, but this might not always be the
3226 case, and some functions for which @code{format} attributes are
3227 appropriate may not be detected.
3229 @item -Wno-multichar
3230 @opindex Wno-multichar
3232 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3233 Usually they indicate a typo in the user's code, as they have
3234 implementation-defined values, and should not be used in portable code.
3236 @item -Wnormalized=<none|id|nfc|nfkc>
3237 @opindex Wnormalized
3240 @cindex character set, input normalization
3241 In ISO C and ISO C++, two identifiers are different if they are
3242 different sequences of characters. However, sometimes when characters
3243 outside the basic ASCII character set are used, you can have two
3244 different character sequences that look the same. To avoid confusion,
3245 the ISO 10646 standard sets out some @dfn{normalization rules} which
3246 when applied ensure that two sequences that look the same are turned into
3247 the same sequence. GCC can warn you if you are using identifiers which
3248 have not been normalized; this option controls that warning.
3250 There are four levels of warning that GCC supports. The default is
3251 @option{-Wnormalized=nfc}, which warns about any identifier which is
3252 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3253 recommended form for most uses.
3255 Unfortunately, there are some characters which ISO C and ISO C++ allow
3256 in identifiers that when turned into NFC aren't allowable as
3257 identifiers. That is, there's no way to use these symbols in portable
3258 ISO C or C++ and have all your identifiers in NFC.
3259 @option{-Wnormalized=id} suppresses the warning for these characters.
3260 It is hoped that future versions of the standards involved will correct
3261 this, which is why this option is not the default.
3263 You can switch the warning off for all characters by writing
3264 @option{-Wnormalized=none}. You would only want to do this if you
3265 were using some other normalization scheme (like ``D''), because
3266 otherwise you can easily create bugs that are literally impossible to see.
3268 Some characters in ISO 10646 have distinct meanings but look identical
3269 in some fonts or display methodologies, especially once formatting has
3270 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3271 LETTER N'', will display just like a regular @code{n} which has been
3272 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3273 normalisation scheme to convert all these into a standard form as
3274 well, and GCC will warn if your code is not in NFKC if you use
3275 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3276 about every identifier that contains the letter O because it might be
3277 confused with the digit 0, and so is not the default, but may be
3278 useful as a local coding convention if the programming environment is
3279 unable to be fixed to display these characters distinctly.
3281 @item -Wno-deprecated-declarations
3282 @opindex Wno-deprecated-declarations
3283 Do not warn about uses of functions, variables, and types marked as
3284 deprecated by using the @code{deprecated} attribute.
3285 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3286 @pxref{Type Attributes}.)
3289 @opindex Wno-overflow
3290 Do not warn about compile-time overflow in constant expressions.
3294 Warn if a structure is given the packed attribute, but the packed
3295 attribute has no effect on the layout or size of the structure.
3296 Such structures may be mis-aligned for little benefit. For
3297 instance, in this code, the variable @code{f.x} in @code{struct bar}
3298 will be misaligned even though @code{struct bar} does not itself
3299 have the packed attribute:
3306 @} __attribute__((packed));
3316 Warn if padding is included in a structure, either to align an element
3317 of the structure or to align the whole structure. Sometimes when this
3318 happens it is possible to rearrange the fields of the structure to
3319 reduce the padding and so make the structure smaller.
3321 @item -Wredundant-decls
3322 @opindex Wredundant-decls
3323 Warn if anything is declared more than once in the same scope, even in
3324 cases where multiple declaration is valid and changes nothing.
3326 @item -Wnested-externs @r{(C only)}
3327 @opindex Wnested-externs
3328 Warn if an @code{extern} declaration is encountered within a function.
3330 @item -Wunreachable-code
3331 @opindex Wunreachable-code
3332 Warn if the compiler detects that code will never be executed.
3334 This option is intended to warn when the compiler detects that at
3335 least a whole line of source code will never be executed, because
3336 some condition is never satisfied or because it is after a
3337 procedure that never returns.
3339 It is possible for this option to produce a warning even though there
3340 are circumstances under which part of the affected line can be executed,
3341 so care should be taken when removing apparently-unreachable code.
3343 For instance, when a function is inlined, a warning may mean that the
3344 line is unreachable in only one inlined copy of the function.
3346 This option is not made part of @option{-Wall} because in a debugging
3347 version of a program there is often substantial code which checks
3348 correct functioning of the program and is, hopefully, unreachable
3349 because the program does work. Another common use of unreachable
3350 code is to provide behavior which is selectable at compile-time.
3354 Warn if a function can not be inlined and it was declared as inline.
3355 Even with this option, the compiler will not warn about failures to
3356 inline functions declared in system headers.
3358 The compiler uses a variety of heuristics to determine whether or not
3359 to inline a function. For example, the compiler takes into account
3360 the size of the function being inlined and the amount of inlining
3361 that has already been done in the current function. Therefore,
3362 seemingly insignificant changes in the source program can cause the
3363 warnings produced by @option{-Winline} to appear or disappear.
3365 @item -Wno-invalid-offsetof @r{(C++ only)}
3366 @opindex Wno-invalid-offsetof
3367 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3368 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3369 to a non-POD type is undefined. In existing C++ implementations,
3370 however, @samp{offsetof} typically gives meaningful results even when
3371 applied to certain kinds of non-POD types. (Such as a simple
3372 @samp{struct} that fails to be a POD type only by virtue of having a
3373 constructor.) This flag is for users who are aware that they are
3374 writing nonportable code and who have deliberately chosen to ignore the
3377 The restrictions on @samp{offsetof} may be relaxed in a future version
3378 of the C++ standard.
3380 @item -Wno-int-to-pointer-cast @r{(C only)}
3381 @opindex Wno-int-to-pointer-cast
3382 Suppress warnings from casts to pointer type of an integer of a
3385 @item -Wno-pointer-to-int-cast @r{(C only)}
3386 @opindex Wno-pointer-to-int-cast
3387 Suppress warnings from casts from a pointer to an integer type of a
3391 @opindex Winvalid-pch
3392 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3393 the search path but can't be used.
3397 @opindex Wno-long-long
3398 Warn if @samp{long long} type is used. This is default. To inhibit
3399 the warning messages, use @option{-Wno-long-long}. Flags
3400 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3401 only when @option{-pedantic} flag is used.
3403 @item -Wvariadic-macros
3404 @opindex Wvariadic-macros
3405 @opindex Wno-variadic-macros
3406 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3407 alternate syntax when in pedantic ISO C99 mode. This is default.
3408 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3410 @item -Wvolatile-register-var
3411 @opindex Wvolatile-register-var
3412 @opindex Wno-volatile-register-var
3413 Warn if a register variable is declared volatile. The volatile
3414 modifier does not inhibit all optimizations that may eliminate reads
3415 and/or writes to register variables.
3417 @item -Wdisabled-optimization
3418 @opindex Wdisabled-optimization
3419 Warn if a requested optimization pass is disabled. This warning does
3420 not generally indicate that there is anything wrong with your code; it
3421 merely indicates that GCC's optimizers were unable to handle the code
3422 effectively. Often, the problem is that your code is too big or too
3423 complex; GCC will refuse to optimize programs when the optimization
3424 itself is likely to take inordinate amounts of time.
3426 @item -Wpointer-sign
3427 @opindex Wpointer-sign
3428 @opindex Wno-pointer-sign
3429 Warn for pointer argument passing or assignment with different signedness.
3430 This option is only supported for C and Objective-C@. It is implied by
3431 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3432 @option{-Wno-pointer-sign}.
3436 Make all warnings into errors.
3440 Make the specified warning into an errors. The specifier for a
3441 warning is appended, for example @option{-Werror=switch} turns the
3442 warnings controlled by @option{-Wswitch} into errors. This switch
3443 takes a negative form, to be used to negate @option{-Werror} for
3444 specific warnings, for example @option{-Wno-error=switch} makes
3445 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3446 is in effect. You can use the @option{-fdiagnostics-show-option}
3447 option to have each controllable warning amended with the option which
3448 controls it, to determine what to use with this option.
3450 Note that specifying @option{-Werror=}@var{foo} automatically implies
3451 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3454 @item -Wstack-protector
3455 @opindex Wstack-protector
3456 This option is only active when @option{-fstack-protector} is active. It
3457 warns about functions that will not be protected against stack smashing.
3459 @item -Wstring-literal-comparison
3460 @opindex Wstring-literal-comparison
3461 Warn about suspicious comparisons to string literal constants. In C,
3462 direct comparisons against the memory address of a string literal, such
3463 as @code{if (x == "abc")}, typically indicate a programmer error, and
3464 even when intentional, result in unspecified behavior and are not portable.
3465 Usually these warnings alert that the programmer intended to use
3466 @code{strcmp}. This warning is enabled by @option{-Wall}.
3468 @item -Woverlength-strings
3469 @opindex Woverlength-strings
3470 Warn about string constants which are longer than the ``minimum
3471 maximum'' length specified in the C standard. Modern compilers
3472 generally allow string constants which are much longer than the
3473 standard's minimum limit, but very portable programs should avoid
3474 using longer strings.
3476 The limit applies @emph{after} string constant concatenation, and does
3477 not count the trailing NUL@. In C89, the limit was 509 characters; in
3478 C99, it was raised to 4095. C++98 does not specify a normative
3479 minimum maximum, so we do not diagnose overlength strings in C++@.
3481 This option is implied by @option{-pedantic}, and can be disabled with
3482 @option{-Wno-overlength-strings}.
3485 @node Debugging Options
3486 @section Options for Debugging Your Program or GCC
3487 @cindex options, debugging
3488 @cindex debugging information options
3490 GCC has various special options that are used for debugging
3491 either your program or GCC:
3496 Produce debugging information in the operating system's native format
3497 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3500 On most systems that use stabs format, @option{-g} enables use of extra
3501 debugging information that only GDB can use; this extra information
3502 makes debugging work better in GDB but will probably make other debuggers
3504 refuse to read the program. If you want to control for certain whether
3505 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3506 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3508 GCC allows you to use @option{-g} with
3509 @option{-O}. The shortcuts taken by optimized code may occasionally
3510 produce surprising results: some variables you declared may not exist
3511 at all; flow of control may briefly move where you did not expect it;
3512 some statements may not be executed because they compute constant
3513 results or their values were already at hand; some statements may
3514 execute in different places because they were moved out of loops.
3516 Nevertheless it proves possible to debug optimized output. This makes
3517 it reasonable to use the optimizer for programs that might have bugs.
3519 The following options are useful when GCC is generated with the
3520 capability for more than one debugging format.
3524 Produce debugging information for use by GDB@. This means to use the
3525 most expressive format available (DWARF 2, stabs, or the native format
3526 if neither of those are supported), including GDB extensions if at all
3531 Produce debugging information in stabs format (if that is supported),
3532 without GDB extensions. This is the format used by DBX on most BSD
3533 systems. On MIPS, Alpha and System V Release 4 systems this option
3534 produces stabs debugging output which is not understood by DBX or SDB@.
3535 On System V Release 4 systems this option requires the GNU assembler.
3537 @item -feliminate-unused-debug-symbols
3538 @opindex feliminate-unused-debug-symbols
3539 Produce debugging information in stabs format (if that is supported),
3540 for only symbols that are actually used.
3542 @item -femit-class-debug-always
3543 Instead of emitting debugging information for a C++ class in only one
3544 object file, emit it in all object files using the class. This option
3545 should be used only with debuggers that are unable to handle the way GCC
3546 normally emits debugging information for classes because using this
3547 option will increase the size of debugging information by as much as a
3552 Produce debugging information in stabs format (if that is supported),
3553 using GNU extensions understood only by the GNU debugger (GDB)@. The
3554 use of these extensions is likely to make other debuggers crash or
3555 refuse to read the program.
3559 Produce debugging information in COFF format (if that is supported).
3560 This is the format used by SDB on most System V systems prior to
3565 Produce debugging information in XCOFF format (if that is supported).
3566 This is the format used by the DBX debugger on IBM RS/6000 systems.
3570 Produce debugging information in XCOFF format (if that is supported),
3571 using GNU extensions understood only by the GNU debugger (GDB)@. The
3572 use of these extensions is likely to make other debuggers crash or
3573 refuse to read the program, and may cause assemblers other than the GNU
3574 assembler (GAS) to fail with an error.
3578 Produce debugging information in DWARF version 2 format (if that is
3579 supported). This is the format used by DBX on IRIX 6. With this
3580 option, GCC uses features of DWARF version 3 when they are useful;
3581 version 3 is upward compatible with version 2, but may still cause
3582 problems for older debuggers.
3586 Produce debugging information in VMS debug format (if that is
3587 supported). This is the format used by DEBUG on VMS systems.
3590 @itemx -ggdb@var{level}
3591 @itemx -gstabs@var{level}
3592 @itemx -gcoff@var{level}
3593 @itemx -gxcoff@var{level}
3594 @itemx -gvms@var{level}
3595 Request debugging information and also use @var{level} to specify how
3596 much information. The default level is 2.
3598 Level 1 produces minimal information, enough for making backtraces in
3599 parts of the program that you don't plan to debug. This includes
3600 descriptions of functions and external variables, but no information
3601 about local variables and no line numbers.
3603 Level 3 includes extra information, such as all the macro definitions
3604 present in the program. Some debuggers support macro expansion when
3605 you use @option{-g3}.
3607 @option{-gdwarf-2} does not accept a concatenated debug level, because
3608 GCC used to support an option @option{-gdwarf} that meant to generate
3609 debug information in version 1 of the DWARF format (which is very
3610 different from version 2), and it would have been too confusing. That
3611 debug format is long obsolete, but the option cannot be changed now.
3612 Instead use an additional @option{-g@var{level}} option to change the
3613 debug level for DWARF2.
3615 @item -feliminate-dwarf2-dups
3616 @opindex feliminate-dwarf2-dups
3617 Compress DWARF2 debugging information by eliminating duplicated
3618 information about each symbol. This option only makes sense when
3619 generating DWARF2 debugging information with @option{-gdwarf-2}.
3621 @cindex @command{prof}
3624 Generate extra code to write profile information suitable for the
3625 analysis program @command{prof}. You must use this option when compiling
3626 the source files you want data about, and you must also use it when
3629 @cindex @command{gprof}
3632 Generate extra code to write profile information suitable for the
3633 analysis program @command{gprof}. You must use this option when compiling
3634 the source files you want data about, and you must also use it when
3639 Makes the compiler print out each function name as it is compiled, and
3640 print some statistics about each pass when it finishes.
3643 @opindex ftime-report
3644 Makes the compiler print some statistics about the time consumed by each
3645 pass when it finishes.
3648 @opindex fmem-report
3649 Makes the compiler print some statistics about permanent memory
3650 allocation when it finishes.
3652 @item -fprofile-arcs
3653 @opindex fprofile-arcs
3654 Add code so that program flow @dfn{arcs} are instrumented. During
3655 execution the program records how many times each branch and call is
3656 executed and how many times it is taken or returns. When the compiled
3657 program exits it saves this data to a file called
3658 @file{@var{auxname}.gcda} for each source file. The data may be used for
3659 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3660 test coverage analysis (@option{-ftest-coverage}). Each object file's
3661 @var{auxname} is generated from the name of the output file, if
3662 explicitly specified and it is not the final executable, otherwise it is
3663 the basename of the source file. In both cases any suffix is removed
3664 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3665 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3666 @xref{Cross-profiling}.
3668 @cindex @command{gcov}
3672 This option is used to compile and link code instrumented for coverage
3673 analysis. The option is a synonym for @option{-fprofile-arcs}
3674 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3675 linking). See the documentation for those options for more details.
3680 Compile the source files with @option{-fprofile-arcs} plus optimization
3681 and code generation options. For test coverage analysis, use the
3682 additional @option{-ftest-coverage} option. You do not need to profile
3683 every source file in a program.
3686 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3687 (the latter implies the former).
3690 Run the program on a representative workload to generate the arc profile
3691 information. This may be repeated any number of times. You can run
3692 concurrent instances of your program, and provided that the file system
3693 supports locking, the data files will be correctly updated. Also
3694 @code{fork} calls are detected and correctly handled (double counting
3698 For profile-directed optimizations, compile the source files again with
3699 the same optimization and code generation options plus
3700 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3701 Control Optimization}).
3704 For test coverage analysis, use @command{gcov} to produce human readable
3705 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3706 @command{gcov} documentation for further information.
3710 With @option{-fprofile-arcs}, for each function of your program GCC
3711 creates a program flow graph, then finds a spanning tree for the graph.
3712 Only arcs that are not on the spanning tree have to be instrumented: the
3713 compiler adds code to count the number of times that these arcs are
3714 executed. When an arc is the only exit or only entrance to a block, the
3715 instrumentation code can be added to the block; otherwise, a new basic
3716 block must be created to hold the instrumentation code.
3719 @item -ftest-coverage
3720 @opindex ftest-coverage
3721 Produce a notes file that the @command{gcov} code-coverage utility
3722 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3723 show program coverage. Each source file's note file is called
3724 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3725 above for a description of @var{auxname} and instructions on how to
3726 generate test coverage data. Coverage data will match the source files
3727 more closely, if you do not optimize.
3729 @item -d@var{letters}
3730 @item -fdump-rtl-@var{pass}
3732 Says to make debugging dumps during compilation at times specified by
3733 @var{letters}. This is used for debugging the RTL-based passes of the
3734 compiler. The file names for most of the dumps are made by appending a
3735 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3736 from the name of the output file, if explicitly specified and it is not
3737 an executable, otherwise it is the basename of the source file.
3739 Most debug dumps can be enabled either passing a letter to the @option{-d}
3740 option, or with a long @option{-fdump-rtl} switch; here are the possible
3741 letters for use in @var{letters} and @var{pass}, and their meanings:
3746 Annotate the assembler output with miscellaneous debugging information.
3749 @itemx -fdump-rtl-bbro
3751 @opindex fdump-rtl-bbro
3752 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3755 @itemx -fdump-rtl-combine
3757 @opindex fdump-rtl-combine
3758 Dump after instruction combination, to the file @file{@var{file}.129r.combine}.
3761 @itemx -fdump-rtl-ce1
3762 @itemx -fdump-rtl-ce2
3764 @opindex fdump-rtl-ce1
3765 @opindex fdump-rtl-ce2
3766 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3767 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3768 and @option{-fdump-rtl-ce2} enable dumping after the second if
3769 conversion, to the file @file{@var{file}.130r.ce2}.
3772 @itemx -fdump-rtl-btl
3773 @itemx -fdump-rtl-dbr
3775 @opindex fdump-rtl-btl
3776 @opindex fdump-rtl-dbr
3777 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3778 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3779 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3780 scheduling, to @file{@var{file}.36.dbr}.
3784 Dump all macro definitions, at the end of preprocessing, in addition to
3788 @itemx -fdump-rtl-ce3
3790 @opindex fdump-rtl-ce3
3791 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3794 @itemx -fdump-rtl-cfg
3795 @itemx -fdump-rtl-life
3797 @opindex fdump-rtl-cfg
3798 @opindex fdump-rtl-life
3799 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3800 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3801 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3802 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3805 @itemx -fdump-rtl-greg
3807 @opindex fdump-rtl-greg
3808 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3811 @itemx -fdump-rtl-gcse
3812 @itemx -fdump-rtl-bypass
3814 @opindex fdump-rtl-gcse
3815 @opindex fdump-rtl-bypass
3816 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3817 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3818 enable dumping after jump bypassing and control flow optimizations, to
3819 @file{@var{file}.115r.bypass}.
3822 @itemx -fdump-rtl-eh
3824 @opindex fdump-rtl-eh
3825 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3828 @itemx -fdump-rtl-sibling
3830 @opindex fdump-rtl-sibling
3831 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3834 @itemx -fdump-rtl-jump
3836 @opindex fdump-rtl-jump
3837 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3840 @itemx -fdump-rtl-stack
3842 @opindex fdump-rtl-stack
3843 Dump after conversion from registers to stack, to @file{@var{file}.152r.stack}.
3846 @itemx -fdump-rtl-lreg
3848 @opindex fdump-rtl-lreg
3849 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3852 @itemx -fdump-rtl-loop2
3854 @opindex fdump-rtl-loop2
3855 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3856 loop optimization pass, to @file{@var{file}.119r.loop2},
3857 @file{@var{file}.120r.loop2_init},
3858 @file{@var{file}.121r.loop2_invariant}, and
3859 @file{@var{file}.125r.loop2_done}.
3862 @itemx -fdump-rtl-sms
3864 @opindex fdump-rtl-sms
3865 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3868 @itemx -fdump-rtl-mach
3870 @opindex fdump-rtl-mach
3871 Dump after performing the machine dependent reorganization pass, to
3872 @file{@var{file}.155r.mach}.
3875 @itemx -fdump-rtl-rnreg
3877 @opindex fdump-rtl-rnreg
3878 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
3881 @itemx -fdump-rtl-regmove
3883 @opindex fdump-rtl-regmove
3884 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
3887 @itemx -fdump-rtl-postreload
3889 @opindex fdump-rtl-postreload
3890 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3893 @itemx -fdump-rtl-expand
3895 @opindex fdump-rtl-expand
3896 Dump after RTL generation, to @file{@var{file}.104r.expand}.
3899 @itemx -fdump-rtl-sched2
3901 @opindex fdump-rtl-sched2
3902 Dump after the second scheduling pass, to @file{@var{file}.150r.sched2}.
3905 @itemx -fdump-rtl-cse
3907 @opindex fdump-rtl-cse
3908 Dump after CSE (including the jump optimization that sometimes follows
3909 CSE), to @file{@var{file}.113r.cse}.
3912 @itemx -fdump-rtl-sched
3914 @opindex fdump-rtl-sched
3915 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3918 @itemx -fdump-rtl-cse2
3920 @opindex fdump-rtl-cse2
3921 Dump after the second CSE pass (including the jump optimization that
3922 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
3925 @itemx -fdump-rtl-tracer
3927 @opindex fdump-rtl-tracer
3928 Dump after running tracer, to @file{@var{file}.118r.tracer}.
3931 @itemx -fdump-rtl-vpt
3932 @itemx -fdump-rtl-vartrack
3934 @opindex fdump-rtl-vpt
3935 @opindex fdump-rtl-vartrack
3936 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3937 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3938 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3939 to @file{@var{file}.154r.vartrack}.
3942 @itemx -fdump-rtl-flow2
3944 @opindex fdump-rtl-flow2
3945 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
3948 @itemx -fdump-rtl-peephole2
3950 @opindex fdump-rtl-peephole2
3951 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
3954 @itemx -fdump-rtl-web
3956 @opindex fdump-rtl-web
3957 Dump after live range splitting, to @file{@var{file}.126r.web}.
3960 @itemx -fdump-rtl-all
3962 @opindex fdump-rtl-all
3963 Produce all the dumps listed above.
3967 Produce a core dump whenever an error occurs.
3971 Print statistics on memory usage, at the end of the run, to
3976 Annotate the assembler output with a comment indicating which
3977 pattern and alternative was used. The length of each instruction is
3982 Dump the RTL in the assembler output as a comment before each instruction.
3983 Also turns on @option{-dp} annotation.
3987 For each of the other indicated dump files (either with @option{-d} or
3988 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3989 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3993 Just generate RTL for a function instead of compiling it. Usually used
3994 with @samp{r} (@option{-fdump-rtl-expand}).
3998 Dump debugging information during parsing, to standard error.
4001 @item -fdump-unnumbered
4002 @opindex fdump-unnumbered
4003 When doing debugging dumps (see @option{-d} option above), suppress instruction
4004 numbers and line number note output. This makes it more feasible to
4005 use diff on debugging dumps for compiler invocations with different
4006 options, in particular with and without @option{-g}.
4008 @item -fdump-translation-unit @r{(C++ only)}
4009 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4010 @opindex fdump-translation-unit
4011 Dump a representation of the tree structure for the entire translation
4012 unit to a file. The file name is made by appending @file{.tu} to the
4013 source file name. If the @samp{-@var{options}} form is used, @var{options}
4014 controls the details of the dump as described for the
4015 @option{-fdump-tree} options.
4017 @item -fdump-class-hierarchy @r{(C++ only)}
4018 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4019 @opindex fdump-class-hierarchy
4020 Dump a representation of each class's hierarchy and virtual function
4021 table layout to a file. The file name is made by appending @file{.class}
4022 to the source file name. If the @samp{-@var{options}} form is used,
4023 @var{options} controls the details of the dump as described for the
4024 @option{-fdump-tree} options.
4026 @item -fdump-ipa-@var{switch}
4028 Control the dumping at various stages of inter-procedural analysis
4029 language tree to a file. The file name is generated by appending a switch
4030 specific suffix to the source file name. The following dumps are possible:
4034 Enables all inter-procedural analysis dumps; currently the only produced
4035 dump is the @samp{cgraph} dump.
4038 Dumps information about call-graph optimization, unused function removal,
4039 and inlining decisions.
4042 @item -fdump-tree-@var{switch}
4043 @itemx -fdump-tree-@var{switch}-@var{options}
4045 Control the dumping at various stages of processing the intermediate
4046 language tree to a file. The file name is generated by appending a switch
4047 specific suffix to the source file name. If the @samp{-@var{options}}
4048 form is used, @var{options} is a list of @samp{-} separated options that
4049 control the details of the dump. Not all options are applicable to all
4050 dumps, those which are not meaningful will be ignored. The following
4051 options are available
4055 Print the address of each node. Usually this is not meaningful as it
4056 changes according to the environment and source file. Its primary use
4057 is for tying up a dump file with a debug environment.
4059 Inhibit dumping of members of a scope or body of a function merely
4060 because that scope has been reached. Only dump such items when they
4061 are directly reachable by some other path. When dumping pretty-printed
4062 trees, this option inhibits dumping the bodies of control structures.
4064 Print a raw representation of the tree. By default, trees are
4065 pretty-printed into a C-like representation.
4067 Enable more detailed dumps (not honored by every dump option).
4069 Enable dumping various statistics about the pass (not honored by every dump
4072 Enable showing basic block boundaries (disabled in raw dumps).
4074 Enable showing virtual operands for every statement.
4076 Enable showing line numbers for statements.
4078 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4080 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4083 The following tree dumps are possible:
4087 Dump before any tree based optimization, to @file{@var{file}.original}.
4090 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4093 Dump after function inlining, to @file{@var{file}.inlined}.
4096 @opindex fdump-tree-gimple
4097 Dump each function before and after the gimplification pass to a file. The
4098 file name is made by appending @file{.gimple} to the source file name.
4101 @opindex fdump-tree-cfg
4102 Dump the control flow graph of each function to a file. The file name is
4103 made by appending @file{.cfg} to the source file name.
4106 @opindex fdump-tree-vcg
4107 Dump the control flow graph of each function to a file in VCG format. The
4108 file name is made by appending @file{.vcg} to the source file name. Note
4109 that if the file contains more than one function, the generated file cannot
4110 be used directly by VCG@. You will need to cut and paste each function's
4111 graph into its own separate file first.
4114 @opindex fdump-tree-ch
4115 Dump each function after copying loop headers. The file name is made by
4116 appending @file{.ch} to the source file name.
4119 @opindex fdump-tree-ssa
4120 Dump SSA related information to a file. The file name is made by appending
4121 @file{.ssa} to the source file name.
4124 @opindex fdump-tree-salias
4125 Dump structure aliasing variable information to a file. This file name
4126 is made by appending @file{.salias} to the source file name.
4129 @opindex fdump-tree-alias
4130 Dump aliasing information for each function. The file name is made by
4131 appending @file{.alias} to the source file name.
4134 @opindex fdump-tree-ccp
4135 Dump each function after CCP@. The file name is made by appending
4136 @file{.ccp} to the source file name.
4139 @opindex fdump-tree-storeccp
4140 Dump each function after STORE-CCP. The file name is made by appending
4141 @file{.storeccp} to the source file name.
4144 @opindex fdump-tree-pre
4145 Dump trees after partial redundancy elimination. The file name is made
4146 by appending @file{.pre} to the source file name.
4149 @opindex fdump-tree-fre
4150 Dump trees after full redundancy elimination. The file name is made
4151 by appending @file{.fre} to the source file name.
4154 @opindex fdump-tree-copyprop
4155 Dump trees after copy propagation. The file name is made
4156 by appending @file{.copyprop} to the source file name.
4158 @item store_copyprop
4159 @opindex fdump-tree-store_copyprop
4160 Dump trees after store copy-propagation. The file name is made
4161 by appending @file{.store_copyprop} to the source file name.
4164 @opindex fdump-tree-dce
4165 Dump each function after dead code elimination. The file name is made by
4166 appending @file{.dce} to the source file name.
4169 @opindex fdump-tree-mudflap
4170 Dump each function after adding mudflap instrumentation. The file name is
4171 made by appending @file{.mudflap} to the source file name.
4174 @opindex fdump-tree-sra
4175 Dump each function after performing scalar replacement of aggregates. The
4176 file name is made by appending @file{.sra} to the source file name.
4179 @opindex fdump-tree-sink
4180 Dump each function after performing code sinking. The file name is made
4181 by appending @file{.sink} to the source file name.
4184 @opindex fdump-tree-dom
4185 Dump each function after applying dominator tree optimizations. The file
4186 name is made by appending @file{.dom} to the source file name.
4189 @opindex fdump-tree-dse
4190 Dump each function after applying dead store elimination. The file
4191 name is made by appending @file{.dse} to the source file name.
4194 @opindex fdump-tree-phiopt
4195 Dump each function after optimizing PHI nodes into straightline code. The file
4196 name is made by appending @file{.phiopt} to the source file name.
4199 @opindex fdump-tree-forwprop
4200 Dump each function after forward propagating single use variables. The file
4201 name is made by appending @file{.forwprop} to the source file name.
4204 @opindex fdump-tree-copyrename
4205 Dump each function after applying the copy rename optimization. The file
4206 name is made by appending @file{.copyrename} to the source file name.
4209 @opindex fdump-tree-nrv
4210 Dump each function after applying the named return value optimization on
4211 generic trees. The file name is made by appending @file{.nrv} to the source
4215 @opindex fdump-tree-vect
4216 Dump each function after applying vectorization of loops. The file name is
4217 made by appending @file{.vect} to the source file name.
4220 @opindex fdump-tree-vrp
4221 Dump each function after Value Range Propagation (VRP). The file name
4222 is made by appending @file{.vrp} to the source file name.
4225 @opindex fdump-tree-all
4226 Enable all the available tree dumps with the flags provided in this option.
4229 @item -ftree-vectorizer-verbose=@var{n}
4230 @opindex ftree-vectorizer-verbose
4231 This option controls the amount of debugging output the vectorizer prints.
4232 This information is written to standard error, unless
4233 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4234 in which case it is output to the usual dump listing file, @file{.vect}.
4235 For @var{n}=0 no diagnostic information is reported.
4236 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4237 and the total number of loops that got vectorized.
4238 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4239 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4240 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4241 level that @option{-fdump-tree-vect-stats} uses.
4242 Higher verbosity levels mean either more information dumped for each
4243 reported loop, or same amount of information reported for more loops:
4244 If @var{n}=3, alignment related information is added to the reports.
4245 If @var{n}=4, data-references related information (e.g. memory dependences,
4246 memory access-patterns) is added to the reports.
4247 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4248 that did not pass the first analysis phase (i.e. may not be countable, or
4249 may have complicated control-flow).
4250 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4251 For @var{n}=7, all the information the vectorizer generates during its
4252 analysis and transformation is reported. This is the same verbosity level
4253 that @option{-fdump-tree-vect-details} uses.
4255 @item -frandom-seed=@var{string}
4256 @opindex frandom-string
4257 This option provides a seed that GCC uses when it would otherwise use
4258 random numbers. It is used to generate certain symbol names
4259 that have to be different in every compiled file. It is also used to
4260 place unique stamps in coverage data files and the object files that
4261 produce them. You can use the @option{-frandom-seed} option to produce
4262 reproducibly identical object files.
4264 The @var{string} should be different for every file you compile.
4266 @item -fsched-verbose=@var{n}
4267 @opindex fsched-verbose
4268 On targets that use instruction scheduling, this option controls the
4269 amount of debugging output the scheduler prints. This information is
4270 written to standard error, unless @option{-dS} or @option{-dR} is
4271 specified, in which case it is output to the usual dump
4272 listing file, @file{.sched} or @file{.sched2} respectively. However
4273 for @var{n} greater than nine, the output is always printed to standard
4276 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4277 same information as @option{-dRS}. For @var{n} greater than one, it
4278 also output basic block probabilities, detailed ready list information
4279 and unit/insn info. For @var{n} greater than two, it includes RTL
4280 at abort point, control-flow and regions info. And for @var{n} over
4281 four, @option{-fsched-verbose} also includes dependence info.
4285 Store the usual ``temporary'' intermediate files permanently; place them
4286 in the current directory and name them based on the source file. Thus,
4287 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4288 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4289 preprocessed @file{foo.i} output file even though the compiler now
4290 normally uses an integrated preprocessor.
4292 When used in combination with the @option{-x} command line option,
4293 @option{-save-temps} is sensible enough to avoid over writing an
4294 input source file with the same extension as an intermediate file.
4295 The corresponding intermediate file may be obtained by renaming the
4296 source file before using @option{-save-temps}.
4300 Report the CPU time taken by each subprocess in the compilation
4301 sequence. For C source files, this is the compiler proper and assembler
4302 (plus the linker if linking is done). The output looks like this:
4309 The first number on each line is the ``user time'', that is time spent
4310 executing the program itself. The second number is ``system time'',
4311 time spent executing operating system routines on behalf of the program.
4312 Both numbers are in seconds.
4314 @item -fvar-tracking
4315 @opindex fvar-tracking
4316 Run variable tracking pass. It computes where variables are stored at each
4317 position in code. Better debugging information is then generated
4318 (if the debugging information format supports this information).
4320 It is enabled by default when compiling with optimization (@option{-Os},
4321 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4322 the debug info format supports it.
4324 @item -print-file-name=@var{library}
4325 @opindex print-file-name
4326 Print the full absolute name of the library file @var{library} that
4327 would be used when linking---and don't do anything else. With this
4328 option, GCC does not compile or link anything; it just prints the
4331 @item -print-multi-directory
4332 @opindex print-multi-directory
4333 Print the directory name corresponding to the multilib selected by any
4334 other switches present in the command line. This directory is supposed
4335 to exist in @env{GCC_EXEC_PREFIX}.
4337 @item -print-multi-lib
4338 @opindex print-multi-lib
4339 Print the mapping from multilib directory names to compiler switches
4340 that enable them. The directory name is separated from the switches by
4341 @samp{;}, and each switch starts with an @samp{@@} instead of the
4342 @samp{-}, without spaces between multiple switches. This is supposed to
4343 ease shell-processing.
4345 @item -print-prog-name=@var{program}
4346 @opindex print-prog-name
4347 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4349 @item -print-libgcc-file-name
4350 @opindex print-libgcc-file-name
4351 Same as @option{-print-file-name=libgcc.a}.
4353 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4354 but you do want to link with @file{libgcc.a}. You can do
4357 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4360 @item -print-search-dirs
4361 @opindex print-search-dirs
4362 Print the name of the configured installation directory and a list of
4363 program and library directories @command{gcc} will search---and don't do anything else.
4365 This is useful when @command{gcc} prints the error message
4366 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4367 To resolve this you either need to put @file{cpp0} and the other compiler
4368 components where @command{gcc} expects to find them, or you can set the environment
4369 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4370 Don't forget the trailing @samp{/}.
4371 @xref{Environment Variables}.
4374 @opindex dumpmachine
4375 Print the compiler's target machine (for example,
4376 @samp{i686-pc-linux-gnu})---and don't do anything else.
4379 @opindex dumpversion
4380 Print the compiler version (for example, @samp{3.0})---and don't do
4385 Print the compiler's built-in specs---and don't do anything else. (This
4386 is used when GCC itself is being built.) @xref{Spec Files}.
4388 @item -feliminate-unused-debug-types
4389 @opindex feliminate-unused-debug-types
4390 Normally, when producing DWARF2 output, GCC will emit debugging
4391 information for all types declared in a compilation
4392 unit, regardless of whether or not they are actually used
4393 in that compilation unit. Sometimes this is useful, such as
4394 if, in the debugger, you want to cast a value to a type that is
4395 not actually used in your program (but is declared). More often,
4396 however, this results in a significant amount of wasted space.
4397 With this option, GCC will avoid producing debug symbol output
4398 for types that are nowhere used in the source file being compiled.
4401 @node Optimize Options
4402 @section Options That Control Optimization
4403 @cindex optimize options
4404 @cindex options, optimization
4406 These options control various sorts of optimizations.
4408 Without any optimization option, the compiler's goal is to reduce the
4409 cost of compilation and to make debugging produce the expected
4410 results. Statements are independent: if you stop the program with a
4411 breakpoint between statements, you can then assign a new value to any
4412 variable or change the program counter to any other statement in the
4413 function and get exactly the results you would expect from the source
4416 Turning on optimization flags makes the compiler attempt to improve
4417 the performance and/or code size at the expense of compilation time
4418 and possibly the ability to debug the program.
4420 The compiler performs optimization based on the knowledge it has of
4421 the program. Optimization levels @option{-O2} and above, in
4422 particular, enable @emph{unit-at-a-time} mode, which allows the
4423 compiler to consider information gained from later functions in
4424 the file when compiling a function. Compiling multiple files at
4425 once to a single output file in @emph{unit-at-a-time} mode allows
4426 the compiler to use information gained from all of the files when
4427 compiling each of them.
4429 Not all optimizations are controlled directly by a flag. Only
4430 optimizations that have a flag are listed.
4437 Optimize. Optimizing compilation takes somewhat more time, and a lot
4438 more memory for a large function.
4440 With @option{-O}, the compiler tries to reduce code size and execution
4441 time, without performing any optimizations that take a great deal of
4444 @option{-O} turns on the following optimization flags:
4445 @gccoptlist{-fdefer-pop @gol
4446 -fdelayed-branch @gol
4447 -fguess-branch-probability @gol
4448 -fcprop-registers @gol
4449 -fif-conversion @gol
4450 -fif-conversion2 @gol
4453 -ftree-dominator-opts @gol
4458 -ftree-copyrename @gol
4463 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4464 where doing so does not interfere with debugging.
4468 Optimize even more. GCC performs nearly all supported optimizations
4469 that do not involve a space-speed tradeoff. The compiler does not
4470 perform loop unrolling or function inlining when you specify @option{-O2}.
4471 As compared to @option{-O}, this option increases both compilation time
4472 and the performance of the generated code.
4474 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4475 also turns on the following optimization flags:
4476 @gccoptlist{-fthread-jumps @gol
4478 -foptimize-sibling-calls @gol
4479 -fcse-follow-jumps -fcse-skip-blocks @gol
4480 -fgcse -fgcse-lm @gol
4481 -fexpensive-optimizations @gol
4482 -frerun-cse-after-loop @gol
4485 -fschedule-insns -fschedule-insns2 @gol
4486 -fsched-interblock -fsched-spec @gol
4488 -fstrict-aliasing @gol
4489 -fdelete-null-pointer-checks @gol
4490 -freorder-blocks -freorder-functions @gol
4491 -funit-at-a-time @gol
4492 -falign-functions -falign-jumps @gol
4493 -falign-loops -falign-labels @gol
4497 Please note the warning under @option{-fgcse} about
4498 invoking @option{-O2} on programs that use computed gotos.
4502 Optimize yet more. @option{-O3} turns on all optimizations specified by
4503 @option{-O2} and also turns on the @option{-finline-functions},
4504 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4508 Do not optimize. This is the default.
4512 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4513 do not typically increase code size. It also performs further
4514 optimizations designed to reduce code size.
4516 @option{-Os} disables the following optimization flags:
4517 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4518 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4519 -fprefetch-loop-arrays -ftree-vect-loop-version}
4521 If you use multiple @option{-O} options, with or without level numbers,
4522 the last such option is the one that is effective.
4525 Options of the form @option{-f@var{flag}} specify machine-independent
4526 flags. Most flags have both positive and negative forms; the negative
4527 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4528 below, only one of the forms is listed---the one you typically will
4529 use. You can figure out the other form by either removing @samp{no-}
4532 The following options control specific optimizations. They are either
4533 activated by @option{-O} options or are related to ones that are. You
4534 can use the following flags in the rare cases when ``fine-tuning'' of
4535 optimizations to be performed is desired.
4538 @item -fno-default-inline
4539 @opindex fno-default-inline
4540 Do not make member functions inline by default merely because they are
4541 defined inside the class scope (C++ only). Otherwise, when you specify
4542 @w{@option{-O}}, member functions defined inside class scope are compiled
4543 inline by default; i.e., you don't need to add @samp{inline} in front of
4544 the member function name.
