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{-fstrength-reduce}, @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 -fno-const-strings @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 -Wpacked -Wpadded @gol
242 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
243 -Wredundant-decls @gol
244 -Wreturn-type -Wsequence-point -Wshadow @gol
245 -Wsign-compare -Wstack-protector @gol
246 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
247 -Wstring-literal-comparison @gol
248 -Wswitch -Wswitch-default -Wswitch-enum @gol
249 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
250 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
251 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
252 -Wunused-value -Wunused-variable -Wvariadic-macros @gol
253 -Wvolatile-register-var -Wwrite-strings}
255 @item C-only Warning Options
256 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
257 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
258 -Wstrict-prototypes -Wtraditional @gol
259 -Wdeclaration-after-statement -Wno-pointer-sign}
261 @item Debugging Options
262 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
263 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
264 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
265 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
266 -fdump-ipa-all -fdump-ipa-cgraph @gol
268 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
269 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
270 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
271 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
273 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
274 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
275 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
276 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
277 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
278 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
279 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-nrv -fdump-tree-vect @gol
282 -fdump-tree-sink @gol
283 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
284 -fdump-tree-salias @gol
285 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
287 -ftree-vectorizer-verbose=@var{n} @gol
288 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
289 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
290 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
291 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
292 -ftest-coverage -ftime-report -fvar-tracking @gol
293 -g -g@var{level} -gcoff -gdwarf-2 @gol
294 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
295 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
296 -print-multi-directory -print-multi-lib @gol
297 -print-prog-name=@var{program} -print-search-dirs -Q @gol
300 @item Optimization Options
301 @xref{Optimize Options,,Options that Control Optimization}.
302 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
303 -falign-labels=@var{n} -falign-loops=@var{n} @gol
304 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
305 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
306 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
307 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
308 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
309 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
310 -fexpensive-optimizations -ffast-math -ffloat-store @gol
311 -fforce-addr -ffunction-sections @gol
312 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
313 -floop-optimize -fcrossjumping -fif-conversion -fif-conversion2 @gol
314 -finline-functions -finline-functions-called-once @gol
315 -finline-limit=@var{n} -fkeep-inline-functions @gol
316 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
317 -fmodulo-sched -fno-branch-count-reg @gol
318 -fno-default-inline -fno-defer-pop -floop-optimize2 -fmove-loop-invariants @gol
319 -fno-function-cse -fno-guess-branch-probability @gol
320 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
321 -funsafe-math-optimizations -funsafe-loop-optimizations -ffinite-math-only @gol
322 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
323 -fomit-frame-pointer -foptimize-register-move @gol
324 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
325 -fprofile-generate -fprofile-use @gol
326 -fregmove -frename-registers @gol
327 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
328 -frerun-cse-after-loop -frerun-loop-opt @gol
329 -frounding-math -frtl-abstract-sequences @gol
330 -fschedule-insns -fschedule-insns2 @gol
331 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
332 -fsched-spec-load-dangerous @gol
333 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
334 -fsched2-use-superblocks @gol
335 -fsched2-use-traces -freschedule-modulo-scheduled-loops @gol
336 -fsignaling-nans -fsingle-precision-constant @gol
337 -fstack-protector -fstack-protector-all @gol
338 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
339 -funroll-all-loops -funroll-loops -fpeel-loops @gol
340 -fsplit-ivs-in-unroller -funswitch-loops @gol
341 -fvariable-expansion-in-unroller @gol
342 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
343 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
344 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
345 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
346 -ftree-vect-loop-version -ftree-salias -fweb @gol
347 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
348 --param @var{name}=@var{value}
349 -O -O0 -O1 -O2 -O3 -Os}
351 @item Preprocessor Options
352 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
353 @gccoptlist{-A@var{question}=@var{answer} @gol
354 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
355 -C -dD -dI -dM -dN @gol
356 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
357 -idirafter @var{dir} @gol
358 -include @var{file} -imacros @var{file} @gol
359 -iprefix @var{file} -iwithprefix @var{dir} @gol
360 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
361 -imultilib @var{dir} -isysroot @var{dir} @gol
362 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
363 -P -fworking-directory -remap @gol
364 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
365 -Xpreprocessor @var{option}}
367 @item Assembler Option
368 @xref{Assembler Options,,Passing Options to the Assembler}.
369 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
372 @xref{Link Options,,Options for Linking}.
373 @gccoptlist{@var{object-file-name} -l@var{library} @gol
374 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
375 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
376 -Wl,@var{option} -Xlinker @var{option} @gol
379 @item Directory Options
380 @xref{Directory Options,,Options for Directory Search}.
381 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
382 -specs=@var{file} -I- --sysroot=@var{dir}}
385 @c I wrote this xref this way to avoid overfull hbox. -- rms
386 @xref{Target Options}.
387 @gccoptlist{-V @var{version} -b @var{machine}}
389 @item Machine Dependent Options
390 @xref{Submodel Options,,Hardware Models and Configurations}.