4546 @item -fno-defer-pop
4547 @opindex fno-defer-pop
4548 Always pop the arguments to each function call as soon as that function
4549 returns. For machines which must pop arguments after a function call,
4550 the compiler normally lets arguments accumulate on the stack for several
4551 function calls and pops them all at once.
4553 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4557 Force memory operands to be copied into registers before doing
4558 arithmetic on them. This produces better code by making all memory
4559 references potential common subexpressions. When they are not common
4560 subexpressions, instruction combination should eliminate the separate
4561 register-load. This option is now a nop and will be removed in 4.2.
4564 @opindex fforce-addr
4565 Force memory address constants to be copied into registers before
4566 doing arithmetic on them.
4568 @item -fomit-frame-pointer
4569 @opindex fomit-frame-pointer
4570 Don't keep the frame pointer in a register for functions that
4571 don't need one. This avoids the instructions to save, set up and
4572 restore frame pointers; it also makes an extra register available
4573 in many functions. @strong{It also makes debugging impossible on
4576 On some machines, such as the VAX, this flag has no effect, because
4577 the standard calling sequence automatically handles the frame pointer
4578 and nothing is saved by pretending it doesn't exist. The
4579 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4580 whether a target machine supports this flag. @xref{Registers,,Register
4581 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4583 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4585 @item -foptimize-sibling-calls
4586 @opindex foptimize-sibling-calls
4587 Optimize sibling and tail recursive calls.
4589 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4593 Don't pay attention to the @code{inline} keyword. Normally this option
4594 is used to keep the compiler from expanding any functions inline.
4595 Note that if you are not optimizing, no functions can be expanded inline.
4597 @item -finline-functions
4598 @opindex finline-functions
4599 Integrate all simple functions into their callers. The compiler
4600 heuristically decides which functions are simple enough to be worth
4601 integrating in this way.
4603 If all calls to a given function are integrated, and the function is
4604 declared @code{static}, then the function is normally not output as
4605 assembler code in its own right.
4607 Enabled at level @option{-O3}.
4609 @item -finline-functions-called-once
4610 @opindex finline-functions-called-once
4611 Consider all @code{static} functions called once for inlining into their
4612 caller even if they are not marked @code{inline}. If a call to a given
4613 function is integrated, then the function is not output as assembler code
4616 Enabled if @option{-funit-at-a-time} is enabled.
4618 @item -fearly-inlining
4619 @opindex fearly-inlining
4620 Inline functions marked by @code{always_inline} and functions whose body seems
4621 smaller than the function call overhead early before doing
4622 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4623 makes profiling significantly cheaper and usually inlining faster on programs
4624 having large chains of nested wrapper functions.
4628 @item -finline-limit=@var{n}
4629 @opindex finline-limit
4630 By default, GCC limits the size of functions that can be inlined. This flag
4631 allows the control of this limit for functions that are explicitly marked as
4632 inline (i.e., marked with the inline keyword or defined within the class
4633 definition in c++). @var{n} is the size of functions that can be inlined in
4634 number of pseudo instructions (not counting parameter handling). The default
4635 value of @var{n} is 600.
4636 Increasing this value can result in more inlined code at
4637 the cost of compilation time and memory consumption. Decreasing usually makes
4638 the compilation faster and less code will be inlined (which presumably
4639 means slower programs). This option is particularly useful for programs that
4640 use inlining heavily such as those based on recursive templates with C++.
4642 Inlining is actually controlled by a number of parameters, which may be
4643 specified individually by using @option{--param @var{name}=@var{value}}.
4644 The @option{-finline-limit=@var{n}} option sets some of these parameters
4648 @item max-inline-insns-single
4649 is set to @var{n}/2.
4650 @item max-inline-insns-auto
4651 is set to @var{n}/2.
4652 @item min-inline-insns
4653 is set to 130 or @var{n}/4, whichever is smaller.
4654 @item max-inline-insns-rtl
4658 See below for a documentation of the individual
4659 parameters controlling inlining.
4661 @emph{Note:} pseudo instruction represents, in this particular context, an
4662 abstract measurement of function's size. In no way does it represent a count
4663 of assembly instructions and as such its exact meaning might change from one
4664 release to an another.
4666 @item -fkeep-inline-functions
4667 @opindex fkeep-inline-functions
4668 In C, emit @code{static} functions that are declared @code{inline}
4669 into the object file, even if the function has been inlined into all
4670 of its callers. This switch does not affect functions using the
4671 @code{extern inline} extension in GNU C@. In C++, emit any and all
4672 inline functions into the object file.
4674 @item -fkeep-static-consts
4675 @opindex fkeep-static-consts
4676 Emit variables declared @code{static const} when optimization isn't turned
4677 on, even if the variables aren't referenced.
4679 GCC enables this option by default. If you want to force the compiler to
4680 check if the variable was referenced, regardless of whether or not
4681 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4683 @item -fmerge-constants
4684 Attempt to merge identical constants (string constants and floating point
4685 constants) across compilation units.
4687 This option is the default for optimized compilation if the assembler and
4688 linker support it. Use @option{-fno-merge-constants} to inhibit this
4691 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4693 @item -fmerge-all-constants
4694 Attempt to merge identical constants and identical variables.
4696 This option implies @option{-fmerge-constants}. In addition to
4697 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4698 arrays or initialized constant variables with integral or floating point
4699 types. Languages like C or C++ require each non-automatic variable to
4700 have distinct location, so using this option will result in non-conforming
4703 @item -fmodulo-sched
4704 @opindex fmodulo-sched
4705 Perform swing modulo scheduling immediately before the first scheduling
4706 pass. This pass looks at innermost loops and reorders their
4707 instructions by overlapping different iterations.
4709 @item -fno-branch-count-reg
4710 @opindex fno-branch-count-reg
4711 Do not use ``decrement and branch'' instructions on a count register,
4712 but instead generate a sequence of instructions that decrement a
4713 register, compare it against zero, then branch based upon the result.
4714 This option is only meaningful on architectures that support such
4715 instructions, which include x86, PowerPC, IA-64 and S/390.
4717 The default is @option{-fbranch-count-reg}.
4719 @item -fno-function-cse
4720 @opindex fno-function-cse
4721 Do not put function addresses in registers; make each instruction that
4722 calls a constant function contain the function's address explicitly.
4724 This option results in less efficient code, but some strange hacks
4725 that alter the assembler output may be confused by the optimizations
4726 performed when this option is not used.
4728 The default is @option{-ffunction-cse}
4730 @item -fno-zero-initialized-in-bss
4731 @opindex fno-zero-initialized-in-bss
4732 If the target supports a BSS section, GCC by default puts variables that
4733 are initialized to zero into BSS@. This can save space in the resulting
4736 This option turns off this behavior because some programs explicitly
4737 rely on variables going to the data section. E.g., so that the
4738 resulting executable can find the beginning of that section and/or make
4739 assumptions based on that.
4741 The default is @option{-fzero-initialized-in-bss}.
4743 @item -fbounds-check
4744 @opindex fbounds-check
4745 For front-ends that support it, generate additional code to check that
4746 indices used to access arrays are within the declared range. This is
4747 currently only supported by the Java and Fortran front-ends, where
4748 this option defaults to true and false respectively.
4750 @item -fmudflap -fmudflapth -fmudflapir
4754 @cindex bounds checking
4756 For front-ends that support it (C and C++), instrument all risky
4757 pointer/array dereferencing operations, some standard library
4758 string/heap functions, and some other associated constructs with
4759 range/validity tests. Modules so instrumented should be immune to
4760 buffer overflows, invalid heap use, and some other classes of C/C++
4761 programming errors. The instrumentation relies on a separate runtime
4762 library (@file{libmudflap}), which will be linked into a program if
4763 @option{-fmudflap} is given at link time. Run-time behavior of the
4764 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4765 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4768 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4769 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4770 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4771 instrumentation should ignore pointer reads. This produces less
4772 instrumentation (and therefore faster execution) and still provides
4773 some protection against outright memory corrupting writes, but allows
4774 erroneously read data to propagate within a program.
4776 @item -fthread-jumps
4777 @opindex fthread-jumps
4778 Perform optimizations where we check to see if a jump branches to a
4779 location where another comparison subsumed by the first is found. If
4780 so, the first branch is redirected to either the destination of the
4781 second branch or a point immediately following it, depending on whether
4782 the condition is known to be true or false.
4784 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4786 @item -fcse-follow-jumps
4787 @opindex fcse-follow-jumps
4788 In common subexpression elimination, scan through jump instructions
4789 when the target of the jump is not reached by any other path. For
4790 example, when CSE encounters an @code{if} statement with an
4791 @code{else} clause, CSE will follow the jump when the condition
4794 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4796 @item -fcse-skip-blocks
4797 @opindex fcse-skip-blocks
4798 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4799 follow jumps which conditionally skip over blocks. When CSE
4800 encounters a simple @code{if} statement with no else clause,
4801 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4802 body of the @code{if}.
4804 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4806 @item -frerun-cse-after-loop
4807 @opindex frerun-cse-after-loop
4808 Re-run common subexpression elimination after loop optimizations has been
4811 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4815 Perform a global common subexpression elimination pass.
4816 This pass also performs global constant and copy propagation.
4818 @emph{Note:} When compiling a program using computed gotos, a GCC
4819 extension, you may get better runtime performance if you disable
4820 the global common subexpression elimination pass by adding
4821 @option{-fno-gcse} to the command line.
4823 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4827 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4828 attempt to move loads which are only killed by stores into themselves. This
4829 allows a loop containing a load/store sequence to be changed to a load outside
4830 the loop, and a copy/store within the loop.
4832 Enabled by default when gcse is enabled.
4836 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4837 global common subexpression elimination. This pass will attempt to move
4838 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4839 loops containing a load/store sequence can be changed to a load before
4840 the loop and a store after the loop.
4842 Not enabled at any optimization level.
4846 When @option{-fgcse-las} is enabled, the global common subexpression
4847 elimination pass eliminates redundant loads that come after stores to the
4848 same memory location (both partial and full redundancies).
4850 Not enabled at any optimization level.
4852 @item -fgcse-after-reload
4853 @opindex fgcse-after-reload
4854 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4855 pass is performed after reload. The purpose of this pass is to cleanup
4858 @item -funsafe-loop-optimizations
4859 @opindex funsafe-loop-optimizations
4860 If given, the loop optimizer will assume that loop indices do not
4861 overflow, and that the loops with nontrivial exit condition are not
4862 infinite. This enables a wider range of loop optimizations even if
4863 the loop optimizer itself cannot prove that these assumptions are valid.
4864 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4865 if it finds this kind of loop.
4867 @item -fcrossjumping
4868 @opindex crossjumping
4869 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4870 resulting code may or may not perform better than without cross-jumping.
4872 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4874 @item -fif-conversion
4875 @opindex if-conversion
4876 Attempt to transform conditional jumps into branch-less equivalents. This
4877 include use of conditional moves, min, max, set flags and abs instructions, and
4878 some tricks doable by standard arithmetics. The use of conditional execution
4879 on chips where it is available is controlled by @code{if-conversion2}.
4881 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4883 @item -fif-conversion2
4884 @opindex if-conversion2
4885 Use conditional execution (where available) to transform conditional jumps into
4886 branch-less equivalents.
4888 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4890 @item -fdelete-null-pointer-checks
4891 @opindex fdelete-null-pointer-checks
4892 Use global dataflow analysis to identify and eliminate useless checks
4893 for null pointers. The compiler assumes that dereferencing a null
4894 pointer would have halted the program. If a pointer is checked after
4895 it has already been dereferenced, it cannot be null.
4897 In some environments, this assumption is not true, and programs can
4898 safely dereference null pointers. Use
4899 @option{-fno-delete-null-pointer-checks} to disable this optimization
4900 for programs which depend on that behavior.
4902 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4904 @item -fexpensive-optimizations
4905 @opindex fexpensive-optimizations
4906 Perform a number of minor optimizations that are relatively expensive.
4908 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4910 @item -foptimize-register-move
4912 @opindex foptimize-register-move
4914 Attempt to reassign register numbers in move instructions and as
4915 operands of other simple instructions in order to maximize the amount of
4916 register tying. This is especially helpful on machines with two-operand
4919 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4922 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4924 @item -fdelayed-branch
4925 @opindex fdelayed-branch
4926 If supported for the target machine, attempt to reorder instructions
4927 to exploit instruction slots available after delayed branch
4930 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4932 @item -fschedule-insns
4933 @opindex fschedule-insns
4934 If supported for the target machine, attempt to reorder instructions to
4935 eliminate execution stalls due to required data being unavailable. This
4936 helps machines that have slow floating point or memory load instructions
4937 by allowing other instructions to be issued until the result of the load
4938 or floating point instruction is required.
4940 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4942 @item -fschedule-insns2
4943 @opindex fschedule-insns2
4944 Similar to @option{-fschedule-insns}, but requests an additional pass of
4945 instruction scheduling after register allocation has been done. This is
4946 especially useful on machines with a relatively small number of
4947 registers and where memory load instructions take more than one cycle.
4949 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4951 @item -fno-sched-interblock
4952 @opindex fno-sched-interblock
4953 Don't schedule instructions across basic blocks. This is normally
4954 enabled by default when scheduling before register allocation, i.e.@:
4955 with @option{-fschedule-insns} or at @option{-O2} or higher.
4957 @item -fno-sched-spec
4958 @opindex fno-sched-spec
4959 Don't allow speculative motion of non-load instructions. This is normally
4960 enabled by default when scheduling before register allocation, i.e.@:
4961 with @option{-fschedule-insns} or at @option{-O2} or higher.
4963 @item -fsched-spec-load
4964 @opindex fsched-spec-load
4965 Allow speculative motion of some load instructions. This only makes
4966 sense when scheduling before register allocation, i.e.@: with
4967 @option{-fschedule-insns} or at @option{-O2} or higher.
4969 @item -fsched-spec-load-dangerous
4970 @opindex fsched-spec-load-dangerous
4971 Allow speculative motion of more load instructions. This only makes
4972 sense when scheduling before register allocation, i.e.@: with
4973 @option{-fschedule-insns} or at @option{-O2} or higher.
4975 @item -fsched-stalled-insns=@var{n}
4976 @opindex fsched-stalled-insns
4977 Define how many insns (if any) can be moved prematurely from the queue
4978 of stalled insns into the ready list, during the second scheduling pass.
4980 @item -fsched-stalled-insns-dep=@var{n}
4981 @opindex fsched-stalled-insns-dep
4982 Define how many insn groups (cycles) will be examined for a dependency
4983 on a stalled insn that is candidate for premature removal from the queue
4984 of stalled insns. Has an effect only during the second scheduling pass,
4985 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4987 @item -fsched2-use-superblocks
4988 @opindex fsched2-use-superblocks
4989 When scheduling after register allocation, do use superblock scheduling
4990 algorithm. Superblock scheduling allows motion across basic block boundaries
4991 resulting on faster schedules. This option is experimental, as not all machine
4992 descriptions used by GCC model the CPU closely enough to avoid unreliable
4993 results from the algorithm.
4995 This only makes sense when scheduling after register allocation, i.e.@: with
4996 @option{-fschedule-insns2} or at @option{-O2} or higher.
4998 @item -fsched2-use-traces
4999 @opindex fsched2-use-traces
5000 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5001 allocation and additionally perform code duplication in order to increase the
5002 size of superblocks using tracer pass. See @option{-ftracer} for details on
5005 This mode should produce faster but significantly longer programs. Also
5006 without @option{-fbranch-probabilities} the traces constructed may not
5007 match the reality and hurt the performance. This only makes
5008 sense when scheduling after register allocation, i.e.@: with
5009 @option{-fschedule-insns2} or at @option{-O2} or higher.
5013 Eliminates redundant extension instructions and move the non redundant
5014 ones to optimal placement using LCM.
5016 @item -freschedule-modulo-scheduled-loops
5017 @opindex fscheduling-in-modulo-scheduled-loops
5018 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5019 we may want to prevent the later scheduling passes from changing its schedule, we use this
5020 option to control that.
5022 @item -fcaller-saves
5023 @opindex fcaller-saves
5024 Enable values to be allocated in registers that will be clobbered by
5025 function calls, by emitting extra instructions to save and restore the
5026 registers around such calls. Such allocation is done only when it
5027 seems to result in better code than would otherwise be produced.
5029 This option is always enabled by default on certain machines, usually
5030 those which have no call-preserved registers to use instead.
5032 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5035 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5036 enabled by default at @option{-O2} and @option{-O3}.
5039 Perform Full Redundancy Elimination (FRE) on trees. The difference
5040 between FRE and PRE is that FRE only considers expressions
5041 that are computed on all paths leading to the redundant computation.
5042 This analysis faster than PRE, though it exposes fewer redundancies.
5043 This flag is enabled by default at @option{-O} and higher.
5045 @item -ftree-copy-prop
5046 Perform copy propagation on trees. This pass eliminates unnecessary
5047 copy operations. This flag is enabled by default at @option{-O} and
5050 @item -ftree-store-copy-prop
5051 Perform copy propagation of memory loads and stores. This pass
5052 eliminates unnecessary copy operations in memory references
5053 (structures, global variables, arrays, etc). This flag is enabled by
5054 default at @option{-O2} and higher.
5057 Perform structural alias analysis on trees. This flag
5058 is enabled by default at @option{-O} and higher.
5061 Perform interprocedural pointer analysis.
5064 Perform forward store motion on trees. This flag is
5065 enabled by default at @option{-O} and higher.
5068 Perform sparse conditional constant propagation (CCP) on trees. This
5069 pass only operates on local scalar variables and is enabled by default
5070 at @option{-O} and higher.
5072 @item -ftree-store-ccp
5073 Perform sparse conditional constant propagation (CCP) on trees. This
5074 pass operates on both local scalar variables and memory stores and
5075 loads (global variables, structures, arrays, etc). This flag is
5076 enabled by default at @option{-O2} and higher.
5079 Perform dead code elimination (DCE) on trees. This flag is enabled by
5080 default at @option{-O} and higher.
5082 @item -ftree-dominator-opts
5083 Perform a variety of simple scalar cleanups (constant/copy
5084 propagation, redundancy elimination, range propagation and expression
5085 simplification) based on a dominator tree traversal. This also
5086 performs jump threading (to reduce jumps to jumps). This flag is
5087 enabled by default at @option{-O} and higher.
5090 Perform loop header copying on trees. This is beneficial since it increases
5091 effectiveness of code motion optimizations. It also saves one jump. This flag
5092 is enabled by default at @option{-O} and higher. It is not enabled
5093 for @option{-Os}, since it usually increases code size.
5095 @item -ftree-loop-optimize
5096 Perform loop optimizations on trees. This flag is enabled by default
5097 at @option{-O} and higher.
5099 @item -ftree-loop-linear
5100 Perform linear loop transformations on tree. This flag can improve cache
5101 performance and allow further loop optimizations to take place.
5103 @item -ftree-loop-im
5104 Perform loop invariant motion on trees. This pass moves only invariants that
5105 would be hard to handle at RTL level (function calls, operations that expand to
5106 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5107 operands of conditions that are invariant out of the loop, so that we can use
5108 just trivial invariantness analysis in loop unswitching. The pass also includes
5111 @item -ftree-loop-ivcanon
5112 Create a canonical counter for number of iterations in the loop for that
5113 determining number of iterations requires complicated analysis. Later
5114 optimizations then may determine the number easily. Useful especially
5115 in connection with unrolling.
5118 Perform induction variable optimizations (strength reduction, induction
5119 variable merging and induction variable elimination) on trees.
5122 Perform scalar replacement of aggregates. This pass replaces structure
5123 references with scalars to prevent committing structures to memory too
5124 early. This flag is enabled by default at @option{-O} and higher.
5126 @item -ftree-copyrename
5127 Perform copy renaming on trees. This pass attempts to rename compiler
5128 temporaries to other variables at copy locations, usually resulting in
5129 variable names which more closely resemble the original variables. This flag
5130 is enabled by default at @option{-O} and higher.
5133 Perform temporary expression replacement during the SSA->normal phase. Single
5134 use/single def temporaries are replaced at their use location with their
5135 defining expression. This results in non-GIMPLE code, but gives the expanders
5136 much more complex trees to work on resulting in better RTL generation. This is
5137 enabled by default at @option{-O} and higher.
5140 Perform live range splitting during the SSA->normal phase. Distinct live
5141 ranges of a variable are split into unique variables, allowing for better
5142 optimization later. This is enabled by default at @option{-O} and higher.
5144 @item -ftree-vectorize
5145 Perform loop vectorization on trees.
5147 @item -ftree-vect-loop-version
5148 @opindex ftree-vect-loop-version
5149 Perform loop versioning when doing loop vectorization on trees. When a loop
5150 appears to be vectorizable except that data alignment or data dependence cannot
5151 be determined at compile time then vectorized and non-vectorized versions of
5152 the loop are generated along with runtime checks for alignment or dependence
5153 to control which version is executed. This option is enabled by default
5154 except at level @option{-Os} where it is disabled.
5157 Perform Value Range Propagation on trees. This is similar to the
5158 constant propagation pass, but instead of values, ranges of values are
5159 propagated. This allows the optimizers to remove unnecessary range
5160 checks like array bound checks and null pointer checks. This is
5161 enabled by default at @option{-O2} and higher. Null pointer check
5162 elimination is only done if @option{-fdelete-null-pointer-checks} is
5167 Perform tail duplication to enlarge superblock size. This transformation
5168 simplifies the control flow of the function allowing other optimizations to do
5171 @item -funroll-loops
5172 @opindex funroll-loops
5173 Unroll loops whose number of iterations can be determined at compile
5174 time or upon entry to the loop. @option{-funroll-loops} implies
5175 @option{-frerun-cse-after-loop}. This option makes code larger,
5176 and may or may not make it run faster.
5178 @item -funroll-all-loops
5179 @opindex funroll-all-loops
5180 Unroll all loops, even if their number of iterations is uncertain when
5181 the loop is entered. This usually makes programs run more slowly.
5182 @option{-funroll-all-loops} implies the same options as
5183 @option{-funroll-loops},
5185 @item -fsplit-ivs-in-unroller
5186 @opindex -fsplit-ivs-in-unroller
5187 Enables expressing of values of induction variables in later iterations
5188 of the unrolled loop using the value in the first iteration. This breaks
5189 long dependency chains, thus improving efficiency of the scheduling passes.
5191 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5192 same effect. However in cases the loop body is more complicated than
5193 a single basic block, this is not reliable. It also does not work at all
5194 on some of the architectures due to restrictions in the CSE pass.
5196 This optimization is enabled by default.
5198 @item -fvariable-expansion-in-unroller
5199 @opindex -fvariable-expansion-in-unroller
5200 With this option, the compiler will create multiple copies of some
5201 local variables when unrolling a loop which can result in superior code.
5203 @item -fprefetch-loop-arrays
5204 @opindex fprefetch-loop-arrays
5205 If supported by the target machine, generate instructions to prefetch
5206 memory to improve the performance of loops that access large arrays.
5208 This option may generate better or worse code; results are highly
5209 dependent on the structure of loops within the source code.
5211 Disabled at level @option{-Os}.
5214 @itemx -fno-peephole2
5215 @opindex fno-peephole
5216 @opindex fno-peephole2
5217 Disable any machine-specific peephole optimizations. The difference
5218 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5219 are implemented in the compiler; some targets use one, some use the
5220 other, a few use both.
5222 @option{-fpeephole} is enabled by default.
5223 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5225 @item -fno-guess-branch-probability
5226 @opindex fno-guess-branch-probability
5227 Do not guess branch probabilities using heuristics.
5229 GCC will use heuristics to guess branch probabilities if they are
5230 not provided by profiling feedback (@option{-fprofile-arcs}). These
5231 heuristics are based on the control flow graph. If some branch probabilities
5232 are specified by @samp{__builtin_expect}, then the heuristics will be
5233 used to guess branch probabilities for the rest of the control flow graph,
5234 taking the @samp{__builtin_expect} info into account. The interactions
5235 between the heuristics and @samp{__builtin_expect} can be complex, and in
5236 some cases, it may be useful to disable the heuristics so that the effects
5237 of @samp{__builtin_expect} are easier to understand.
5239 The default is @option{-fguess-branch-probability} at levels
5240 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5242 @item -freorder-blocks
5243 @opindex freorder-blocks
5244 Reorder basic blocks in the compiled function in order to reduce number of
5245 taken branches and improve code locality.
5247 Enabled at levels @option{-O2}, @option{-O3}.
5249 @item -freorder-blocks-and-partition
5250 @opindex freorder-blocks-and-partition
5251 In addition to reordering basic blocks in the compiled function, in order
5252 to reduce number of taken branches, partitions hot and cold basic blocks
5253 into separate sections of the assembly and .o files, to improve
5254 paging and cache locality performance.
5256 This optimization is automatically turned off in the presence of
5257 exception handling, for linkonce sections, for functions with a user-defined
5258 section attribute and on any architecture that does not support named
5261 @item -freorder-functions
5262 @opindex freorder-functions
5263 Reorder functions in the object file in order to
5264 improve code locality. This is implemented by using special
5265 subsections @code{.text.hot} for most frequently executed functions and
5266 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5267 the linker so object file format must support named sections and linker must
5268 place them in a reasonable way.
5270 Also profile feedback must be available in to make this option effective. See
5271 @option{-fprofile-arcs} for details.
5273 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5275 @item -fstrict-aliasing
5276 @opindex fstrict-aliasing
5277 Allows the compiler to assume the strictest aliasing rules applicable to
5278 the language being compiled. For C (and C++), this activates
5279 optimizations based on the type of expressions. In particular, an
5280 object of one type is assumed never to reside at the same address as an
5281 object of a different type, unless the types are almost the same. For
5282 example, an @code{unsigned int} can alias an @code{int}, but not a
5283 @code{void*} or a @code{double}. A character type may alias any other
5286 Pay special attention to code like this:
5299 The practice of reading from a different union member than the one most
5300 recently written to (called ``type-punning'') is common. Even with
5301 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5302 is accessed through the union type. So, the code above will work as
5303 expected. However, this code might not:
5314 Every language that wishes to perform language-specific alias analysis
5315 should define a function that computes, given an @code{tree}
5316 node, an alias set for the node. Nodes in different alias sets are not
5317 allowed to alias. For an example, see the C front-end function
5318 @code{c_get_alias_set}.
5320 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5322 @item -falign-functions
5323 @itemx -falign-functions=@var{n}
5324 @opindex falign-functions
5325 Align the start of functions to the next power-of-two greater than
5326 @var{n}, skipping up to @var{n} bytes. For instance,
5327 @option{-falign-functions=32} aligns functions to the next 32-byte
5328 boundary, but @option{-falign-functions=24} would align to the next
5329 32-byte boundary only if this can be done by skipping 23 bytes or less.
5331 @option{-fno-align-functions} and @option{-falign-functions=1} are
5332 equivalent and mean that functions will not be aligned.
5334 Some assemblers only support this flag when @var{n} is a power of two;
5335 in that case, it is rounded up.
5337 If @var{n} is not specified or is zero, use a machine-dependent default.
5339 Enabled at levels @option{-O2}, @option{-O3}.
5341 @item -falign-labels
5342 @itemx -falign-labels=@var{n}
5343 @opindex falign-labels
5344 Align all branch targets to a power-of-two boundary, skipping up to
5345 @var{n} bytes like @option{-falign-functions}. This option can easily
5346 make code slower, because it must insert dummy operations for when the
5347 branch target is reached in the usual flow of the code.
5349 @option{-fno-align-labels} and @option{-falign-labels=1} are
5350 equivalent and mean that labels will not be aligned.
5352 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5353 are greater than this value, then their values are used instead.
5355 If @var{n} is not specified or is zero, use a machine-dependent default
5356 which is very likely to be @samp{1}, meaning no alignment.
5358 Enabled at levels @option{-O2}, @option{-O3}.
5361 @itemx -falign-loops=@var{n}
5362 @opindex falign-loops
5363 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5364 like @option{-falign-functions}. The hope is that the loop will be
5365 executed many times, which will make up for any execution of the dummy
5368 @option{-fno-align-loops} and @option{-falign-loops=1} are
5369 equivalent and mean that loops will not be aligned.
5371 If @var{n} is not specified or is zero, use a machine-dependent default.
5373 Enabled at levels @option{-O2}, @option{-O3}.
5376 @itemx -falign-jumps=@var{n}
5377 @opindex falign-jumps
5378 Align branch targets to a power-of-two boundary, for branch targets
5379 where the targets can only be reached by jumping, skipping up to @var{n}
5380 bytes like @option{-falign-functions}. In this case, no dummy operations
5383 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5384 equivalent and mean that loops will not be aligned.
5386 If @var{n} is not specified or is zero, use a machine-dependent default.
5388 Enabled at levels @option{-O2}, @option{-O3}.
5390 @item -funit-at-a-time
5391 @opindex funit-at-a-time
5392 Parse the whole compilation unit before starting to produce code.
5393 This allows some extra optimizations to take place but consumes
5394 more memory (in general). There are some compatibility issues
5395 with @emph{unit-at-a-time} mode:
5398 enabling @emph{unit-at-a-time} mode may change the order
5399 in which functions, variables, and top-level @code{asm} statements
5400 are emitted, and will likely break code relying on some particular
5401 ordering. The majority of such top-level @code{asm} statements,
5402 though, can be replaced by @code{section} attributes. The
5403 @option{fno-toplevel-reorder} option may be used to keep the ordering
5404 used in the input file, at the cost of some optimizations.
5407 @emph{unit-at-a-time} mode removes unreferenced static variables
5408 and functions. This may result in undefined references
5409 when an @code{asm} statement refers directly to variables or functions
5410 that are otherwise unused. In that case either the variable/function
5411 shall be listed as an operand of the @code{asm} statement operand or,
5412 in the case of top-level @code{asm} statements the attribute @code{used}
5413 shall be used on the declaration.
5416 Static functions now can use non-standard passing conventions that
5417 may break @code{asm} statements calling functions directly. Again,
5418 attribute @code{used} will prevent this behavior.
5421 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5422 but this scheme may not be supported by future releases of GCC@.
5424 Enabled at levels @option{-O2}, @option{-O3}.
5426 @item -fno-toplevel-reorder
5427 Do not reorder top-level functions, variables, and @code{asm}
5428 statements. Output them in the same order that they appear in the
5429 input file. When this option is used, unreferenced static variables
5430 will not be removed. This option is intended to support existing code
5431 which relies on a particular ordering. For new code, it is better to
5436 Constructs webs as commonly used for register allocation purposes and assign
5437 each web individual pseudo register. This allows the register allocation pass
5438 to operate on pseudos directly, but also strengthens several other optimization
5439 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5440 however, make debugging impossible, since variables will no longer stay in a
5443 Enabled by default with @option{-funroll-loops}.
5445 @item -fwhole-program
5446 @opindex fwhole-program
5447 Assume that the current compilation unit represents whole program being
5448 compiled. All public functions and variables with the exception of @code{main}
5449 and those merged by attribute @code{externally_visible} become static functions
5450 and in a affect gets more aggressively optimized by interprocedural optimizers.
5451 While this option is equivalent to proper use of @code{static} keyword for
5452 programs consisting of single file, in combination with option
5453 @option{--combine} this flag can be used to compile most of smaller scale C
5454 programs since the functions and variables become local for the whole combined
5455 compilation unit, not for the single source file itself.
5458 @item -fno-cprop-registers
5459 @opindex fno-cprop-registers
5460 After register allocation and post-register allocation instruction splitting,
5461 we perform a copy-propagation pass to try to reduce scheduling dependencies
5462 and occasionally eliminate the copy.
5464 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5466 @item -fprofile-generate
5467 @opindex fprofile-generate
5469 Enable options usually used for instrumenting application to produce
5470 profile useful for later recompilation with profile feedback based
5471 optimization. You must use @option{-fprofile-generate} both when
5472 compiling and when linking your program.
5474 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5477 @opindex fprofile-use
5478 Enable profile feedback directed optimizations, and optimizations
5479 generally profitable only with profile feedback available.
5481 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5482 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5486 The following options control compiler behavior regarding floating
5487 point arithmetic. These options trade off between speed and
5488 correctness. All must be specifically enabled.
5492 @opindex ffloat-store
5493 Do not store floating point variables in registers, and inhibit other
5494 options that might change whether a floating point value is taken from a
5497 @cindex floating point precision
5498 This option prevents undesirable excess precision on machines such as
5499 the 68000 where the floating registers (of the 68881) keep more
5500 precision than a @code{double} is supposed to have. Similarly for the
5501 x86 architecture. For most programs, the excess precision does only
5502 good, but a few programs rely on the precise definition of IEEE floating
5503 point. Use @option{-ffloat-store} for such programs, after modifying
5504 them to store all pertinent intermediate computations into variables.
5508 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5509 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5510 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5511 and @option{fcx-limited-range}.
5513 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5515 This option should never be turned on by any @option{-O} option since
5516 it can result in incorrect output for programs which depend on
5517 an exact implementation of IEEE or ISO rules/specifications for
5520 @item -fno-math-errno
5521 @opindex fno-math-errno
5522 Do not set ERRNO after calling math functions that are executed
5523 with a single instruction, e.g., sqrt. A program that relies on
5524 IEEE exceptions for math error handling may want to use this flag
5525 for speed while maintaining IEEE arithmetic compatibility.
5527 This option should never be turned on by any @option{-O} option since
5528 it can result in incorrect output for programs which depend on
5529 an exact implementation of IEEE or ISO rules/specifications for
5532 The default is @option{-fmath-errno}.
5534 On Darwin systems, the math library never sets @code{errno}. There is therefore
5535 no reason for the compiler to consider the possibility that it might,
5536 and @option{-fno-math-errno} is the default.
5538 @item -funsafe-math-optimizations
5539 @opindex funsafe-math-optimizations
5540 Allow optimizations for floating-point arithmetic that (a) assume
5541 that arguments and results are valid and (b) may violate IEEE or
5542 ANSI standards. When used at link-time, it may include libraries
5543 or startup files that change the default FPU control word or other
5544 similar optimizations.
5546 This option should never be turned on by any @option{-O} option since
5547 it can result in incorrect output for programs which depend on
5548 an exact implementation of IEEE or ISO rules/specifications for
5551 The default is @option{-fno-unsafe-math-optimizations}.
5553 @item -ffinite-math-only
5554 @opindex ffinite-math-only
5555 Allow optimizations for floating-point arithmetic that assume
5556 that arguments and results are not NaNs or +-Infs.
5558 This option should never be turned on by any @option{-O} option since
5559 it can result in incorrect output for programs which depend on
5560 an exact implementation of IEEE or ISO rules/specifications.
5562 The default is @option{-fno-finite-math-only}.
5564 @item -fno-trapping-math
5565 @opindex fno-trapping-math
5566 Compile code assuming that floating-point operations cannot generate
5567 user-visible traps. These traps include division by zero, overflow,
5568 underflow, inexact result and invalid operation. This option implies
5569 @option{-fno-signaling-nans}. Setting this option may allow faster
5570 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5572 This option should never be turned on by any @option{-O} option since
5573 it can result in incorrect output for programs which depend on
5574 an exact implementation of IEEE or ISO rules/specifications for
5577 The default is @option{-ftrapping-math}.
5579 @item -frounding-math
5580 @opindex frounding-math
5581 Disable transformations and optimizations that assume default floating
5582 point rounding behavior. This is round-to-zero for all floating point
5583 to integer conversions, and round-to-nearest for all other arithmetic
5584 truncations. This option should be specified for programs that change
5585 the FP rounding mode dynamically, or that may be executed with a
5586 non-default rounding mode. This option disables constant folding of
5587 floating point expressions at compile-time (which may be affected by
5588 rounding mode) and arithmetic transformations that are unsafe in the
5589 presence of sign-dependent rounding modes.
5591 The default is @option{-fno-rounding-math}.
5593 This option is experimental and does not currently guarantee to
5594 disable all GCC optimizations that are affected by rounding mode.
5595 Future versions of GCC may provide finer control of this setting
5596 using C99's @code{FENV_ACCESS} pragma. This command line option
5597 will be used to specify the default state for @code{FENV_ACCESS}.
5599 @item -frtl-abstract-sequences
5600 @opindex frtl-abstract-sequences
5601 It is a size optimization method. This option is to find identical
5602 sequences of code, which can be turned into pseudo-procedures and
5603 then replace all occurrences with calls to the newly created
5604 subroutine. It is kind of an opposite of @option{-finline-functions}.
5605 This optimization runs at RTL level.
5607 @item -fsignaling-nans
5608 @opindex fsignaling-nans
5609 Compile code assuming that IEEE signaling NaNs may generate user-visible
5610 traps during floating-point operations. Setting this option disables
5611 optimizations that may change the number of exceptions visible with
5612 signaling NaNs. This option implies @option{-ftrapping-math}.
5614 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5617 The default is @option{-fno-signaling-nans}.
5619 This option is experimental and does not currently guarantee to
5620 disable all GCC optimizations that affect signaling NaN behavior.
5622 @item -fsingle-precision-constant
5623 @opindex fsingle-precision-constant
5624 Treat floating point constant as single precision constant instead of
5625 implicitly converting it to double precision constant.
5627 @item -fcx-limited-range
5628 @itemx -fno-cx-limited-range
5629 @opindex fcx-limited-range
5630 @opindex fno-cx-limited-range
5631 When enabled, this option states that a range reduction step is not
5632 needed when performing complex division. The default is
5633 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5635 This option controls the default setting of the ISO C99
5636 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5641 The following options control optimizations that may improve
5642 performance, but are not enabled by any @option{-O} options. This
5643 section includes experimental options that may produce broken code.
5646 @item -fbranch-probabilities
5647 @opindex fbranch-probabilities
5648 After running a program compiled with @option{-fprofile-arcs}
5649 (@pxref{Debugging Options,, Options for Debugging Your Program or
5650 @command{gcc}}), you can compile it a second time using
5651 @option{-fbranch-probabilities}, to improve optimizations based on
5652 the number of times each branch was taken. When the program
5653 compiled with @option{-fprofile-arcs} exits it saves arc execution
5654 counts to a file called @file{@var{sourcename}.gcda} for each source
5655 file The information in this data file is very dependent on the
5656 structure of the generated code, so you must use the same source code
5657 and the same optimization options for both compilations.
5659 With @option{-fbranch-probabilities}, GCC puts a
5660 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5661 These can be used to improve optimization. Currently, they are only
5662 used in one place: in @file{reorg.c}, instead of guessing which path a
5663 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5664 exactly determine which path is taken more often.
5666 @item -fprofile-values
5667 @opindex fprofile-values
5668 If combined with @option{-fprofile-arcs}, it adds code so that some
5669 data about values of expressions in the program is gathered.
5671 With @option{-fbranch-probabilities}, it reads back the data gathered
5672 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5673 notes to instructions for their later usage in optimizations.
5675 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5679 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5680 a code to gather information about values of expressions.
5682 With @option{-fbranch-probabilities}, it reads back the data gathered
5683 and actually performs the optimizations based on them.
5684 Currently the optimizations include specialization of division operation
5685 using the knowledge about the value of the denominator.
5687 @item -frename-registers
5688 @opindex frename-registers
5689 Attempt to avoid false dependencies in scheduled code by making use
5690 of registers left over after register allocation. This optimization
5691 will most benefit processors with lots of registers. Depending on the
5692 debug information format adopted by the target, however, it can
5693 make debugging impossible, since variables will no longer stay in
5694 a ``home register''.
5696 Enabled by default with @option{-funroll-loops}.
5700 Perform tail duplication to enlarge superblock size. This transformation
5701 simplifies the control flow of the function allowing other optimizations to do
5704 Enabled with @option{-fprofile-use}.
5706 @item -funroll-loops
5707 @opindex funroll-loops
5708 Unroll loops whose number of iterations can be determined at compile time or
5709 upon entry to the loop. @option{-funroll-loops} implies
5710 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5711 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5712 small constant number of iterations). This option makes code larger, and may
5713 or may not make it run faster.
5715 Enabled with @option{-fprofile-use}.
5717 @item -funroll-all-loops
5718 @opindex funroll-all-loops
5719 Unroll all loops, even if their number of iterations is uncertain when
5720 the loop is entered. This usually makes programs run more slowly.
5721 @option{-funroll-all-loops} implies the same options as
5722 @option{-funroll-loops}.
5725 @opindex fpeel-loops
5726 Peels the loops for that there is enough information that they do not
5727 roll much (from profile feedback). It also turns on complete loop peeling
5728 (i.e.@: complete removal of loops with small constant number of iterations).
5730 Enabled with @option{-fprofile-use}.
5732 @item -fmove-loop-invariants
5733 @opindex fmove-loop-invariants
5734 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5735 at level @option{-O1}
5737 @item -funswitch-loops
5738 @opindex funswitch-loops
5739 Move branches with loop invariant conditions out of the loop, with duplicates
5740 of the loop on both branches (modified according to result of the condition).
5742 @item -ffunction-sections
5743 @itemx -fdata-sections
5744 @opindex ffunction-sections
5745 @opindex fdata-sections
5746 Place each function or data item into its own section in the output
5747 file if the target supports arbitrary sections. The name of the
5748 function or the name of the data item determines the section's name
5751 Use these options on systems where the linker can perform optimizations
5752 to improve locality of reference in the instruction space. Most systems
5753 using the ELF object format and SPARC processors running Solaris 2 have
5754 linkers with such optimizations. AIX may have these optimizations in
5757 Only use these options when there are significant benefits from doing
5758 so. When you specify these options, the assembler and linker will
5759 create larger object and executable files and will also be slower.
5760 You will not be able to use @code{gprof} on all systems if you
5761 specify this option and you may have problems with debugging if
5762 you specify both this option and @option{-g}.
5764 @item -fbranch-target-load-optimize
5765 @opindex fbranch-target-load-optimize
5766 Perform branch target register load optimization before prologue / epilogue
5768 The use of target registers can typically be exposed only during reload,
5769 thus hoisting loads out of loops and doing inter-block scheduling needs
5770 a separate optimization pass.
5772 @item -fbranch-target-load-optimize2
5773 @opindex fbranch-target-load-optimize2
5774 Perform branch target register load optimization after prologue / epilogue
5777 @item -fbtr-bb-exclusive
5778 @opindex fbtr-bb-exclusive
5779 When performing branch target register load optimization, don't reuse
5780 branch target registers in within any basic block.
5782 @item -fstack-protector
5783 Emit extra code to check for buffer overflows, such as stack smashing
5784 attacks. This is done by adding a guard variable to functions with
5785 vulnerable objects. This includes functions that call alloca, and
5786 functions with buffers larger than 8 bytes. The guards are initialized
5787 when a function is entered and then checked when the function exits.
5788 If a guard check fails, an error message is printed and the program exits.
5790 @item -fstack-protector-all
5791 Like @option{-fstack-protector} except that all functions are protected.
5793 @item -fsection-anchors
5794 @opindex fsection-anchors
5795 Try to reduce the number of symbolic address calculations by using
5796 shared ``anchor'' symbols to address nearby objects. This transformation
5797 can help to reduce the number of GOT entries and GOT accesses on some
5800 For example, the implementation of the following function @code{foo}:
5804 int foo (void) @{ return a + b + c; @}
5807 would usually calculate the addresses of all three variables, but if you
5808 compile it with @option{-fsection-anchors}, it will access the variables
5809 from a common anchor point instead. The effect is similar to the
5810 following pseudocode (which isn't valid C):
5815 register int *xr = &x;
5816 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5820 Not all targets support this option.
5822 @item --param @var{name}=@var{value}
5824 In some places, GCC uses various constants to control the amount of
5825 optimization that is done. For example, GCC will not inline functions
5826 that contain more that a certain number of instructions. You can
5827 control some of these constants on the command-line using the
5828 @option{--param} option.
5830 The names of specific parameters, and the meaning of the values, are
5831 tied to the internals of the compiler, and are subject to change
5832 without notice in future releases.
5834 In each case, the @var{value} is an integer. The allowable choices for
5835 @var{name} are given in the following table:
5838 @item salias-max-implicit-fields
5839 The maximum number of fields in a variable without direct
5840 structure accesses for which structure aliasing will consider trying
5841 to track each field. The default is 5
5843 @item salias-max-array-elements
5844 The maximum number of elements an array can have and its elements
5845 still be tracked individually by structure aliasing. The default is 4
5847 @item sra-max-structure-size
5848 The maximum structure size, in bytes, at which the scalar replacement
5849 of aggregates (SRA) optimization will perform block copies. The
5850 default value, 0, implies that GCC will select the most appropriate
5853 @item sra-field-structure-ratio
5854 The threshold ratio (as a percentage) between instantiated fields and
5855 the complete structure size. We say that if the ratio of the number
5856 of bytes in instantiated fields to the number of bytes in the complete
5857 structure exceeds this parameter, then block copies are not used. The
5860 @item max-crossjump-edges
5861 The maximum number of incoming edges to consider for crossjumping.
5862 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5863 the number of edges incoming to each block. Increasing values mean
5864 more aggressive optimization, making the compile time increase with
5865 probably small improvement in executable size.
5867 @item min-crossjump-insns
5868 The minimum number of instructions which must be matched at the end
5869 of two blocks before crossjumping will be performed on them. This
5870 value is ignored in the case where all instructions in the block being
5871 crossjumped from are matched. The default value is 5.
5873 @item max-grow-copy-bb-insns
5874 The maximum code size expansion factor when copying basic blocks
5875 instead of jumping. The expansion is relative to a jump instruction.
5876 The default value is 8.
5878 @item max-goto-duplication-insns
5879 The maximum number of instructions to duplicate to a block that jumps
5880 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5881 passes, GCC factors computed gotos early in the compilation process,
5882 and unfactors them as late as possible. Only computed jumps at the
5883 end of a basic blocks with no more than max-goto-duplication-insns are
5884 unfactored. The default value is 8.
5886 @item max-delay-slot-insn-search
5887 The maximum number of instructions to consider when looking for an
5888 instruction to fill a delay slot. If more than this arbitrary number of
5889 instructions is searched, the time savings from filling the delay slot
5890 will be minimal so stop searching. Increasing values mean more
5891 aggressive optimization, making the compile time increase with probably
5892 small improvement in executable run time.
5894 @item max-delay-slot-live-search
5895 When trying to fill delay slots, the maximum number of instructions to
5896 consider when searching for a block with valid live register
5897 information. Increasing this arbitrarily chosen value means more
5898 aggressive optimization, increasing the compile time. This parameter
5899 should be removed when the delay slot code is rewritten to maintain the
5902 @item max-gcse-memory
5903 The approximate maximum amount of memory that will be allocated in
5904 order to perform the global common subexpression elimination
5905 optimization. If more memory than specified is required, the
5906 optimization will not be done.
5908 @item max-gcse-passes
5909 The maximum number of passes of GCSE to run. The default is 1.
5911 @item max-pending-list-length
5912 The maximum number of pending dependencies scheduling will allow
5913 before flushing the current state and starting over. Large functions
5914 with few branches or calls can create excessively large lists which
5915 needlessly consume memory and resources.
5917 @item max-inline-insns-single
5918 Several parameters control the tree inliner used in gcc.
5919 This number sets the maximum number of instructions (counted in GCC's
5920 internal representation) in a single function that the tree inliner
5921 will consider for inlining. This only affects functions declared
5922 inline and methods implemented in a class declaration (C++).
5923 The default value is 450.
5925 @item max-inline-insns-auto
5926 When you use @option{-finline-functions} (included in @option{-O3}),
5927 a lot of functions that would otherwise not be considered for inlining
5928 by the compiler will be investigated. To those functions, a different
5929 (more restrictive) limit compared to functions declared inline can
5931 The default value is 90.
5933 @item large-function-insns
5934 The limit specifying really large functions. For functions larger than this
5935 limit after inlining inlining is constrained by
5936 @option{--param large-function-growth}. This parameter is useful primarily
5937 to avoid extreme compilation time caused by non-linear algorithms used by the
5939 This parameter is ignored when @option{-funit-at-a-time} is not used.
5940 The default value is 2700.
5942 @item large-function-growth
5943 Specifies maximal growth of large function caused by inlining in percents.
5944 This parameter is ignored when @option{-funit-at-a-time} is not used.
5945 The default value is 100 which limits large function growth to 2.0 times
5948 @item large-unit-insns
5949 The limit specifying large translation unit. Growth caused by inlining of
5950 units larger than this limit is limited by @option{--param inline-unit-growth}.
5951 For small units this might be too tight (consider unit consisting of function A
5952 that is inline and B that just calls A three time. If B is small relative to
5953 A, the growth of unit is 300\% and yet such inlining is very sane. For very
5954 large units consisting of small inlininable functions however the overall unit
5955 growth limit is needed to avoid exponential explosion of code size. Thus for
5956 smaller units, the size is increased to @option{--param large-unit-insns}
5957 before applying @option{--param inline-unit-growth}. The default is 10000
5959 @item inline-unit-growth
5960 Specifies maximal overall growth of the compilation unit caused by inlining.
5961 This parameter is ignored when @option{-funit-at-a-time} is not used.
5962 The default value is 50 which limits unit growth to 1.5 times the original
5965 @item max-inline-insns-recursive
5966 @itemx max-inline-insns-recursive-auto
5967 Specifies maximum number of instructions out-of-line copy of self recursive inline
5968 function can grow into by performing recursive inlining.
5970 For functions declared inline @option{--param max-inline-insns-recursive} is
5971 taken into acount. For function not declared inline, recursive inlining
5972 happens only when @option{-finline-functions} (included in @option{-O3}) is
5973 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5974 default value is 450.
5976 @item max-inline-recursive-depth
5977 @itemx max-inline-recursive-depth-auto
5978 Specifies maximum recursion depth used by the recursive inlining.
5980 For functions declared inline @option{--param max-inline-recursive-depth} is
5981 taken into acount. For function not declared inline, recursive inlining
5982 happens only when @option{-finline-functions} (included in @option{-O3}) is
5983 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5984 default value is 450.
5986 @item min-inline-recursive-probability
5987 Recursive inlining is profitable only for function having deep recursion
5988 in average and can hurt for function having little recursion depth by
5989 increasing the prologue size or complexity of function body to other
5992 When profile feedback is available (see @option{-fprofile-generate}) the actual
5993 recursion depth can be guessed from probability that function will recurse via
5994 given call expression. This parameter limits inlining only to call expression
5995 whose probability exceeds given threshold (in percents). The default value is
5998 @item inline-call-cost
5999 Specify cost of call instruction relative to simple arithmetics operations
6000 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6001 functions and at the same time increases size of leaf function that is believed to
6002 reduce function size by being inlined. In effect it increases amount of
6003 inlining for code having large abstraction penalty (many functions that just
6004 pass the arguments to other functions) and decrease inlining for code with low
6005 abstraction penalty. The default value is 16.
6007 @item max-unrolled-insns
6008 The maximum number of instructions that a loop should have if that loop
6009 is unrolled, and if the loop is unrolled, it determines how many times
6010 the loop code is unrolled.
6012 @item max-average-unrolled-insns
6013 The maximum number of instructions biased by probabilities of their execution
6014 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6015 it determines how many times the loop code is unrolled.
6017 @item max-unroll-times
6018 The maximum number of unrollings of a single loop.
6020 @item max-peeled-insns
6021 The maximum number of instructions that a loop should have if that loop
6022 is peeled, and if the loop is peeled, it determines how many times
6023 the loop code is peeled.
6025 @item max-peel-times
6026 The maximum number of peelings of a single loop.
6028 @item max-completely-peeled-insns
6029 The maximum number of insns of a completely peeled loop.
6031 @item max-completely-peel-times
6032 The maximum number of iterations of a loop to be suitable for complete peeling.
6034 @item max-unswitch-insns
6035 The maximum number of insns of an unswitched loop.
6037 @item max-unswitch-level
6038 The maximum number of branches unswitched in a single loop.
6041 The minimum cost of an expensive expression in the loop invariant motion.
6043 @item iv-consider-all-candidates-bound
6044 Bound on number of candidates for induction variables below that
6045 all candidates are considered for each use in induction variable
6046 optimizations. Only the most relevant candidates are considered
6047 if there are more candidates, to avoid quadratic time complexity.
6049 @item iv-max-considered-uses
6050 The induction variable optimizations give up on loops that contain more
6051 induction variable uses.
6053 @item iv-always-prune-cand-set-bound
6054 If number of candidates in the set is smaller than this value,
6055 we always try to remove unnecessary ivs from the set during its
6056 optimization when a new iv is added to the set.
6058 @item scev-max-expr-size
6059 Bound on size of expressions used in the scalar evolutions analyzer.
6060 Large expressions slow the analyzer.
6062 @item vect-max-version-checks
6063 The maximum number of runtime checks that can be performed when doing
6064 loop versioning in the vectorizer. See option ftree-vect-loop-version
6065 for more information.
6067 @item max-iterations-to-track
6069 The maximum number of iterations of a loop the brute force algorithm
6070 for analysis of # of iterations of the loop tries to evaluate.
6072 @item hot-bb-count-fraction
6073 Select fraction of the maximal count of repetitions of basic block in program
6074 given basic block needs to have to be considered hot.
6076 @item hot-bb-frequency-fraction
6077 Select fraction of the maximal frequency of executions of basic block in
6078 function given basic block needs to have to be considered hot
6080 @item max-predicted-iterations
6081 The maximum number of loop iterations we predict statically. This is useful
6082 in cases where function contain single loop with known bound and other loop
6083 with unknown. We predict the known number of iterations correctly, while
6084 the unknown number of iterations average to roughly 10. This means that the
6085 loop without bounds would appear artificially cold relative to the other one.
6087 @item tracer-dynamic-coverage
6088 @itemx tracer-dynamic-coverage-feedback
6090 This value is used to limit superblock formation once the given percentage of
6091 executed instructions is covered. This limits unnecessary code size
6094 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6095 feedback is available. The real profiles (as opposed to statically estimated
6096 ones) are much less balanced allowing the threshold to be larger value.
6098 @item tracer-max-code-growth
6099 Stop tail duplication once code growth has reached given percentage. This is
6100 rather hokey argument, as most of the duplicates will be eliminated later in
6101 cross jumping, so it may be set to much higher values than is the desired code
6104 @item tracer-min-branch-ratio
6106 Stop reverse growth when the reverse probability of best edge is less than this
6107 threshold (in percent).
6109 @item tracer-min-branch-ratio
6110 @itemx tracer-min-branch-ratio-feedback
6112 Stop forward growth if the best edge do have probability lower than this
6115 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6116 compilation for profile feedback and one for compilation without. The value
6117 for compilation with profile feedback needs to be more conservative (higher) in
6118 order to make tracer effective.
6120 @item max-cse-path-length
6122 Maximum number of basic blocks on path that cse considers. The default is 10.
6125 The maximum instructions CSE process before flushing. The default is 1000.
6127 @item global-var-threshold
6129 Counts the number of function calls (@var{n}) and the number of
6130 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
6131 single artificial variable will be created to represent all the
6132 call-clobbered variables at function call sites. This artificial
6133 variable will then be made to alias every call-clobbered variable.
6134 (done as @code{int * size_t} on the host machine; beware overflow).
6136 @item max-aliased-vops
6138 Maximum number of virtual operands allowed to represent aliases
6139 before triggering the alias grouping heuristic. Alias grouping
6140 reduces compile times and memory consumption needed for aliasing at
6141 the expense of precision loss in alias information.
6143 @item ggc-min-expand
6145 GCC uses a garbage collector to manage its own memory allocation. This
6146 parameter specifies the minimum percentage by which the garbage
6147 collector's heap should be allowed to expand between collections.
6148 Tuning this may improve compilation speed; it has no effect on code
6151 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6152 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6153 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6154 GCC is not able to calculate RAM on a particular platform, the lower
6155 bound of 30% is used. Setting this parameter and
6156 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6157 every opportunity. This is extremely slow, but can be useful for
6160 @item ggc-min-heapsize
6162 Minimum size of the garbage collector's heap before it begins bothering
6163 to collect garbage. The first collection occurs after the heap expands
6164 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6165 tuning this may improve compilation speed, and has no effect on code
6168 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6169 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6170 with a lower bound of 4096 (four megabytes) and an upper bound of
6171 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6172 particular platform, the lower bound is used. Setting this parameter
6173 very large effectively disables garbage collection. Setting this
6174 parameter and @option{ggc-min-expand} to zero causes a full collection
6175 to occur at every opportunity.
6177 @item max-reload-search-insns
6178 The maximum number of instruction reload should look backward for equivalent
6179 register. Increasing values mean more aggressive optimization, making the
6180 compile time increase with probably slightly better performance. The default
6183 @item max-cselib-memory-location
6184 The maximum number of memory locations cselib should take into acount.
6185 Increasing values mean more aggressive optimization, making the compile time
6186 increase with probably slightly better performance. The default value is 500.
6188 @item max-flow-memory-location
6189 Similar as @option{max-cselib-memory-location} but for dataflow liveness.
6190 The default value is 100.
6192 @item reorder-blocks-duplicate
6193 @itemx reorder-blocks-duplicate-feedback
6195 Used by basic block reordering pass to decide whether to use unconditional
6196 branch or duplicate the code on its destination. Code is duplicated when its
6197 estimated size is smaller than this value multiplied by the estimated size of
6198 unconditional jump in the hot spots of the program.
6200 The @option{reorder-block-duplicate-feedback} is used only when profile
6201 feedback is available and may be set to higher values than
6202 @option{reorder-block-duplicate} since information about the hot spots is more
6205 @item max-sched-ready-insns
6206 The maximum number of instructions ready to be issued the scheduler should
6207 consider at any given time during the first scheduling pass. Increasing
6208 values mean more thorough searches, making the compilation time increase
6209 with probably little benefit. The default value is 100.
6211 @item max-sched-region-blocks
6212 The maximum number of blocks in a region to be considered for
6213 interblock scheduling. The default value is 10.
6215 @item max-sched-region-insns
6216 The maximum number of insns in a region to be considered for
6217 interblock scheduling. The default value is 100.
6220 The minimum probability (in percents) of reaching a source block
6221 for interblock speculative scheduling. The default value is 40.
6223 @item max-sched-extend-regions-iters
6224 The maximum number of iterations through CFG to extend regions.
6225 0 - disable region extension,
6226 N - do at most N iterations.
6227 The default value is 2.
6229 @item max-sched-insn-conflict-delay
6230 The maximum conflict delay for an insn to be considered for speculative motion.
6231 The default value is 3.
6233 @item sched-spec-prob-cutoff
6234 The minimal probability of speculation success (in percents), so that
6235 speculative insn will be scheduled.
6236 The default value is 40.
6238 @item max-last-value-rtl
6240 The maximum size measured as number of RTLs that can be recorded in an expression
6241 in combiner for a pseudo register as last known value of that register. The default
6244 @item integer-share-limit
6245 Small integer constants can use a shared data structure, reducing the
6246 compiler's memory usage and increasing its speed. This sets the maximum
6247 value of a shared integer constant's. The default value is 256.
6249 @item min-virtual-mappings
6250 Specifies the minimum number of virtual mappings in the incremental
6251 SSA updater that should be registered to trigger the virtual mappings
6252 heuristic defined by virtual-mappings-ratio. The default value is
6255 @item virtual-mappings-ratio
6256 If the number of virtual mappings is virtual-mappings-ratio bigger
6257 than the number of virtual symbols to be updated, then the incremental
6258 SSA updater switches to a full update for those symbols. The default
6261 @item ssp-buffer-size
6262 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6263 protection when @option{-fstack-protection} is used.
6265 @item max-jump-thread-duplication-stmts
6266 Maximum number of statements allowed in a block that needs to be
6267 duplicated when threading jumps.
6269 @item max-fields-for-field-sensitive
6270 Maximum number of fields in a structure we will treat in
6271 a field sensitive manner during pointer analysis.
6276 @node Preprocessor Options
6277 @section Options Controlling the Preprocessor
6278 @cindex preprocessor options
6279 @cindex options, preprocessor
6281 These options control the C preprocessor, which is run on each C source
6282 file before actual compilation.
6284 If you use the @option{-E} option, nothing is done except preprocessing.
6285 Some of these options make sense only together with @option{-E} because
6286 they cause the preprocessor output to be unsuitable for actual
6291 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6292 and pass @var{option} directly through to the preprocessor. If
6293 @var{option} contains commas, it is split into multiple options at the
6294 commas. However, many options are modified, translated or interpreted
6295 by the compiler driver before being passed to the preprocessor, and
6296 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6297 interface is undocumented and subject to change, so whenever possible
6298 you should avoid using @option{-Wp} and let the driver handle the
6301 @item -Xpreprocessor @var{option}
6302 @opindex preprocessor
6303 Pass @var{option} as an option to the preprocessor. You can use this to
6304 supply system-specific preprocessor options which GCC does not know how to
6307 If you want to pass an option that takes an argument, you must use
6308 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6311 @include cppopts.texi
6313 @node Assembler Options
6314 @section Passing Options to the Assembler
6316 @c prevent bad page break with this line
6317 You can pass options to the assembler.
6320 @item -Wa,@var{option}
6322 Pass @var{option} as an option to the assembler. If @var{option}
6323 contains commas, it is split into multiple options at the commas.
6325 @item -Xassembler @var{option}
6327 Pass @var{option} as an option to the assembler. You can use this to
6328 supply system-specific assembler options which GCC does not know how to
6331 If you want to pass an option that takes an argument, you must use
6332 @option{-Xassembler} twice, once for the option and once for the argument.
6337 @section Options for Linking
6338 @cindex link options
6339 @cindex options, linking
6341 These options come into play when the compiler links object files into
6342 an executable output file. They are meaningless if the compiler is
6343 not doing a link step.
6347 @item @var{object-file-name}
6348 A file name that does not end in a special recognized suffix is
6349 considered to name an object file or library. (Object files are
6350 distinguished from libraries by the linker according to the file
6351 contents.) If linking is done, these object files are used as input
6360 If any of these options is used, then the linker is not run, and
6361 object file names should not be used as arguments. @xref{Overall
6365 @item -l@var{library}
6366 @itemx -l @var{library}
6368 Search the library named @var{library} when linking. (The second
6369 alternative with the library as a separate argument is only for
6370 POSIX compliance and is not recommended.)
6372 It makes a difference where in the command you write this option; the
6373 linker searches and processes libraries and object files in the order they
6374 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6375 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6376 to functions in @samp{z}, those functions may not be loaded.
6378 The linker searches a standard list of directories for the library,
6379 which is actually a file named @file{lib@var{library}.a}. The linker
6380 then uses this file as if it had been specified precisely by name.
6382 The directories searched include several standard system directories
6383 plus any that you specify with @option{-L}.
6385 Normally the files found this way are library files---archive files
6386 whose members are object files. The linker handles an archive file by
6387 scanning through it for members which define symbols that have so far
6388 been referenced but not defined. But if the file that is found is an
6389 ordinary object file, it is linked in the usual fashion. The only
6390 difference between using an @option{-l} option and specifying a file name
6391 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6392 and searches several directories.
6396 You need this special case of the @option{-l} option in order to
6397 link an Objective-C or Objective-C++ program.
6400 @opindex nostartfiles
6401 Do not use the standard system startup files when linking.
6402 The standard system libraries are used normally, unless @option{-nostdlib}
6403 or @option{-nodefaultlibs} is used.
6405 @item -nodefaultlibs
6406 @opindex nodefaultlibs
6407 Do not use the standard system libraries when linking.
6408 Only the libraries you specify will be passed to the linker.
6409 The standard startup files are used normally, unless @option{-nostartfiles}
6410 is used. The compiler may generate calls to @code{memcmp},
6411 @code{memset}, @code{memcpy} and @code{memmove}.
6412 These entries are usually resolved by entries in
6413 libc. These entry points should be supplied through some other
6414 mechanism when this option is specified.
6418 Do not use the standard system startup files or libraries when linking.
6419 No startup files and only the libraries you specify will be passed to
6420 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6421 @code{memcpy} and @code{memmove}.
6422 These entries are usually resolved by entries in
6423 libc. These entry points should be supplied through some other
6424 mechanism when this option is specified.
6426 @cindex @option{-lgcc}, use with @option{-nostdlib}
6427 @cindex @option{-nostdlib} and unresolved references
6428 @cindex unresolved references and @option{-nostdlib}
6429 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6430 @cindex @option{-nodefaultlibs} and unresolved references
6431 @cindex unresolved references and @option{-nodefaultlibs}
6432 One of the standard libraries bypassed by @option{-nostdlib} and
6433 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6434 that GCC uses to overcome shortcomings of particular machines, or special
6435 needs for some languages.
6436 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6437 Collection (GCC) Internals},
6438 for more discussion of @file{libgcc.a}.)
6439 In most cases, you need @file{libgcc.a} even when you want to avoid
6440 other standard libraries. In other words, when you specify @option{-nostdlib}
6441 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6442 This ensures that you have no unresolved references to internal GCC
6443 library subroutines. (For example, @samp{__main}, used to ensure C++
6444 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6445 GNU Compiler Collection (GCC) Internals}.)
6449 Produce a position independent executable on targets which support it.
6450 For predictable results, you must also specify the same set of options
6451 that were used to generate code (@option{-fpie}, @option{-fPIE},
6452 or model suboptions) when you specify this option.
6456 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6457 that support it. This instructs the linker to add all symbols, not
6458 only used ones, to the dynamic symbol table. This option is needed
6459 for some uses of @code{dlopen} or to allow obtaining backtraces
6460 from within a program.
6464 Remove all symbol table and relocation information from the executable.
6468 On systems that support dynamic linking, this prevents linking with the shared
6469 libraries. On other systems, this option has no effect.
6473 Produce a shared object which can then be linked with other objects to
6474 form an executable. Not all systems support this option. For predictable
6475 results, you must also specify the same set of options that were used to
6476 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6477 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6478 needs to build supplementary stub code for constructors to work. On
6479 multi-libbed systems, @samp{gcc -shared} must select the correct support
6480 libraries to link against. Failing to supply the correct flags may lead
6481 to subtle defects. Supplying them in cases where they are not necessary
6484 @item -shared-libgcc
6485 @itemx -static-libgcc
6486 @opindex shared-libgcc
6487 @opindex static-libgcc
6488 On systems that provide @file{libgcc} as a shared library, these options
6489 force the use of either the shared or static version respectively.
6490 If no shared version of @file{libgcc} was built when the compiler was
6491 configured, these options have no effect.
6493 There are several situations in which an application should use the
6494 shared @file{libgcc} instead of the static version. The most common
6495 of these is when the application wishes to throw and catch exceptions
6496 across different shared libraries. In that case, each of the libraries
6497 as well as the application itself should use the shared @file{libgcc}.
6499 Therefore, the G++ and GCJ drivers automatically add
6500 @option{-shared-libgcc} whenever you build a shared library or a main
6501 executable, because C++ and Java programs typically use exceptions, so
6502 this is the right thing to do.
6504 If, instead, you use the GCC driver to create shared libraries, you may
6505 find that they will not always be linked with the shared @file{libgcc}.
6506 If GCC finds, at its configuration time, that you have a non-GNU linker
6507 or a GNU linker that does not support option @option{--eh-frame-hdr},
6508 it will link the shared version of @file{libgcc} into shared libraries
6509 by default. Otherwise, it will take advantage of the linker and optimize
6510 away the linking with the shared version of @file{libgcc}, linking with
6511 the static version of libgcc by default. This allows exceptions to
6512 propagate through such shared libraries, without incurring relocation
6513 costs at library load time.
6515 However, if a library or main executable is supposed to throw or catch
6516 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6517 for the languages used in the program, or using the option
6518 @option{-shared-libgcc}, such that it is linked with the shared
6523 Bind references to global symbols when building a shared object. Warn
6524 about any unresolved references (unless overridden by the link editor
6525 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6528 @item -Xlinker @var{option}
6530 Pass @var{option} as an option to the linker. You can use this to
6531 supply system-specific linker options which GCC does not know how to
6534 If you want to pass an option that takes an argument, you must use
6535 @option{-Xlinker} twice, once for the option and once for the argument.
6536 For example, to pass @option{-assert definitions}, you must write
6537 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6538 @option{-Xlinker "-assert definitions"}, because this passes the entire
6539 string as a single argument, which is not what the linker expects.
6541 @item -Wl,@var{option}
6543 Pass @var{option} as an option to the linker. If @var{option} contains
6544 commas, it is split into multiple options at the commas.
6546 @item -u @var{symbol}
6548 Pretend the symbol @var{symbol} is undefined, to force linking of
6549 library modules to define it. You can use @option{-u} multiple times with
6550 different symbols to force loading of additional library modules.
6553 @node Directory Options
6554 @section Options for Directory Search
6555 @cindex directory options
6556 @cindex options, directory search
6559 These options specify directories to search for header files, for
6560 libraries and for parts of the compiler:
6565 Add the directory @var{dir} to the head of the list of directories to be
6566 searched for header files. This can be used to override a system header
6567 file, substituting your own version, since these directories are
6568 searched before the system header file directories. However, you should
6569 not use this option to add directories that contain vendor-supplied
6570 system header files (use @option{-isystem} for that). If you use more than
6571 one @option{-I} option, the directories are scanned in left-to-right
6572 order; the standard system directories come after.
6574 If a standard system include directory, or a directory specified with
6575 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6576 option will be ignored. The directory will still be searched but as a
6577 system directory at its normal position in the system include chain.
6578 This is to ensure that GCC's procedure to fix buggy system headers and
6579 the ordering for the include_next directive are not inadvertently changed.
6580 If you really need to change the search order for system directories,
6581 use the @option{-nostdinc} and/or @option{-isystem} options.
6583 @item -iquote@var{dir}
6585 Add the directory @var{dir} to the head of the list of directories to
6586 be searched for header files only for the case of @samp{#include
6587 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6588 otherwise just like @option{-I}.
6592 Add directory @var{dir} to the list of directories to be searched
6595 @item -B@var{prefix}
6597 This option specifies where to find the executables, libraries,
6598 include files, and data files of the compiler itself.
6600 The compiler driver program runs one or more of the subprograms
6601 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6602 @var{prefix} as a prefix for each program it tries to run, both with and
6603 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6605 For each subprogram to be run, the compiler driver first tries the
6606 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6607 was not specified, the driver tries two standard prefixes, which are
6608 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6609 those results in a file name that is found, the unmodified program
6610 name is searched for using the directories specified in your
6611 @env{PATH} environment variable.
6613 The compiler will check to see if the path provided by the @option{-B}
6614 refers to a directory, and if necessary it will add a directory
6615 separator character at the end of the path.
6617 @option{-B} prefixes that effectively specify directory names also apply
6618 to libraries in the linker, because the compiler translates these
6619 options into @option{-L} options for the linker. They also apply to
6620 includes files in the preprocessor, because the compiler translates these
6621 options into @option{-isystem} options for the preprocessor. In this case,
6622 the compiler appends @samp{include} to the prefix.
6624 The run-time support file @file{libgcc.a} can also be searched for using
6625 the @option{-B} prefix, if needed. If it is not found there, the two
6626 standard prefixes above are tried, and that is all. The file is left
6627 out of the link if it is not found by those means.
6629 Another way to specify a prefix much like the @option{-B} prefix is to use
6630 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6633 As a special kludge, if the path provided by @option{-B} is
6634 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6635 9, then it will be replaced by @file{[dir/]include}. This is to help
6636 with boot-strapping the compiler.
6638 @item -specs=@var{file}
6640 Process @var{file} after the compiler reads in the standard @file{specs}
6641 file, in order to override the defaults that the @file{gcc} driver
6642 program uses when determining what switches to pass to @file{cc1},
6643 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6644 @option{-specs=@var{file}} can be specified on the command line, and they
6645 are processed in order, from left to right.
6647 @item --sysroot=@var{dir}
6649 Use @var{dir} as the logical root directory for headers and libraries.
6650 For example, if the compiler would normally search for headers in
6651 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6652 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6654 If you use both this option and the @option{-isysroot} option, then
6655 the @option{--sysroot} option will apply to libraries, but the
6656 @option{-isysroot} option will apply to header files.
6658 The GNU linker (beginning with version 2.16) has the necessary support
6659 for this option. If your linker does not support this option, the
6660 header file aspect of @option{--sysroot} will still work, but the
6661 library aspect will not.
6665 This option has been deprecated. Please use @option{-iquote} instead for
6666 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6667 Any directories you specify with @option{-I} options before the @option{-I-}
6668 option are searched only for the case of @samp{#include "@var{file}"};
6669 they are not searched for @samp{#include <@var{file}>}.
6671 If additional directories are specified with @option{-I} options after
6672 the @option{-I-}, these directories are searched for all @samp{#include}
6673 directives. (Ordinarily @emph{all} @option{-I} directories are used
6676 In addition, the @option{-I-} option inhibits the use of the current
6677 directory (where the current input file came from) as the first search
6678 directory for @samp{#include "@var{file}"}. There is no way to
6679 override this effect of @option{-I-}. With @option{-I.} you can specify
6680 searching the directory which was current when the compiler was
6681 invoked. That is not exactly the same as what the preprocessor does
6682 by default, but it is often satisfactory.
6684 @option{-I-} does not inhibit the use of the standard system directories
6685 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6692 @section Specifying subprocesses and the switches to pass to them
6695 @command{gcc} is a driver program. It performs its job by invoking a
6696 sequence of other programs to do the work of compiling, assembling and
6697 linking. GCC interprets its command-line parameters and uses these to
6698 deduce which programs it should invoke, and which command-line options
6699 it ought to place on their command lines. This behavior is controlled
6700 by @dfn{spec strings}. In most cases there is one spec string for each
6701 program that GCC can invoke, but a few programs have multiple spec
6702 strings to control their behavior. The spec strings built into GCC can
6703 be overridden by using the @option{-specs=} command-line switch to specify
6706 @dfn{Spec files} are plaintext files that are used to construct spec
6707 strings. They consist of a sequence of directives separated by blank
6708 lines. The type of directive is determined by the first non-whitespace
6709 character on the line and it can be one of the following:
6712 @item %@var{command}
6713 Issues a @var{command} to the spec file processor. The commands that can
6717 @item %include <@var{file}>
6719 Search for @var{file} and insert its text at the current point in the
6722 @item %include_noerr <@var{file}>
6723 @cindex %include_noerr
6724 Just like @samp{%include}, but do not generate an error message if the include
6725 file cannot be found.
6727 @item %rename @var{old_name} @var{new_name}
6729 Rename the spec string @var{old_name} to @var{new_name}.
6733 @item *[@var{spec_name}]:
6734 This tells the compiler to create, override or delete the named spec
6735 string. All lines after this directive up to the next directive or
6736 blank line are considered to be the text for the spec string. If this
6737 results in an empty string then the spec will be deleted. (Or, if the
6738 spec did not exist, then nothing will happened.) Otherwise, if the spec
6739 does not currently exist a new spec will be created. If the spec does
6740 exist then its contents will be overridden by the text of this
6741 directive, unless the first character of that text is the @samp{+}
6742 character, in which case the text will be appended to the spec.
6744 @item [@var{suffix}]:
6745 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6746 and up to the next directive or blank line are considered to make up the
6747 spec string for the indicated suffix. When the compiler encounters an
6748 input file with the named suffix, it will processes the spec string in
6749 order to work out how to compile that file. For example:
6756 This says that any input file whose name ends in @samp{.ZZ} should be
6757 passed to the program @samp{z-compile}, which should be invoked with the
6758 command-line switch @option{-input} and with the result of performing the
6759 @samp{%i} substitution. (See below.)
6761 As an alternative to providing a spec string, the text that follows a
6762 suffix directive can be one of the following:
6765 @item @@@var{language}
6766 This says that the suffix is an alias for a known @var{language}. This is
6767 similar to using the @option{-x} command-line switch to GCC to specify a
6768 language explicitly. For example:
6775 Says that .ZZ files are, in fact, C++ source files.
6778 This causes an error messages saying:
6781 @var{name} compiler not installed on this system.
6785 GCC already has an extensive list of suffixes built into it.
6786 This directive will add an entry to the end of the list of suffixes, but
6787 since the list is searched from the end backwards, it is effectively
6788 possible to override earlier entries using this technique.
6792 GCC has the following spec strings built into it. Spec files can
6793 override these strings or create their own. Note that individual
6794 targets can also add their own spec strings to this list.
6797 asm Options to pass to the assembler
6798 asm_final Options to pass to the assembler post-processor
6799 cpp Options to pass to the C preprocessor
6800 cc1 Options to pass to the C compiler
6801 cc1plus Options to pass to the C++ compiler
6802 endfile Object files to include at the end of the link
6803 link Options to pass to the linker
6804 lib Libraries to include on the command line to the linker
6805 libgcc Decides which GCC support library to pass to the linker
6806 linker Sets the name of the linker
6807 predefines Defines to be passed to the C preprocessor
6808 signed_char Defines to pass to CPP to say whether @code{char} is signed
6810 startfile Object files to include at the start of the link
6813 Here is a small example of a spec file:
6819 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6822 This example renames the spec called @samp{lib} to @samp{old_lib} and
6823 then overrides the previous definition of @samp{lib} with a new one.
6824 The new definition adds in some extra command-line options before
6825 including the text of the old definition.
6827 @dfn{Spec strings} are a list of command-line options to be passed to their
6828 corresponding program. In addition, the spec strings can contain
6829 @samp{%}-prefixed sequences to substitute variable text or to
6830 conditionally insert text into the command line. Using these constructs
6831 it is possible to generate quite complex command lines.
6833 Here is a table of all defined @samp{%}-sequences for spec
6834 strings. Note that spaces are not generated automatically around the
6835 results of expanding these sequences. Therefore you can concatenate them
6836 together or combine them with constant text in a single argument.
6840 Substitute one @samp{%} into the program name or argument.
6843 Substitute the name of the input file being processed.
6846 Substitute the basename of the input file being processed.
6847 This is the substring up to (and not including) the last period
6848 and not including the directory.
6851 This is the same as @samp{%b}, but include the file suffix (text after
6855 Marks the argument containing or following the @samp{%d} as a
6856 temporary file name, so that that file will be deleted if GCC exits
6857 successfully. Unlike @samp{%g}, this contributes no text to the
6860 @item %g@var{suffix}
6861 Substitute a file name that has suffix @var{suffix} and is chosen
6862 once per compilation, and mark the argument in the same way as
6863 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6864 name is now chosen in a way that is hard to predict even when previously
6865 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6866 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6867 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6868 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6869 was simply substituted with a file name chosen once per compilation,
6870 without regard to any appended suffix (which was therefore treated
6871 just like ordinary text), making such attacks more likely to succeed.
6873 @item %u@var{suffix}
6874 Like @samp{%g}, but generates a new temporary file name even if
6875 @samp{%u@var{suffix}} was already seen.
6877 @item %U@var{suffix}
6878 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6879 new one if there is no such last file name. In the absence of any
6880 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6881 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6882 would involve the generation of two distinct file names, one
6883 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6884 simply substituted with a file name chosen for the previous @samp{%u},
6885 without regard to any appended suffix.
6887 @item %j@var{suffix}
6888 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6889 writable, and if save-temps is off; otherwise, substitute the name
6890 of a temporary file, just like @samp{%u}. This temporary file is not
6891 meant for communication between processes, but rather as a junk
6894 @item %|@var{suffix}
6895 @itemx %m@var{suffix}
6896 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6897 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6898 all. These are the two most common ways to instruct a program that it
6899 should read from standard input or write to standard output. If you
6900 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6901 construct: see for example @file{f/lang-specs.h}.
6903 @item %.@var{SUFFIX}
6904 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6905 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6906 terminated by the next space or %.
6909 Marks the argument containing or following the @samp{%w} as the
6910 designated output file of this compilation. This puts the argument
6911 into the sequence of arguments that @samp{%o} will substitute later.
6914 Substitutes the names of all the output files, with spaces
6915 automatically placed around them. You should write spaces
6916 around the @samp{%o} as well or the results are undefined.
6917 @samp{%o} is for use in the specs for running the linker.
6918 Input files whose names have no recognized suffix are not compiled
6919 at all, but they are included among the output files, so they will
6923 Substitutes the suffix for object files. Note that this is
6924 handled specially when it immediately follows @samp{%g, %u, or %U},
6925 because of the need for those to form complete file names. The
6926 handling is such that @samp{%O} is treated exactly as if it had already
6927 been substituted, except that @samp{%g, %u, and %U} do not currently
6928 support additional @var{suffix} characters following @samp{%O} as they would
6929 following, for example, @samp{.o}.
6932 Substitutes the standard macro predefinitions for the
6933 current target machine. Use this when running @code{cpp}.
6936 Like @samp{%p}, but puts @samp{__} before and after the name of each
6937 predefined macro, except for macros that start with @samp{__} or with
6938 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6942 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6943 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
6944 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6945 and @option{-imultilib} as necessary.
6948 Current argument is the name of a library or startup file of some sort.
6949 Search for that file in a standard list of directories and substitute
6950 the full name found.
6953 Print @var{str} as an error message. @var{str} is terminated by a newline.
6954 Use this when inconsistent options are detected.
6957 Substitute the contents of spec string @var{name} at this point.
6960 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6962 @item %x@{@var{option}@}
6963 Accumulate an option for @samp{%X}.
6966 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6970 Output the accumulated assembler options specified by @option{-Wa}.
6973 Output the accumulated preprocessor options specified by @option{-Wp}.
6976 Process the @code{asm} spec. This is used to compute the
6977 switches to be passed to the assembler.