391 @c This list is ordered alphanumerically by subsection name.
392 @c Try and put the significant identifier (CPU or system) first,
393 @c so users have a clue at guessing where the ones they want will be.
396 @gccoptlist{-EB -EL @gol
397 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
398 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
401 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
402 -mabi=@var{name} @gol
403 -mapcs-stack-check -mno-apcs-stack-check @gol
404 -mapcs-float -mno-apcs-float @gol
405 -mapcs-reentrant -mno-apcs-reentrant @gol
406 -msched-prolog -mno-sched-prolog @gol
407 -mlittle-endian -mbig-endian -mwords-little-endian @gol
408 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
409 -mthumb-interwork -mno-thumb-interwork @gol
410 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
411 -mstructure-size-boundary=@var{n} @gol
412 -mabort-on-noreturn @gol
413 -mlong-calls -mno-long-calls @gol
414 -msingle-pic-base -mno-single-pic-base @gol
415 -mpic-register=@var{reg} @gol
416 -mnop-fun-dllimport @gol
417 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
418 -mpoke-function-name @gol
420 -mtpcs-frame -mtpcs-leaf-frame @gol
421 -mcaller-super-interworking -mcallee-super-interworking @gol
425 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
426 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
428 @emph{Blackfin Options}
429 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
430 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
431 -mlow-64k -mno-low64k -mid-shared-library @gol
432 -mno-id-shared-library -mshared-library-id=@var{n} @gol
433 -mlong-calls -mno-long-calls}
436 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
437 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
438 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
439 -mstack-align -mdata-align -mconst-align @gol
440 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
441 -melf -maout -melinux -mlinux -sim -sim2 @gol
442 -mmul-bug-workaround -mno-mul-bug-workaround}
445 @gccoptlist{-mmac -mpush-args}
447 @emph{Darwin Options}
448 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
449 -arch_only -bind_at_load -bundle -bundle_loader @gol
450 -client_name -compatibility_version -current_version @gol
452 -dependency-file -dylib_file -dylinker_install_name @gol
453 -dynamic -dynamiclib -exported_symbols_list @gol
454 -filelist -flat_namespace -force_cpusubtype_ALL @gol
455 -force_flat_namespace -headerpad_max_install_names @gol
456 -image_base -init -install_name -keep_private_externs @gol
457 -multi_module -multiply_defined -multiply_defined_unused @gol
458 -noall_load -no_dead_strip_inits_and_terms @gol
459 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
460 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
461 -private_bundle -read_only_relocs -sectalign @gol
462 -sectobjectsymbols -whyload -seg1addr @gol
463 -sectcreate -sectobjectsymbols -sectorder @gol
464 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
465 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
466 -segprot -segs_read_only_addr -segs_read_write_addr @gol
467 -single_module -static -sub_library -sub_umbrella @gol
468 -twolevel_namespace -umbrella -undefined @gol
469 -unexported_symbols_list -weak_reference_mismatches @gol
470 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
473 @emph{DEC Alpha Options}
474 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
475 -mieee -mieee-with-inexact -mieee-conformant @gol
476 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
477 -mtrap-precision=@var{mode} -mbuild-constants @gol
478 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
479 -mbwx -mmax -mfix -mcix @gol
480 -mfloat-vax -mfloat-ieee @gol
481 -mexplicit-relocs -msmall-data -mlarge-data @gol
482 -msmall-text -mlarge-text @gol
483 -mmemory-latency=@var{time}}
485 @emph{DEC Alpha/VMS Options}
486 @gccoptlist{-mvms-return-codes}
489 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
490 -mhard-float -msoft-float @gol
491 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
492 -mdouble -mno-double @gol
493 -mmedia -mno-media -mmuladd -mno-muladd @gol
494 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
495 -mlinked-fp -mlong-calls -malign-labels @gol
496 -mlibrary-pic -macc-4 -macc-8 @gol
497 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
498 -moptimize-membar -mno-optimize-membar @gol
499 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
500 -mvliw-branch -mno-vliw-branch @gol
501 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
502 -mno-nested-cond-exec -mtomcat-stats @gol
506 @emph{H8/300 Options}
507 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
510 @gccoptlist{-march=@var{architecture-type} @gol
511 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
512 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
513 -mfixed-range=@var{register-range} @gol
514 -mjump-in-delay -mlinker-opt -mlong-calls @gol
515 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
516 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
517 -mno-jump-in-delay -mno-long-load-store @gol
518 -mno-portable-runtime -mno-soft-float @gol
519 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
520 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
521 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
522 -munix=@var{unix-std} -nolibdld -static -threads}
524 @emph{i386 and x86-64 Options}
525 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
526 -mfpmath=@var{unit} @gol
527 -masm=@var{dialect} -mno-fancy-math-387 @gol
528 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
529 -mno-wide-multiply -mrtd -malign-double @gol