6980 Process the @code{asm_final} spec. This is a spec string for
6981 passing switches to an assembler post-processor, if such a program is
6985 Process the @code{link} spec. This is the spec for computing the
6986 command line passed to the linker. Typically it will make use of the
6987 @samp{%L %G %S %D and %E} sequences.
6990 Dump out a @option{-L} option for each directory that GCC believes might
6991 contain startup files. If the target supports multilibs then the
6992 current multilib directory will be prepended to each of these paths.
6995 Process the @code{lib} spec. This is a spec string for deciding which
6996 libraries should be included on the command line to the linker.
6999 Process the @code{libgcc} spec. This is a spec string for deciding
7000 which GCC support library should be included on the command line to the linker.
7003 Process the @code{startfile} spec. This is a spec for deciding which
7004 object files should be the first ones passed to the linker. Typically
7005 this might be a file named @file{crt0.o}.
7008 Process the @code{endfile} spec. This is a spec string that specifies
7009 the last object files that will be passed to the linker.
7012 Process the @code{cpp} spec. This is used to construct the arguments
7013 to be passed to the C preprocessor.
7016 Process the @code{cc1} spec. This is used to construct the options to be
7017 passed to the actual C compiler (@samp{cc1}).
7020 Process the @code{cc1plus} spec. This is used to construct the options to be
7021 passed to the actual C++ compiler (@samp{cc1plus}).
7024 Substitute the variable part of a matched option. See below.
7025 Note that each comma in the substituted string is replaced by
7029 Remove all occurrences of @code{-S} from the command line. Note---this
7030 command is position dependent. @samp{%} commands in the spec string
7031 before this one will see @code{-S}, @samp{%} commands in the spec string
7032 after this one will not.
7034 @item %:@var{function}(@var{args})
7035 Call the named function @var{function}, passing it @var{args}.
7036 @var{args} is first processed as a nested spec string, then split
7037 into an argument vector in the usual fashion. The function returns
7038 a string which is processed as if it had appeared literally as part
7039 of the current spec.
7041 The following built-in spec functions are provided:
7044 @item @code{if-exists}
7045 The @code{if-exists} spec function takes one argument, an absolute
7046 pathname to a file. If the file exists, @code{if-exists} returns the
7047 pathname. Here is a small example of its usage:
7051 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7054 @item @code{if-exists-else}
7055 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7056 spec function, except that it takes two arguments. The first argument is
7057 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7058 returns the pathname. If it does not exist, it returns the second argument.
7059 This way, @code{if-exists-else} can be used to select one file or another,
7060 based on the existence of the first. Here is a small example of its usage:
7064 crt0%O%s %:if-exists(crti%O%s) \
7065 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7068 @item @code{replace-outfile}
7069 The @code{replace-outfile} spec function takes two arguments. It looks for the
7070 first argument in the outfiles array and replaces it with the second argument. Here
7071 is a small example of its usage:
7074 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7080 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7081 If that switch was not specified, this substitutes nothing. Note that
7082 the leading dash is omitted when specifying this option, and it is
7083 automatically inserted if the substitution is performed. Thus the spec
7084 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7085 and would output the command line option @option{-foo}.
7087 @item %W@{@code{S}@}
7088 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7091 @item %@{@code{S}*@}
7092 Substitutes all the switches specified to GCC whose names start
7093 with @code{-S}, but which also take an argument. This is used for
7094 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7095 GCC considers @option{-o foo} as being
7096 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7097 text, including the space. Thus two arguments would be generated.
7099 @item %@{@code{S}*&@code{T}*@}
7100 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7101 (the order of @code{S} and @code{T} in the spec is not significant).
7102 There can be any number of ampersand-separated variables; for each the
7103 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7105 @item %@{@code{S}:@code{X}@}
7106 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7108 @item %@{!@code{S}:@code{X}@}
7109 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7111 @item %@{@code{S}*:@code{X}@}
7112 Substitutes @code{X} if one or more switches whose names start with
7113 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7114 once, no matter how many such switches appeared. However, if @code{%*}
7115 appears somewhere in @code{X}, then @code{X} will be substituted once
7116 for each matching switch, with the @code{%*} replaced by the part of
7117 that switch that matched the @code{*}.
7119 @item %@{.@code{S}:@code{X}@}
7120 Substitutes @code{X}, if processing a file with suffix @code{S}.
7122 @item %@{!.@code{S}:@code{X}@}
7123 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7125 @item %@{@code{S}|@code{P}:@code{X}@}
7126 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7127 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7128 although they have a stronger binding than the @samp{|}. If @code{%*}
7129 appears in @code{X}, all of the alternatives must be starred, and only
7130 the first matching alternative is substituted.
7132 For example, a spec string like this:
7135 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7138 will output the following command-line options from the following input
7139 command-line options:
7144 -d fred.c -foo -baz -boggle
7145 -d jim.d -bar -baz -boggle
7148 @item %@{S:X; T:Y; :D@}
7150 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7151 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7152 be as many clauses as you need. This may be combined with @code{.},
7153 @code{!}, @code{|}, and @code{*} as needed.
7158 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7159 construct may contain other nested @samp{%} constructs or spaces, or
7160 even newlines. They are processed as usual, as described above.
7161 Trailing white space in @code{X} is ignored. White space may also
7162 appear anywhere on the left side of the colon in these constructs,
7163 except between @code{.} or @code{*} and the corresponding word.
7165 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7166 handled specifically in these constructs. If another value of
7167 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7168 @option{-W} switch is found later in the command line, the earlier
7169 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7170 just one letter, which passes all matching options.
7172 The character @samp{|} at the beginning of the predicate text is used to
7173 indicate that a command should be piped to the following command, but
7174 only if @option{-pipe} is specified.
7176 It is built into GCC which switches take arguments and which do not.
7177 (You might think it would be useful to generalize this to allow each
7178 compiler's spec to say which switches take arguments. But this cannot
7179 be done in a consistent fashion. GCC cannot even decide which input
7180 files have been specified without knowing which switches take arguments,
7181 and it must know which input files to compile in order to tell which
7184 GCC also knows implicitly that arguments starting in @option{-l} are to be
7185 treated as compiler output files, and passed to the linker in their
7186 proper position among the other output files.
7188 @c man begin OPTIONS
7190 @node Target Options
7191 @section Specifying Target Machine and Compiler Version
7192 @cindex target options
7193 @cindex cross compiling
7194 @cindex specifying machine version
7195 @cindex specifying compiler version and target machine
7196 @cindex compiler version, specifying
7197 @cindex target machine, specifying
7199 The usual way to run GCC is to run the executable called @file{gcc}, or
7200 @file{<machine>-gcc} when cross-compiling, or
7201 @file{<machine>-gcc-<version>} to run a version other than the one that
7202 was installed last. Sometimes this is inconvenient, so GCC provides
7203 options that will switch to another cross-compiler or version.
7206 @item -b @var{machine}
7208 The argument @var{machine} specifies the target machine for compilation.
7210 The value to use for @var{machine} is the same as was specified as the
7211 machine type when configuring GCC as a cross-compiler. For
7212 example, if a cross-compiler was configured with @samp{configure
7213 arm-elf}, meaning to compile for an arm processor with elf binaries,
7214 then you would specify @option{-b arm-elf} to run that cross compiler.
7215 Because there are other options beginning with @option{-b}, the
7216 configuration must contain a hyphen.
7218 @item -V @var{version}
7220 The argument @var{version} specifies which version of GCC to run.
7221 This is useful when multiple versions are installed. For example,
7222 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7225 The @option{-V} and @option{-b} options work by running the
7226 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7227 use them if you can just run that directly.
7229 @node Submodel Options
7230 @section Hardware Models and Configurations
7231 @cindex submodel options
7232 @cindex specifying hardware config
7233 @cindex hardware models and configurations, specifying
7234 @cindex machine dependent options
7236 Earlier we discussed the standard option @option{-b} which chooses among
7237 different installed compilers for completely different target
7238 machines, such as VAX vs.@: 68000 vs.@: 80386.
7240 In addition, each of these target machine types can have its own
7241 special options, starting with @samp{-m}, to choose among various
7242 hardware models or configurations---for example, 68010 vs 68020,
7243 floating coprocessor or none. A single installed version of the
7244 compiler can compile for any model or configuration, according to the
7247 Some configurations of the compiler also support additional special
7248 options, usually for compatibility with other compilers on the same
7251 @c This list is ordered alphanumerically by subsection name.
7252 @c It should be the same order and spelling as these options are listed
7253 @c in Machine Dependent Options
7259 * Blackfin Options::
7263 * DEC Alpha Options::
7264 * DEC Alpha/VMS Options::
7266 * GNU/Linux Options::
7269 * i386 and x86-64 Options::
7282 * RS/6000 and PowerPC Options::
7283 * S/390 and zSeries Options::
7286 * System V Options::
7287 * TMS320C3x/C4x Options::
7291 * Xstormy16 Options::
7297 @subsection ARC Options
7300 These options are defined for ARC implementations:
7305 Compile code for little endian mode. This is the default.
7309 Compile code for big endian mode.
7312 @opindex mmangle-cpu
7313 Prepend the name of the cpu to all public symbol names.
7314 In multiple-processor systems, there are many ARC variants with different
7315 instruction and register set characteristics. This flag prevents code
7316 compiled for one cpu to be linked with code compiled for another.
7317 No facility exists for handling variants that are ``almost identical''.
7318 This is an all or nothing option.
7320 @item -mcpu=@var{cpu}
7322 Compile code for ARC variant @var{cpu}.
7323 Which variants are supported depend on the configuration.
7324 All variants support @option{-mcpu=base}, this is the default.
7326 @item -mtext=@var{text-section}
7327 @itemx -mdata=@var{data-section}
7328 @itemx -mrodata=@var{readonly-data-section}
7332 Put functions, data, and readonly data in @var{text-section},
7333 @var{data-section}, and @var{readonly-data-section} respectively
7334 by default. This can be overridden with the @code{section} attribute.
7335 @xref{Variable Attributes}.
7340 @subsection ARM Options
7343 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7347 @item -mabi=@var{name}
7349 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7350 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7353 @opindex mapcs-frame
7354 Generate a stack frame that is compliant with the ARM Procedure Call
7355 Standard for all functions, even if this is not strictly necessary for
7356 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7357 with this option will cause the stack frames not to be generated for
7358 leaf functions. The default is @option{-mno-apcs-frame}.
7362 This is a synonym for @option{-mapcs-frame}.
7365 @c not currently implemented
7366 @item -mapcs-stack-check
7367 @opindex mapcs-stack-check
7368 Generate code to check the amount of stack space available upon entry to
7369 every function (that actually uses some stack space). If there is
7370 insufficient space available then either the function
7371 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7372 called, depending upon the amount of stack space required. The run time
7373 system is required to provide these functions. The default is
7374 @option{-mno-apcs-stack-check}, since this produces smaller code.
7376 @c not currently implemented
7378 @opindex mapcs-float
7379 Pass floating point arguments using the float point registers. This is
7380 one of the variants of the APCS@. This option is recommended if the
7381 target hardware has a floating point unit or if a lot of floating point
7382 arithmetic is going to be performed by the code. The default is
7383 @option{-mno-apcs-float}, since integer only code is slightly increased in
7384 size if @option{-mapcs-float} is used.
7386 @c not currently implemented
7387 @item -mapcs-reentrant
7388 @opindex mapcs-reentrant
7389 Generate reentrant, position independent code. The default is
7390 @option{-mno-apcs-reentrant}.
7393 @item -mthumb-interwork
7394 @opindex mthumb-interwork
7395 Generate code which supports calling between the ARM and Thumb
7396 instruction sets. Without this option the two instruction sets cannot
7397 be reliably used inside one program. The default is
7398 @option{-mno-thumb-interwork}, since slightly larger code is generated
7399 when @option{-mthumb-interwork} is specified.
7401 @item -mno-sched-prolog
7402 @opindex mno-sched-prolog
7403 Prevent the reordering of instructions in the function prolog, or the
7404 merging of those instruction with the instructions in the function's
7405 body. This means that all functions will start with a recognizable set
7406 of instructions (or in fact one of a choice from a small set of
7407 different function prologues), and this information can be used to
7408 locate the start if functions inside an executable piece of code. The
7409 default is @option{-msched-prolog}.
7412 @opindex mhard-float
7413 Generate output containing floating point instructions. This is the
7417 @opindex msoft-float
7418 Generate output containing library calls for floating point.
7419 @strong{Warning:} the requisite libraries are not available for all ARM
7420 targets. Normally the facilities of the machine's usual C compiler are
7421 used, but this cannot be done directly in cross-compilation. You must make
7422 your own arrangements to provide suitable library functions for
7425 @option{-msoft-float} changes the calling convention in the output file;
7426 therefore, it is only useful if you compile @emph{all} of a program with
7427 this option. In particular, you need to compile @file{libgcc.a}, the
7428 library that comes with GCC, with @option{-msoft-float} in order for
7431 @item -mfloat-abi=@var{name}
7433 Specifies which ABI to use for floating point values. Permissible values
7434 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7436 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7437 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7438 of floating point instructions, but still uses the soft-float calling
7441 @item -mlittle-endian
7442 @opindex mlittle-endian
7443 Generate code for a processor running in little-endian mode. This is
7444 the default for all standard configurations.
7447 @opindex mbig-endian
7448 Generate code for a processor running in big-endian mode; the default is
7449 to compile code for a little-endian processor.
7451 @item -mwords-little-endian
7452 @opindex mwords-little-endian
7453 This option only applies when generating code for big-endian processors.
7454 Generate code for a little-endian word order but a big-endian byte
7455 order. That is, a byte order of the form @samp{32107654}. Note: this
7456 option should only be used if you require compatibility with code for
7457 big-endian ARM processors generated by versions of the compiler prior to
7460 @item -mcpu=@var{name}
7462 This specifies the name of the target ARM processor. GCC uses this name
7463 to determine what kind of instructions it can emit when generating
7464 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7465 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7466 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7467 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7468 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7469 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7470 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7471 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7472 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7473 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7474 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7475 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7476 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7477 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7480 @itemx -mtune=@var{name}
7482 This option is very similar to the @option{-mcpu=} option, except that
7483 instead of specifying the actual target processor type, and hence
7484 restricting which instructions can be used, it specifies that GCC should
7485 tune the performance of the code as if the target were of the type
7486 specified in this option, but still choosing the instructions that it
7487 will generate based on the cpu specified by a @option{-mcpu=} option.
7488 For some ARM implementations better performance can be obtained by using
7491 @item -march=@var{name}
7493 This specifies the name of the target ARM architecture. GCC uses this
7494 name to determine what kind of instructions it can emit when generating
7495 assembly code. This option can be used in conjunction with or instead
7496 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7497 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7498 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7499 @samp{iwmmxt}, @samp{ep9312}.
7501 @item -mfpu=@var{name}
7502 @itemx -mfpe=@var{number}
7503 @itemx -mfp=@var{number}
7507 This specifies what floating point hardware (or hardware emulation) is
7508 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7509 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7510 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7511 with older versions of GCC@.
7513 If @option{-msoft-float} is specified this specifies the format of
7514 floating point values.
7516 @item -mstructure-size-boundary=@var{n}
7517 @opindex mstructure-size-boundary
7518 The size of all structures and unions will be rounded up to a multiple
7519 of the number of bits set by this option. Permissible values are 8, 32
7520 and 64. The default value varies for different toolchains. For the COFF
7521 targeted toolchain the default value is 8. A value of 64 is only allowed
7522 if the underlying ABI supports it.
7524 Specifying the larger number can produce faster, more efficient code, but
7525 can also increase the size of the program. Different values are potentially
7526 incompatible. Code compiled with one value cannot necessarily expect to
7527 work with code or libraries compiled with another value, if they exchange
7528 information using structures or unions.
7530 @item -mabort-on-noreturn
7531 @opindex mabort-on-noreturn
7532 Generate a call to the function @code{abort} at the end of a
7533 @code{noreturn} function. It will be executed if the function tries to
7537 @itemx -mno-long-calls
7538 @opindex mlong-calls
7539 @opindex mno-long-calls
7540 Tells the compiler to perform function calls by first loading the
7541 address of the function into a register and then performing a subroutine
7542 call on this register. This switch is needed if the target function
7543 will lie outside of the 64 megabyte addressing range of the offset based
7544 version of subroutine call instruction.
7546 Even if this switch is enabled, not all function calls will be turned
7547 into long calls. The heuristic is that static functions, functions
7548 which have the @samp{short-call} attribute, functions that are inside
7549 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7550 definitions have already been compiled within the current compilation
7551 unit, will not be turned into long calls. The exception to this rule is
7552 that weak function definitions, functions with the @samp{long-call}
7553 attribute or the @samp{section} attribute, and functions that are within
7554 the scope of a @samp{#pragma long_calls} directive, will always be
7555 turned into long calls.
7557 This feature is not enabled by default. Specifying
7558 @option{-mno-long-calls} will restore the default behavior, as will
7559 placing the function calls within the scope of a @samp{#pragma
7560 long_calls_off} directive. Note these switches have no effect on how
7561 the compiler generates code to handle function calls via function
7564 @item -mnop-fun-dllimport
7565 @opindex mnop-fun-dllimport
7566 Disable support for the @code{dllimport} attribute.
7568 @item -msingle-pic-base
7569 @opindex msingle-pic-base
7570 Treat the register used for PIC addressing as read-only, rather than
7571 loading it in the prologue for each function. The run-time system is
7572 responsible for initializing this register with an appropriate value
7573 before execution begins.
7575 @item -mpic-register=@var{reg}
7576 @opindex mpic-register
7577 Specify the register to be used for PIC addressing. The default is R10
7578 unless stack-checking is enabled, when R9 is used.
7580 @item -mcirrus-fix-invalid-insns
7581 @opindex mcirrus-fix-invalid-insns
7582 @opindex mno-cirrus-fix-invalid-insns
7583 Insert NOPs into the instruction stream to in order to work around
7584 problems with invalid Maverick instruction combinations. This option
7585 is only valid if the @option{-mcpu=ep9312} option has been used to
7586 enable generation of instructions for the Cirrus Maverick floating
7587 point co-processor. This option is not enabled by default, since the
7588 problem is only present in older Maverick implementations. The default
7589 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7592 @item -mpoke-function-name
7593 @opindex mpoke-function-name
7594 Write the name of each function into the text section, directly
7595 preceding the function prologue. The generated code is similar to this:
7599 .ascii "arm_poke_function_name", 0
7602 .word 0xff000000 + (t1 - t0)
7603 arm_poke_function_name
7605 stmfd sp!, @{fp, ip, lr, pc@}
7609 When performing a stack backtrace, code can inspect the value of
7610 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7611 location @code{pc - 12} and the top 8 bits are set, then we know that
7612 there is a function name embedded immediately preceding this location
7613 and has length @code{((pc[-3]) & 0xff000000)}.
7617 Generate code for the 16-bit Thumb instruction set. The default is to
7618 use the 32-bit ARM instruction set.
7621 @opindex mtpcs-frame
7622 Generate a stack frame that is compliant with the Thumb Procedure Call
7623 Standard for all non-leaf functions. (A leaf function is one that does
7624 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7626 @item -mtpcs-leaf-frame
7627 @opindex mtpcs-leaf-frame
7628 Generate a stack frame that is compliant with the Thumb Procedure Call
7629 Standard for all leaf functions. (A leaf function is one that does
7630 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7632 @item -mcallee-super-interworking
7633 @opindex mcallee-super-interworking
7634 Gives all externally visible functions in the file being compiled an ARM
7635 instruction set header which switches to Thumb mode before executing the
7636 rest of the function. This allows these functions to be called from
7637 non-interworking code.
7639 @item -mcaller-super-interworking
7640 @opindex mcaller-super-interworking
7641 Allows calls via function pointers (including virtual functions) to
7642 execute correctly regardless of whether the target code has been
7643 compiled for interworking or not. There is a small overhead in the cost
7644 of executing a function pointer if this option is enabled.
7646 @item -mtp=@var{name}
7648 Specify the access model for the thread local storage pointer. The valid
7649 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7650 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7651 (supported in the arm6k architecture), and @option{auto}, which uses the
7652 best available method for the selected processor. The default setting is
7658 @subsection AVR Options
7661 These options are defined for AVR implementations:
7664 @item -mmcu=@var{mcu}
7666 Specify ATMEL AVR instruction set or MCU type.
7668 Instruction set avr1 is for the minimal AVR core, not supported by the C
7669 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7670 attiny11, attiny12, attiny15, attiny28).
7672 Instruction set avr2 (default) is for the classic AVR core with up to
7673 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7674 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7675 at90c8534, at90s8535).
7677 Instruction set avr3 is for the classic AVR core with up to 128K program
7678 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7680 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7681 memory space (MCU types: atmega8, atmega83, atmega85).
7683 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7684 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7685 atmega64, atmega128, at43usb355, at94k).
7689 Output instruction sizes to the asm file.
7691 @item -minit-stack=@var{N}
7692 @opindex minit-stack
7693 Specify the initial stack address, which may be a symbol or numeric value,
7694 @samp{__stack} is the default.
7696 @item -mno-interrupts
7697 @opindex mno-interrupts
7698 Generated code is not compatible with hardware interrupts.
7699 Code size will be smaller.
7701 @item -mcall-prologues
7702 @opindex mcall-prologues
7703 Functions prologues/epilogues expanded as call to appropriate
7704 subroutines. Code size will be smaller.
7706 @item -mno-tablejump
7707 @opindex mno-tablejump
7708 Do not generate tablejump insns which sometimes increase code size.
7711 @opindex mtiny-stack
7712 Change only the low 8 bits of the stack pointer.
7716 Assume int to be 8 bit integer. This affects the sizes of all types: A
7717 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7718 and long long will be 4 bytes. Please note that this option does not
7719 comply to the C standards, but it will provide you with smaller code
7723 @node Blackfin Options
7724 @subsection Blackfin Options
7725 @cindex Blackfin Options
7728 @item -momit-leaf-frame-pointer
7729 @opindex momit-leaf-frame-pointer
7730 Don't keep the frame pointer in a register for leaf functions. This
7731 avoids the instructions to save, set up and restore frame pointers and
7732 makes an extra register available in leaf functions. The option
7733 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7734 which might make debugging harder.
7736 @item -mspecld-anomaly
7737 @opindex mspecld-anomaly
7738 When enabled, the compiler will ensure that the generated code does not
7739 contain speculative loads after jump instructions. This option is enabled
7742 @item -mno-specld-anomaly
7743 @opindex mno-specld-anomaly
7744 Don't generate extra code to prevent speculative loads from occurring.
7746 @item -mcsync-anomaly
7747 @opindex mcsync-anomaly
7748 When enabled, the compiler will ensure that the generated code does not
7749 contain CSYNC or SSYNC instructions too soon after conditional branches.
7750 This option is enabled by default.
7752 @item -mno-csync-anomaly
7753 @opindex mno-csync-anomaly
7754 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7755 occurring too soon after a conditional branch.
7759 When enabled, the compiler is free to take advantage of the knowledge that
7760 the entire program fits into the low 64k of memory.
7763 @opindex mno-low-64k
7764 Assume that the program is arbitrarily large. This is the default.
7766 @item -mid-shared-library
7767 @opindex mid-shared-library
7768 Generate code that supports shared libraries via the library ID method.
7769 This allows for execute in place and shared libraries in an environment
7770 without virtual memory management. This option implies @option{-fPIC}.
7772 @item -mno-id-shared-library
7773 @opindex mno-id-shared-library
7774 Generate code that doesn't assume ID based shared libraries are being used.
7775 This is the default.
7777 @item -mshared-library-id=n
7778 @opindex mshared-library-id
7779 Specified the identification number of the ID based shared library being
7780 compiled. Specifying a value of 0 will generate more compact code, specifying
7781 other values will force the allocation of that number to the current
7782 library but is no more space or time efficient than omitting this option.
7785 @itemx -mno-long-calls
7786 @opindex mlong-calls
7787 @opindex mno-long-calls
7788 Tells the compiler to perform function calls by first loading the
7789 address of the function into a register and then performing a subroutine
7790 call on this register. This switch is needed if the target function
7791 will lie outside of the 24 bit addressing range of the offset based
7792 version of subroutine call instruction.
7794 This feature is not enabled by default. Specifying
7795 @option{-mno-long-calls} will restore the default behavior. Note these
7796 switches have no effect on how the compiler generates code to handle
7797 function calls via function pointers.
7801 @subsection CRIS Options
7802 @cindex CRIS Options
7804 These options are defined specifically for the CRIS ports.
7807 @item -march=@var{architecture-type}
7808 @itemx -mcpu=@var{architecture-type}
7811 Generate code for the specified architecture. The choices for
7812 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7813 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7814 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7817 @item -mtune=@var{architecture-type}
7819 Tune to @var{architecture-type} everything applicable about the generated
7820 code, except for the ABI and the set of available instructions. The
7821 choices for @var{architecture-type} are the same as for
7822 @option{-march=@var{architecture-type}}.
7824 @item -mmax-stack-frame=@var{n}
7825 @opindex mmax-stack-frame
7826 Warn when the stack frame of a function exceeds @var{n} bytes.
7828 @item -melinux-stacksize=@var{n}
7829 @opindex melinux-stacksize
7830 Only available with the @samp{cris-axis-aout} target. Arranges for
7831 indications in the program to the kernel loader that the stack of the
7832 program should be set to @var{n} bytes.
7838 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7839 @option{-march=v3} and @option{-march=v8} respectively.
7841 @item -mmul-bug-workaround
7842 @itemx -mno-mul-bug-workaround
7843 @opindex mmul-bug-workaround
7844 @opindex mno-mul-bug-workaround
7845 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7846 models where it applies. This option is active by default.
7850 Enable CRIS-specific verbose debug-related information in the assembly
7851 code. This option also has the effect to turn off the @samp{#NO_APP}
7852 formatted-code indicator to the assembler at the beginning of the
7857 Do not use condition-code results from previous instruction; always emit
7858 compare and test instructions before use of condition codes.
7860 @item -mno-side-effects
7861 @opindex mno-side-effects
7862 Do not emit instructions with side-effects in addressing modes other than
7866 @itemx -mno-stack-align
7868 @itemx -mno-data-align
7869 @itemx -mconst-align
7870 @itemx -mno-const-align
7871 @opindex mstack-align
7872 @opindex mno-stack-align
7873 @opindex mdata-align
7874 @opindex mno-data-align
7875 @opindex mconst-align
7876 @opindex mno-const-align
7877 These options (no-options) arranges (eliminate arrangements) for the
7878 stack-frame, individual data and constants to be aligned for the maximum
7879 single data access size for the chosen CPU model. The default is to
7880 arrange for 32-bit alignment. ABI details such as structure layout are
7881 not affected by these options.
7889 Similar to the stack- data- and const-align options above, these options
7890 arrange for stack-frame, writable data and constants to all be 32-bit,
7891 16-bit or 8-bit aligned. The default is 32-bit alignment.
7893 @item -mno-prologue-epilogue
7894 @itemx -mprologue-epilogue
7895 @opindex mno-prologue-epilogue
7896 @opindex mprologue-epilogue
7897 With @option{-mno-prologue-epilogue}, the normal function prologue and
7898 epilogue that sets up the stack-frame are omitted and no return
7899 instructions or return sequences are generated in the code. Use this
7900 option only together with visual inspection of the compiled code: no
7901 warnings or errors are generated when call-saved registers must be saved,
7902 or storage for local variable needs to be allocated.
7908 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7909 instruction sequences that load addresses for functions from the PLT part
7910 of the GOT rather than (traditional on other architectures) calls to the
7911 PLT@. The default is @option{-mgotplt}.
7915 Legacy no-op option only recognized with the cris-axis-aout target.
7919 Legacy no-op option only recognized with the cris-axis-elf and
7920 cris-axis-linux-gnu targets.
7924 Only recognized with the cris-axis-aout target, where it selects a
7925 GNU/linux-like multilib, include files and instruction set for
7930 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7934 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7935 to link with input-output functions from a simulator library. Code,
7936 initialized data and zero-initialized data are allocated consecutively.
7940 Like @option{-sim}, but pass linker options to locate initialized data at
7941 0x40000000 and zero-initialized data at 0x80000000.
7945 @subsection CRX Options
7948 These options are defined specifically for the CRX ports.
7954 Enable the use of multiply-accumulate instructions. Disabled by default.
7958 Push instructions will be used to pass outgoing arguments when functions
7959 are called. Enabled by default.
7962 @node Darwin Options
7963 @subsection Darwin Options
7964 @cindex Darwin options
7966 These options are defined for all architectures running the Darwin operating
7969 FSF GCC on Darwin does not create ``fat'' object files; it will create
7970 an object file for the single architecture that it was built to
7971 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7972 @option{-arch} options are used; it does so by running the compiler or
7973 linker multiple times and joining the results together with
7976 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7977 @samp{i686}) is determined by the flags that specify the ISA
7978 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7979 @option{-force_cpusubtype_ALL} option can be used to override this.
7981 The Darwin tools vary in their behavior when presented with an ISA
7982 mismatch. The assembler, @file{as}, will only permit instructions to
7983 be used that are valid for the subtype of the file it is generating,
7984 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7985 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7986 and print an error if asked to create a shared library with a less
7987 restrictive subtype than its input files (for instance, trying to put
7988 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7989 for executables, @file{ld}, will quietly give the executable the most
7990 restrictive subtype of any of its input files.
7995 Add the framework directory @var{dir} to the head of the list of
7996 directories to be searched for header files. These directories are
7997 interleaved with those specified by @option{-I} options and are
7998 scanned in a left-to-right order.
8000 A framework directory is a directory with frameworks in it. A
8001 framework is a directory with a @samp{"Headers"} and/or
8002 @samp{"PrivateHeaders"} directory contained directly in it that ends
8003 in @samp{".framework"}. The name of a framework is the name of this
8004 directory excluding the @samp{".framework"}. Headers associated with
8005 the framework are found in one of those two directories, with
8006 @samp{"Headers"} being searched first. A subframework is a framework
8007 directory that is in a framework's @samp{"Frameworks"} directory.
8008 Includes of subframework headers can only appear in a header of a
8009 framework that contains the subframework, or in a sibling subframework
8010 header. Two subframeworks are siblings if they occur in the same
8011 framework. A subframework should not have the same name as a
8012 framework, a warning will be issued if this is violated. Currently a
8013 subframework cannot have subframeworks, in the future, the mechanism
8014 may be extended to support this. The standard frameworks can be found
8015 in @samp{"/System/Library/Frameworks"} and
8016 @samp{"/Library/Frameworks"}. An example include looks like
8017 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8018 the name of the framework and header.h is found in the
8019 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8023 Emit debugging information for symbols that are used. For STABS
8024 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8025 This is by default ON@.
8029 Emit debugging information for all symbols and types.
8031 @item -mmacosx-version-min=@var{version}
8032 The earliest version of MacOS X that this executable will run on
8033 is @var{version}. Typical values of @var{version} include @code{10.1},
8034 @code{10.2}, and @code{10.3.9}.
8036 The default for this option is to make choices that seem to be most
8039 @item -mone-byte-bool
8040 @opindex -mone-byte-bool
8041 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8042 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8043 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8044 option has no effect on x86.
8046 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8047 to generate code that is not binary compatible with code generated
8048 without that switch. Using this switch may require recompiling all
8049 other modules in a program, including system libraries. Use this
8050 switch to conform to a non-default data model.
8052 @item -mfix-and-continue
8053 @itemx -ffix-and-continue
8054 @itemx -findirect-data
8055 @opindex mfix-and-continue
8056 @opindex ffix-and-continue
8057 @opindex findirect-data
8058 Generate code suitable for fast turn around development. Needed to
8059 enable gdb to dynamically load @code{.o} files into already running
8060 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8061 are provided for backwards compatibility.
8065 Loads all members of static archive libraries.
8066 See man ld(1) for more information.
8068 @item -arch_errors_fatal
8069 @opindex arch_errors_fatal
8070 Cause the errors having to do with files that have the wrong architecture
8074 @opindex bind_at_load
8075 Causes the output file to be marked such that the dynamic linker will
8076 bind all undefined references when the file is loaded or launched.
8080 Produce a Mach-o bundle format file.
8081 See man ld(1) for more information.
8083 @item -bundle_loader @var{executable}
8084 @opindex bundle_loader
8085 This option specifies the @var{executable} that will be loading the build
8086 output file being linked. See man ld(1) for more information.
8089 @opindex -dynamiclib
8090 When passed this option, GCC will produce a dynamic library instead of
8091 an executable when linking, using the Darwin @file{libtool} command.
8093 @item -force_cpusubtype_ALL
8094 @opindex -force_cpusubtype_ALL
8095 This causes GCC's output file to have the @var{ALL} subtype, instead of
8096 one controlled by the @option{-mcpu} or @option{-march} option.
8098 @item -allowable_client @var{client_name}
8100 @itemx -compatibility_version
8101 @itemx -current_version
8103 @itemx -dependency-file
8105 @itemx -dylinker_install_name
8107 @itemx -exported_symbols_list
8109 @itemx -flat_namespace
8110 @itemx -force_flat_namespace
8111 @itemx -headerpad_max_install_names
8114 @itemx -install_name
8115 @itemx -keep_private_externs
8116 @itemx -multi_module
8117 @itemx -multiply_defined
8118 @itemx -multiply_defined_unused
8120 @itemx -no_dead_strip_inits_and_terms
8121 @itemx -nofixprebinding
8124 @itemx -noseglinkedit
8125 @itemx -pagezero_size
8127 @itemx -prebind_all_twolevel_modules
8128 @itemx -private_bundle
8129 @itemx -read_only_relocs
8131 @itemx -sectobjectsymbols
8135 @itemx -sectobjectsymbols
8138 @itemx -segs_read_only_addr
8139 @itemx -segs_read_write_addr
8140 @itemx -seg_addr_table
8141 @itemx -seg_addr_table_filename
8144 @itemx -segs_read_only_addr
8145 @itemx -segs_read_write_addr
8146 @itemx -single_module
8149 @itemx -sub_umbrella
8150 @itemx -twolevel_namespace
8153 @itemx -unexported_symbols_list
8154 @itemx -weak_reference_mismatches
8157 @opindex allowable_client
8158 @opindex client_name
8159 @opindex compatibility_version
8160 @opindex current_version
8162 @opindex dependency-file
8164 @opindex dylinker_install_name
8166 @opindex exported_symbols_list
8168 @opindex flat_namespace
8169 @opindex force_flat_namespace
8170 @opindex headerpad_max_install_names
8173 @opindex install_name
8174 @opindex keep_private_externs
8175 @opindex multi_module
8176 @opindex multiply_defined
8177 @opindex multiply_defined_unused
8179 @opindex no_dead_strip_inits_and_terms
8180 @opindex nofixprebinding
8181 @opindex nomultidefs
8183 @opindex noseglinkedit
8184 @opindex pagezero_size
8186 @opindex prebind_all_twolevel_modules
8187 @opindex private_bundle
8188 @opindex read_only_relocs
8190 @opindex sectobjectsymbols
8194 @opindex sectobjectsymbols
8197 @opindex segs_read_only_addr
8198 @opindex segs_read_write_addr
8199 @opindex seg_addr_table
8200 @opindex seg_addr_table_filename
8201 @opindex seglinkedit
8203 @opindex segs_read_only_addr
8204 @opindex segs_read_write_addr
8205 @opindex single_module
8207 @opindex sub_library
8208 @opindex sub_umbrella
8209 @opindex twolevel_namespace
8212 @opindex unexported_symbols_list
8213 @opindex weak_reference_mismatches
8214 @opindex whatsloaded
8216 These options are passed to the Darwin linker. The Darwin linker man page
8217 describes them in detail.
8220 @node DEC Alpha Options
8221 @subsection DEC Alpha Options
8223 These @samp{-m} options are defined for the DEC Alpha implementations:
8226 @item -mno-soft-float
8228 @opindex mno-soft-float
8229 @opindex msoft-float
8230 Use (do not use) the hardware floating-point instructions for
8231 floating-point operations. When @option{-msoft-float} is specified,
8232 functions in @file{libgcc.a} will be used to perform floating-point
8233 operations. Unless they are replaced by routines that emulate the
8234 floating-point operations, or compiled in such a way as to call such
8235 emulations routines, these routines will issue floating-point
8236 operations. If you are compiling for an Alpha without floating-point
8237 operations, you must ensure that the library is built so as not to call
8240 Note that Alpha implementations without floating-point operations are
8241 required to have floating-point registers.
8246 @opindex mno-fp-regs
8247 Generate code that uses (does not use) the floating-point register set.
8248 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8249 register set is not used, floating point operands are passed in integer
8250 registers as if they were integers and floating-point results are passed
8251 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8252 so any function with a floating-point argument or return value called by code
8253 compiled with @option{-mno-fp-regs} must also be compiled with that
8256 A typical use of this option is building a kernel that does not use,
8257 and hence need not save and restore, any floating-point registers.
8261 The Alpha architecture implements floating-point hardware optimized for
8262 maximum performance. It is mostly compliant with the IEEE floating
8263 point standard. However, for full compliance, software assistance is
8264 required. This option generates code fully IEEE compliant code
8265 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8266 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8267 defined during compilation. The resulting code is less efficient but is
8268 able to correctly support denormalized numbers and exceptional IEEE
8269 values such as not-a-number and plus/minus infinity. Other Alpha
8270 compilers call this option @option{-ieee_with_no_inexact}.
8272 @item -mieee-with-inexact
8273 @opindex mieee-with-inexact
8274 This is like @option{-mieee} except the generated code also maintains
8275 the IEEE @var{inexact-flag}. Turning on this option causes the
8276 generated code to implement fully-compliant IEEE math. In addition to
8277 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8278 macro. On some Alpha implementations the resulting code may execute
8279 significantly slower than the code generated by default. Since there is
8280 very little code that depends on the @var{inexact-flag}, you should
8281 normally not specify this option. Other Alpha compilers call this
8282 option @option{-ieee_with_inexact}.
8284 @item -mfp-trap-mode=@var{trap-mode}
8285 @opindex mfp-trap-mode
8286 This option controls what floating-point related traps are enabled.
8287 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8288 The trap mode can be set to one of four values:
8292 This is the default (normal) setting. The only traps that are enabled
8293 are the ones that cannot be disabled in software (e.g., division by zero
8297 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8301 Like @samp{su}, but the instructions are marked to be safe for software
8302 completion (see Alpha architecture manual for details).
8305 Like @samp{su}, but inexact traps are enabled as well.
8308 @item -mfp-rounding-mode=@var{rounding-mode}
8309 @opindex mfp-rounding-mode
8310 Selects the IEEE rounding mode. Other Alpha compilers call this option
8311 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8316 Normal IEEE rounding mode. Floating point numbers are rounded towards
8317 the nearest machine number or towards the even machine number in case
8321 Round towards minus infinity.
8324 Chopped rounding mode. Floating point numbers are rounded towards zero.
8327 Dynamic rounding mode. A field in the floating point control register
8328 (@var{fpcr}, see Alpha architecture reference manual) controls the
8329 rounding mode in effect. The C library initializes this register for
8330 rounding towards plus infinity. Thus, unless your program modifies the
8331 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8334 @item -mtrap-precision=@var{trap-precision}
8335 @opindex mtrap-precision
8336 In the Alpha architecture, floating point traps are imprecise. This
8337 means without software assistance it is impossible to recover from a
8338 floating trap and program execution normally needs to be terminated.
8339 GCC can generate code that can assist operating system trap handlers
8340 in determining the exact location that caused a floating point trap.
8341 Depending on the requirements of an application, different levels of
8342 precisions can be selected:
8346 Program precision. This option is the default and means a trap handler
8347 can only identify which program caused a floating point exception.
8350 Function precision. The trap handler can determine the function that
8351 caused a floating point exception.
8354 Instruction precision. The trap handler can determine the exact
8355 instruction that caused a floating point exception.
8358 Other Alpha compilers provide the equivalent options called
8359 @option{-scope_safe} and @option{-resumption_safe}.
8361 @item -mieee-conformant
8362 @opindex mieee-conformant
8363 This option marks the generated code as IEEE conformant. You must not
8364 use this option unless you also specify @option{-mtrap-precision=i} and either
8365 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8366 is to emit the line @samp{.eflag 48} in the function prologue of the
8367 generated assembly file. Under DEC Unix, this has the effect that
8368 IEEE-conformant math library routines will be linked in.
8370 @item -mbuild-constants
8371 @opindex mbuild-constants
8372 Normally GCC examines a 32- or 64-bit integer constant to
8373 see if it can construct it from smaller constants in two or three
8374 instructions. If it cannot, it will output the constant as a literal and
8375 generate code to load it from the data segment at runtime.
8377 Use this option to require GCC to construct @emph{all} integer constants
8378 using code, even if it takes more instructions (the maximum is six).
8380 You would typically use this option to build a shared library dynamic
8381 loader. Itself a shared library, it must relocate itself in memory
8382 before it can find the variables and constants in its own data segment.
8388 Select whether to generate code to be assembled by the vendor-supplied
8389 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8407 Indicate whether GCC should generate code to use the optional BWX,
8408 CIX, FIX and MAX instruction sets. The default is to use the instruction
8409 sets supported by the CPU type specified via @option{-mcpu=} option or that
8410 of the CPU on which GCC was built if none was specified.
8415 @opindex mfloat-ieee
8416 Generate code that uses (does not use) VAX F and G floating point
8417 arithmetic instead of IEEE single and double precision.
8419 @item -mexplicit-relocs
8420 @itemx -mno-explicit-relocs
8421 @opindex mexplicit-relocs
8422 @opindex mno-explicit-relocs
8423 Older Alpha assemblers provided no way to generate symbol relocations
8424 except via assembler macros. Use of these macros does not allow
8425 optimal instruction scheduling. GNU binutils as of version 2.12
8426 supports a new syntax that allows the compiler to explicitly mark
8427 which relocations should apply to which instructions. This option
8428 is mostly useful for debugging, as GCC detects the capabilities of
8429 the assembler when it is built and sets the default accordingly.
8433 @opindex msmall-data
8434 @opindex mlarge-data
8435 When @option{-mexplicit-relocs} is in effect, static data is
8436 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8437 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8438 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8439 16-bit relocations off of the @code{$gp} register. This limits the
8440 size of the small data area to 64KB, but allows the variables to be
8441 directly accessed via a single instruction.
8443 The default is @option{-mlarge-data}. With this option the data area
8444 is limited to just below 2GB@. Programs that require more than 2GB of
8445 data must use @code{malloc} or @code{mmap} to allocate the data in the
8446 heap instead of in the program's data segment.
8448 When generating code for shared libraries, @option{-fpic} implies
8449 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8453 @opindex msmall-text
8454 @opindex mlarge-text
8455 When @option{-msmall-text} is used, the compiler assumes that the
8456 code of the entire program (or shared library) fits in 4MB, and is
8457 thus reachable with a branch instruction. When @option{-msmall-data}
8458 is used, the compiler can assume that all local symbols share the
8459 same @code{$gp} value, and thus reduce the number of instructions
8460 required for a function call from 4 to 1.
8462 The default is @option{-mlarge-text}.
8464 @item -mcpu=@var{cpu_type}
8466 Set the instruction set and instruction scheduling parameters for
8467 machine type @var{cpu_type}. You can specify either the @samp{EV}
8468 style name or the corresponding chip number. GCC supports scheduling
8469 parameters for the EV4, EV5 and EV6 family of processors and will
8470 choose the default values for the instruction set from the processor
8471 you specify. If you do not specify a processor type, GCC will default
8472 to the processor on which the compiler was built.
8474 Supported values for @var{cpu_type} are
8480 Schedules as an EV4 and has no instruction set extensions.
8484 Schedules as an EV5 and has no instruction set extensions.
8488 Schedules as an EV5 and supports the BWX extension.
8493 Schedules as an EV5 and supports the BWX and MAX extensions.
8497 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8501 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8504 @item -mtune=@var{cpu_type}
8506 Set only the instruction scheduling parameters for machine type
8507 @var{cpu_type}. The instruction set is not changed.
8509 @item -mmemory-latency=@var{time}
8510 @opindex mmemory-latency
8511 Sets the latency the scheduler should assume for typical memory
8512 references as seen by the application. This number is highly
8513 dependent on the memory access patterns used by the application
8514 and the size of the external cache on the machine.
8516 Valid options for @var{time} are
8520 A decimal number representing clock cycles.
8526 The compiler contains estimates of the number of clock cycles for
8527 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8528 (also called Dcache, Scache, and Bcache), as well as to main memory.
8529 Note that L3 is only valid for EV5.
8534 @node DEC Alpha/VMS Options
8535 @subsection DEC Alpha/VMS Options
8537 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8540 @item -mvms-return-codes
8541 @opindex mvms-return-codes
8542 Return VMS condition codes from main. The default is to return POSIX
8543 style condition (e.g.@ error) codes.
8547 @subsection FRV Options
8554 Only use the first 32 general purpose registers.
8559 Use all 64 general purpose registers.
8564 Use only the first 32 floating point registers.
8569 Use all 64 floating point registers
8572 @opindex mhard-float
8574 Use hardware instructions for floating point operations.
8577 @opindex msoft-float
8579 Use library routines for floating point operations.
8584 Dynamically allocate condition code registers.
8589 Do not try to dynamically allocate condition code registers, only
8590 use @code{icc0} and @code{fcc0}.
8595 Change ABI to use double word insns.
8600 Do not use double word instructions.
8605 Use floating point double instructions.
8610 Do not use floating point double instructions.
8615 Use media instructions.
8620 Do not use media instructions.
8625 Use multiply and add/subtract instructions.
8630 Do not use multiply and add/subtract instructions.
8635 Select the FDPIC ABI, that uses function descriptors to represent
8636 pointers to functions. Without any PIC/PIE-related options, it
8637 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8638 assumes GOT entries and small data are within a 12-bit range from the
8639 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8640 are computed with 32 bits.
8643 @opindex minline-plt
8645 Enable inlining of PLT entries in function calls to functions that are
8646 not known to bind locally. It has no effect without @option{-mfdpic}.
8647 It's enabled by default if optimizing for speed and compiling for
8648 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8649 optimization option such as @option{-O3} or above is present in the
8655 Assume a large TLS segment when generating thread-local code.
8660 Do not assume a large TLS segment when generating thread-local code.
8665 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8666 that is known to be in read-only sections. It's enabled by default,
8667 except for @option{-fpic} or @option{-fpie}: even though it may help
8668 make the global offset table smaller, it trades 1 instruction for 4.
8669 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8670 one of which may be shared by multiple symbols, and it avoids the need
8671 for a GOT entry for the referenced symbol, so it's more likely to be a
8672 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8674 @item -multilib-library-pic
8675 @opindex multilib-library-pic
8677 Link with the (library, not FD) pic libraries. It's implied by
8678 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8679 @option{-fpic} without @option{-mfdpic}. You should never have to use
8685 Follow the EABI requirement of always creating a frame pointer whenever
8686 a stack frame is allocated. This option is enabled by default and can
8687 be disabled with @option{-mno-linked-fp}.
8690 @opindex mlong-calls
8692 Use indirect addressing to call functions outside the current
8693 compilation unit. This allows the functions to be placed anywhere
8694 within the 32-bit address space.
8696 @item -malign-labels
8697 @opindex malign-labels
8699 Try to align labels to an 8-byte boundary by inserting nops into the
8700 previous packet. This option only has an effect when VLIW packing
8701 is enabled. It doesn't create new packets; it merely adds nops to
8705 @opindex mlibrary-pic
8707 Generate position-independent EABI code.
8712 Use only the first four media accumulator registers.
8717 Use all eight media accumulator registers.
8722 Pack VLIW instructions.
8727 Do not pack VLIW instructions.
8732 Do not mark ABI switches in e_flags.
8737 Enable the use of conditional-move instructions (default).
8739 This switch is mainly for debugging the compiler and will likely be removed
8740 in a future version.
8742 @item -mno-cond-move
8743 @opindex mno-cond-move
8745 Disable the use of conditional-move instructions.
8747 This switch is mainly for debugging the compiler and will likely be removed
8748 in a future version.
8753 Enable the use of conditional set instructions (default).
8755 This switch is mainly for debugging the compiler and will likely be removed
8756 in a future version.
8761 Disable the use of conditional set instructions.
8763 This switch is mainly for debugging the compiler and will likely be removed
8764 in a future version.
8769 Enable the use of conditional execution (default).
8771 This switch is mainly for debugging the compiler and will likely be removed
8772 in a future version.
8774 @item -mno-cond-exec
8775 @opindex mno-cond-exec
8777 Disable the use of conditional execution.
8779 This switch is mainly for debugging the compiler and will likely be removed
8780 in a future version.
8783 @opindex mvliw-branch
8785 Run a pass to pack branches into VLIW instructions (default).
8787 This switch is mainly for debugging the compiler and will likely be removed
8788 in a future version.
8790 @item -mno-vliw-branch
8791 @opindex mno-vliw-branch
8793 Do not run a pass to pack branches into VLIW instructions.
8795 This switch is mainly for debugging the compiler and will likely be removed
8796 in a future version.
8798 @item -mmulti-cond-exec
8799 @opindex mmulti-cond-exec
8801 Enable optimization of @code{&&} and @code{||} in conditional execution
8804 This switch is mainly for debugging the compiler and will likely be removed
8805 in a future version.
8807 @item -mno-multi-cond-exec
8808 @opindex mno-multi-cond-exec
8810 Disable optimization of @code{&&} and @code{||} in conditional execution.
8812 This switch is mainly for debugging the compiler and will likely be removed
8813 in a future version.
8815 @item -mnested-cond-exec
8816 @opindex mnested-cond-exec
8818 Enable nested conditional execution optimizations (default).
8820 This switch is mainly for debugging the compiler and will likely be removed
8821 in a future version.
8823 @item -mno-nested-cond-exec
8824 @opindex mno-nested-cond-exec
8826 Disable nested conditional execution optimizations.
8828 This switch is mainly for debugging the compiler and will likely be removed
8829 in a future version.
8831 @item -moptimize-membar
8832 @opindex moptimize-membar
8834 This switch removes redundant @code{membar} instructions from the
8835 compiler generated code. It is enabled by default.
8837 @item -mno-optimize-membar
8838 @opindex mno-optimize-membar
8840 This switch disables the automatic removal of redundant @code{membar}
8841 instructions from the generated code.
8843 @item -mtomcat-stats
8844 @opindex mtomcat-stats
8846 Cause gas to print out tomcat statistics.
8848 @item -mcpu=@var{cpu}
8851 Select the processor type for which to generate code. Possible values are
8852 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8853 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8857 @node GNU/Linux Options
8858 @subsection GNU/Linux Options
8860 These @samp{-m} options are defined for GNU/Linux targets:
8865 Use the GNU C library instead of uClibc. This is the default except
8866 on @samp{*-*-linux-*uclibc*} targets.
8870 Use uClibc instead of the GNU C library. This is the default on
8871 @samp{*-*-linux-*uclibc*} targets.
8874 @node H8/300 Options
8875 @subsection H8/300 Options
8877 These @samp{-m} options are defined for the H8/300 implementations:
8882 Shorten some address references at link time, when possible; uses the
8883 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8884 ld, Using ld}, for a fuller description.
8888 Generate code for the H8/300H@.
8892 Generate code for the H8S@.
8896 Generate code for the H8S and H8/300H in the normal mode. This switch
8897 must be used either with @option{-mh} or @option{-ms}.
8901 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8905 Make @code{int} data 32 bits by default.
8909 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8910 The default for the H8/300H and H8S is to align longs and floats on 4
8912 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8913 This option has no effect on the H8/300.
8917 @subsection HPPA Options
8918 @cindex HPPA Options
8920 These @samp{-m} options are defined for the HPPA family of computers:
8923 @item -march=@var{architecture-type}
8925 Generate code for the specified architecture. The choices for
8926 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8927 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8928 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8929 architecture option for your machine. Code compiled for lower numbered
8930 architectures will run on higher numbered architectures, but not the
8934 @itemx -mpa-risc-1-1
8935 @itemx -mpa-risc-2-0
8936 @opindex mpa-risc-1-0
8937 @opindex mpa-risc-1-1
8938 @opindex mpa-risc-2-0
8939 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8942 @opindex mbig-switch
8943 Generate code suitable for big switch tables. Use this option only if
8944 the assembler/linker complain about out of range branches within a switch
8947 @item -mjump-in-delay
8948 @opindex mjump-in-delay
8949 Fill delay slots of function calls with unconditional jump instructions
8950 by modifying the return pointer for the function call to be the target
8951 of the conditional jump.
8953 @item -mdisable-fpregs
8954 @opindex mdisable-fpregs
8955 Prevent floating point registers from being used in any manner. This is
8956 necessary for compiling kernels which perform lazy context switching of
8957 floating point registers. If you use this option and attempt to perform
8958 floating point operations, the compiler will abort.
8960 @item -mdisable-indexing
8961 @opindex mdisable-indexing
8962 Prevent the compiler from using indexing address modes. This avoids some
8963 rather obscure problems when compiling MIG generated code under MACH@.
8965 @item -mno-space-regs
8966 @opindex mno-space-regs
8967 Generate code that assumes the target has no space registers. This allows
8968 GCC to generate faster indirect calls and use unscaled index address modes.
8970 Such code is suitable for level 0 PA systems and kernels.
8972 @item -mfast-indirect-calls
8973 @opindex mfast-indirect-calls
8974 Generate code that assumes calls never cross space boundaries. This
8975 allows GCC to emit code which performs faster indirect calls.
8977 This option will not work in the presence of shared libraries or nested
8980 @item -mfixed-range=@var{register-range}
8981 @opindex mfixed-range
8982 Generate code treating the given register range as fixed registers.
8983 A fixed register is one that the register allocator can not use. This is
8984 useful when compiling kernel code. A register range is specified as
8985 two registers separated by a dash. Multiple register ranges can be
8986 specified separated by a comma.
8988 @item -mlong-load-store
8989 @opindex mlong-load-store
8990 Generate 3-instruction load and store sequences as sometimes required by
8991 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8994 @item -mportable-runtime
8995 @opindex mportable-runtime
8996 Use the portable calling conventions proposed by HP for ELF systems.
9000 Enable the use of assembler directives only GAS understands.
9002 @item -mschedule=@var{cpu-type}
9004 Schedule code according to the constraints for the machine type
9005 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9006 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9007 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9008 proper scheduling option for your machine. The default scheduling is
9012 @opindex mlinker-opt
9013 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9014 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9015 linkers in which they give bogus error messages when linking some programs.
9018 @opindex msoft-float
9019 Generate output containing library calls for floating point.
9020 @strong{Warning:} the requisite libraries are not available for all HPPA
9021 targets. Normally the facilities of the machine's usual C compiler are
9022 used, but this cannot be done directly in cross-compilation. You must make
9023 your own arrangements to provide suitable library functions for
9024 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9025 does provide software floating point support.
9027 @option{-msoft-float} changes the calling convention in the output file;
9028 therefore, it is only useful if you compile @emph{all} of a program with
9029 this option. In particular, you need to compile @file{libgcc.a}, the
9030 library that comes with GCC, with @option{-msoft-float} in order for
9035 Generate the predefine, @code{_SIO}, for server IO@. The default is
9036 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9037 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9038 options are available under HP-UX and HI-UX@.
9042 Use GNU ld specific options. This passes @option{-shared} to ld when
9043 building a shared library. It is the default when GCC is configured,
9044 explicitly or implicitly, with the GNU linker. This option does not
9045 have any affect on which ld is called, it only changes what parameters
9046 are passed to that ld. The ld that is called is determined by the
9047 @option{--with-ld} configure option, GCC's program search path, and
9048 finally by the user's @env{PATH}. The linker used by GCC can be printed
9049 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9050 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9054 Use HP ld specific options. This passes @option{-b} to ld when building
9055 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9056 links. It is the default when GCC is configured, explicitly or
9057 implicitly, with the HP linker. This option does not have any affect on
9058 which ld is called, it only changes what parameters are passed to that
9059 ld. The ld that is called is determined by the @option{--with-ld}
9060 configure option, GCC's program search path, and finally by the user's
9061 @env{PATH}. The linker used by GCC can be printed using @samp{which
9062 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9063 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9066 @opindex mno-long-calls
9067 Generate code that uses long call sequences. This ensures that a call
9068 is always able to reach linker generated stubs. The default is to generate
9069 long calls only when the distance from the call site to the beginning
9070 of the function or translation unit, as the case may be, exceeds a
9071 predefined limit set by the branch type being used. The limits for
9072 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9073 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9076 Distances are measured from the beginning of functions when using the
9077 @option{-ffunction-sections} option, or when using the @option{-mgas}
9078 and @option{-mno-portable-runtime} options together under HP-UX with
9081 It is normally not desirable to use this option as it will degrade
9082 performance. However, it may be useful in large applications,
9083 particularly when partial linking is used to build the application.
9085 The types of long calls used depends on the capabilities of the
9086 assembler and linker, and the type of code being generated. The
9087 impact on systems that support long absolute calls, and long pic
9088 symbol-difference or pc-relative calls should be relatively small.
9089 However, an indirect call is used on 32-bit ELF systems in pic code
9090 and it is quite long.
9092 @item -munix=@var{unix-std}
9094 Generate compiler predefines and select a startfile for the specified
9095 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9096 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9097 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9098 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9099 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9102 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9103 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9104 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9105 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9106 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9107 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9109 It is @emph{important} to note that this option changes the interfaces
9110 for various library routines. It also affects the operational behavior
9111 of the C library. Thus, @emph{extreme} care is needed in using this
9114 Library code that is intended to operate with more than one UNIX
9115 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9116 as appropriate. Most GNU software doesn't provide this capability.
9120 Suppress the generation of link options to search libdld.sl when the
9121 @option{-static} option is specified on HP-UX 10 and later.
9125 The HP-UX implementation of setlocale in libc has a dependency on
9126 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9127 when the @option{-static} option is specified, special link options
9128 are needed to resolve this dependency.
9130 On HP-UX 10 and later, the GCC driver adds the necessary options to
9131 link with libdld.sl when the @option{-static} option is specified.
9132 This causes the resulting binary to be dynamic. On the 64-bit port,
9133 the linkers generate dynamic binaries by default in any case. The
9134 @option{-nolibdld} option can be used to prevent the GCC driver from
9135 adding these link options.
9139 Add support for multithreading with the @dfn{dce thread} library
9140 under HP-UX@. This option sets flags for both the preprocessor and
9144 @node i386 and x86-64 Options
9145 @subsection Intel 386 and AMD x86-64 Options
9146 @cindex i386 Options
9147 @cindex x86-64 Options
9148 @cindex Intel 386 Options
9149 @cindex AMD x86-64 Options
9151 These @samp{-m} options are defined for the i386 and x86-64 family of
9155 @item -mtune=@var{cpu-type}
9157 Tune to @var{cpu-type} everything applicable about the generated code, except
9158 for the ABI and the set of available instructions. The choices for
9162 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9163 If you know the CPU on which your code will run, then you should use
9164 the corresponding @option{-mtune} option instead of
9165 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9166 of your application will have, then you should use this option.
9168 As new processors are deployed in the marketplace, the behavior of this
9169 option will change. Therefore, if you upgrade to a newer version of
9170 GCC, the code generated option will change to reflect the processors
9171 that were most common when that version of GCC was released.
9173 There is no @option{-march=generic} option because @option{-march}
9174 indicates the instruction set the compiler can use, and there is no
9175 generic instruction set applicable to all processors. In contrast,
9176 @option{-mtune} indicates the processor (or, in this case, collection of
9177 processors) for which the code is optimized.
9179 This selects the CPU to tune for at compilation time by determining
9180 the processor type of the compiling machine. Using @option{-mtune=native}
9181 will produce code optimized for the local machine under the constraints
9182 of the selected instruction set. Using @option{-march=native} will
9183 enable all instruction subsets supported by the local machine (hence
9184 the result might not run on different machines).
9186 Original Intel's i386 CPU@.
9188 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9190 Intel Pentium CPU with no MMX support.
9192 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9194 Intel PentiumPro CPU@.
9196 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9197 instruction set will be used, so the code will run on all i686 family chips.
9199 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9200 @item pentium3, pentium3m
9201 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9204 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9205 support. Used by Centrino notebooks.
9206 @item pentium4, pentium4m
9207 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9209 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9212 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9213 SSE2 and SSE3 instruction set support.
9215 AMD K6 CPU with MMX instruction set support.
9217 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9218 @item athlon, athlon-tbird
9219 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9221 @item athlon-4, athlon-xp, athlon-mp
9222 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9223 instruction set support.
9224 @item k8, opteron, athlon64, athlon-fx
9225 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9226 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9228 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9231 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9232 instruction set support.
9234 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9235 implemented for this chip.)
9237 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9238 implemented for this chip.)
9241 While picking a specific @var{cpu-type} will schedule things appropriately
9242 for that particular chip, the compiler will not generate any code that
9243 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9246 @item -march=@var{cpu-type}
9248 Generate instructions for the machine type @var{cpu-type}. The choices
9249 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9250 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9252 @item -mcpu=@var{cpu-type}
9254 A deprecated synonym for @option{-mtune}.
9263 @opindex mpentiumpro
9264 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9265 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9266 These synonyms are deprecated.
9268 @item -mfpmath=@var{unit}
9270 Generate floating point arithmetics for selected unit @var{unit}. The choices
9275 Use the standard 387 floating point coprocessor present majority of chips and
9276 emulated otherwise. Code compiled with this option will run almost everywhere.
9277 The temporary results are computed in 80bit precision instead of precision
9278 specified by the type resulting in slightly different results compared to most
9279 of other chips. See @option{-ffloat-store} for more detailed description.
9281 This is the default choice for i386 compiler.
9284 Use scalar floating point instructions present in the SSE instruction set.
9285 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9286 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9287 instruction set supports only single precision arithmetics, thus the double and
9288 extended precision arithmetics is still done using 387. Later version, present
9289 only in Pentium4 and the future AMD x86-64 chips supports double precision
9292 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9293 or @option{-msse2} switches to enable SSE extensions and make this option
9294 effective. For the x86-64 compiler, these extensions are enabled by default.
9296 The resulting code should be considerably faster in the majority of cases and avoid
9297 the numerical instability problems of 387 code, but may break some existing
9298 code that expects temporaries to be 80bit.
9300 This is the default choice for the x86-64 compiler.
9303 Attempt to utilize both instruction sets at once. This effectively double the
9304 amount of available registers and on chips with separate execution units for
9305 387 and SSE the execution resources too. Use this option with care, as it is
9306 still experimental, because the GCC register allocator does not model separate
9307 functional units well resulting in instable performance.
9310 @item -masm=@var{dialect}
9311 @opindex masm=@var{dialect}
9312 Output asm instructions using selected @var{dialect}. Supported
9313 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9314 not support @samp{intel}.
9319 @opindex mno-ieee-fp
9320 Control whether or not the compiler uses IEEE floating point
9321 comparisons. These handle correctly the case where the result of a
9322 comparison is unordered.
9325 @opindex msoft-float
9326 Generate output containing library calls for floating point.
9327 @strong{Warning:} the requisite libraries are not part of GCC@.
9328 Normally the facilities of the machine's usual C compiler are used, but
9329 this can't be done directly in cross-compilation. You must make your
9330 own arrangements to provide suitable library functions for
9333 On machines where a function returns floating point results in the 80387
9334 register stack, some floating point opcodes may be emitted even if
9335 @option{-msoft-float} is used.
9337 @item -mno-fp-ret-in-387
9338 @opindex mno-fp-ret-in-387
9339 Do not use the FPU registers for return values of functions.
9341 The usual calling convention has functions return values of types
9342 @code{float} and @code{double} in an FPU register, even if there
9343 is no FPU@. The idea is that the operating system should emulate
9346 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9347 in ordinary CPU registers instead.
9349 @item -mno-fancy-math-387
9350 @opindex mno-fancy-math-387
9351 Some 387 emulators do not support the @code{sin}, @code{cos} and
9352 @code{sqrt} instructions for the 387. Specify this option to avoid
9353 generating those instructions. This option is the default on FreeBSD,
9354 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9355 indicates that the target cpu will always have an FPU and so the
9356 instruction will not need emulation. As of revision 2.6.1, these
9357 instructions are not generated unless you also use the
9358 @option{-funsafe-math-optimizations} switch.
9360 @item -malign-double
9361 @itemx -mno-align-double
9362 @opindex malign-double
9363 @opindex mno-align-double
9364 Control whether GCC aligns @code{double}, @code{long double}, and
9365 @code{long long} variables on a two word boundary or a one word
9366 boundary. Aligning @code{double} variables on a two word boundary will
9367 produce code that runs somewhat faster on a @samp{Pentium} at the
9368 expense of more memory.
9370 @strong{Warning:} if you use the @option{-malign-double} switch,
9371 structures containing the above types will be aligned differently than
9372 the published application binary interface specifications for the 386
9373 and will not be binary compatible with structures in code compiled
9374 without that switch.
9376 @item -m96bit-long-double
9377 @itemx -m128bit-long-double
9378 @opindex m96bit-long-double
9379 @opindex m128bit-long-double
9380 These switches control the size of @code{long double} type. The i386
9381 application binary interface specifies the size to be 96 bits,
9382 so @option{-m96bit-long-double} is the default in 32 bit mode.
9384 Modern architectures (Pentium and newer) would prefer @code{long double}
9385 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9386 conforming to the ABI, this would not be possible. So specifying a
9387 @option{-m128bit-long-double} will align @code{long double}
9388 to a 16 byte boundary by padding the @code{long double} with an additional
9391 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9392 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9394 Notice that neither of these options enable any extra precision over the x87
9395 standard of 80 bits for a @code{long double}.
9397 @strong{Warning:} if you override the default value for your target ABI, the
9398 structures and arrays containing @code{long double} variables will change
9399 their size as well as function calling convention for function taking
9400 @code{long double} will be modified. Hence they will not be binary
9401 compatible with arrays or structures in code compiled without that switch.
9403 @item -mmlarge-data-threshold=@var{number}
9404 @opindex mlarge-data-threshold=@var{number}
9405 When @option{-mcmodel=medium} is specified, the data greater than
9406 @var{threshold} are placed in large data section. This value must be the
9407 same across all object linked into the binary and defaults to 65535.
9410 @itemx -mno-svr3-shlib
9411 @opindex msvr3-shlib
9412 @opindex mno-svr3-shlib
9413 Control whether GCC places uninitialized local variables into the
9414 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9415 into @code{bss}. These options are meaningful only on System V Release 3.
9419 Use a different function-calling convention, in which functions that
9420 take a fixed number of arguments return with the @code{ret} @var{num}
9421 instruction, which pops their arguments while returning. This saves one
9422 instruction in the caller since there is no need to pop the arguments
9425 You can specify that an individual function is called with this calling
9426 sequence with the function attribute @samp{stdcall}. You can also
9427 override the @option{-mrtd} option by using the function attribute
9428 @samp{cdecl}. @xref{Function Attributes}.
9430 @strong{Warning:} this calling convention is incompatible with the one
9431 normally used on Unix, so you cannot use it if you need to call
9432 libraries compiled with the Unix compiler.
9434 Also, you must provide function prototypes for all functions that
9435 take variable numbers of arguments (including @code{printf});
9436 otherwise incorrect code will be generated for calls to those
9439 In addition, seriously incorrect code will result if you call a
9440 function with too many arguments. (Normally, extra arguments are
9441 harmlessly ignored.)
9443 @item -mregparm=@var{num}
9445 Control how many registers are used to pass integer arguments. By
9446 default, no registers are used to pass arguments, and at most 3
9447 registers can be used. You can control this behavior for a specific
9448 function by using the function attribute @samp{regparm}.
9449 @xref{Function Attributes}.
9451 @strong{Warning:} if you use this switch, and
9452 @var{num} is nonzero, then you must build all modules with the same
9453 value, including any libraries. This includes the system libraries and
9457 @opindex msseregparm
9458 Use SSE register passing conventions for float and double arguments
9459 and return values. You can control this behavior for a specific
9460 function by using the function attribute @samp{sseregparm}.
9461 @xref{Function Attributes}.
9463 @strong{Warning:} if you use this switch then you must build all
9464 modules with the same value, including any libraries. This includes
9465 the system libraries and startup modules.
9467 @item -mpreferred-stack-boundary=@var{num}
9468 @opindex mpreferred-stack-boundary
9469 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9470 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9471 the default is 4 (16 bytes or 128 bits), except when optimizing for code
9472 size (@option{-Os}), in which case the default is the minimum correct
9473 alignment (4 bytes for x86, and 8 bytes for x86-64).
9475 On Pentium and PentiumPro, @code{double} and @code{long double} values
9476 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9477 suffer significant run time performance penalties. On Pentium III, the
9478 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
9479 penalties if it is not 16 byte aligned.
9481 To ensure proper alignment of this values on the stack, the stack boundary
9482 must be as aligned as that required by any value stored on the stack.
9483 Further, every function must be generated such that it keeps the stack
9484 aligned. Thus calling a function compiled with a higher preferred
9485 stack boundary from a function compiled with a lower preferred stack
9486 boundary will most likely misalign the stack. It is recommended that
9487 libraries that use callbacks always use the default setting.
9489 This extra alignment does consume extra stack space, and generally
9490 increases code size. Code that is sensitive to stack space usage, such
9491 as embedded systems and operating system kernels, may want to reduce the
9492 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9510 These switches enable or disable the use of instructions in the MMX,
9511 SSE, SSE2 or 3DNow! extended instruction sets. These extensions are
9512 also available as built-in functions: see @ref{X86 Built-in Functions},
9513 for details of the functions enabled and disabled by these switches.
9515 To have SSE/SSE2 instructions generated automatically from floating-point
9516 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9518 These options will enable GCC to use these extended instructions in
9519 generated code, even without @option{-mfpmath=sse}. Applications which
9520 perform runtime CPU detection must compile separate files for each
9521 supported architecture, using the appropriate flags. In particular,
9522 the file containing the CPU detection code should be compiled without
9527 Use special versions of certain libm routines that come with an SSE
9528 ABI and an SSE implementation. Useful together with @option{-mfpmath=sse}
9529 to avoid moving values between SSE registers and the x87 FP stack.
9532 @itemx -mno-push-args
9534 @opindex mno-push-args
9535 Use PUSH operations to store outgoing parameters. This method is shorter
9536 and usually equally fast as method using SUB/MOV operations and is enabled
9537 by default. In some cases disabling it may improve performance because of
9538 improved scheduling and reduced dependencies.
9540 @item -maccumulate-outgoing-args
9541 @opindex maccumulate-outgoing-args
9542 If enabled, the maximum amount of space required for outgoing arguments will be
9543 computed in the function prologue. This is faster on most modern CPUs
9544 because of reduced dependencies, improved scheduling and reduced stack usage
9545 when preferred stack boundary is not equal to 2. The drawback is a notable
9546 increase in code size. This switch implies @option{-mno-push-args}.
9550 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9551 on thread-safe exception handling must compile and link all code with the
9552 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9553 @option{-D_MT}; when linking, it links in a special thread helper library
9554 @option{-lmingwthrd} which cleans up per thread exception handling data.
9556 @item -mno-align-stringops
9557 @opindex mno-align-stringops
9558 Do not align destination of inlined string operations. This switch reduces
9559 code size and improves performance in case the destination is already aligned,
9560 but GCC doesn't know about it.
9562 @item -minline-all-stringops
9563 @opindex minline-all-stringops
9564 By default GCC inlines string operations only when destination is known to be
9565 aligned at least to 4 byte boundary. This enables more inlining, increase code
9566 size, but may improve performance of code that depends on fast memcpy, strlen
9567 and memset for short lengths.
9569 @item -momit-leaf-frame-pointer
9570 @opindex momit-leaf-frame-pointer
9571 Don't keep the frame pointer in a register for leaf functions. This
9572 avoids the instructions to save, set up and restore frame pointers and
9573 makes an extra register available in leaf functions. The option
9574 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9575 which might make debugging harder.
9577 @item -mtls-direct-seg-refs
9578 @itemx -mno-tls-direct-seg-refs
9579 @opindex mtls-direct-seg-refs
9580 Controls whether TLS variables may be accessed with offsets from the
9581 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9582 or whether the thread base pointer must be added. Whether or not this
9583 is legal depends on the operating system, and whether it maps the
9584 segment to cover the entire TLS area.
9586 For systems that use GNU libc, the default is on.
9589 These @samp{-m} switches are supported in addition to the above
9590 on AMD x86-64 processors in 64-bit environments.
9597 Generate code for a 32-bit or 64-bit environment.
9598 The 32-bit environment sets int, long and pointer to 32 bits and
9599 generates code that runs on any i386 system.
9600 The 64-bit environment sets int to 32 bits and long and pointer
9601 to 64 bits and generates code for AMD's x86-64 architecture.
9604 @opindex no-red-zone
9605 Do not use a so called red zone for x86-64 code. The red zone is mandated
9606 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9607 stack pointer that will not be modified by signal or interrupt handlers
9608 and therefore can be used for temporary data without adjusting the stack
9609 pointer. The flag @option{-mno-red-zone} disables this red zone.
9611 @item -mcmodel=small
9612 @opindex mcmodel=small
9613 Generate code for the small code model: the program and its symbols must
9614 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9615 Programs can be statically or dynamically linked. This is the default
9618 @item -mcmodel=kernel
9619 @opindex mcmodel=kernel
9620 Generate code for the kernel code model. The kernel runs in the
9621 negative 2 GB of the address space.
9622 This model has to be used for Linux kernel code.
9624 @item -mcmodel=medium
9625 @opindex mcmodel=medium
9626 Generate code for the medium model: The program is linked in the lower 2
9627 GB of the address space but symbols can be located anywhere in the
9628 address space. Programs can be statically or dynamically linked, but
9629 building of shared libraries are not supported with the medium model.
9631 @item -mcmodel=large
9632 @opindex mcmodel=large
9633 Generate code for the large model: This model makes no assumptions
9634 about addresses and sizes of sections. Currently GCC does not implement
9639 @subsection IA-64 Options
9640 @cindex IA-64 Options
9642 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9646 @opindex mbig-endian
9647 Generate code for a big endian target. This is the default for HP-UX@.
9649 @item -mlittle-endian
9650 @opindex mlittle-endian
9651 Generate code for a little endian target. This is the default for AIX5
9658 Generate (or don't) code for the GNU assembler. This is the default.
9659 @c Also, this is the default if the configure option @option{--with-gnu-as}
9666 Generate (or don't) code for the GNU linker. This is the default.
9667 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9672 Generate code that does not use a global pointer register. The result
9673 is not position independent code, and violates the IA-64 ABI@.
9675 @item -mvolatile-asm-stop
9676 @itemx -mno-volatile-asm-stop
9677 @opindex mvolatile-asm-stop
9678 @opindex mno-volatile-asm-stop
9679 Generate (or don't) a stop bit immediately before and after volatile asm
9682 @item -mregister-names
9683 @itemx -mno-register-names
9684 @opindex mregister-names
9685 @opindex mno-register-names
9686 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9687 the stacked registers. This may make assembler output more readable.
9693 Disable (or enable) optimizations that use the small data section. This may
9694 be useful for working around optimizer bugs.
9697 @opindex mconstant-gp
9698 Generate code that uses a single constant global pointer value. This is
9699 useful when compiling kernel code.
9703 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9704 This is useful when compiling firmware code.
9706 @item -minline-float-divide-min-latency
9707 @opindex minline-float-divide-min-latency
9708 Generate code for inline divides of floating point values
9709 using the minimum latency algorithm.
9711 @item -minline-float-divide-max-throughput
9712 @opindex minline-float-divide-max-throughput
9713 Generate code for inline divides of floating point values
9714 using the maximum throughput algorithm.
9716 @item -minline-int-divide-min-latency
9717 @opindex minline-int-divide-min-latency
9718 Generate code for inline divides of integer values
9719 using the minimum latency algorithm.
9721 @item -minline-int-divide-max-throughput
9722 @opindex minline-int-divide-max-throughput
9723 Generate code for inline divides of integer values
9724 using the maximum throughput algorithm.
9726 @item -minline-sqrt-min-latency
9727 @opindex minline-sqrt-min-latency
9728 Generate code for inline square roots
9729 using the minimum latency algorithm.
9731 @item -minline-sqrt-max-throughput
9732 @opindex minline-sqrt-max-throughput
9733 Generate code for inline square roots
9734 using the maximum throughput algorithm.
9736 @item -mno-dwarf2-asm
9738 @opindex mno-dwarf2-asm
9739 @opindex mdwarf2-asm
9740 Don't (or do) generate assembler code for the DWARF2 line number debugging
9741 info. This may be useful when not using the GNU assembler.
9743 @item -mearly-stop-bits
9744 @itemx -mno-early-stop-bits
9745 @opindex mearly-stop-bits
9746 @opindex mno-early-stop-bits
9747 Allow stop bits to be placed earlier than immediately preceding the
9748 instruction that triggered the stop bit. This can improve instruction
9749 scheduling, but does not always do so.
9751 @item -mfixed-range=@var{register-range}
9752 @opindex mfixed-range
9753 Generate code treating the given register range as fixed registers.
9754 A fixed register is one that the register allocator can not use. This is
9755 useful when compiling kernel code. A register range is specified as
9756 two registers separated by a dash. Multiple register ranges can be
9757 specified separated by a comma.
9759 @item -mtls-size=@var{tls-size}
9761 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9764 @item -mtune=@var{cpu-type}
9766 Tune the instruction scheduling for a particular CPU, Valid values are
9767 itanium, itanium1, merced, itanium2, and mckinley.
9773 Add support for multithreading using the POSIX threads library. This
9774 option sets flags for both the preprocessor and linker. It does
9775 not affect the thread safety of object code produced by the compiler or
9776 that of libraries supplied with it. These are HP-UX specific flags.
9782 Generate code for a 32-bit or 64-bit environment.
9783 The 32-bit environment sets int, long and pointer to 32 bits.
9784 The 64-bit environment sets int to 32 bits and long and pointer
9785 to 64 bits. These are HP-UX specific flags.
9787 @item -mno-sched-br-data-spec
9788 @itemx -msched-br-data-spec
9789 @opindex -mno-sched-br-data-spec
9790 @opindex -msched-br-data-spec
9791 (Dis/En)able data speculative scheduling before reload.
9792 This will result in generation of the ld.a instructions and
9793 the corresponding check instructions (ld.c / chk.a).
9794 The default is 'disable'.
9796 @item -msched-ar-data-spec
9797 @itemx -mno-sched-ar-data-spec
9798 @opindex -msched-ar-data-spec
9799 @opindex -mno-sched-ar-data-spec
9800 (En/Dis)able data speculative scheduling after reload.
9801 This will result in generation of the ld.a instructions and
9802 the corresponding check instructions (ld.c / chk.a).
9803 The default is 'enable'.
9805 @item -mno-sched-control-spec
9806 @itemx -msched-control-spec
9807 @opindex -mno-sched-control-spec
9808 @opindex -msched-control-spec
9809 (Dis/En)able control speculative scheduling. This feature is
9810 available only during region scheduling (i.e. before reload).
9811 This will result in generation of the ld.s instructions and
9812 the corresponding check instructions chk.s .
9813 The default is 'disable'.
9815 @item -msched-br-in-data-spec
9816 @itemx -mno-sched-br-in-data-spec
9817 @opindex -msched-br-in-data-spec
9818 @opindex -mno-sched-br-in-data-spec
9819 (En/Dis)able speculative scheduling of the instructions that
9820 are dependent on the data speculative loads before reload.
9821 This is effective only with @option{-msched-br-data-spec} enabled.
9822 The default is 'enable'.
9824 @item -msched-ar-in-data-spec
9825 @itemx -mno-sched-ar-in-data-spec
9826 @opindex -msched-ar-in-data-spec
9827 @opindex -mno-sched-ar-in-data-spec
9828 (En/Dis)able speculative scheduling of the instructions that
9829 are dependent on the data speculative loads after reload.
9830 This is effective only with @option{-msched-ar-data-spec} enabled.
9831 The default is 'enable'.
9833 @item -msched-in-control-spec
9834 @itemx -mno-sched-in-control-spec
9835 @opindex -msched-in-control-spec
9836 @opindex -mno-sched-in-control-spec
9837 (En/Dis)able speculative scheduling of the instructions that
9838 are dependent on the control speculative loads.
9839 This is effective only with @option{-msched-control-spec} enabled.
9840 The default is 'enable'.
9843 @itemx -mno-sched-ldc
9844 @opindex -msched-ldc
9845 @opindex -mno-sched-ldc
9846 (En/Dis)able use of simple data speculation checks ld.c .
9847 If disabled, only chk.a instructions will be emitted to check
9848 data speculative loads.
9849 The default is 'enable'.
9851 @item -mno-sched-control-ldc
9852 @itemx -msched-control-ldc
9853 @opindex -mno-sched-control-ldc
9854 @opindex -msched-control-ldc
9855 (Dis/En)able use of ld.c instructions to check control speculative loads.
9856 If enabled, in case of control speculative load with no speculatively
9857 scheduled dependent instructions this load will be emitted as ld.sa and
9858 ld.c will be used to check it.
9859 The default is 'disable'.
9861 @item -mno-sched-spec-verbose
9862 @itemx -msched-spec-verbose
9863 @opindex -mno-sched-spec-verbose
9864 @opindex -msched-spec-verbose
9865 (Dis/En)able printing of the information about speculative motions.