530 -mpreferred-stack-boundary=@var{num} @gol
531 -mmmx -msse -msse2 -msse3 -m3dnow @gol
532 -mthreads -mno-align-stringops -minline-all-stringops @gol
533 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
534 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
535 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
536 -mcmodel=@var{code-model} @gol
537 -m32 -m64 -mlarge-data-threshold=@var{num}}
540 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
541 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
542 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
543 -minline-float-divide-max-throughput @gol
544 -minline-int-divide-min-latency @gol
545 -minline-int-divide-max-throughput @gol
546 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
547 -mno-dwarf2-asm -mearly-stop-bits @gol
548 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
549 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64}
551 @emph{M32R/D Options}
552 @gccoptlist{-m32r2 -m32rx -m32r @gol
554 -malign-loops -mno-align-loops @gol
555 -missue-rate=@var{number} @gol
556 -mbranch-cost=@var{number} @gol
557 -mmodel=@var{code-size-model-type} @gol
558 -msdata=@var{sdata-type} @gol
559 -mno-flush-func -mflush-func=@var{name} @gol
560 -mno-flush-trap -mflush-trap=@var{number} @gol
564 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
566 @emph{M680x0 Options}
567 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
568 -m68060 -mcpu32 -m5200 -mcfv4e -m68881 -mbitfield @gol
569 -mc68000 -mc68020 @gol
570 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
571 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
572 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
574 @emph{M68hc1x Options}
575 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
576 -mauto-incdec -minmax -mlong-calls -mshort @gol
577 -msoft-reg-count=@var{count}}
580 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
581 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
582 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
583 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
584 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
587 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
588 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
589 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
590 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
591 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
592 -mdsp -mpaired-single -mips3d @gol
593 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
594 -G@var{num} -membedded-data -mno-embedded-data @gol
595 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
596 -msplit-addresses -mno-split-addresses @gol
597 -mexplicit-relocs -mno-explicit-relocs @gol
598 -mcheck-zero-division -mno-check-zero-division @gol
599 -mdivide-traps -mdivide-breaks @gol
600 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
601 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
602 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
603 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
604 -mfix-sb1 -mno-fix-sb1 @gol
605 -mflush-func=@var{func} -mno-flush-func @gol
606 -mbranch-likely -mno-branch-likely @gol
607 -mfp-exceptions -mno-fp-exceptions @gol
608 -mvr4130-align -mno-vr4130-align}
611 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
612 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
613 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
614 -mno-base-addresses -msingle-exit -mno-single-exit}
616 @emph{MN10300 Options}
617 @gccoptlist{-mmult-bug -mno-mult-bug @gol
618 -mam33 -mno-am33 @gol
619 -mam33-2 -mno-am33-2 @gol
620 -mreturn-pointer-on-d0 @gol
624 @gccoptlist{-mno-crt0 -mbacc -msim @gol
625 -march=@var{cpu-type} }
627 @emph{PDP-11 Options}
628 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
629 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
630 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
631 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
632 -mbranch-expensive -mbranch-cheap @gol
633 -msplit -mno-split -munix-asm -mdec-asm}
635 @emph{PowerPC Options}
636 See RS/6000 and PowerPC Options.
638 @emph{RS/6000 and PowerPC Options}
639 @gccoptlist{-mcpu=@var{cpu-type} @gol
640 -mtune=@var{cpu-type} @gol
641 -mpower -mno-power -mpower2 -mno-power2 @gol
642 -mpowerpc -mpowerpc64 -mno-powerpc @gol
643 -maltivec -mno-altivec @gol
644 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
645 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
646 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
647 -mnew-mnemonics -mold-mnemonics @gol
648 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
649 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
650 -malign-power -malign-natural @gol
651 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
652 -mstring -mno-string -mupdate -mno-update @gol
653 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
654 -mstrict-align -mno-strict-align -mrelocatable @gol
655 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
656 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
657 -mdynamic-no-pic -maltivec -mswdiv @gol
658 -mprioritize-restricted-insns=@var{priority} @gol
659 -msched-costly-dep=@var{dependence_type} @gol
660 -minsert-sched-nops=@var{scheme} @gol
661 -mcall-sysv -mcall-netbsd @gol
662 -maix-struct-return -msvr4-struct-return @gol
663 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
664 -misel -mno-isel @gol
665 -misel=yes -misel=no @gol
667 -mspe=yes -mspe=no @gol
668 -mvrsave -mno-vrsave @gol
669 -mmulhw -mno-mulhw @gol
670 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
671 -mprototype -mno-prototype @gol
672 -msim -mmvme -mads -myellowknife -memb -msdata @gol
673 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
675 @emph{S/390 and zSeries Options}
676 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
677 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
678 -mpacked-stack -mno-packed-stack @gol
679 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
680 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
681 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
682 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
685 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
686 -m4-nofpu -m4-single-only -m4-single -m4 @gol
687 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
688 -m5-64media -m5-64media-nofpu @gol
689 -m5-32media -m5-32media-nofpu @gol
690 -m5-compact -m5-compact-nofpu @gol
691 -mb -ml -mdalign -mrelax @gol
692 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
693 -mieee -misize -mpadstruct -mspace @gol
694 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
695 -mdivsi3_libfunc=@var{name} @gol
696 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
700 @gccoptlist{-mcpu=@var{cpu-type} @gol
701 -mtune=@var{cpu-type} @gol
702 -mcmodel=@var{code-model} @gol
703 -m32 -m64 -mapp-regs -mno-app-regs @gol
704 -mfaster-structs -mno-faster-structs @gol
705 -mfpu -mno-fpu -mhard-float -msoft-float @gol
706 -mhard-quad-float -msoft-quad-float @gol
707 -mimpure-text -mno-impure-text -mlittle-endian @gol
708 -mstack-bias -mno-stack-bias @gol
709 -munaligned-doubles -mno-unaligned-doubles @gol
710 -mv8plus -mno-v8plus -mvis -mno-vis
713 @emph{System V Options}
714 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
716 @emph{TMS320C3x/C4x Options}
717 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
718 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
719 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
720 -mparallel-insns -mparallel-mpy -mpreserve-float}
723 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
724 -mprolog-function -mno-prolog-function -mspace @gol
725 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
726 -mapp-regs -mno-app-regs @gol
727 -mdisable-callt -mno-disable-callt @gol
733 @gccoptlist{-mg -mgnu -munix}
735 @emph{x86-64 Options}
736 See i386 and x86-64 Options.
738 @emph{Xstormy16 Options}
741 @emph{Xtensa Options}
742 @gccoptlist{-mconst16 -mno-const16 @gol
743 -mfused-madd -mno-fused-madd @gol
744 -mtext-section-literals -mno-text-section-literals @gol
745 -mtarget-align -mno-target-align @gol
746 -mlongcalls -mno-longcalls}
748 @emph{zSeries Options}
749 See S/390 and zSeries Options.
751 @item Code Generation Options
752 @xref{Code Gen Options,,Options for Code Generation Conventions}.
753 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
754 -ffixed-@var{reg} -fexceptions @gol
755 -fnon-call-exceptions -funwind-tables @gol
756 -fasynchronous-unwind-tables @gol
757 -finhibit-size-directive -finstrument-functions @gol
758 -fno-common -fno-ident @gol
759 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
760 -fno-jump-tables @gol
761 -freg-struct-return -fshared-data -fshort-enums @gol
762 -fshort-double -fshort-wchar @gol
763 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
764 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
765 -fargument-alias -fargument-noalias @gol
766 -fargument-noalias-global -fleading-underscore @gol
767 -ftls-model=@var{model} @gol
768 -ftrapv -fwrapv -fbounds-check @gol
769 -fvisibility -fopenmp}
773 * Overall Options:: Controlling the kind of output:
774 an executable, object files, assembler files,
775 or preprocessed source.
776 * C Dialect Options:: Controlling the variant of C language compiled.
777 * C++ Dialect Options:: Variations on C++.
778 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
780 * Language Independent Options:: Controlling how diagnostics should be
782 * Warning Options:: How picky should the compiler be?
783 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
784 * Optimize Options:: How much optimization?
785 * Preprocessor Options:: Controlling header files and macro definitions.
786 Also, getting dependency information for Make.
787 * Assembler Options:: Passing options to the assembler.
788 * Link Options:: Specifying libraries and so on.
789 * Directory Options:: Where to find header files and libraries.
790 Where to find the compiler executable files.
791 * Spec Files:: How to pass switches to sub-processes.
792 * Target Options:: Running a cross-compiler, or an old version of GCC.
795 @node Overall Options
796 @section Options Controlling the Kind of Output
798 Compilation can involve up to four stages: preprocessing, compilation
799 proper, assembly and linking, always in that order. GCC is capable of
800 preprocessing and compiling several files either into several
801 assembler input files, or into one assembler input file; then each
802 assembler input file produces an object file, and linking combines all
803 the object files (those newly compiled, and those specified as input)
804 into an executable file.
806 @cindex file name suffix
807 For any given input file, the file name suffix determines what kind of
812 C source code which must be preprocessed.
815 C source code which should not be preprocessed.
818 C++ source code which should not be preprocessed.
821 Objective-C source code. Note that you must link with the @file{libobjc}
822 library to make an Objective-C program work.
825 Objective-C source code which should not be preprocessed.
829 Objective-C++ source code. Note that you must link with the @file{libobjc}
830 library to make an Objective-C++ program work. Note that @samp{.M} refers
831 to a literal capital M@.