9867 @item -mno-sched-prefer-non-data-spec-insns
9868 @itemx -msched-prefer-non-data-spec-insns
9869 @opindex -mno-sched-prefer-non-data-spec-insns
9870 @opindex -msched-prefer-non-data-spec-insns
9871 If enabled, data speculative instructions will be chosen for schedule
9872 only if there are no other choices at the moment. This will make
9873 the use of the data speculation much more conservative.
9874 The default is 'disable'.
9876 @item -mno-sched-prefer-non-control-spec-insns
9877 @itemx -msched-prefer-non-control-spec-insns
9878 @opindex -mno-sched-prefer-non-control-spec-insns
9879 @opindex -msched-prefer-non-control-spec-insns
9880 If enabled, control speculative instructions will be chosen for schedule
9881 only if there are no other choices at the moment. This will make
9882 the use of the control speculation much more conservative.
9883 The default is 'disable'.
9885 @item -mno-sched-count-spec-in-critical-path
9886 @itemx -msched-count-spec-in-critical-path
9887 @opindex -mno-sched-count-spec-in-critical-path
9888 @opindex -msched-count-spec-in-critical-path
9889 If enabled, speculative dependencies will be considered during
9890 computation of the instructions priorities. This will make the use of the
9891 speculation a bit more conservative.
9892 The default is 'disable'.
9897 @subsection M32C Options
9898 @cindex M32C options
9901 @item -mcpu=@var{name}
9903 Select the CPU for which code is generated. @var{name} may be one of
9904 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9905 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9910 Specifies that the program will be run on the simulator. This causes
9911 an alternate runtime library to be linked in which supports, for
9912 example, file I/O. You must not use this option when generating
9913 programs that will run on real hardware; you must provide your own
9914 runtime library for whatever I/O functions are needed.
9916 @item -memregs=@var{number}
9918 Specifies the number of memory-based pseudo-registers GCC will use
9919 during code generation. These pseudo-registers will be used like real
9920 registers, so there is a tradeoff between GCC's ability to fit the
9921 code into available registers, and the performance penalty of using
9922 memory instead of registers. Note that all modules in a program must
9923 be compiled with the same value for this option. Because of that, you
9924 must not use this option with the default runtime libraries gcc
9929 @node M32R/D Options
9930 @subsection M32R/D Options
9931 @cindex M32R/D options
9933 These @option{-m} options are defined for Renesas M32R/D architectures:
9938 Generate code for the M32R/2@.
9942 Generate code for the M32R/X@.
9946 Generate code for the M32R@. This is the default.
9949 @opindex mmodel=small
9950 Assume all objects live in the lower 16MB of memory (so that their addresses
9951 can be loaded with the @code{ld24} instruction), and assume all subroutines
9952 are reachable with the @code{bl} instruction.
9953 This is the default.
9955 The addressability of a particular object can be set with the
9956 @code{model} attribute.
9958 @item -mmodel=medium
9959 @opindex mmodel=medium
9960 Assume objects may be anywhere in the 32-bit address space (the compiler
9961 will generate @code{seth/add3} instructions to load their addresses), and
9962 assume all subroutines are reachable with the @code{bl} instruction.
9965 @opindex mmodel=large
9966 Assume objects may be anywhere in the 32-bit address space (the compiler
9967 will generate @code{seth/add3} instructions to load their addresses), and
9968 assume subroutines may not be reachable with the @code{bl} instruction
9969 (the compiler will generate the much slower @code{seth/add3/jl}
9970 instruction sequence).
9973 @opindex msdata=none
9974 Disable use of the small data area. Variables will be put into
9975 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
9976 @code{section} attribute has been specified).
9977 This is the default.
9979 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
9980 Objects may be explicitly put in the small data area with the
9981 @code{section} attribute using one of these sections.
9984 @opindex msdata=sdata
9985 Put small global and static data in the small data area, but do not
9986 generate special code to reference them.
9990 Put small global and static data in the small data area, and generate
9991 special instructions to reference them.
9995 @cindex smaller data references
9996 Put global and static objects less than or equal to @var{num} bytes
9997 into the small data or bss sections instead of the normal data or bss
9998 sections. The default value of @var{num} is 8.
9999 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10000 for this option to have any effect.
10002 All modules should be compiled with the same @option{-G @var{num}} value.
10003 Compiling with different values of @var{num} may or may not work; if it
10004 doesn't the linker will give an error message---incorrect code will not be
10009 Makes the M32R specific code in the compiler display some statistics
10010 that might help in debugging programs.
10012 @item -malign-loops
10013 @opindex malign-loops
10014 Align all loops to a 32-byte boundary.
10016 @item -mno-align-loops
10017 @opindex mno-align-loops
10018 Do not enforce a 32-byte alignment for loops. This is the default.
10020 @item -missue-rate=@var{number}
10021 @opindex missue-rate=@var{number}
10022 Issue @var{number} instructions per cycle. @var{number} can only be 1
10025 @item -mbranch-cost=@var{number}
10026 @opindex mbranch-cost=@var{number}
10027 @var{number} can only be 1 or 2. If it is 1 then branches will be
10028 preferred over conditional code, if it is 2, then the opposite will
10031 @item -mflush-trap=@var{number}
10032 @opindex mflush-trap=@var{number}
10033 Specifies the trap number to use to flush the cache. The default is
10034 12. Valid numbers are between 0 and 15 inclusive.
10036 @item -mno-flush-trap
10037 @opindex mno-flush-trap
10038 Specifies that the cache cannot be flushed by using a trap.
10040 @item -mflush-func=@var{name}
10041 @opindex mflush-func=@var{name}
10042 Specifies the name of the operating system function to call to flush
10043 the cache. The default is @emph{_flush_cache}, but a function call
10044 will only be used if a trap is not available.
10046 @item -mno-flush-func
10047 @opindex mno-flush-func
10048 Indicates that there is no OS function for flushing the cache.
10052 @node M680x0 Options
10053 @subsection M680x0 Options
10054 @cindex M680x0 options
10056 These are the @samp{-m} options defined for the 68000 series. The default
10057 values for these options depends on which style of 68000 was selected when
10058 the compiler was configured; the defaults for the most common choices are
10066 Generate output for a 68000. This is the default
10067 when the compiler is configured for 68000-based systems.
10069 Use this option for microcontrollers with a 68000 or EC000 core,
10070 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10076 Generate output for a 68020. This is the default
10077 when the compiler is configured for 68020-based systems.
10081 Generate output containing 68881 instructions for floating point.
10082 This is the default for most 68020 systems unless @option{--nfp} was
10083 specified when the compiler was configured.
10087 Generate output for a 68030. This is the default when the compiler is
10088 configured for 68030-based systems.
10092 Generate output for a 68040. This is the default when the compiler is
10093 configured for 68040-based systems.
10095 This option inhibits the use of 68881/68882 instructions that have to be
10096 emulated by software on the 68040. Use this option if your 68040 does not
10097 have code to emulate those instructions.
10101 Generate output for a 68060. This is the default when the compiler is
10102 configured for 68060-based systems.
10104 This option inhibits the use of 68020 and 68881/68882 instructions that
10105 have to be emulated by software on the 68060. Use this option if your 68060
10106 does not have code to emulate those instructions.
10110 Generate output for a CPU32. This is the default
10111 when the compiler is configured for CPU32-based systems.
10113 Use this option for microcontrollers with a
10114 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10115 68336, 68340, 68341, 68349 and 68360.
10119 Generate output for a 520X ``coldfire'' family cpu. This is the default
10120 when the compiler is configured for 520X-based systems.
10122 Use this option for microcontroller with a 5200 core, including
10123 the MCF5202, MCF5203, MCF5204 and MCF5202.
10127 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
10128 This includes use of hardware floating point instructions.
10132 Generate output for a 68040, without using any of the new instructions.
10133 This results in code which can run relatively efficiently on either a
10134 68020/68881 or a 68030 or a 68040. The generated code does use the
10135 68881 instructions that are emulated on the 68040.
10139 Generate output for a 68060, without using any of the new instructions.
10140 This results in code which can run relatively efficiently on either a
10141 68020/68881 or a 68030 or a 68040. The generated code does use the
10142 68881 instructions that are emulated on the 68060.
10145 @opindex msoft-float
10146 Generate output containing library calls for floating point.
10147 @strong{Warning:} the requisite libraries are not available for all m68k
10148 targets. Normally the facilities of the machine's usual C compiler are
10149 used, but this can't be done directly in cross-compilation. You must
10150 make your own arrangements to provide suitable library functions for
10151 cross-compilation. The embedded targets @samp{m68k-*-aout} and
10152 @samp{m68k-*-coff} do provide software floating point support.
10156 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10157 Additionally, parameters passed on the stack are also aligned to a
10158 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10161 @opindex mnobitfield
10162 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10163 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10167 Do use the bit-field instructions. The @option{-m68020} option implies
10168 @option{-mbitfield}. This is the default if you use a configuration
10169 designed for a 68020.
10173 Use a different function-calling convention, in which functions
10174 that take a fixed number of arguments return with the @code{rtd}
10175 instruction, which pops their arguments while returning. This
10176 saves one instruction in the caller since there is no need to pop
10177 the arguments there.
10179 This calling convention is incompatible with the one normally
10180 used on Unix, so you cannot use it if you need to call libraries
10181 compiled with the Unix compiler.
10183 Also, you must provide function prototypes for all functions that
10184 take variable numbers of arguments (including @code{printf});
10185 otherwise incorrect code will be generated for calls to those
10188 In addition, seriously incorrect code will result if you call a
10189 function with too many arguments. (Normally, extra arguments are
10190 harmlessly ignored.)
10192 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10193 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10196 @itemx -mno-align-int
10197 @opindex malign-int
10198 @opindex mno-align-int
10199 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10200 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10201 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10202 Aligning variables on 32-bit boundaries produces code that runs somewhat
10203 faster on processors with 32-bit busses at the expense of more memory.
10205 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10206 align structures containing the above types differently than
10207 most published application binary interface specifications for the m68k.
10211 Use the pc-relative addressing mode of the 68000 directly, instead of
10212 using a global offset table. At present, this option implies @option{-fpic},
10213 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10214 not presently supported with @option{-mpcrel}, though this could be supported for
10215 68020 and higher processors.
10217 @item -mno-strict-align
10218 @itemx -mstrict-align
10219 @opindex mno-strict-align
10220 @opindex mstrict-align
10221 Do not (do) assume that unaligned memory references will be handled by
10225 Generate code that allows the data segment to be located in a different
10226 area of memory from the text segment. This allows for execute in place in
10227 an environment without virtual memory management. This option implies
10230 @item -mno-sep-data
10231 Generate code that assumes that the data segment follows the text segment.
10232 This is the default.
10234 @item -mid-shared-library
10235 Generate code that supports shared libraries via the library ID method.
10236 This allows for execute in place and shared libraries in an environment
10237 without virtual memory management. This option implies @option{-fPIC}.
10239 @item -mno-id-shared-library
10240 Generate code that doesn't assume ID based shared libraries are being used.
10241 This is the default.
10243 @item -mshared-library-id=n
10244 Specified the identification number of the ID based shared library being
10245 compiled. Specifying a value of 0 will generate more compact code, specifying
10246 other values will force the allocation of that number to the current
10247 library but is no more space or time efficient than omitting this option.
10251 @node M68hc1x Options
10252 @subsection M68hc1x Options
10253 @cindex M68hc1x options
10255 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10256 microcontrollers. The default values for these options depends on
10257 which style of microcontroller was selected when the compiler was configured;
10258 the defaults for the most common choices are given below.
10265 Generate output for a 68HC11. This is the default
10266 when the compiler is configured for 68HC11-based systems.
10272 Generate output for a 68HC12. This is the default
10273 when the compiler is configured for 68HC12-based systems.
10279 Generate output for a 68HCS12.
10281 @item -mauto-incdec
10282 @opindex mauto-incdec
10283 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10290 Enable the use of 68HC12 min and max instructions.
10293 @itemx -mno-long-calls
10294 @opindex mlong-calls
10295 @opindex mno-long-calls
10296 Treat all calls as being far away (near). If calls are assumed to be
10297 far away, the compiler will use the @code{call} instruction to
10298 call a function and the @code{rtc} instruction for returning.
10302 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10304 @item -msoft-reg-count=@var{count}
10305 @opindex msoft-reg-count
10306 Specify the number of pseudo-soft registers which are used for the
10307 code generation. The maximum number is 32. Using more pseudo-soft
10308 register may or may not result in better code depending on the program.
10309 The default is 4 for 68HC11 and 2 for 68HC12.
10313 @node MCore Options
10314 @subsection MCore Options
10315 @cindex MCore options
10317 These are the @samp{-m} options defined for the Motorola M*Core
10323 @itemx -mno-hardlit
10325 @opindex mno-hardlit
10326 Inline constants into the code stream if it can be done in two
10327 instructions or less.
10333 Use the divide instruction. (Enabled by default).
10335 @item -mrelax-immediate
10336 @itemx -mno-relax-immediate
10337 @opindex mrelax-immediate
10338 @opindex mno-relax-immediate
10339 Allow arbitrary sized immediates in bit operations.
10341 @item -mwide-bitfields
10342 @itemx -mno-wide-bitfields
10343 @opindex mwide-bitfields
10344 @opindex mno-wide-bitfields
10345 Always treat bit-fields as int-sized.
10347 @item -m4byte-functions
10348 @itemx -mno-4byte-functions
10349 @opindex m4byte-functions
10350 @opindex mno-4byte-functions
10351 Force all functions to be aligned to a four byte boundary.
10353 @item -mcallgraph-data
10354 @itemx -mno-callgraph-data
10355 @opindex mcallgraph-data
10356 @opindex mno-callgraph-data
10357 Emit callgraph information.
10360 @itemx -mno-slow-bytes
10361 @opindex mslow-bytes
10362 @opindex mno-slow-bytes
10363 Prefer word access when reading byte quantities.
10365 @item -mlittle-endian
10366 @itemx -mbig-endian
10367 @opindex mlittle-endian
10368 @opindex mbig-endian
10369 Generate code for a little endian target.
10375 Generate code for the 210 processor.
10379 @subsection MIPS Options
10380 @cindex MIPS options
10386 Generate big-endian code.
10390 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10393 @item -march=@var{arch}
10395 Generate code that will run on @var{arch}, which can be the name of a
10396 generic MIPS ISA, or the name of a particular processor.
10398 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10399 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10400 The processor names are:
10401 @samp{4kc}, @samp{4km}, @samp{4kp},
10402 @samp{5kc}, @samp{5kf},
10404 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10407 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10408 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10409 @samp{rm7000}, @samp{rm9000},
10412 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10413 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10414 The special value @samp{from-abi} selects the
10415 most compatible architecture for the selected ABI (that is,
10416 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10418 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10419 (for example, @samp{-march=r2k}). Prefixes are optional, and
10420 @samp{vr} may be written @samp{r}.
10422 GCC defines two macros based on the value of this option. The first
10423 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10424 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10425 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10426 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10427 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10429 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10430 above. In other words, it will have the full prefix and will not
10431 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10432 the macro names the resolved architecture (either @samp{"mips1"} or
10433 @samp{"mips3"}). It names the default architecture when no
10434 @option{-march} option is given.
10436 @item -mtune=@var{arch}
10438 Optimize for @var{arch}. Among other things, this option controls
10439 the way instructions are scheduled, and the perceived cost of arithmetic
10440 operations. The list of @var{arch} values is the same as for
10443 When this option is not used, GCC will optimize for the processor
10444 specified by @option{-march}. By using @option{-march} and
10445 @option{-mtune} together, it is possible to generate code that will
10446 run on a family of processors, but optimize the code for one
10447 particular member of that family.
10449 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10450 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10451 @samp{-march} ones described above.
10455 Equivalent to @samp{-march=mips1}.
10459 Equivalent to @samp{-march=mips2}.
10463 Equivalent to @samp{-march=mips3}.
10467 Equivalent to @samp{-march=mips4}.
10471 Equivalent to @samp{-march=mips32}.
10475 Equivalent to @samp{-march=mips32r2}.
10479 Equivalent to @samp{-march=mips64}.
10484 @opindex mno-mips16
10485 Generate (do not generate) MIPS16 code. If GCC is targetting a
10486 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10498 Generate code for the given ABI@.
10500 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10501 generates 64-bit code when you select a 64-bit architecture, but you
10502 can use @option{-mgp32} to get 32-bit code instead.
10504 For information about the O64 ABI, see
10505 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10508 @itemx -mno-abicalls
10510 @opindex mno-abicalls
10511 Generate (do not generate) code that is suitable for SVR4-style
10512 dynamic objects. @option{-mabicalls} is the default for SVR4-based
10517 Generate (do not generate) code that is fully position-independent,
10518 and that can therefore be linked into shared libraries. This option
10519 only affects @option{-mabicalls}.
10521 All @option{-mabicalls} code has traditionally been position-independent,
10522 regardless of options like @option{-fPIC} and @option{-fpic}. However,
10523 as an extension, the GNU toolchain allows executables to use absolute
10524 accesses for locally-binding symbols. It can also use shorter GP
10525 initialization sequences and generate direct calls to locally-defined
10526 functions. This mode is selected by @option{-mno-shared}.
10528 @option{-mno-shared} depends on binutils 2.16 or higher and generates
10529 objects that can only be linked by the GNU linker. However, the option
10530 does not affect the ABI of the final executable; it only affects the ABI
10531 of relocatable objects. Using @option{-mno-shared} will generally make
10532 executables both smaller and quicker.
10534 @option{-mshared} is the default.
10540 Lift (do not lift) the usual restrictions on the size of the global
10543 GCC normally uses a single instruction to load values from the GOT@.
10544 While this is relatively efficient, it will only work if the GOT
10545 is smaller than about 64k. Anything larger will cause the linker
10546 to report an error such as:
10548 @cindex relocation truncated to fit (MIPS)
10550 relocation truncated to fit: R_MIPS_GOT16 foobar
10553 If this happens, you should recompile your code with @option{-mxgot}.
10554 It should then work with very large GOTs, although it will also be
10555 less efficient, since it will take three instructions to fetch the
10556 value of a global symbol.
10558 Note that some linkers can create multiple GOTs. If you have such a
10559 linker, you should only need to use @option{-mxgot} when a single object
10560 file accesses more than 64k's worth of GOT entries. Very few do.
10562 These options have no effect unless GCC is generating position
10567 Assume that general-purpose registers are 32 bits wide.
10571 Assume that general-purpose registers are 64 bits wide.
10575 Assume that floating-point registers are 32 bits wide.
10579 Assume that floating-point registers are 64 bits wide.
10582 @opindex mhard-float
10583 Use floating-point coprocessor instructions.
10586 @opindex msoft-float
10587 Do not use floating-point coprocessor instructions. Implement
10588 floating-point calculations using library calls instead.
10590 @item -msingle-float
10591 @opindex msingle-float
10592 Assume that the floating-point coprocessor only supports single-precision
10595 @itemx -mdouble-float
10596 @opindex mdouble-float
10597 Assume that the floating-point coprocessor supports double-precision
10598 operations. This is the default.
10604 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10606 @itemx -mpaired-single
10607 @itemx -mno-paired-single
10608 @opindex mpaired-single
10609 @opindex mno-paired-single
10610 Use (do not use) paired-single floating-point instructions.
10611 @xref{MIPS Paired-Single Support}. This option can only be used
10612 when generating 64-bit code and requires hardware floating-point
10613 support to be enabled.
10618 @opindex mno-mips3d
10619 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10620 The option @option{-mips3d} implies @option{-mpaired-single}.
10624 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10625 an explanation of the default and the way that the pointer size is
10630 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10632 The default size of @code{int}s, @code{long}s and pointers depends on
10633 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10634 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10635 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10636 or the same size as integer registers, whichever is smaller.
10642 Assume (do not assume) that all symbols have 32-bit values, regardless
10643 of the selected ABI@. This option is useful in combination with
10644 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10645 to generate shorter and faster references to symbolic addresses.
10649 @cindex smaller data references (MIPS)
10650 @cindex gp-relative references (MIPS)
10651 Put global and static items less than or equal to @var{num} bytes into
10652 the small data or bss section instead of the normal data or bss section.
10653 This allows the data to be accessed using a single instruction.
10655 All modules should be compiled with the same @option{-G @var{num}}
10658 @item -membedded-data
10659 @itemx -mno-embedded-data
10660 @opindex membedded-data
10661 @opindex mno-embedded-data
10662 Allocate variables to the read-only data section first if possible, then
10663 next in the small data section if possible, otherwise in data. This gives
10664 slightly slower code than the default, but reduces the amount of RAM required
10665 when executing, and thus may be preferred for some embedded systems.
10667 @item -muninit-const-in-rodata
10668 @itemx -mno-uninit-const-in-rodata
10669 @opindex muninit-const-in-rodata
10670 @opindex mno-uninit-const-in-rodata
10671 Put uninitialized @code{const} variables in the read-only data section.
10672 This option is only meaningful in conjunction with @option{-membedded-data}.
10674 @item -msplit-addresses
10675 @itemx -mno-split-addresses
10676 @opindex msplit-addresses
10677 @opindex mno-split-addresses
10678 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10679 relocation operators. This option has been superseded by
10680 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10682 @item -mexplicit-relocs
10683 @itemx -mno-explicit-relocs
10684 @opindex mexplicit-relocs
10685 @opindex mno-explicit-relocs
10686 Use (do not use) assembler relocation operators when dealing with symbolic
10687 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10688 is to use assembler macros instead.
10690 @option{-mexplicit-relocs} is the default if GCC was configured
10691 to use an assembler that supports relocation operators.
10693 @item -mcheck-zero-division
10694 @itemx -mno-check-zero-division
10695 @opindex mcheck-zero-division
10696 @opindex mno-check-zero-division
10697 Trap (do not trap) on integer division by zero. The default is
10698 @option{-mcheck-zero-division}.
10700 @item -mdivide-traps
10701 @itemx -mdivide-breaks
10702 @opindex mdivide-traps
10703 @opindex mdivide-breaks
10704 MIPS systems check for division by zero by generating either a
10705 conditional trap or a break instruction. Using traps results in
10706 smaller code, but is only supported on MIPS II and later. Also, some
10707 versions of the Linux kernel have a bug that prevents trap from
10708 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10709 allow conditional traps on architectures that support them and
10710 @option{-mdivide-breaks} to force the use of breaks.
10712 The default is usually @option{-mdivide-traps}, but this can be
10713 overridden at configure time using @option{--with-divide=breaks}.
10714 Divide-by-zero checks can be completely disabled using
10715 @option{-mno-check-zero-division}.
10720 @opindex mno-memcpy
10721 Force (do not force) the use of @code{memcpy()} for non-trivial block
10722 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10723 most constant-sized copies.
10726 @itemx -mno-long-calls
10727 @opindex mlong-calls
10728 @opindex mno-long-calls
10729 Disable (do not disable) use of the @code{jal} instruction. Calling
10730 functions using @code{jal} is more efficient but requires the caller
10731 and callee to be in the same 256 megabyte segment.
10733 This option has no effect on abicalls code. The default is
10734 @option{-mno-long-calls}.
10740 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10741 instructions, as provided by the R4650 ISA@.
10744 @itemx -mno-fused-madd
10745 @opindex mfused-madd
10746 @opindex mno-fused-madd
10747 Enable (disable) use of the floating point multiply-accumulate
10748 instructions, when they are available. The default is
10749 @option{-mfused-madd}.
10751 When multiply-accumulate instructions are used, the intermediate
10752 product is calculated to infinite precision and is not subject to
10753 the FCSR Flush to Zero bit. This may be undesirable in some
10758 Tell the MIPS assembler to not run its preprocessor over user
10759 assembler files (with a @samp{.s} suffix) when assembling them.
10762 @itemx -mno-fix-r4000
10763 @opindex mfix-r4000
10764 @opindex mno-fix-r4000
10765 Work around certain R4000 CPU errata:
10768 A double-word or a variable shift may give an incorrect result if executed
10769 immediately after starting an integer division.
10771 A double-word or a variable shift may give an incorrect result if executed
10772 while an integer multiplication is in progress.
10774 An integer division may give an incorrect result if started in a delay slot
10775 of a taken branch or a jump.
10779 @itemx -mno-fix-r4400
10780 @opindex mfix-r4400
10781 @opindex mno-fix-r4400
10782 Work around certain R4400 CPU errata:
10785 A double-word or a variable shift may give an incorrect result if executed
10786 immediately after starting an integer division.
10790 @itemx -mno-fix-vr4120
10791 @opindex mfix-vr4120
10792 Work around certain VR4120 errata:
10795 @code{dmultu} does not always produce the correct result.
10797 @code{div} and @code{ddiv} do not always produce the correct result if one
10798 of the operands is negative.
10800 The workarounds for the division errata rely on special functions in
10801 @file{libgcc.a}. At present, these functions are only provided by
10802 the @code{mips64vr*-elf} configurations.
10804 Other VR4120 errata require a nop to be inserted between certain pairs of
10805 instructions. These errata are handled by the assembler, not by GCC itself.
10808 @opindex mfix-vr4130
10809 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10810 workarounds are implemented by the assembler rather than by GCC,
10811 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10812 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10813 instructions are available instead.
10816 @itemx -mno-fix-sb1
10818 Work around certain SB-1 CPU core errata.
10819 (This flag currently works around the SB-1 revision 2
10820 ``F1'' and ``F2'' floating point errata.)
10822 @item -mflush-func=@var{func}
10823 @itemx -mno-flush-func
10824 @opindex mflush-func
10825 Specifies the function to call to flush the I and D caches, or to not
10826 call any such function. If called, the function must take the same
10827 arguments as the common @code{_flush_func()}, that is, the address of the
10828 memory range for which the cache is being flushed, the size of the
10829 memory range, and the number 3 (to flush both caches). The default
10830 depends on the target GCC was configured for, but commonly is either
10831 @samp{_flush_func} or @samp{__cpu_flush}.
10833 @item -mbranch-likely
10834 @itemx -mno-branch-likely
10835 @opindex mbranch-likely
10836 @opindex mno-branch-likely
10837 Enable or disable use of Branch Likely instructions, regardless of the
10838 default for the selected architecture. By default, Branch Likely
10839 instructions may be generated if they are supported by the selected
10840 architecture. An exception is for the MIPS32 and MIPS64 architectures
10841 and processors which implement those architectures; for those, Branch
10842 Likely instructions will not be generated by default because the MIPS32
10843 and MIPS64 architectures specifically deprecate their use.
10845 @item -mfp-exceptions
10846 @itemx -mno-fp-exceptions
10847 @opindex mfp-exceptions
10848 Specifies whether FP exceptions are enabled. This affects how we schedule
10849 FP instructions for some processors. The default is that FP exceptions are
10852 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10853 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10856 @item -mvr4130-align
10857 @itemx -mno-vr4130-align
10858 @opindex mvr4130-align
10859 The VR4130 pipeline is two-way superscalar, but can only issue two
10860 instructions together if the first one is 8-byte aligned. When this
10861 option is enabled, GCC will align pairs of instructions that it
10862 thinks should execute in parallel.
10864 This option only has an effect when optimizing for the VR4130.
10865 It normally makes code faster, but at the expense of making it bigger.
10866 It is enabled by default at optimization level @option{-O3}.
10870 @subsection MMIX Options
10871 @cindex MMIX Options
10873 These options are defined for the MMIX:
10877 @itemx -mno-libfuncs
10879 @opindex mno-libfuncs
10880 Specify that intrinsic library functions are being compiled, passing all
10881 values in registers, no matter the size.
10884 @itemx -mno-epsilon
10886 @opindex mno-epsilon
10887 Generate floating-point comparison instructions that compare with respect
10888 to the @code{rE} epsilon register.
10890 @item -mabi=mmixware
10892 @opindex mabi-mmixware
10894 Generate code that passes function parameters and return values that (in
10895 the called function) are seen as registers @code{$0} and up, as opposed to
10896 the GNU ABI which uses global registers @code{$231} and up.
10898 @item -mzero-extend
10899 @itemx -mno-zero-extend
10900 @opindex mzero-extend
10901 @opindex mno-zero-extend
10902 When reading data from memory in sizes shorter than 64 bits, use (do not
10903 use) zero-extending load instructions by default, rather than
10904 sign-extending ones.
10907 @itemx -mno-knuthdiv
10909 @opindex mno-knuthdiv
10910 Make the result of a division yielding a remainder have the same sign as
10911 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10912 remainder follows the sign of the dividend. Both methods are
10913 arithmetically valid, the latter being almost exclusively used.
10915 @item -mtoplevel-symbols
10916 @itemx -mno-toplevel-symbols
10917 @opindex mtoplevel-symbols
10918 @opindex mno-toplevel-symbols
10919 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10920 code can be used with the @code{PREFIX} assembly directive.
10924 Generate an executable in the ELF format, rather than the default
10925 @samp{mmo} format used by the @command{mmix} simulator.
10927 @item -mbranch-predict
10928 @itemx -mno-branch-predict
10929 @opindex mbranch-predict
10930 @opindex mno-branch-predict
10931 Use (do not use) the probable-branch instructions, when static branch
10932 prediction indicates a probable branch.
10934 @item -mbase-addresses
10935 @itemx -mno-base-addresses
10936 @opindex mbase-addresses
10937 @opindex mno-base-addresses
10938 Generate (do not generate) code that uses @emph{base addresses}. Using a
10939 base address automatically generates a request (handled by the assembler
10940 and the linker) for a constant to be set up in a global register. The
10941 register is used for one or more base address requests within the range 0
10942 to 255 from the value held in the register. The generally leads to short
10943 and fast code, but the number of different data items that can be
10944 addressed is limited. This means that a program that uses lots of static
10945 data may require @option{-mno-base-addresses}.
10947 @item -msingle-exit
10948 @itemx -mno-single-exit
10949 @opindex msingle-exit
10950 @opindex mno-single-exit
10951 Force (do not force) generated code to have a single exit point in each
10955 @node MN10300 Options
10956 @subsection MN10300 Options
10957 @cindex MN10300 options
10959 These @option{-m} options are defined for Matsushita MN10300 architectures:
10964 Generate code to avoid bugs in the multiply instructions for the MN10300
10965 processors. This is the default.
10967 @item -mno-mult-bug
10968 @opindex mno-mult-bug
10969 Do not generate code to avoid bugs in the multiply instructions for the
10970 MN10300 processors.
10974 Generate code which uses features specific to the AM33 processor.
10978 Do not generate code which uses features specific to the AM33 processor. This
10981 @item -mreturn-pointer-on-d0
10982 @opindex mreturn-pointer-on-d0
10983 When generating a function which returns a pointer, return the pointer
10984 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
10985 only in a0, and attempts to call such functions without a prototype
10986 would result in errors. Note that this option is on by default; use
10987 @option{-mno-return-pointer-on-d0} to disable it.
10991 Do not link in the C run-time initialization object file.
10995 Indicate to the linker that it should perform a relaxation optimization pass
10996 to shorten branches, calls and absolute memory addresses. This option only
10997 has an effect when used on the command line for the final link step.
10999 This option makes symbolic debugging impossible.
11003 @subsection MT Options
11006 These @option{-m} options are defined for Morpho MT architectures:
11010 @item -march=@var{cpu-type}
11012 Generate code that will run on @var{cpu-type}, which is the name of a system
11013 representing a certain processor type. Possible values for
11014 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11015 @samp{ms1-16-003} and @samp{ms2}.
11017 When this option is not used, the default is @option{-march=ms1-16-002}.
11021 Use byte loads and stores when generating code.
11025 Do not use byte loads and stores when generating code.
11029 Use simulator runtime
11033 Do not link in the C run-time initialization object file
11034 @file{crti.o}. Other run-time initialization and termination files
11035 such as @file{startup.o} and @file{exit.o} are still included on the
11036 linker command line.
11040 @node PDP-11 Options
11041 @subsection PDP-11 Options
11042 @cindex PDP-11 Options
11044 These options are defined for the PDP-11:
11049 Use hardware FPP floating point. This is the default. (FIS floating
11050 point on the PDP-11/40 is not supported.)
11053 @opindex msoft-float
11054 Do not use hardware floating point.
11058 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11062 Return floating-point results in memory. This is the default.
11066 Generate code for a PDP-11/40.
11070 Generate code for a PDP-11/45. This is the default.
11074 Generate code for a PDP-11/10.
11076 @item -mbcopy-builtin
11077 @opindex bcopy-builtin
11078 Use inline @code{movmemhi} patterns for copying memory. This is the
11083 Do not use inline @code{movmemhi} patterns for copying memory.
11089 Use 16-bit @code{int}. This is the default.
11095 Use 32-bit @code{int}.
11098 @itemx -mno-float32
11100 @opindex mno-float32
11101 Use 64-bit @code{float}. This is the default.
11104 @itemx -mno-float64
11106 @opindex mno-float64
11107 Use 32-bit @code{float}.
11111 Use @code{abshi2} pattern. This is the default.
11115 Do not use @code{abshi2} pattern.
11117 @item -mbranch-expensive
11118 @opindex mbranch-expensive
11119 Pretend that branches are expensive. This is for experimenting with
11120 code generation only.
11122 @item -mbranch-cheap
11123 @opindex mbranch-cheap
11124 Do not pretend that branches are expensive. This is the default.
11128 Generate code for a system with split I&D@.
11132 Generate code for a system without split I&D@. This is the default.
11136 Use Unix assembler syntax. This is the default when configured for
11137 @samp{pdp11-*-bsd}.
11141 Use DEC assembler syntax. This is the default when configured for any
11142 PDP-11 target other than @samp{pdp11-*-bsd}.
11145 @node PowerPC Options
11146 @subsection PowerPC Options
11147 @cindex PowerPC options
11149 These are listed under @xref{RS/6000 and PowerPC Options}.
11151 @node RS/6000 and PowerPC Options
11152 @subsection IBM RS/6000 and PowerPC Options
11153 @cindex RS/6000 and PowerPC Options
11154 @cindex IBM RS/6000 and PowerPC Options
11156 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11163 @itemx -mno-powerpc
11164 @itemx -mpowerpc-gpopt
11165 @itemx -mno-powerpc-gpopt
11166 @itemx -mpowerpc-gfxopt
11167 @itemx -mno-powerpc-gfxopt
11169 @itemx -mno-powerpc64
11173 @itemx -mno-popcntb
11179 @opindex mno-power2
11181 @opindex mno-powerpc
11182 @opindex mpowerpc-gpopt
11183 @opindex mno-powerpc-gpopt
11184 @opindex mpowerpc-gfxopt
11185 @opindex mno-powerpc-gfxopt
11186 @opindex mpowerpc64
11187 @opindex mno-powerpc64
11191 @opindex mno-popcntb
11194 GCC supports two related instruction set architectures for the
11195 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11196 instructions supported by the @samp{rios} chip set used in the original
11197 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11198 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11199 the IBM 4xx, 6xx, and follow-on microprocessors.
11201 Neither architecture is a subset of the other. However there is a
11202 large common subset of instructions supported by both. An MQ
11203 register is included in processors supporting the POWER architecture.
11205 You use these options to specify which instructions are available on the
11206 processor you are using. The default value of these options is
11207 determined when configuring GCC@. Specifying the
11208 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11209 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11210 rather than the options listed above.
11212 The @option{-mpower} option allows GCC to generate instructions that
11213 are found only in the POWER architecture and to use the MQ register.
11214 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11215 to generate instructions that are present in the POWER2 architecture but
11216 not the original POWER architecture.
11218 The @option{-mpowerpc} option allows GCC to generate instructions that
11219 are found only in the 32-bit subset of the PowerPC architecture.
11220 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11221 GCC to use the optional PowerPC architecture instructions in the
11222 General Purpose group, including floating-point square root. Specifying
11223 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11224 use the optional PowerPC architecture instructions in the Graphics
11225 group, including floating-point select.
11227 The @option{-mmfcrf} option allows GCC to generate the move from
11228 condition register field instruction implemented on the POWER4
11229 processor and other processors that support the PowerPC V2.01
11231 The @option{-mpopcntb} option allows GCC to generate the popcount and
11232 double precision FP reciprocal estimate instruction implemented on the
11233 POWER5 processor and other processors that support the PowerPC V2.02
11235 The @option{-mfprnd} option allows GCC to generate the FP round to
11236 integer instructions implemented on the POWER5+ processor and other
11237 processors that support the PowerPC V2.03 architecture.
11239 The @option{-mpowerpc64} option allows GCC to generate the additional
11240 64-bit instructions that are found in the full PowerPC64 architecture
11241 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11242 @option{-mno-powerpc64}.
11244 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11245 will use only the instructions in the common subset of both
11246 architectures plus some special AIX common-mode calls, and will not use
11247 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11248 permits GCC to use any instruction from either architecture and to
11249 allow use of the MQ register; specify this for the Motorola MPC601.
11251 @item -mnew-mnemonics
11252 @itemx -mold-mnemonics
11253 @opindex mnew-mnemonics
11254 @opindex mold-mnemonics
11255 Select which mnemonics to use in the generated assembler code. With
11256 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11257 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11258 assembler mnemonics defined for the POWER architecture. Instructions
11259 defined in only one architecture have only one mnemonic; GCC uses that
11260 mnemonic irrespective of which of these options is specified.
11262 GCC defaults to the mnemonics appropriate for the architecture in
11263 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11264 value of these option. Unless you are building a cross-compiler, you
11265 should normally not specify either @option{-mnew-mnemonics} or
11266 @option{-mold-mnemonics}, but should instead accept the default.
11268 @item -mcpu=@var{cpu_type}
11270 Set architecture type, register usage, choice of mnemonics, and
11271 instruction scheduling parameters for machine type @var{cpu_type}.
11272 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11273 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11274 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11275 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11276 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11277 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11278 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11279 @samp{power4}, @samp{power5}, @samp{power5+},
11280 @samp{common}, @samp{powerpc}, @samp{powerpc64},
11281 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11283 @option{-mcpu=common} selects a completely generic processor. Code
11284 generated under this option will run on any POWER or PowerPC processor.
11285 GCC will use only the instructions in the common subset of both
11286 architectures, and will not use the MQ register. GCC assumes a generic
11287 processor model for scheduling purposes.
11289 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11290 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11291 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11292 types, with an appropriate, generic processor model assumed for
11293 scheduling purposes.
11295 The other options specify a specific processor. Code generated under
11296 those options will run best on that processor, and may not run at all on
11299 The @option{-mcpu} options automatically enable or disable the
11300 following options: @option{-maltivec}, @option{-mfprnd},
11301 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11302 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11303 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11304 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw}, @option{dlmzb}.
11305 The particular options
11306 set for any particular CPU will vary between compiler versions,
11307 depending on what setting seems to produce optimal code for that CPU;
11308 it doesn't necessarily reflect the actual hardware's capabilities. If
11309 you wish to set an individual option to a particular value, you may
11310 specify it after the @option{-mcpu} option, like @samp{-mcpu=970
11313 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11314 not enabled or disabled by the @option{-mcpu} option at present because
11315 AIX does not have full support for these options. You may still
11316 enable or disable them individually if you're sure it'll work in your
11319 @item -mtune=@var{cpu_type}
11321 Set the instruction scheduling parameters for machine type
11322 @var{cpu_type}, but do not set the architecture type, register usage, or
11323 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11324 values for @var{cpu_type} are used for @option{-mtune} as for
11325 @option{-mcpu}. If both are specified, the code generated will use the
11326 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11327 scheduling parameters set by @option{-mtune}.
11333 Generate code to compute division as reciprocal estimate and iterative
11334 refinement, creating opportunities for increased throughput. This
11335 feature requires: optional PowerPC Graphics instruction set for single
11336 precision and FRE instruction for double precision, assuming divides
11337 cannot generate user-visible traps, and the domain values not include
11338 Infinities, denormals or zero denominator.
11341 @itemx -mno-altivec
11343 @opindex mno-altivec
11344 Generate code that uses (does not use) AltiVec instructions, and also
11345 enable the use of built-in functions that allow more direct access to
11346 the AltiVec instruction set. You may also need to set
11347 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11353 @opindex mno-vrsave
11354 Generate VRSAVE instructions when generating AltiVec code.
11357 @opindex msecure-plt
11358 Generate code that allows ld and ld.so to build executables and shared
11359 libraries with non-exec .plt and .got sections. This is a PowerPC
11360 32-bit SYSV ABI option.
11364 Generate code that uses a BSS .plt section that ld.so fills in, and
11365 requires .plt and .got sections that are both writable and executable.
11366 This is a PowerPC 32-bit SYSV ABI option.
11372 This switch enables or disables the generation of ISEL instructions.
11374 @item -misel=@var{yes/no}
11375 This switch has been deprecated. Use @option{-misel} and
11376 @option{-mno-isel} instead.
11382 This switch enables or disables the generation of SPE simd
11385 @item -mspe=@var{yes/no}
11386 This option has been deprecated. Use @option{-mspe} and
11387 @option{-mno-spe} instead.