834 Objective-C++ source code which should not be preprocessed.
837 C, C++, Objective-C or Objective-C++ header file to be turned into a
842 @itemx @var{file}.cxx
843 @itemx @var{file}.cpp
844 @itemx @var{file}.CPP
845 @itemx @var{file}.c++
847 C++ source code which must be preprocessed. Note that in @samp{.cxx},
848 the last two letters must both be literally @samp{x}. Likewise,
849 @samp{.C} refers to a literal capital C@.
853 Objective-C++ source code which must be preprocessed.
856 Objective-C++ source code which should not be preprocessed.
860 C++ header file to be turned into a precompiled header.
863 @itemx @var{file}.for
864 @itemx @var{file}.FOR
865 Fixed form Fortran source code which should not be preprocessed.
868 @itemx @var{file}.fpp
869 @itemx @var{file}.FPP
870 Fixed form Fortran source code which must be preprocessed (with the traditional
874 @itemx @var{file}.f95
875 Free form Fortran source code which should not be preprocessed.
878 @itemx @var{file}.F95
879 Free form Fortran source code which must be preprocessed (with the
880 traditional preprocessor).
882 @c FIXME: Descriptions of Java file types.
889 Ada source code file which contains a library unit declaration (a
890 declaration of a package, subprogram, or generic, or a generic
891 instantiation), or a library unit renaming declaration (a package,
892 generic, or subprogram renaming declaration). Such files are also
895 @itemx @var{file}.adb
896 Ada source code file containing a library unit body (a subprogram or
897 package body). Such files are also called @dfn{bodies}.
899 @c GCC also knows about some suffixes for languages not yet included:
910 Assembler code which must be preprocessed.
913 An object file to be fed straight into linking.
914 Any file name with no recognized suffix is treated this way.
918 You can specify the input language explicitly with the @option{-x} option:
921 @item -x @var{language}
922 Specify explicitly the @var{language} for the following input files
923 (rather than letting the compiler choose a default based on the file
924 name suffix). This option applies to all following input files until
925 the next @option{-x} option. Possible values for @var{language} are:
927 c c-header c-cpp-output
928 c++ c++-header c++-cpp-output
929 objective-c objective-c-header objective-c-cpp-output
930 objective-c++ objective-c++-header objective-c++-cpp-output
931 assembler assembler-with-cpp
940 Turn off any specification of a language, so that subsequent files are
941 handled according to their file name suffixes (as they are if @option{-x}
942 has not been used at all).
944 @item -pass-exit-codes
945 @opindex pass-exit-codes
946 Normally the @command{gcc} program will exit with the code of 1 if any
947 phase of the compiler returns a non-success return code. If you specify
948 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
949 numerically highest error produced by any phase that returned an error
953 If you only want some of the stages of compilation, you can use
954 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
955 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
956 @command{gcc} is to stop. Note that some combinations (for example,
957 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
962 Compile or assemble the source files, but do not link. The linking
963 stage simply is not done. The ultimate output is in the form of an
964 object file for each source file.
966 By default, the object file name for a source file is made by replacing
967 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
969 Unrecognized input files, not requiring compilation or assembly, are
974 Stop after the stage of compilation proper; do not assemble. The output
975 is in the form of an assembler code file for each non-assembler input
978 By default, the assembler file name for a source file is made by
979 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
981 Input files that don't require compilation are ignored.
985 Stop after the preprocessing stage; do not run the compiler proper. The
986 output is in the form of preprocessed source code, which is sent to the
989 Input files which don't require preprocessing are ignored.
991 @cindex output file option
994 Place output in file @var{file}. This applies regardless to whatever
995 sort of output is being produced, whether it be an executable file,
996 an object file, an assembler file or preprocessed C code.
998 If @option{-o} is not specified, the default is to put an executable
999 file in @file{a.out}, the object file for
1000 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1001 assembler file in @file{@var{source}.s}, a precompiled header file in
1002 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1007 Print (on standard error output) the commands executed to run the stages
1008 of compilation. Also print the version number of the compiler driver
1009 program and of the preprocessor and the compiler proper.
1013 Like @option{-v} except the commands are not executed and all command
1014 arguments are quoted. This is useful for shell scripts to capture the
1015 driver-generated command lines.
1019 Use pipes rather than temporary files for communication between the
1020 various stages of compilation. This fails to work on some systems where
1021 the assembler is unable to read from a pipe; but the GNU assembler has
1026 If you are compiling multiple source files, this option tells the driver
1027 to pass all the source files to the compiler at once (for those
1028 languages for which the compiler can handle this). This will allow
1029 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1030 language for which this is supported is C@. If you pass source files for
1031 multiple languages to the driver, using this option, the driver will invoke
1032 the compiler(s) that support IMA once each, passing each compiler all the
1033 source files appropriate for it. For those languages that do not support
1034 IMA this option will be ignored, and the compiler will be invoked once for
1035 each source file in that language. If you use this option in conjunction
1036 with @option{-save-temps}, the compiler will generate multiple
1038 (one for each source file), but only one (combined) @file{.o} or
1043 Print (on the standard output) a description of the command line options
1044 understood by @command{gcc}. If the @option{-v} option is also specified
1045 then @option{--help} will also be passed on to the various processes
1046 invoked by @command{gcc}, so that they can display the command line options
1047 they accept. If the @option{-Wextra} option is also specified then command
1048 line options which have no documentation associated with them will also
1052 @opindex target-help
1053 Print (on the standard output) a description of target specific command
1054 line options for each tool.