11389 @item -mfloat-gprs=@var{yes/single/double/no}
11390 @itemx -mfloat-gprs
11391 @opindex mfloat-gprs
11392 This switch enables or disables the generation of floating point
11393 operations on the general purpose registers for architectures that
11396 The argument @var{yes} or @var{single} enables the use of
11397 single-precision floating point operations.
11399 The argument @var{double} enables the use of single and
11400 double-precision floating point operations.
11402 The argument @var{no} disables floating point operations on the
11403 general purpose registers.
11405 This option is currently only available on the MPC854x.
11411 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11412 targets (including GNU/Linux). The 32-bit environment sets int, long
11413 and pointer to 32 bits and generates code that runs on any PowerPC
11414 variant. The 64-bit environment sets int to 32 bits and long and
11415 pointer to 64 bits, and generates code for PowerPC64, as for
11416 @option{-mpowerpc64}.
11419 @itemx -mno-fp-in-toc
11420 @itemx -mno-sum-in-toc
11421 @itemx -mminimal-toc
11423 @opindex mno-fp-in-toc
11424 @opindex mno-sum-in-toc
11425 @opindex mminimal-toc
11426 Modify generation of the TOC (Table Of Contents), which is created for
11427 every executable file. The @option{-mfull-toc} option is selected by
11428 default. In that case, GCC will allocate at least one TOC entry for
11429 each unique non-automatic variable reference in your program. GCC
11430 will also place floating-point constants in the TOC@. However, only
11431 16,384 entries are available in the TOC@.
11433 If you receive a linker error message that saying you have overflowed
11434 the available TOC space, you can reduce the amount of TOC space used
11435 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11436 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11437 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11438 generate code to calculate the sum of an address and a constant at
11439 run-time instead of putting that sum into the TOC@. You may specify one
11440 or both of these options. Each causes GCC to produce very slightly
11441 slower and larger code at the expense of conserving TOC space.
11443 If you still run out of space in the TOC even when you specify both of
11444 these options, specify @option{-mminimal-toc} instead. This option causes
11445 GCC to make only one TOC entry for every file. When you specify this
11446 option, GCC will produce code that is slower and larger but which
11447 uses extremely little TOC space. You may wish to use this option
11448 only on files that contain less frequently executed code.
11454 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11455 @code{long} type, and the infrastructure needed to support them.
11456 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11457 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11458 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11461 @itemx -mno-xl-compat
11462 @opindex mxl-compat
11463 @opindex mno-xl-compat
11464 Produce code that conforms more closely to IBM XL compiler semantics
11465 when using AIX-compatible ABI. Pass floating-point arguments to
11466 prototyped functions beyond the register save area (RSA) on the stack
11467 in addition to argument FPRs. Do not assume that most significant
11468 double in 128-bit long double value is properly rounded when comparing
11469 values and converting to double. Use XL symbol names for long double
11472 The AIX calling convention was extended but not initially documented to
11473 handle an obscure K&R C case of calling a function that takes the
11474 address of its arguments with fewer arguments than declared. IBM XL
11475 compilers access floating point arguments which do not fit in the
11476 RSA from the stack when a subroutine is compiled without
11477 optimization. Because always storing floating-point arguments on the
11478 stack is inefficient and rarely needed, this option is not enabled by
11479 default and only is necessary when calling subroutines compiled by IBM
11480 XL compilers without optimization.
11484 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11485 application written to use message passing with special startup code to
11486 enable the application to run. The system must have PE installed in the
11487 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11488 must be overridden with the @option{-specs=} option to specify the
11489 appropriate directory location. The Parallel Environment does not
11490 support threads, so the @option{-mpe} option and the @option{-pthread}
11491 option are incompatible.
11493 @item -malign-natural
11494 @itemx -malign-power
11495 @opindex malign-natural
11496 @opindex malign-power
11497 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11498 @option{-malign-natural} overrides the ABI-defined alignment of larger
11499 types, such as floating-point doubles, on their natural size-based boundary.
11500 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11501 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11503 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11507 @itemx -mhard-float
11508 @opindex msoft-float
11509 @opindex mhard-float
11510 Generate code that does not use (uses) the floating-point register set.
11511 Software floating point emulation is provided if you use the
11512 @option{-msoft-float} option, and pass the option to GCC when linking.
11515 @itemx -mno-multiple
11517 @opindex mno-multiple
11518 Generate code that uses (does not use) the load multiple word
11519 instructions and the store multiple word instructions. These
11520 instructions are generated by default on POWER systems, and not
11521 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11522 endian PowerPC systems, since those instructions do not work when the
11523 processor is in little endian mode. The exceptions are PPC740 and
11524 PPC750 which permit the instructions usage in little endian mode.
11529 @opindex mno-string
11530 Generate code that uses (does not use) the load string instructions
11531 and the store string word instructions to save multiple registers and
11532 do small block moves. These instructions are generated by default on
11533 POWER systems, and not generated on PowerPC systems. Do not use
11534 @option{-mstring} on little endian PowerPC systems, since those
11535 instructions do not work when the processor is in little endian mode.
11536 The exceptions are PPC740 and PPC750 which permit the instructions
11537 usage in little endian mode.
11542 @opindex mno-update
11543 Generate code that uses (does not use) the load or store instructions
11544 that update the base register to the address of the calculated memory
11545 location. These instructions are generated by default. If you use
11546 @option{-mno-update}, there is a small window between the time that the
11547 stack pointer is updated and the address of the previous frame is
11548 stored, which means code that walks the stack frame across interrupts or
11549 signals may get corrupted data.
11552 @itemx -mno-fused-madd
11553 @opindex mfused-madd
11554 @opindex mno-fused-madd
11555 Generate code that uses (does not use) the floating point multiply and
11556 accumulate instructions. These instructions are generated by default if
11557 hardware floating is used.
11563 Generate code that uses (does not use) the half-word multiply and
11564 multiply-accumulate instructions on the IBM 405 and 440 processors.
11565 These instructions are generated by default when targetting those
11572 Generate code that uses (does not use) the string-search @samp{dlmzb}
11573 instruction on the IBM 405 and 440 processors. This instruction is
11574 generated by default when targetting those processors.
11576 @item -mno-bit-align
11578 @opindex mno-bit-align
11579 @opindex mbit-align
11580 On System V.4 and embedded PowerPC systems do not (do) force structures
11581 and unions that contain bit-fields to be aligned to the base type of the
11584 For example, by default a structure containing nothing but 8
11585 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11586 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11587 the structure would be aligned to a 1 byte boundary and be one byte in
11590 @item -mno-strict-align
11591 @itemx -mstrict-align
11592 @opindex mno-strict-align
11593 @opindex mstrict-align
11594 On System V.4 and embedded PowerPC systems do not (do) assume that
11595 unaligned memory references will be handled by the system.
11597 @item -mrelocatable
11598 @itemx -mno-relocatable
11599 @opindex mrelocatable
11600 @opindex mno-relocatable
11601 On embedded PowerPC systems generate code that allows (does not allow)
11602 the program to be relocated to a different address at runtime. If you
11603 use @option{-mrelocatable} on any module, all objects linked together must
11604 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11606 @item -mrelocatable-lib
11607 @itemx -mno-relocatable-lib
11608 @opindex mrelocatable-lib
11609 @opindex mno-relocatable-lib
11610 On embedded PowerPC systems generate code that allows (does not allow)
11611 the program to be relocated to a different address at runtime. Modules
11612 compiled with @option{-mrelocatable-lib} can be linked with either modules
11613 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11614 with modules compiled with the @option{-mrelocatable} options.
11620 On System V.4 and embedded PowerPC systems do not (do) assume that
11621 register 2 contains a pointer to a global area pointing to the addresses
11622 used in the program.
11625 @itemx -mlittle-endian
11627 @opindex mlittle-endian
11628 On System V.4 and embedded PowerPC systems compile code for the
11629 processor in little endian mode. The @option{-mlittle-endian} option is
11630 the same as @option{-mlittle}.
11633 @itemx -mbig-endian
11635 @opindex mbig-endian
11636 On System V.4 and embedded PowerPC systems compile code for the
11637 processor in big endian mode. The @option{-mbig-endian} option is
11638 the same as @option{-mbig}.
11640 @item -mdynamic-no-pic
11641 @opindex mdynamic-no-pic
11642 On Darwin and Mac OS X systems, compile code so that it is not
11643 relocatable, but that its external references are relocatable. The
11644 resulting code is suitable for applications, but not shared
11647 @item -mprioritize-restricted-insns=@var{priority}
11648 @opindex mprioritize-restricted-insns
11649 This option controls the priority that is assigned to
11650 dispatch-slot restricted instructions during the second scheduling
11651 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11652 @var{no/highest/second-highest} priority to dispatch slot restricted
11655 @item -msched-costly-dep=@var{dependence_type}
11656 @opindex msched-costly-dep
11657 This option controls which dependences are considered costly
11658 by the target during instruction scheduling. The argument
11659 @var{dependence_type} takes one of the following values:
11660 @var{no}: no dependence is costly,
11661 @var{all}: all dependences are costly,
11662 @var{true_store_to_load}: a true dependence from store to load is costly,
11663 @var{store_to_load}: any dependence from store to load is costly,
11664 @var{number}: any dependence which latency >= @var{number} is costly.
11666 @item -minsert-sched-nops=@var{scheme}
11667 @opindex minsert-sched-nops
11668 This option controls which nop insertion scheme will be used during
11669 the second scheduling pass. The argument @var{scheme} takes one of the
11671 @var{no}: Don't insert nops.
11672 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11673 according to the scheduler's grouping.
11674 @var{regroup_exact}: Insert nops to force costly dependent insns into
11675 separate groups. Insert exactly as many nops as needed to force an insn
11676 to a new group, according to the estimated processor grouping.
11677 @var{number}: Insert nops to force costly dependent insns into
11678 separate groups. Insert @var{number} nops to force an insn to a new group.
11681 @opindex mcall-sysv
11682 On System V.4 and embedded PowerPC systems compile code using calling
11683 conventions that adheres to the March 1995 draft of the System V
11684 Application Binary Interface, PowerPC processor supplement. This is the
11685 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11687 @item -mcall-sysv-eabi
11688 @opindex mcall-sysv-eabi
11689 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11691 @item -mcall-sysv-noeabi
11692 @opindex mcall-sysv-noeabi
11693 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11695 @item -mcall-solaris
11696 @opindex mcall-solaris
11697 On System V.4 and embedded PowerPC systems compile code for the Solaris
11701 @opindex mcall-linux
11702 On System V.4 and embedded PowerPC systems compile code for the
11703 Linux-based GNU system.
11707 On System V.4 and embedded PowerPC systems compile code for the
11708 Hurd-based GNU system.
11710 @item -mcall-netbsd
11711 @opindex mcall-netbsd
11712 On System V.4 and embedded PowerPC systems compile code for the
11713 NetBSD operating system.
11715 @item -maix-struct-return
11716 @opindex maix-struct-return
11717 Return all structures in memory (as specified by the AIX ABI)@.
11719 @item -msvr4-struct-return
11720 @opindex msvr4-struct-return
11721 Return structures smaller than 8 bytes in registers (as specified by the
11724 @item -mabi=@var{abi-type}
11726 Extend the current ABI with a particular extension, or remove such extension.
11727 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11728 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
11732 Extend the current ABI with SPE ABI extensions. This does not change
11733 the default ABI, instead it adds the SPE ABI extensions to the current
11737 @opindex mabi=no-spe
11738 Disable Booke SPE ABI extensions for the current ABI@.
11740 @item -mabi=ibmlongdouble
11741 @opindex mabi=ibmlongdouble
11742 Change the current ABI to use IBM extended precision long double.
11743 This is a PowerPC 32-bit SYSV ABI option.
11745 @item -mabi=ieeelongdouble
11746 @opindex mabi=ieeelongdouble
11747 Change the current ABI to use IEEE extended precision long double.
11748 This is a PowerPC 32-bit Linux ABI option.
11751 @itemx -mno-prototype
11752 @opindex mprototype
11753 @opindex mno-prototype
11754 On System V.4 and embedded PowerPC systems assume that all calls to
11755 variable argument functions are properly prototyped. Otherwise, the
11756 compiler must insert an instruction before every non prototyped call to
11757 set or clear bit 6 of the condition code register (@var{CR}) to
11758 indicate whether floating point values were passed in the floating point
11759 registers in case the function takes a variable arguments. With
11760 @option{-mprototype}, only calls to prototyped variable argument functions
11761 will set or clear the bit.
11765 On embedded PowerPC systems, assume that the startup module is called
11766 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11767 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11772 On embedded PowerPC systems, assume that the startup module is called
11773 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11778 On embedded PowerPC systems, assume that the startup module is called
11779 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11782 @item -myellowknife
11783 @opindex myellowknife
11784 On embedded PowerPC systems, assume that the startup module is called
11785 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11790 On System V.4 and embedded PowerPC systems, specify that you are
11791 compiling for a VxWorks system.
11795 Specify that you are compiling for the WindISS simulation environment.
11799 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11800 header to indicate that @samp{eabi} extended relocations are used.
11806 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11807 Embedded Applications Binary Interface (eabi) which is a set of
11808 modifications to the System V.4 specifications. Selecting @option{-meabi}
11809 means that the stack is aligned to an 8 byte boundary, a function
11810 @code{__eabi} is called to from @code{main} to set up the eabi
11811 environment, and the @option{-msdata} option can use both @code{r2} and
11812 @code{r13} to point to two separate small data areas. Selecting
11813 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11814 do not call an initialization function from @code{main}, and the
11815 @option{-msdata} option will only use @code{r13} to point to a single
11816 small data area. The @option{-meabi} option is on by default if you
11817 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11820 @opindex msdata=eabi
11821 On System V.4 and embedded PowerPC systems, put small initialized
11822 @code{const} global and static data in the @samp{.sdata2} section, which
11823 is pointed to by register @code{r2}. Put small initialized
11824 non-@code{const} global and static data in the @samp{.sdata} section,
11825 which is pointed to by register @code{r13}. Put small uninitialized
11826 global and static data in the @samp{.sbss} section, which is adjacent to
11827 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11828 incompatible with the @option{-mrelocatable} option. The
11829 @option{-msdata=eabi} option also sets the @option{-memb} option.
11832 @opindex msdata=sysv
11833 On System V.4 and embedded PowerPC systems, put small global and static
11834 data in the @samp{.sdata} section, which is pointed to by register
11835 @code{r13}. Put small uninitialized global and static data in the
11836 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11837 The @option{-msdata=sysv} option is incompatible with the
11838 @option{-mrelocatable} option.
11840 @item -msdata=default
11842 @opindex msdata=default
11844 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11845 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11846 same as @option{-msdata=sysv}.
11849 @opindex msdata-data
11850 On System V.4 and embedded PowerPC systems, put small global
11851 data in the @samp{.sdata} section. Put small uninitialized global
11852 data in the @samp{.sbss} section. Do not use register @code{r13}
11853 to address small data however. This is the default behavior unless
11854 other @option{-msdata} options are used.
11858 @opindex msdata=none
11860 On embedded PowerPC systems, put all initialized global and static data
11861 in the @samp{.data} section, and all uninitialized data in the
11862 @samp{.bss} section.
11866 @cindex smaller data references (PowerPC)
11867 @cindex .sdata/.sdata2 references (PowerPC)
11868 On embedded PowerPC systems, put global and static items less than or
11869 equal to @var{num} bytes into the small data or bss sections instead of
11870 the normal data or bss section. By default, @var{num} is 8. The
11871 @option{-G @var{num}} switch is also passed to the linker.
11872 All modules should be compiled with the same @option{-G @var{num}} value.
11875 @itemx -mno-regnames
11877 @opindex mno-regnames
11878 On System V.4 and embedded PowerPC systems do (do not) emit register
11879 names in the assembly language output using symbolic forms.
11882 @itemx -mno-longcall
11884 @opindex mno-longcall
11885 Default to making all function calls indirectly, using a register, so
11886 that functions which reside further than 32 megabytes (33,554,432
11887 bytes) from the current location can be called. This setting can be
11888 overridden by the @code{shortcall} function attribute, or by
11889 @code{#pragma longcall(0)}.
11891 Some linkers are capable of detecting out-of-range calls and generating
11892 glue code on the fly. On these systems, long calls are unnecessary and
11893 generate slower code. As of this writing, the AIX linker can do this,
11894 as can the GNU linker for PowerPC/64. It is planned to add this feature
11895 to the GNU linker for 32-bit PowerPC systems as well.
11897 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11898 callee, L42'', plus a ``branch island'' (glue code). The two target
11899 addresses represent the callee and the ``branch island''. The
11900 Darwin/PPC linker will prefer the first address and generate a ``bl
11901 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11902 otherwise, the linker will generate ``bl L42'' to call the ``branch
11903 island''. The ``branch island'' is appended to the body of the
11904 calling function; it computes the full 32-bit address of the callee
11907 On Mach-O (Darwin) systems, this option directs the compiler emit to
11908 the glue for every direct call, and the Darwin linker decides whether
11909 to use or discard it.
11911 In the future, we may cause GCC to ignore all longcall specifications
11912 when the linker is known to generate glue.
11916 Adds support for multithreading with the @dfn{pthreads} library.
11917 This option sets flags for both the preprocessor and linker.
11921 @node S/390 and zSeries Options
11922 @subsection S/390 and zSeries Options
11923 @cindex S/390 and zSeries Options
11925 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
11929 @itemx -msoft-float
11930 @opindex mhard-float
11931 @opindex msoft-float
11932 Use (do not use) the hardware floating-point instructions and registers
11933 for floating-point operations. When @option{-msoft-float} is specified,
11934 functions in @file{libgcc.a} will be used to perform floating-point
11935 operations. When @option{-mhard-float} is specified, the compiler
11936 generates IEEE floating-point instructions. This is the default.
11938 @item -mlong-double-64
11939 @itemx -mlong-double-128
11940 @opindex mlong-double-64
11941 @opindex mlong-double-128
11942 These switches control the size of @code{long double} type. A size
11943 of 64bit makes the @code{long double} type equivalent to the @code{double}
11944 type. This is the default.
11947 @itemx -mno-backchain
11948 @opindex mbackchain
11949 @opindex mno-backchain
11950 Store (do not store) the address of the caller's frame as backchain pointer
11951 into the callee's stack frame.
11952 A backchain may be needed to allow debugging using tools that do not understand
11953 DWARF-2 call frame information.
11954 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
11955 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
11956 the backchain is placed into the topmost word of the 96/160 byte register
11959 In general, code compiled with @option{-mbackchain} is call-compatible with
11960 code compiled with @option{-mmo-backchain}; however, use of the backchain
11961 for debugging purposes usually requires that the whole binary is built with
11962 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
11963 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11964 to build a linux kernel use @option{-msoft-float}.
11966 The default is to not maintain the backchain.
11968 @item -mpacked-stack
11969 @item -mno-packed-stack
11970 @opindex mpacked-stack
11971 @opindex mno-packed-stack
11972 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
11973 specified, the compiler uses the all fields of the 96/160 byte register save
11974 area only for their default purpose; unused fields still take up stack space.
11975 When @option{-mpacked-stack} is specified, register save slots are densely
11976 packed at the top of the register save area; unused space is reused for other
11977 purposes, allowing for more efficient use of the available stack space.
11978 However, when @option{-mbackchain} is also in effect, the topmost word of
11979 the save area is always used to store the backchain, and the return address
11980 register is always saved two words below the backchain.
11982 As long as the stack frame backchain is not used, code generated with
11983 @option{-mpacked-stack} is call-compatible with code generated with
11984 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
11985 S/390 or zSeries generated code that uses the stack frame backchain at run
11986 time, not just for debugging purposes. Such code is not call-compatible
11987 with code compiled with @option{-mpacked-stack}. Also, note that the
11988 combination of @option{-mbackchain},
11989 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11990 to build a linux kernel use @option{-msoft-float}.
11992 The default is to not use the packed stack layout.
11995 @itemx -mno-small-exec
11996 @opindex msmall-exec
11997 @opindex mno-small-exec
11998 Generate (or do not generate) code using the @code{bras} instruction
11999 to do subroutine calls.
12000 This only works reliably if the total executable size does not
12001 exceed 64k. The default is to use the @code{basr} instruction instead,
12002 which does not have this limitation.
12008 When @option{-m31} is specified, generate code compliant to the
12009 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12010 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12011 particular to generate 64-bit instructions. For the @samp{s390}
12012 targets, the default is @option{-m31}, while the @samp{s390x}
12013 targets default to @option{-m64}.
12019 When @option{-mzarch} is specified, generate code using the
12020 instructions available on z/Architecture.
12021 When @option{-mesa} is specified, generate code using the
12022 instructions available on ESA/390. Note that @option{-mesa} is
12023 not possible with @option{-m64}.
12024 When generating code compliant to the GNU/Linux for S/390 ABI,
12025 the default is @option{-mesa}. When generating code compliant
12026 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12032 Generate (or do not generate) code using the @code{mvcle} instruction
12033 to perform block moves. When @option{-mno-mvcle} is specified,
12034 use a @code{mvc} loop instead. This is the default unless optimizing for
12041 Print (or do not print) additional debug information when compiling.
12042 The default is to not print debug information.
12044 @item -march=@var{cpu-type}
12046 Generate code that will run on @var{cpu-type}, which is the name of a system
12047 representing a certain processor type. Possible values for
12048 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12049 When generating code using the instructions available on z/Architecture,
12050 the default is @option{-march=z900}. Otherwise, the default is
12051 @option{-march=g5}.
12053 @item -mtune=@var{cpu-type}
12055 Tune to @var{cpu-type} everything applicable about the generated code,
12056 except for the ABI and the set of available instructions.
12057 The list of @var{cpu-type} values is the same as for @option{-march}.
12058 The default is the value used for @option{-march}.
12061 @itemx -mno-tpf-trace
12062 @opindex mtpf-trace
12063 @opindex mno-tpf-trace
12064 Generate code that adds (does not add) in TPF OS specific branches to trace
12065 routines in the operating system. This option is off by default, even
12066 when compiling for the TPF OS@.
12069 @itemx -mno-fused-madd
12070 @opindex mfused-madd
12071 @opindex mno-fused-madd
12072 Generate code that uses (does not use) the floating point multiply and
12073 accumulate instructions. These instructions are generated by default if
12074 hardware floating point is used.
12076 @item -mwarn-framesize=@var{framesize}
12077 @opindex mwarn-framesize
12078 Emit a warning if the current function exceeds the given frame size. Because
12079 this is a compile time check it doesn't need to be a real problem when the program
12080 runs. It is intended to identify functions which most probably cause
12081 a stack overflow. It is useful to be used in an environment with limited stack
12082 size e.g.@: the linux kernel.
12084 @item -mwarn-dynamicstack
12085 @opindex mwarn-dynamicstack
12086 Emit a warning if the function calls alloca or uses dynamically
12087 sized arrays. This is generally a bad idea with a limited stack size.
12089 @item -mstack-guard=@var{stack-guard}
12090 @item -mstack-size=@var{stack-size}
12091 @opindex mstack-guard
12092 @opindex mstack-size
12093 These arguments always have to be used in conjunction. If they are present the s390
12094 back end emits additional instructions in the function prologue which trigger a trap
12095 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12096 (remember that the stack on s390 grows downward). These options are intended to
12097 be used to help debugging stack overflow problems. The additionally emitted code
12098 causes only little overhead and hence can also be used in production like systems
12099 without greater performance degradation. The given values have to be exact
12100 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12102 In order to be efficient the extra code makes the assumption that the stack starts
12103 at an address aligned to the value given by @var{stack-size}.
12107 @subsection SH Options
12109 These @samp{-m} options are defined for the SH implementations:
12114 Generate code for the SH1.
12118 Generate code for the SH2.
12121 Generate code for the SH2e.
12125 Generate code for the SH3.
12129 Generate code for the SH3e.
12133 Generate code for the SH4 without a floating-point unit.
12135 @item -m4-single-only
12136 @opindex m4-single-only
12137 Generate code for the SH4 with a floating-point unit that only
12138 supports single-precision arithmetic.
12142 Generate code for the SH4 assuming the floating-point unit is in
12143 single-precision mode by default.
12147 Generate code for the SH4.
12151 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12152 floating-point unit is not used.
12154 @item -m4a-single-only
12155 @opindex m4a-single-only
12156 Generate code for the SH4a, in such a way that no double-precision
12157 floating point operations are used.
12160 @opindex m4a-single
12161 Generate code for the SH4a assuming the floating-point unit is in
12162 single-precision mode by default.
12166 Generate code for the SH4a.
12170 Same as @option{-m4a-nofpu}, except that it implicitly passes
12171 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12172 instructions at the moment.
12176 Compile code for the processor in big endian mode.
12180 Compile code for the processor in little endian mode.
12184 Align doubles at 64-bit boundaries. Note that this changes the calling
12185 conventions, and thus some functions from the standard C library will
12186 not work unless you recompile it first with @option{-mdalign}.
12190 Shorten some address references at link time, when possible; uses the
12191 linker option @option{-relax}.
12195 Use 32-bit offsets in @code{switch} tables. The default is to use
12200 Enable the use of the instruction @code{fmovd}.
12204 Comply with the calling conventions defined by Renesas.
12208 Comply with the calling conventions defined by Renesas.
12212 Comply with the calling conventions defined for GCC before the Renesas
12213 conventions were available. This option is the default for all
12214 targets of the SH toolchain except for @samp{sh-symbianelf}.
12217 @opindex mnomacsave
12218 Mark the @code{MAC} register as call-clobbered, even if
12219 @option{-mhitachi} is given.
12223 Increase IEEE-compliance of floating-point code.
12224 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12225 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12226 comparisons of NANs / infinities incurs extra overhead in every
12227 floating point comparison, therefore the default is set to
12228 @option{-ffinite-math-only}.
12232 Dump instruction size and location in the assembly code.
12235 @opindex mpadstruct
12236 This option is deprecated. It pads structures to multiple of 4 bytes,
12237 which is incompatible with the SH ABI@.
12241 Optimize for space instead of speed. Implied by @option{-Os}.
12244 @opindex mprefergot
12245 When generating position-independent code, emit function calls using
12246 the Global Offset Table instead of the Procedure Linkage Table.
12250 Generate a library function call to invalidate instruction cache
12251 entries, after fixing up a trampoline. This library function call
12252 doesn't assume it can write to the whole memory address space. This
12253 is the default when the target is @code{sh-*-linux*}.
12255 @item -multcost=@var{number}
12256 @opindex multcost=@var{number}
12257 Set the cost to assume for a multiply insn.
12259 @item -mdiv=@var{strategy}
12260 @opindex mdiv=@var{strategy}
12261 Set the division strategy to use for SHmedia code. @var{strategy} must be
12262 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12263 inv:call2, inv:fp .
12264 "fp" performs the operation in floating point. This has a very high latency,
12265 but needs only a few instructions, so it might be a good choice if
12266 your code has enough easily exploitable ILP to allow the compiler to
12267 schedule the floating point instructions together with other instructions.
12268 Division by zero causes a floating point exception.
12269 "inv" uses integer operations to calculate the inverse of the divisor,
12270 and then multiplies the dividend with the inverse. This strategy allows
12271 cse and hoisting of the inverse calculation. Division by zero calculates
12272 an unspecified result, but does not trap.
12273 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12274 have been found, or if the entire operation has been hoisted to the same
12275 place, the last stages of the inverse calculation are intertwined with the
12276 final multiply to reduce the overall latency, at the expense of using a few
12277 more instructions, and thus offering fewer scheduling opportunities with
12279 "call" calls a library function that usually implements the inv:minlat
12281 This gives high code density for m5-*media-nofpu compilations.
12282 "call2" uses a different entry point of the same library function, where it
12283 assumes that a pointer to a lookup table has already been set up, which
12284 exposes the pointer load to cse / code hoisting optimizations.
12285 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12286 code generation, but if the code stays unoptimized, revert to the "call",
12287 "call2", or "fp" strategies, respectively. Note that the
12288 potentially-trapping side effect of division by zero is carried by a
12289 separate instruction, so it is possible that all the integer instructions
12290 are hoisted out, but the marker for the side effect stays where it is.
12291 A recombination to fp operations or a call is not possible in that case.
12292 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12293 that the inverse calculation was nor separated from the multiply, they speed
12294 up division where the dividend fits into 20 bits (plus sign where applicable),
12295 by inserting a test to skip a number of operations in this case; this test
12296 slows down the case of larger dividends. inv20u assumes the case of a such
12297 a small dividend to be unlikely, and inv20l assumes it to be likely.
12299 @item -mdivsi3_libfunc=@var{name}
12300 @opindex mdivsi3_libfunc=@var{name}
12301 Set the name of the library function used for 32 bit signed division to
12302 @var{name}. This only affect the name used in the call and inv:call
12303 division strategies, and the compiler will still expect the same
12304 sets of input/output/clobbered registers as if this option was not present.
12306 @item -madjust-unroll
12307 @opindex madjust-unroll
12308 Throttle unrolling to avoid thrashing target registers.
12309 This option only has an effect if the gcc code base supports the
12310 TARGET_ADJUST_UNROLL_MAX target hook.
12312 @item -mindexed-addressing
12313 @opindex mindexed-addressing
12314 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12315 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12316 semantics for the indexed addressing mode. The architecture allows the
12317 implementation of processors with 64 bit MMU, which the OS could use to
12318 get 32 bit addressing, but since no current hardware implementation supports
12319 this or any other way to make the indexed addressing mode safe to use in
12320 the 32 bit ABI, the default is -mno-indexed-addressing.
12322 @item -mgettrcost=@var{number}
12323 @opindex mgettrcost=@var{number}
12324 Set the cost assumed for the gettr instruction to @var{number}.
12325 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12329 Assume pt* instructions won't trap. This will generally generate better
12330 scheduled code, but is unsafe on current hardware. The current architecture
12331 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12332 This has the unintentional effect of making it unsafe to schedule ptabs /
12333 ptrel before a branch, or hoist it out of a loop. For example,
12334 __do_global_ctors, a part of libgcc that runs constructors at program
12335 startup, calls functions in a list which is delimited by -1. With the
12336 -mpt-fixed option, the ptabs will be done before testing against -1.
12337 That means that all the constructors will be run a bit quicker, but when
12338 the loop comes to the end of the list, the program crashes because ptabs
12339 loads -1 into a target register. Since this option is unsafe for any
12340 hardware implementing the current architecture specification, the default
12341 is -mno-pt-fixed. Unless the user specifies a specific cost with
12342 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12343 this deters register allocation using target registers for storing
12346 @item -minvalid-symbols
12347 @opindex minvalid-symbols
12348 Assume symbols might be invalid. Ordinary function symbols generated by
12349 the compiler will always be valid to load with movi/shori/ptabs or
12350 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12351 to generate symbols that will cause ptabs / ptrel to trap.
12352 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12353 It will then prevent cross-basic-block cse, hoisting and most scheduling
12354 of symbol loads. The default is @option{-mno-invalid-symbols}.
12357 @node SPARC Options
12358 @subsection SPARC Options
12359 @cindex SPARC options
12361 These @samp{-m} options are supported on the SPARC:
12364 @item -mno-app-regs
12366 @opindex mno-app-regs
12368 Specify @option{-mapp-regs} to generate output using the global registers
12369 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12372 To be fully SVR4 ABI compliant at the cost of some performance loss,
12373 specify @option{-mno-app-regs}. You should compile libraries and system
12374 software with this option.
12377 @itemx -mhard-float
12379 @opindex mhard-float
12380 Generate output containing floating point instructions. This is the
12384 @itemx -msoft-float
12386 @opindex msoft-float
12387 Generate output containing library calls for floating point.
12388 @strong{Warning:} the requisite libraries are not available for all SPARC
12389 targets. Normally the facilities of the machine's usual C compiler are
12390 used, but this cannot be done directly in cross-compilation. You must make
12391 your own arrangements to provide suitable library functions for
12392 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12393 @samp{sparclite-*-*} do provide software floating point support.
12395 @option{-msoft-float} changes the calling convention in the output file;
12396 therefore, it is only useful if you compile @emph{all} of a program with
12397 this option. In particular, you need to compile @file{libgcc.a}, the
12398 library that comes with GCC, with @option{-msoft-float} in order for
12401 @item -mhard-quad-float
12402 @opindex mhard-quad-float
12403 Generate output containing quad-word (long double) floating point
12406 @item -msoft-quad-float
12407 @opindex msoft-quad-float
12408 Generate output containing library calls for quad-word (long double)
12409 floating point instructions. The functions called are those specified
12410 in the SPARC ABI@. This is the default.
12412 As of this writing, there are no SPARC implementations that have hardware
12413 support for the quad-word floating point instructions. They all invoke
12414 a trap handler for one of these instructions, and then the trap handler
12415 emulates the effect of the instruction. Because of the trap handler overhead,
12416 this is much slower than calling the ABI library routines. Thus the
12417 @option{-msoft-quad-float} option is the default.
12419 @item -mno-unaligned-doubles
12420 @itemx -munaligned-doubles
12421 @opindex mno-unaligned-doubles
12422 @opindex munaligned-doubles
12423 Assume that doubles have 8 byte alignment. This is the default.
12425 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12426 alignment only if they are contained in another type, or if they have an
12427 absolute address. Otherwise, it assumes they have 4 byte alignment.
12428 Specifying this option avoids some rare compatibility problems with code
12429 generated by other compilers. It is not the default because it results
12430 in a performance loss, especially for floating point code.
12432 @item -mno-faster-structs
12433 @itemx -mfaster-structs
12434 @opindex mno-faster-structs
12435 @opindex mfaster-structs
12436 With @option{-mfaster-structs}, the compiler assumes that structures
12437 should have 8 byte alignment. This enables the use of pairs of
12438 @code{ldd} and @code{std} instructions for copies in structure
12439 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12440 However, the use of this changed alignment directly violates the SPARC
12441 ABI@. Thus, it's intended only for use on targets where the developer
12442 acknowledges that their resulting code will not be directly in line with
12443 the rules of the ABI@.
12445 @item -mimpure-text
12446 @opindex mimpure-text
12447 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12448 the compiler to not pass @option{-z text} to the linker when linking a
12449 shared object. Using this option, you can link position-dependent
12450 code into a shared object.
12452 @option{-mimpure-text} suppresses the ``relocations remain against
12453 allocatable but non-writable sections'' linker error message.
12454 However, the necessary relocations will trigger copy-on-write, and the
12455 shared object is not actually shared across processes. Instead of
12456 using @option{-mimpure-text}, you should compile all source code with
12457 @option{-fpic} or @option{-fPIC}.
12459 This option is only available on SunOS and Solaris.
12461 @item -mcpu=@var{cpu_type}
12463 Set the instruction set, register set, and instruction scheduling parameters
12464 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12465 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12466 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12467 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12468 @samp{ultrasparc3}, and @samp{niagara}.
12470 Default instruction scheduling parameters are used for values that select
12471 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12472 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12474 Here is a list of each supported architecture and their supported
12479 v8: supersparc, hypersparc
12480 sparclite: f930, f934, sparclite86x
12482 v9: ultrasparc, ultrasparc3, niagara
12485 By default (unless configured otherwise), GCC generates code for the V7
12486 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12487 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12488 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12489 SPARCStation 1, 2, IPX etc.
12491 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12492 architecture. The only difference from V7 code is that the compiler emits
12493 the integer multiply and integer divide instructions which exist in SPARC-V8
12494 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12495 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12498 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12499 the SPARC architecture. This adds the integer multiply, integer divide step
12500 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12501 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12502 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12503 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12504 MB86934 chip, which is the more recent SPARClite with FPU@.
12506 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12507 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12508 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12509 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12510 optimizes it for the TEMIC SPARClet chip.
12512 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12513 architecture. This adds 64-bit integer and floating-point move instructions,
12514 3 additional floating-point condition code registers and conditional move
12515 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12516 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12517 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12518 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12519 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12520 Sun UltraSPARC T1 chips.
12522 @item -mtune=@var{cpu_type}
12524 Set the instruction scheduling parameters for machine type
12525 @var{cpu_type}, but do not set the instruction set or register set that the
12526 option @option{-mcpu=@var{cpu_type}} would.
12528 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12529 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12530 that select a particular cpu implementation. Those are @samp{cypress},
12531 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12532 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12533 @samp{ultrasparc3}, and @samp{niagara}.
12538 @opindex mno-v8plus
12539 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12540 difference from the V8 ABI is that the global and out registers are
12541 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12542 mode for all SPARC-V9 processors.
12548 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12549 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12552 These @samp{-m} options are supported in addition to the above
12553 on SPARC-V9 processors in 64-bit environments:
12556 @item -mlittle-endian
12557 @opindex mlittle-endian
12558 Generate code for a processor running in little-endian mode. It is only
12559 available for a few configurations and most notably not on Solaris and Linux.
12565 Generate code for a 32-bit or 64-bit environment.
12566 The 32-bit environment sets int, long and pointer to 32 bits.
12567 The 64-bit environment sets int to 32 bits and long and pointer
12570 @item -mcmodel=medlow
12571 @opindex mcmodel=medlow
12572 Generate code for the Medium/Low code model: 64-bit addresses, programs
12573 must be linked in the low 32 bits of memory. Programs can be statically
12574 or dynamically linked.
12576 @item -mcmodel=medmid
12577 @opindex mcmodel=medmid
12578 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12579 must be linked in the low 44 bits of memory, the text and data segments must
12580 be less than 2GB in size and the data segment must be located within 2GB of
12583 @item -mcmodel=medany
12584 @opindex mcmodel=medany
12585 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12586 may be linked anywhere in memory, the text and data segments must be less
12587 than 2GB in size and the data segment must be located within 2GB of the
12590 @item -mcmodel=embmedany
12591 @opindex mcmodel=embmedany
12592 Generate code for the Medium/Anywhere code model for embedded systems:
12593 64-bit addresses, the text and data segments must be less than 2GB in
12594 size, both starting anywhere in memory (determined at link time). The
12595 global register %g4 points to the base of the data segment. Programs
12596 are statically linked and PIC is not supported.
12599 @itemx -mno-stack-bias
12600 @opindex mstack-bias
12601 @opindex mno-stack-bias
12602 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12603 frame pointer if present, are offset by @minus{}2047 which must be added back
12604 when making stack frame references. This is the default in 64-bit mode.
12605 Otherwise, assume no such offset is present.
12608 These switches are supported in addition to the above on Solaris:
12613 Add support for multithreading using the Solaris threads library. This
12614 option sets flags for both the preprocessor and linker. This option does
12615 not affect the thread safety of object code produced by the compiler or
12616 that of libraries supplied with it.
12620 Add support for multithreading using the POSIX threads library. This
12621 option sets flags for both the preprocessor and linker. This option does
12622 not affect the thread safety of object code produced by the compiler or
12623 that of libraries supplied with it.
12627 This is a synonym for @option{-pthreads}.
12630 @node System V Options
12631 @subsection Options for System V
12633 These additional options are available on System V Release 4 for
12634 compatibility with other compilers on those systems:
12639 Create a shared object.
12640 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12644 Identify the versions of each tool used by the compiler, in a
12645 @code{.ident} assembler directive in the output.
12649 Refrain from adding @code{.ident} directives to the output file (this is
12652 @item -YP,@var{dirs}
12654 Search the directories @var{dirs}, and no others, for libraries
12655 specified with @option{-l}.
12657 @item -Ym,@var{dir}
12659 Look in the directory @var{dir} to find the M4 preprocessor.
12660 The assembler uses this option.
12661 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12662 @c the generic assembler that comes with Solaris takes just -Ym.
12665 @node TMS320C3x/C4x Options
12666 @subsection TMS320C3x/C4x Options
12667 @cindex TMS320C3x/C4x Options
12669 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12673 @item -mcpu=@var{cpu_type}
12675 Set the instruction set, register set, and instruction scheduling
12676 parameters for machine type @var{cpu_type}. Supported values for
12677 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12678 @samp{c44}. The default is @samp{c40} to generate code for the
12683 @itemx -msmall-memory
12685 @opindex mbig-memory
12687 @opindex msmall-memory
12689 Generates code for the big or small memory model. The small memory
12690 model assumed that all data fits into one 64K word page. At run-time
12691 the data page (DP) register must be set to point to the 64K page
12692 containing the .bss and .data program sections. The big memory model is
12693 the default and requires reloading of the DP register for every direct
12700 Allow (disallow) allocation of general integer operands into the block
12701 count register BK@.
12707 Enable (disable) generation of code using decrement and branch,
12708 DBcond(D), instructions. This is enabled by default for the C4x. To be
12709 on the safe side, this is disabled for the C3x, since the maximum
12710 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12711 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12712 that it can utilize the decrement and branch instruction, but will give
12713 up if there is more than one memory reference in the loop. Thus a loop
12714 where the loop counter is decremented can generate slightly more
12715 efficient code, in cases where the RPTB instruction cannot be utilized.