1058 Display the version number and copyrights of the invoked GCC@.
1060 @include @value{srcdir}/../libiberty/at-file.texi
1064 @section Compiling C++ Programs
1066 @cindex suffixes for C++ source
1067 @cindex C++ source file suffixes
1068 C++ source files conventionally use one of the suffixes @samp{.C},
1069 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1070 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1071 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1072 files with these names and compiles them as C++ programs even if you
1073 call the compiler the same way as for compiling C programs (usually
1074 with the name @command{gcc}).
1078 However, C++ programs often require class libraries as well as a
1079 compiler that understands the C++ language---and under some
1080 circumstances, you might want to compile programs or header files from
1081 standard input, or otherwise without a suffix that flags them as C++
1082 programs. You might also like to precompile a C header file with a
1083 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1084 program that calls GCC with the default language set to C++, and
1085 automatically specifies linking against the C++ library. On many
1086 systems, @command{g++} is also installed with the name @command{c++}.
1088 @cindex invoking @command{g++}
1089 When you compile C++ programs, you may specify many of the same
1090 command-line options that you use for compiling programs in any
1091 language; or command-line options meaningful for C and related
1092 languages; or options that are meaningful only for C++ programs.
1093 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1094 explanations of options for languages related to C@.
1095 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1096 explanations of options that are meaningful only for C++ programs.
1098 @node C Dialect Options
1099 @section Options Controlling C Dialect
1100 @cindex dialect options
1101 @cindex language dialect options
1102 @cindex options, dialect
1104 The following options control the dialect of C (or languages derived
1105 from C, such as C++, Objective-C and Objective-C++) that the compiler
1109 @cindex ANSI support
1113 In C mode, support all ISO C90 programs. In C++ mode,
1114 remove GNU extensions that conflict with ISO C++.
1116 This turns off certain features of GCC that are incompatible with ISO
1117 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1118 such as the @code{asm} and @code{typeof} keywords, and
1119 predefined macros such as @code{unix} and @code{vax} that identify the
1120 type of system you are using. It also enables the undesirable and
1121 rarely used ISO trigraph feature. For the C compiler,
1122 it disables recognition of C++ style @samp{//} comments as well as
1123 the @code{inline} keyword.
1125 The alternate keywords @code{__asm__}, @code{__extension__},
1126 @code{__inline__} and @code{__typeof__} continue to work despite
1127 @option{-ansi}. You would not want to use them in an ISO C program, of
1128 course, but it is useful to put them in header files that might be included
1129 in compilations done with @option{-ansi}. Alternate predefined macros
1130 such as @code{__unix__} and @code{__vax__} are also available, with or
1131 without @option{-ansi}.
1133 The @option{-ansi} option does not cause non-ISO programs to be
1134 rejected gratuitously. For that, @option{-pedantic} is required in
1135 addition to @option{-ansi}. @xref{Warning Options}.
1137 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1138 option is used. Some header files may notice this macro and refrain
1139 from declaring certain functions or defining certain macros that the
1140 ISO standard doesn't call for; this is to avoid interfering with any
1141 programs that might use these names for other things.
1143 Functions which would normally be built in but do not have semantics
1144 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1145 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1146 built-in functions provided by GCC}, for details of the functions
1151 Determine the language standard. This option is currently only
1152 supported when compiling C or C++. A value for this option must be
1153 provided; possible values are
1158 ISO C90 (same as @option{-ansi}).
1160 @item iso9899:199409
1161 ISO C90 as modified in amendment 1.
1167 ISO C99. Note that this standard is not yet fully supported; see
1168 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1169 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1172 Default, ISO C90 plus GNU extensions (including some C99 features).
1176 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1177 this will become the default. The name @samp{gnu9x} is deprecated.
1180 The 1998 ISO C++ standard plus amendments.
1183 The same as @option{-std=c++98} plus GNU extensions. This is the
1184 default for C++ code.
1187 Even when this option is not specified, you can still use some of the
1188 features of newer standards in so far as they do not conflict with
1189 previous C standards. For example, you may use @code{__restrict__} even
1190 when @option{-std=c99} is not specified.
1192 The @option{-std} options specifying some version of ISO C have the same
1193 effects as @option{-ansi}, except that features that were not in ISO C90
1194 but are in the specified version (for example, @samp{//} comments and
1195 the @code{inline} keyword in ISO C99) are not disabled.
1197 @xref{Standards,,Language Standards Supported by GCC}, for details of
1198 these standard versions.
1200 @item -aux-info @var{filename}
1202 Output to the given filename prototyped declarations for all functions
1203 declared and/or defined in a translation unit, including those in header
1204 files. This option is silently ignored in any language other than C@.
1206 Besides declarations, the file indicates, in comments, the origin of
1207 each declaration (source file and line), whether the declaration was
1208 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1209 @samp{O} for old, respectively, in the first character after the line
1210 number and the colon), and whether it came from a declaration or a
1211 definition (@samp{C} or @samp{F}, respectively, in the following
1212 character). In the case of function definitions, a K&R-style list of
1213 arguments followed by their declarations is also provided, inside
1214 comments, after the declaration.
1218 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1219 keyword, so that code can use these words as identifiers. You can use
1220 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1221 instead. @option{-ansi} implies @option{-fno-asm}.
1223 In C++, this switch only affects the @code{typeof} keyword, since
1224 @code{asm} and @code{inline} are standard keywords. You may want to
1225 use the @option{-fno-gnu-keywords} flag instead, which has the same
1226 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1227 switch only affects the @code{asm} and @code{typeof} keywords, since
1228 @code{inline} is a standard keyword in ISO C99.
1231 @itemx -fno-builtin-@var{function}
1232 @opindex fno-builtin
1233 @cindex built-in functions
1234 Don't recognize built-in functions that do not begin with
1235 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1236 functions provided by GCC}, for details of the functions affected,
1237 including those which are not built-in functions when @option{-ansi} or
1238 @option{-std} options for strict ISO C conformance are used because they
1239 do not have an ISO standard meaning.
1241 GCC normally generates special code to handle certain built-in functions
1242 more efficiently; for instance, calls to @code{alloca} may become single
1243 instructions that adjust the stack directly, and calls to @code{memcpy}
1244 may become inline copy loops. The resulting code is often both smaller
1245 and faster, but since the function calls no longer appear as such, you
1246 cannot set a breakpoint on those calls, nor can you change the behavior
1247 of the functions by linking with a different library. In addition,
1248 when a function is recognized as a built-in function, GCC may use
1249 information about that function to warn about problems with calls to
1250 that function, or to generate more efficient code, even if the
1251 resulting code still contains calls to that function. For example,
1252 warnings are given with @option{-Wformat} for bad calls to
1253 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1254 known not to modify global memory.
1256 With the @option{-fno-builtin-@var{function}} option
1257 only the built-in function @var{function} is
1258 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1259 function is named this is not built-in in this version of GCC, this
1260 option is ignored. There is no corresponding
1261 @option{-fbuiltin-@var{function}} option; if you wish to enable
1262 built-in functions selectively when using @option{-fno-builtin} or
1263 @option{-ffreestanding}, you may define macros such as:
1266 #define abs(n) __builtin_abs ((n))
1267 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1272 @cindex hosted environment
1274 Assert that compilation takes place in a hosted environment. This implies
1275 @option{-fbuiltin}. A hosted environment is one in which the
1276 entire standard library is available, and in which @code{main} has a return
1277 type of @code{int}. Examples are nearly everything except a kernel.
1278 This is equivalent to @option{-fno-freestanding}.
1280 @item -ffreestanding
1281 @opindex ffreestanding
1282 @cindex hosted environment
1284 Assert that compilation takes place in a freestanding environment. This
1285 implies @option{-fno-builtin}. A freestanding environment
1286 is one in which the standard library may not exist, and program startup may
1287 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1288 This is equivalent to @option{-fno-hosted}.
1290 @xref{Standards,,Language Standards Supported by GCC}, for details of
1291 freestanding and hosted environments.
1293 @item -fms-extensions
1294 @opindex fms-extensions
1295 Accept some non-standard constructs used in Microsoft header files.
1297 Some cases of unnamed fields in structures and unions are only
1298 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1299 fields within structs/unions}, for details.
1303 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1304 options for strict ISO C conformance) implies @option{-trigraphs}.
1306 @item -no-integrated-cpp
1307 @opindex no-integrated-cpp
1308 Performs a compilation in two passes: preprocessing and compiling. This
1309 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1310 @option{-B} option. The user supplied compilation step can then add in
1311 an additional preprocessing step after normal preprocessing but before
1312 compiling. The default is to use the integrated cpp (internal cpp)
1314 The semantics of this option will change if "cc1", "cc1plus", and
1315 "cc1obj" are merged.
1317 @cindex traditional C language
1318 @cindex C language, traditional
1320 @itemx -traditional-cpp
1321 @opindex traditional-cpp
1322 @opindex traditional
1323 Formerly, these options caused GCC to attempt to emulate a pre-standard
1324 C compiler. They are now only supported with the @option{-E} switch.
1325 The preprocessor continues to support a pre-standard mode. See the GNU
1326 CPP manual for details.
1328 @item -fcond-mismatch
1329 @opindex fcond-mismatch
1330 Allow conditional expressions with mismatched types in the second and
1331 third arguments. The value of such an expression is void. This option
1332 is not supported for C++.