12717 @item -mdp-isr-reload
12719 @opindex mdp-isr-reload
12721 Force the DP register to be saved on entry to an interrupt service
12722 routine (ISR), reloaded to point to the data section, and restored on
12723 exit from the ISR@. This should not be required unless someone has
12724 violated the small memory model by modifying the DP register, say within
12731 For the C3x use the 24-bit MPYI instruction for integer multiplies
12732 instead of a library call to guarantee 32-bit results. Note that if one
12733 of the operands is a constant, then the multiplication will be performed
12734 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12735 then squaring operations are performed inline instead of a library call.
12738 @itemx -mno-fast-fix
12740 @opindex mno-fast-fix
12741 The C3x/C4x FIX instruction to convert a floating point value to an
12742 integer value chooses the nearest integer less than or equal to the
12743 floating point value rather than to the nearest integer. Thus if the
12744 floating point number is negative, the result will be incorrectly
12745 truncated an additional code is necessary to detect and correct this
12746 case. This option can be used to disable generation of the additional
12747 code required to correct the result.
12753 Enable (disable) generation of repeat block sequences using the RPTB
12754 instruction for zero overhead looping. The RPTB construct is only used
12755 for innermost loops that do not call functions or jump across the loop
12756 boundaries. There is no advantage having nested RPTB loops due to the
12757 overhead required to save and restore the RC, RS, and RE registers.
12758 This is enabled by default with @option{-O2}.
12760 @item -mrpts=@var{count}
12764 Enable (disable) the use of the single instruction repeat instruction
12765 RPTS@. If a repeat block contains a single instruction, and the loop
12766 count can be guaranteed to be less than the value @var{count}, GCC will
12767 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12768 then a RPTS will be emitted even if the loop count cannot be determined
12769 at compile time. Note that the repeated instruction following RPTS does
12770 not have to be reloaded from memory each iteration, thus freeing up the
12771 CPU buses for operands. However, since interrupts are blocked by this
12772 instruction, it is disabled by default.
12774 @item -mloop-unsigned
12775 @itemx -mno-loop-unsigned
12776 @opindex mloop-unsigned
12777 @opindex mno-loop-unsigned
12778 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12779 is @math{2^{31} + 1} since these instructions test if the iteration count is
12780 negative to terminate the loop. If the iteration count is unsigned
12781 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12782 exceeded. This switch allows an unsigned iteration count.
12786 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12787 with. This also enforces compatibility with the API employed by the TI
12788 C3x C compiler. For example, long doubles are passed as structures
12789 rather than in floating point registers.
12795 Generate code that uses registers (stack) for passing arguments to functions.
12796 By default, arguments are passed in registers where possible rather
12797 than by pushing arguments on to the stack.
12799 @item -mparallel-insns
12800 @itemx -mno-parallel-insns
12801 @opindex mparallel-insns
12802 @opindex mno-parallel-insns
12803 Allow the generation of parallel instructions. This is enabled by
12804 default with @option{-O2}.
12806 @item -mparallel-mpy
12807 @itemx -mno-parallel-mpy
12808 @opindex mparallel-mpy
12809 @opindex mno-parallel-mpy
12810 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12811 provided @option{-mparallel-insns} is also specified. These instructions have
12812 tight register constraints which can pessimize the code generation
12813 of large functions.
12818 @subsection V850 Options
12819 @cindex V850 Options
12821 These @samp{-m} options are defined for V850 implementations:
12825 @itemx -mno-long-calls
12826 @opindex mlong-calls
12827 @opindex mno-long-calls
12828 Treat all calls as being far away (near). If calls are assumed to be
12829 far away, the compiler will always load the functions address up into a
12830 register, and call indirect through the pointer.
12836 Do not optimize (do optimize) basic blocks that use the same index
12837 pointer 4 or more times to copy pointer into the @code{ep} register, and
12838 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12839 option is on by default if you optimize.
12841 @item -mno-prolog-function
12842 @itemx -mprolog-function
12843 @opindex mno-prolog-function
12844 @opindex mprolog-function
12845 Do not use (do use) external functions to save and restore registers
12846 at the prologue and epilogue of a function. The external functions
12847 are slower, but use less code space if more than one function saves
12848 the same number of registers. The @option{-mprolog-function} option
12849 is on by default if you optimize.
12853 Try to make the code as small as possible. At present, this just turns
12854 on the @option{-mep} and @option{-mprolog-function} options.
12856 @item -mtda=@var{n}
12858 Put static or global variables whose size is @var{n} bytes or less into
12859 the tiny data area that register @code{ep} points to. The tiny data
12860 area can hold up to 256 bytes in total (128 bytes for byte references).
12862 @item -msda=@var{n}
12864 Put static or global variables whose size is @var{n} bytes or less into
12865 the small data area that register @code{gp} points to. The small data
12866 area can hold up to 64 kilobytes.
12868 @item -mzda=@var{n}
12870 Put static or global variables whose size is @var{n} bytes or less into
12871 the first 32 kilobytes of memory.
12875 Specify that the target processor is the V850.
12878 @opindex mbig-switch
12879 Generate code suitable for big switch tables. Use this option only if
12880 the assembler/linker complain about out of range branches within a switch
12885 This option will cause r2 and r5 to be used in the code generated by
12886 the compiler. This setting is the default.
12888 @item -mno-app-regs
12889 @opindex mno-app-regs
12890 This option will cause r2 and r5 to be treated as fixed registers.
12894 Specify that the target processor is the V850E1. The preprocessor
12895 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
12896 this option is used.
12900 Specify that the target processor is the V850E@. The preprocessor
12901 constant @samp{__v850e__} will be defined if this option is used.
12903 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
12904 are defined then a default target processor will be chosen and the
12905 relevant @samp{__v850*__} preprocessor constant will be defined.
12907 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
12908 defined, regardless of which processor variant is the target.
12910 @item -mdisable-callt
12911 @opindex mdisable-callt
12912 This option will suppress generation of the CALLT instruction for the
12913 v850e and v850e1 flavors of the v850 architecture. The default is
12914 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
12919 @subsection VAX Options
12920 @cindex VAX options
12922 These @samp{-m} options are defined for the VAX:
12927 Do not output certain jump instructions (@code{aobleq} and so on)
12928 that the Unix assembler for the VAX cannot handle across long
12933 Do output those jump instructions, on the assumption that you
12934 will assemble with the GNU assembler.
12938 Output code for g-format floating point numbers instead of d-format.
12941 @node x86-64 Options
12942 @subsection x86-64 Options
12943 @cindex x86-64 options
12945 These are listed under @xref{i386 and x86-64 Options}.
12947 @node Xstormy16 Options
12948 @subsection Xstormy16 Options
12949 @cindex Xstormy16 Options
12951 These options are defined for Xstormy16:
12956 Choose startup files and linker script suitable for the simulator.
12959 @node Xtensa Options
12960 @subsection Xtensa Options
12961 @cindex Xtensa Options
12963 These options are supported for Xtensa targets:
12967 @itemx -mno-const16
12969 @opindex mno-const16
12970 Enable or disable use of @code{CONST16} instructions for loading
12971 constant values. The @code{CONST16} instruction is currently not a
12972 standard option from Tensilica. When enabled, @code{CONST16}
12973 instructions are always used in place of the standard @code{L32R}
12974 instructions. The use of @code{CONST16} is enabled by default only if
12975 the @code{L32R} instruction is not available.
12978 @itemx -mno-fused-madd
12979 @opindex mfused-madd
12980 @opindex mno-fused-madd
12981 Enable or disable use of fused multiply/add and multiply/subtract
12982 instructions in the floating-point option. This has no effect if the
12983 floating-point option is not also enabled. Disabling fused multiply/add
12984 and multiply/subtract instructions forces the compiler to use separate
12985 instructions for the multiply and add/subtract operations. This may be
12986 desirable in some cases where strict IEEE 754-compliant results are
12987 required: the fused multiply add/subtract instructions do not round the
12988 intermediate result, thereby producing results with @emph{more} bits of
12989 precision than specified by the IEEE standard. Disabling fused multiply
12990 add/subtract instructions also ensures that the program output is not
12991 sensitive to the compiler's ability to combine multiply and add/subtract
12994 @item -mtext-section-literals
12995 @itemx -mno-text-section-literals
12996 @opindex mtext-section-literals
12997 @opindex mno-text-section-literals
12998 Control the treatment of literal pools. The default is
12999 @option{-mno-text-section-literals}, which places literals in a separate
13000 section in the output file. This allows the literal pool to be placed
13001 in a data RAM/ROM, and it also allows the linker to combine literal
13002 pools from separate object files to remove redundant literals and
13003 improve code size. With @option{-mtext-section-literals}, the literals
13004 are interspersed in the text section in order to keep them as close as
13005 possible to their references. This may be necessary for large assembly
13008 @item -mtarget-align
13009 @itemx -mno-target-align
13010 @opindex mtarget-align
13011 @opindex mno-target-align
13012 When this option is enabled, GCC instructs the assembler to
13013 automatically align instructions to reduce branch penalties at the
13014 expense of some code density. The assembler attempts to widen density
13015 instructions to align branch targets and the instructions following call
13016 instructions. If there are not enough preceding safe density
13017 instructions to align a target, no widening will be performed. The
13018 default is @option{-mtarget-align}. These options do not affect the
13019 treatment of auto-aligned instructions like @code{LOOP}, which the
13020 assembler will always align, either by widening density instructions or
13021 by inserting no-op instructions.
13024 @itemx -mno-longcalls
13025 @opindex mlongcalls
13026 @opindex mno-longcalls
13027 When this option is enabled, GCC instructs the assembler to translate
13028 direct calls to indirect calls unless it can determine that the target
13029 of a direct call is in the range allowed by the call instruction. This
13030 translation typically occurs for calls to functions in other source
13031 files. Specifically, the assembler translates a direct @code{CALL}
13032 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13033 The default is @option{-mno-longcalls}. This option should be used in
13034 programs where the call target can potentially be out of range. This
13035 option is implemented in the assembler, not the compiler, so the
13036 assembly code generated by GCC will still show direct call
13037 instructions---look at the disassembled object code to see the actual
13038 instructions. Note that the assembler will use an indirect call for
13039 every cross-file call, not just those that really will be out of range.
13042 @node zSeries Options
13043 @subsection zSeries Options
13044 @cindex zSeries options
13046 These are listed under @xref{S/390 and zSeries Options}.
13048 @node Code Gen Options
13049 @section Options for Code Generation Conventions
13050 @cindex code generation conventions
13051 @cindex options, code generation
13052 @cindex run-time options
13054 These machine-independent options control the interface conventions
13055 used in code generation.
13057 Most of them have both positive and negative forms; the negative form
13058 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13059 one of the forms is listed---the one which is not the default. You
13060 can figure out the other form by either removing @samp{no-} or adding
13064 @item -fbounds-check
13065 @opindex fbounds-check
13066 For front-ends that support it, generate additional code to check that
13067 indices used to access arrays are within the declared range. This is
13068 currently only supported by the Java and Fortran 77 front-ends, where
13069 this option defaults to true and false respectively.
13073 This option generates traps for signed overflow on addition, subtraction,
13074 multiplication operations.
13078 This option instructs the compiler to assume that signed arithmetic
13079 overflow of addition, subtraction and multiplication wraps around
13080 using twos-complement representation. This flag enables some optimizations
13081 and disables others. This option is enabled by default for the Java
13082 front-end, as required by the Java language specification.
13085 @opindex fexceptions
13086 Enable exception handling. Generates extra code needed to propagate
13087 exceptions. For some targets, this implies GCC will generate frame
13088 unwind information for all functions, which can produce significant data
13089 size overhead, although it does not affect execution. If you do not
13090 specify this option, GCC will enable it by default for languages like
13091 C++ which normally require exception handling, and disable it for
13092 languages like C that do not normally require it. However, you may need
13093 to enable this option when compiling C code that needs to interoperate
13094 properly with exception handlers written in C++. You may also wish to
13095 disable this option if you are compiling older C++ programs that don't
13096 use exception handling.
13098 @item -fnon-call-exceptions
13099 @opindex fnon-call-exceptions
13100 Generate code that allows trapping instructions to throw exceptions.
13101 Note that this requires platform-specific runtime support that does
13102 not exist everywhere. Moreover, it only allows @emph{trapping}
13103 instructions to throw exceptions, i.e.@: memory references or floating
13104 point instructions. It does not allow exceptions to be thrown from
13105 arbitrary signal handlers such as @code{SIGALRM}.
13107 @item -funwind-tables
13108 @opindex funwind-tables
13109 Similar to @option{-fexceptions}, except that it will just generate any needed
13110 static data, but will not affect the generated code in any other way.
13111 You will normally not enable this option; instead, a language processor
13112 that needs this handling would enable it on your behalf.
13114 @item -fasynchronous-unwind-tables
13115 @opindex fasynchronous-unwind-tables
13116 Generate unwind table in dwarf2 format, if supported by target machine. The
13117 table is exact at each instruction boundary, so it can be used for stack
13118 unwinding from asynchronous events (such as debugger or garbage collector).
13120 @item -fpcc-struct-return
13121 @opindex fpcc-struct-return
13122 Return ``short'' @code{struct} and @code{union} values in memory like
13123 longer ones, rather than in registers. This convention is less
13124 efficient, but it has the advantage of allowing intercallability between
13125 GCC-compiled files and files compiled with other compilers, particularly
13126 the Portable C Compiler (pcc).
13128 The precise convention for returning structures in memory depends
13129 on the target configuration macros.
13131 Short structures and unions are those whose size and alignment match
13132 that of some integer type.
13134 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13135 switch is not binary compatible with code compiled with the
13136 @option{-freg-struct-return} switch.
13137 Use it to conform to a non-default application binary interface.
13139 @item -freg-struct-return
13140 @opindex freg-struct-return
13141 Return @code{struct} and @code{union} values in registers when possible.
13142 This is more efficient for small structures than
13143 @option{-fpcc-struct-return}.
13145 If you specify neither @option{-fpcc-struct-return} nor
13146 @option{-freg-struct-return}, GCC defaults to whichever convention is
13147 standard for the target. If there is no standard convention, GCC
13148 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13149 the principal compiler. In those cases, we can choose the standard, and
13150 we chose the more efficient register return alternative.
13152 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13153 switch is not binary compatible with code compiled with the
13154 @option{-fpcc-struct-return} switch.
13155 Use it to conform to a non-default application binary interface.
13157 @item -fshort-enums
13158 @opindex fshort-enums
13159 Allocate to an @code{enum} type only as many bytes as it needs for the
13160 declared range of possible values. Specifically, the @code{enum} type
13161 will be equivalent to the smallest integer type which has enough room.
13163 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13164 code that is not binary compatible with code generated without that switch.
13165 Use it to conform to a non-default application binary interface.
13167 @item -fshort-double
13168 @opindex fshort-double
13169 Use the same size for @code{double} as for @code{float}.
13171 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13172 code that is not binary compatible with code generated without that switch.
13173 Use it to conform to a non-default application binary interface.
13175 @item -fshort-wchar
13176 @opindex fshort-wchar
13177 Override the underlying type for @samp{wchar_t} to be @samp{short
13178 unsigned int} instead of the default for the target. This option is
13179 useful for building programs to run under WINE@.
13181 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13182 code that is not binary compatible with code generated without that switch.
13183 Use it to conform to a non-default application binary interface.
13186 @opindex fno-common
13187 In C, allocate even uninitialized global variables in the data section of the
13188 object file, rather than generating them as common blocks. This has the
13189 effect that if the same variable is declared (without @code{extern}) in
13190 two different compilations, you will get an error when you link them.
13191 The only reason this might be useful is if you wish to verify that the
13192 program will work on other systems which always work this way.
13196 Ignore the @samp{#ident} directive.
13198 @item -finhibit-size-directive
13199 @opindex finhibit-size-directive
13200 Don't output a @code{.size} assembler directive, or anything else that
13201 would cause trouble if the function is split in the middle, and the
13202 two halves are placed at locations far apart in memory. This option is
13203 used when compiling @file{crtstuff.c}; you should not need to use it
13206 @item -fverbose-asm
13207 @opindex fverbose-asm
13208 Put extra commentary information in the generated assembly code to
13209 make it more readable. This option is generally only of use to those
13210 who actually need to read the generated assembly code (perhaps while
13211 debugging the compiler itself).
13213 @option{-fno-verbose-asm}, the default, causes the
13214 extra information to be omitted and is useful when comparing two assembler
13219 @cindex global offset table
13221 Generate position-independent code (PIC) suitable for use in a shared
13222 library, if supported for the target machine. Such code accesses all
13223 constant addresses through a global offset table (GOT)@. The dynamic
13224 loader resolves the GOT entries when the program starts (the dynamic
13225 loader is not part of GCC; it is part of the operating system). If
13226 the GOT size for the linked executable exceeds a machine-specific
13227 maximum size, you get an error message from the linker indicating that
13228 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13229 instead. (These maximums are 8k on the SPARC and 32k
13230 on the m68k and RS/6000. The 386 has no such limit.)
13232 Position-independent code requires special support, and therefore works
13233 only on certain machines. For the 386, GCC supports PIC for System V
13234 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13235 position-independent.
13237 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13242 If supported for the target machine, emit position-independent code,
13243 suitable for dynamic linking and avoiding any limit on the size of the
13244 global offset table. This option makes a difference on the m68k,
13245 PowerPC and SPARC@.
13247 Position-independent code requires special support, and therefore works
13248 only on certain machines.
13250 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13257 These options are similar to @option{-fpic} and @option{-fPIC}, but
13258 generated position independent code can be only linked into executables.
13259 Usually these options are used when @option{-pie} GCC option will be
13260 used during linking.
13262 @item -fno-jump-tables
13263 @opindex fno-jump-tables
13264 Do not use jump tables for switch statements even where it would be
13265 more efficient than other code generation strategies. This option is
13266 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13267 building code which forms part of a dynamic linker and cannot
13268 reference the address of a jump table. On some targets, jump tables
13269 do not require a GOT and this option is not needed.
13271 @item -ffixed-@var{reg}
13273 Treat the register named @var{reg} as a fixed register; generated code
13274 should never refer to it (except perhaps as a stack pointer, frame
13275 pointer or in some other fixed role).
13277 @var{reg} must be the name of a register. The register names accepted
13278 are machine-specific and are defined in the @code{REGISTER_NAMES}
13279 macro in the machine description macro file.
13281 This flag does not have a negative form, because it specifies a
13284 @item -fcall-used-@var{reg}
13285 @opindex fcall-used
13286 Treat the register named @var{reg} as an allocable register that is
13287 clobbered by function calls. It may be allocated for temporaries or
13288 variables that do not live across a call. Functions compiled this way
13289 will not save and restore the register @var{reg}.
13291 It is an error to used this flag with the frame pointer or stack pointer.
13292 Use of this flag for other registers that have fixed pervasive roles in
13293 the machine's execution model will produce disastrous results.
13295 This flag does not have a negative form, because it specifies a
13298 @item -fcall-saved-@var{reg}
13299 @opindex fcall-saved
13300 Treat the register named @var{reg} as an allocable register saved by
13301 functions. It may be allocated even for temporaries or variables that
13302 live across a call. Functions compiled this way will save and restore
13303 the register @var{reg} if they use it.
13305 It is an error to used this flag with the frame pointer or stack pointer.
13306 Use of this flag for other registers that have fixed pervasive roles in
13307 the machine's execution model will produce disastrous results.
13309 A different sort of disaster will result from the use of this flag for
13310 a register in which function values may be returned.
13312 This flag does not have a negative form, because it specifies a
13315 @item -fpack-struct[=@var{n}]
13316 @opindex fpack-struct
13317 Without a value specified, pack all structure members together without
13318 holes. When a value is specified (which must be a small power of two), pack
13319 structure members according to this value, representing the maximum
13320 alignment (that is, objects with default alignment requirements larger than
13321 this will be output potentially unaligned at the next fitting location.
13323 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13324 code that is not binary compatible with code generated without that switch.
13325 Additionally, it makes the code suboptimal.
13326 Use it to conform to a non-default application binary interface.
13328 @item -finstrument-functions
13329 @opindex finstrument-functions
13330 Generate instrumentation calls for entry and exit to functions. Just
13331 after function entry and just before function exit, the following
13332 profiling functions will be called with the address of the current
13333 function and its call site. (On some platforms,
13334 @code{__builtin_return_address} does not work beyond the current
13335 function, so the call site information may not be available to the
13336 profiling functions otherwise.)
13339 void __cyg_profile_func_enter (void *this_fn,
13341 void __cyg_profile_func_exit (void *this_fn,
13345 The first argument is the address of the start of the current function,
13346 which may be looked up exactly in the symbol table.
13348 This instrumentation is also done for functions expanded inline in other
13349 functions. The profiling calls will indicate where, conceptually, the
13350 inline function is entered and exited. This means that addressable
13351 versions of such functions must be available. If all your uses of a
13352 function are expanded inline, this may mean an additional expansion of
13353 code size. If you use @samp{extern inline} in your C code, an
13354 addressable version of such functions must be provided. (This is
13355 normally the case anyways, but if you get lucky and the optimizer always
13356 expands the functions inline, you might have gotten away without
13357 providing static copies.)
13359 A function may be given the attribute @code{no_instrument_function}, in
13360 which case this instrumentation will not be done. This can be used, for
13361 example, for the profiling functions listed above, high-priority
13362 interrupt routines, and any functions from which the profiling functions
13363 cannot safely be called (perhaps signal handlers, if the profiling
13364 routines generate output or allocate memory).
13366 @item -fstack-check
13367 @opindex fstack-check
13368 Generate code to verify that you do not go beyond the boundary of the
13369 stack. You should specify this flag if you are running in an
13370 environment with multiple threads, but only rarely need to specify it in
13371 a single-threaded environment since stack overflow is automatically
13372 detected on nearly all systems if there is only one stack.
13374 Note that this switch does not actually cause checking to be done; the
13375 operating system must do that. The switch causes generation of code
13376 to ensure that the operating system sees the stack being extended.
13378 @item -fstack-limit-register=@var{reg}
13379 @itemx -fstack-limit-symbol=@var{sym}
13380 @itemx -fno-stack-limit
13381 @opindex fstack-limit-register
13382 @opindex fstack-limit-symbol
13383 @opindex fno-stack-limit
13384 Generate code to ensure that the stack does not grow beyond a certain value,
13385 either the value of a register or the address of a symbol. If the stack
13386 would grow beyond the value, a signal is raised. For most targets,
13387 the signal is raised before the stack overruns the boundary, so
13388 it is possible to catch the signal without taking special precautions.
13390 For instance, if the stack starts at absolute address @samp{0x80000000}
13391 and grows downwards, you can use the flags
13392 @option{-fstack-limit-symbol=__stack_limit} and
13393 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13394 of 128KB@. Note that this may only work with the GNU linker.
13396 @cindex aliasing of parameters
13397 @cindex parameters, aliased
13398 @item -fargument-alias
13399 @itemx -fargument-noalias
13400 @itemx -fargument-noalias-global
13401 @itemx -fargument-noalias-anything
13402 @opindex fargument-alias
13403 @opindex fargument-noalias
13404 @opindex fargument-noalias-global
13405 @opindex fargument-noalias-anything
13406 Specify the possible relationships among parameters and between
13407 parameters and global data.
13409 @option{-fargument-alias} specifies that arguments (parameters) may
13410 alias each other and may alias global storage.@*
13411 @option{-fargument-noalias} specifies that arguments do not alias
13412 each other, but may alias global storage.@*
13413 @option{-fargument-noalias-global} specifies that arguments do not
13414 alias each other and do not alias global storage.
13415 @option{-fargument-noalias-anything} specifies that arguments do not
13416 alias any other storage.
13418 Each language will automatically use whatever option is required by
13419 the language standard. You should not need to use these options yourself.
13421 @item -fleading-underscore
13422 @opindex fleading-underscore
13423 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13424 change the way C symbols are represented in the object file. One use
13425 is to help link with legacy assembly code.
13427 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13428 generate code that is not binary compatible with code generated without that
13429 switch. Use it to conform to a non-default application binary interface.
13430 Not all targets provide complete support for this switch.
13432 @item -ftls-model=@var{model}
13433 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13434 The @var{model} argument should be one of @code{global-dynamic},
13435 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13437 The default without @option{-fpic} is @code{initial-exec}; with
13438 @option{-fpic} the default is @code{global-dynamic}.
13440 @item -fvisibility=@var{default|internal|hidden|protected}
13441 @opindex fvisibility
13442 Set the default ELF image symbol visibility to the specified option---all
13443 symbols will be marked with this unless overridden within the code.
13444 Using this feature can very substantially improve linking and
13445 load times of shared object libraries, produce more optimized
13446 code, provide near-perfect API export and prevent symbol clashes.
13447 It is @strong{strongly} recommended that you use this in any shared objects
13450 Despite the nomenclature, @code{default} always means public ie;
13451 available to be linked against from outside the shared object.
13452 @code{protected} and @code{internal} are pretty useless in real-world
13453 usage so the only other commonly used option will be @code{hidden}.
13454 The default if @option{-fvisibility} isn't specified is
13455 @code{default}, i.e., make every
13456 symbol public---this causes the same behavior as previous versions of
13459 A good explanation of the benefits offered by ensuring ELF
13460 symbols have the correct visibility is given by ``How To Write
13461 Shared Libraries'' by Ulrich Drepper (which can be found at
13462 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13463 solution made possible by this option to marking things hidden when
13464 the default is public is to make the default hidden and mark things
13465 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13466 and @code{__attribute__ ((visibility("default")))} instead of
13467 @code{__declspec(dllexport)} you get almost identical semantics with
13468 identical syntax. This is a great boon to those working with
13469 cross-platform projects.
13471 For those adding visibility support to existing code, you may find
13472 @samp{#pragma GCC visibility} of use. This works by you enclosing
13473 the declarations you wish to set visibility for with (for example)
13474 @samp{#pragma GCC visibility push(hidden)} and
13475 @samp{#pragma GCC visibility pop}.
13476 Bear in mind that symbol visibility should be viewed @strong{as
13477 part of the API interface contract} and thus all new code should
13478 always specify visibility when it is not the default ie; declarations
13479 only for use within the local DSO should @strong{always} be marked explicitly
13480 as hidden as so to avoid PLT indirection overheads---making this
13481 abundantly clear also aids readability and self-documentation of the code.
13482 Note that due to ISO C++ specification requirements, operator new and
13483 operator delete must always be of default visibility.
13485 Be aware that headers from outside your project, in particular system
13486 headers and headers from any other library you use, may not be
13487 expecting to be compiled with visibility other than the default. You
13488 may need to explicitly say @samp{#pragma GCC visibility push(default)}
13489 before including any such headers.
13491 An overview of these techniques, their benefits and how to use them
13492 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13496 @cindex openmp parallel
13497 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
13498 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
13499 compiler generates parallel code according to the OpenMP Application
13500 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
13506 @node Environment Variables
13507 @section Environment Variables Affecting GCC
13508 @cindex environment variables
13510 @c man begin ENVIRONMENT
13511 This section describes several environment variables that affect how GCC
13512 operates. Some of them work by specifying directories or prefixes to use
13513 when searching for various kinds of files. Some are used to specify other
13514 aspects of the compilation environment.
13516 Note that you can also specify places to search using options such as
13517 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13518 take precedence over places specified using environment variables, which
13519 in turn take precedence over those specified by the configuration of GCC@.
13520 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13521 GNU Compiler Collection (GCC) Internals}.
13526 @c @itemx LC_COLLATE
13528 @c @itemx LC_MONETARY
13529 @c @itemx LC_NUMERIC
13534 @c @findex LC_COLLATE
13535 @findex LC_MESSAGES
13536 @c @findex LC_MONETARY
13537 @c @findex LC_NUMERIC
13541 These environment variables control the way that GCC uses
13542 localization information that allow GCC to work with different
13543 national conventions. GCC inspects the locale categories
13544 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13545 so. These locale categories can be set to any value supported by your
13546 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13547 Kingdom encoded in UTF-8.
13549 The @env{LC_CTYPE} environment variable specifies character
13550 classification. GCC uses it to determine the character boundaries in
13551 a string; this is needed for some multibyte encodings that contain quote
13552 and escape characters that would otherwise be interpreted as a string
13555 The @env{LC_MESSAGES} environment variable specifies the language to
13556 use in diagnostic messages.
13558 If the @env{LC_ALL} environment variable is set, it overrides the value
13559 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13560 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13561 environment variable. If none of these variables are set, GCC
13562 defaults to traditional C English behavior.
13566 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13567 files. GCC uses temporary files to hold the output of one stage of
13568 compilation which is to be used as input to the next stage: for example,
13569 the output of the preprocessor, which is the input to the compiler
13572 @item GCC_EXEC_PREFIX
13573 @findex GCC_EXEC_PREFIX
13574 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13575 names of the subprograms executed by the compiler. No slash is added
13576 when this prefix is combined with the name of a subprogram, but you can
13577 specify a prefix that ends with a slash if you wish.
13579 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13580 an appropriate prefix to use based on the pathname it was invoked with.
13582 If GCC cannot find the subprogram using the specified prefix, it
13583 tries looking in the usual places for the subprogram.
13585 The default value of @env{GCC_EXEC_PREFIX} is
13586 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13587 of @code{prefix} when you ran the @file{configure} script.
13589 Other prefixes specified with @option{-B} take precedence over this prefix.
13591 This prefix is also used for finding files such as @file{crt0.o} that are
13594 In addition, the prefix is used in an unusual way in finding the
13595 directories to search for header files. For each of the standard
13596 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13597 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13598 replacing that beginning with the specified prefix to produce an
13599 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13600 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13601 These alternate directories are searched first; the standard directories
13604 @item COMPILER_PATH
13605 @findex COMPILER_PATH
13606 The value of @env{COMPILER_PATH} is a colon-separated list of
13607 directories, much like @env{PATH}. GCC tries the directories thus
13608 specified when searching for subprograms, if it can't find the
13609 subprograms using @env{GCC_EXEC_PREFIX}.
13612 @findex LIBRARY_PATH
13613 The value of @env{LIBRARY_PATH} is a colon-separated list of
13614 directories, much like @env{PATH}. When configured as a native compiler,
13615 GCC tries the directories thus specified when searching for special
13616 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13617 using GCC also uses these directories when searching for ordinary
13618 libraries for the @option{-l} option (but directories specified with
13619 @option{-L} come first).
13623 @cindex locale definition
13624 This variable is used to pass locale information to the compiler. One way in
13625 which this information is used is to determine the character set to be used
13626 when character literals, string literals and comments are parsed in C and C++.
13627 When the compiler is configured to allow multibyte characters,
13628 the following values for @env{LANG} are recognized:
13632 Recognize JIS characters.
13634 Recognize SJIS characters.
13636 Recognize EUCJP characters.
13639 If @env{LANG} is not defined, or if it has some other value, then the
13640 compiler will use mblen and mbtowc as defined by the default locale to
13641 recognize and translate multibyte characters.
13645 Some additional environments variables affect the behavior of the
13648 @include cppenv.texi
13652 @node Precompiled Headers
13653 @section Using Precompiled Headers
13654 @cindex precompiled headers
13655 @cindex speed of compilation
13657 Often large projects have many header files that are included in every
13658 source file. The time the compiler takes to process these header files
13659 over and over again can account for nearly all of the time required to
13660 build the project. To make builds faster, GCC allows users to
13661 `precompile' a header file; then, if builds can use the precompiled
13662 header file they will be much faster.
13664 To create a precompiled header file, simply compile it as you would any
13665 other file, if necessary using the @option{-x} option to make the driver
13666 treat it as a C or C++ header file. You will probably want to use a
13667 tool like @command{make} to keep the precompiled header up-to-date when
13668 the headers it contains change.
13670 A precompiled header file will be searched for when @code{#include} is
13671 seen in the compilation. As it searches for the included file
13672 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13673 compiler looks for a precompiled header in each directory just before it
13674 looks for the include file in that directory. The name searched for is
13675 the name specified in the @code{#include} with @samp{.gch} appended. If
13676 the precompiled header file can't be used, it is ignored.
13678 For instance, if you have @code{#include "all.h"}, and you have
13679 @file{all.h.gch} in the same directory as @file{all.h}, then the
13680 precompiled header file will be used if possible, and the original
13681 header will be used otherwise.
13683 Alternatively, you might decide to put the precompiled header file in a
13684 directory and use @option{-I} to ensure that directory is searched
13685 before (or instead of) the directory containing the original header.
13686 Then, if you want to check that the precompiled header file is always
13687 used, you can put a file of the same name as the original header in this
13688 directory containing an @code{#error} command.
13690 This also works with @option{-include}. So yet another way to use
13691 precompiled headers, good for projects not designed with precompiled
13692 header files in mind, is to simply take most of the header files used by
13693 a project, include them from another header file, precompile that header
13694 file, and @option{-include} the precompiled header. If the header files
13695 have guards against multiple inclusion, they will be skipped because
13696 they've already been included (in the precompiled header).
13698 If you need to precompile the same header file for different
13699 languages, targets, or compiler options, you can instead make a
13700 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13701 header in the directory, perhaps using @option{-o}. It doesn't matter
13702 what you call the files in the directory, every precompiled header in
13703 the directory will be considered. The first precompiled header
13704 encountered in the directory that is valid for this compilation will
13705 be used; they're searched in no particular order.
13707 There are many other possibilities, limited only by your imagination,
13708 good sense, and the constraints of your build system.
13710 A precompiled header file can be used only when these conditions apply:
13714 Only one precompiled header can be used in a particular compilation.
13717 A precompiled header can't be used once the first C token is seen. You
13718 can have preprocessor directives before a precompiled header; you can
13719 even include a precompiled header from inside another header, so long as
13720 there are no C tokens before the @code{#include}.
13723 The precompiled header file must be produced for the same language as
13724 the current compilation. You can't use a C precompiled header for a C++
13728 The precompiled header file must have been produced by the same compiler
13729 binary as the current compilation is using.
13732 Any macros defined before the precompiled header is included must
13733 either be defined in the same way as when the precompiled header was
13734 generated, or must not affect the precompiled header, which usually
13735 means that they don't appear in the precompiled header at all.
13737 The @option{-D} option is one way to define a macro before a
13738 precompiled header is included; using a @code{#define} can also do it.
13739 There are also some options that define macros implicitly, like
13740 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13743 @item If debugging information is output when using the precompiled
13744 header, using @option{-g} or similar, the same kind of debugging information
13745 must have been output when building the precompiled header. However,
13746 a precompiled header built using @option{-g} can be used in a compilation
13747 when no debugging information is being output.
13749 @item The same @option{-m} options must generally be used when building
13750 and using the precompiled header. @xref{Submodel Options},
13751 for any cases where this rule is relaxed.
13753 @item Each of the following options must be the same when building and using
13754 the precompiled header:
13756 @gccoptlist{-fexceptions -funit-at-a-time}
13759 Some other command-line options starting with @option{-f},
13760 @option{-p}, or @option{-O} must be defined in the same way as when
13761 the precompiled header was generated. At present, it's not clear
13762 which options are safe to change and which are not; the safest choice
13763 is to use exactly the same options when generating and using the
13764 precompiled header. The following are known to be safe:
13766 @gccoptlist{-fmessage-length= -fpreprocessed
13767 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13768 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13773 For all of these except the last, the compiler will automatically
13774 ignore the precompiled header if the conditions aren't met. If you
13775 find an option combination that doesn't work and doesn't cause the
13776 precompiled header to be ignored, please consider filing a bug report,
13779 If you do use differing options when generating and using the
13780 precompiled header, the actual behavior will be a mixture of the
13781 behavior for the options. For instance, if you use @option{-g} to
13782 generate the precompiled header but not when using it, you may or may
13783 not get debugging information for routines in the precompiled header.
13785 @node Running Protoize
13786 @section Running Protoize
13788 The program @code{protoize} is an optional part of GCC@. You can use
13789 it to add prototypes to a program, thus converting the program to ISO
13790 C in one respect. The companion program @code{unprotoize} does the
13791 reverse: it removes argument types from any prototypes that are found.
13793 When you run these programs, you must specify a set of source files as
13794 command line arguments. The conversion programs start out by compiling
13795 these files to see what functions they define. The information gathered
13796 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13798 After scanning comes actual conversion. The specified files are all
13799 eligible to be converted; any files they include (whether sources or
13800 just headers) are eligible as well.
13802 But not all the eligible files are converted. By default,
13803 @code{protoize} and @code{unprotoize} convert only source and header
13804 files in the current directory. You can specify additional directories
13805 whose files should be converted with the @option{-d @var{directory}}
13806 option. You can also specify particular files to exclude with the
13807 @option{-x @var{file}} option. A file is converted if it is eligible, its
13808 directory name matches one of the specified directory names, and its
13809 name within the directory has not been excluded.
13811 Basic conversion with @code{protoize} consists of rewriting most
13812 function definitions and function declarations to specify the types of
13813 the arguments. The only ones not rewritten are those for varargs
13816 @code{protoize} optionally inserts prototype declarations at the
13817 beginning of the source file, to make them available for any calls that
13818 precede the function's definition. Or it can insert prototype
13819 declarations with block scope in the blocks where undeclared functions
13822 Basic conversion with @code{unprotoize} consists of rewriting most
13823 function declarations to remove any argument types, and rewriting
13824 function definitions to the old-style pre-ISO form.
13826 Both conversion programs print a warning for any function declaration or
13827 definition that they can't convert. You can suppress these warnings
13830 The output from @code{protoize} or @code{unprotoize} replaces the
13831 original source file. The original file is renamed to a name ending
13832 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13833 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13834 for DOS) file already exists, then the source file is simply discarded.
13836 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13837 scan the program and collect information about the functions it uses.
13838 So neither of these programs will work until GCC is installed.
13840 Here is a table of the options you can use with @code{protoize} and
13841 @code{unprotoize}. Each option works with both programs unless
13845 @item -B @var{directory}
13846 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13847 usual directory (normally @file{/usr/local/lib}). This file contains
13848 prototype information about standard system functions. This option
13849 applies only to @code{protoize}.
13851 @item -c @var{compilation-options}
13852 Use @var{compilation-options} as the options when running @command{gcc} to
13853 produce the @samp{.X} files. The special option @option{-aux-info} is
13854 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13856 Note that the compilation options must be given as a single argument to
13857 @code{protoize} or @code{unprotoize}. If you want to specify several
13858 @command{gcc} options, you must quote the entire set of compilation options
13859 to make them a single word in the shell.
13861 There are certain @command{gcc} arguments that you cannot use, because they
13862 would produce the wrong kind of output. These include @option{-g},
13863 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13864 the @var{compilation-options}, they are ignored.
13867 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13868 systems) instead of @samp{.c}. This is convenient if you are converting
13869 a C program to C++. This option applies only to @code{protoize}.
13872 Add explicit global declarations. This means inserting explicit
13873 declarations at the beginning of each source file for each function
13874 that is called in the file and was not declared. These declarations
13875 precede the first function definition that contains a call to an
13876 undeclared function. This option applies only to @code{protoize}.
13878 @item -i @var{string}
13879 Indent old-style parameter declarations with the string @var{string}.
13880 This option applies only to @code{protoize}.
13882 @code{unprotoize} converts prototyped function definitions to old-style
13883 function definitions, where the arguments are declared between the
13884 argument list and the initial @samp{@{}. By default, @code{unprotoize}
13885 uses five spaces as the indentation. If you want to indent with just
13886 one space instead, use @option{-i " "}.
13889 Keep the @samp{.X} files. Normally, they are deleted after conversion
13893 Add explicit local declarations. @code{protoize} with @option{-l} inserts
13894 a prototype declaration for each function in each block which calls the
13895 function without any declaration. This option applies only to
13899 Make no real changes. This mode just prints information about the conversions
13900 that would have been done without @option{-n}.
13903 Make no @samp{.save} files. The original files are simply deleted.
13904 Use this option with caution.
13906 @item -p @var{program}
13907 Use the program @var{program} as the compiler. Normally, the name
13908 @file{gcc} is used.
13911 Work quietly. Most warnings are suppressed.
13914 Print the version number, just like @option{-v} for @command{gcc}.
13917 If you need special compiler options to compile one of your program's
13918 source files, then you should generate that file's @samp{.X} file
13919 specially, by running @command{gcc} on that source file with the
13920 appropriate options and the option @option{-aux-info}. Then run
13921 @code{protoize} on the entire set of files. @code{protoize} will use
13922 the existing @samp{.X} file because it is newer than the source file.
13926 gcc -Dfoo=bar file1.c -aux-info file1.X
13931 You need to include the special files along with the rest in the
13932 @code{protoize} command, even though their @samp{.X} files already
13933 exist, because otherwise they won't get converted.
13935 @xref{Protoize Caveats}, for more information on how to use
13936 @code{protoize} successfully.