1334 @item -funsigned-char
1335 @opindex funsigned-char
1336 Let the type @code{char} be unsigned, like @code{unsigned char}.
1338 Each kind of machine has a default for what @code{char} should
1339 be. It is either like @code{unsigned char} by default or like
1340 @code{signed char} by default.
1342 Ideally, a portable program should always use @code{signed char} or
1343 @code{unsigned char} when it depends on the signedness of an object.
1344 But many programs have been written to use plain @code{char} and
1345 expect it to be signed, or expect it to be unsigned, depending on the
1346 machines they were written for. This option, and its inverse, let you
1347 make such a program work with the opposite default.
1349 The type @code{char} is always a distinct type from each of
1350 @code{signed char} or @code{unsigned char}, even though its behavior
1351 is always just like one of those two.
1354 @opindex fsigned-char
1355 Let the type @code{char} be signed, like @code{signed char}.
1357 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1358 the negative form of @option{-funsigned-char}. Likewise, the option
1359 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1361 @item -fsigned-bitfields
1362 @itemx -funsigned-bitfields
1363 @itemx -fno-signed-bitfields
1364 @itemx -fno-unsigned-bitfields
1365 @opindex fsigned-bitfields
1366 @opindex funsigned-bitfields
1367 @opindex fno-signed-bitfields
1368 @opindex fno-unsigned-bitfields
1369 These options control whether a bit-field is signed or unsigned, when the
1370 declaration does not use either @code{signed} or @code{unsigned}. By
1371 default, such a bit-field is signed, because this is consistent: the
1372 basic integer types such as @code{int} are signed types.
1375 @node C++ Dialect Options
1376 @section Options Controlling C++ Dialect
1378 @cindex compiler options, C++
1379 @cindex C++ options, command line
1380 @cindex options, C++
1381 This section describes the command-line options that are only meaningful
1382 for C++ programs; but you can also use most of the GNU compiler options
1383 regardless of what language your program is in. For example, you
1384 might compile a file @code{firstClass.C} like this:
1387 g++ -g -frepo -O -c firstClass.C
1391 In this example, only @option{-frepo} is an option meant
1392 only for C++ programs; you can use the other options with any
1393 language supported by GCC@.
1395 Here is a list of options that are @emph{only} for compiling C++ programs:
1399 @item -fabi-version=@var{n}
1400 @opindex fabi-version
1401 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1402 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1403 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1404 the version that conforms most closely to the C++ ABI specification.
1405 Therefore, the ABI obtained using version 0 will change as ABI bugs
1408 The default is version 2.
1410 @item -fno-access-control
1411 @opindex fno-access-control
1412 Turn off all access checking. This switch is mainly useful for working
1413 around bugs in the access control code.
1417 Check that the pointer returned by @code{operator new} is non-null
1418 before attempting to modify the storage allocated. This check is
1419 normally unnecessary because the C++ standard specifies that
1420 @code{operator new} will only return @code{0} if it is declared
1421 @samp{throw()}, in which case the compiler will always check the
1422 return value even without this option. In all other cases, when
1423 @code{operator new} has a non-empty exception specification, memory
1424 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1425 @samp{new (nothrow)}.
1427 @item -fconserve-space
1428 @opindex fconserve-space
1429 Put uninitialized or runtime-initialized global variables into the
1430 common segment, as C does. This saves space in the executable at the
1431 cost of not diagnosing duplicate definitions. If you compile with this
1432 flag and your program mysteriously crashes after @code{main()} has
1433 completed, you may have an object that is being destroyed twice because
1434 two definitions were merged.
1436 This option is no longer useful on most targets, now that support has
1437 been added for putting variables into BSS without making them common.
1439 @item -ffriend-injection
1440 @opindex ffriend-injection
1441 Inject friend functions into the enclosing namespace, so that they are
1442 visible outside the scope of the class in which they are declared.
1443 Friend functions were documented to work this way in the old Annotated
1444 C++ Reference Manual, and versions of G++ before 4.1 always worked
1445 that way. However, in ISO C++ a friend function which is not declared
1446 in an enclosing scope can only be found using argument dependent
1447 lookup. This option causes friends to be injected as they were in
1450 This option is for compatibility, and may be removed in a future
1453 @item -fno-const-strings
1454 @opindex fno-const-strings
1455 Give string constants type @code{char *} instead of type @code{const
1456 char *}. By default, G++ uses type @code{const char *} as required by
1457 the standard. Even if you use @option{-fno-const-strings}, you cannot
1458 actually modify the value of a string constant.
1460 This option might be removed in a future release of G++. For maximum
1461 portability, you should structure your code so that it works with
1462 string constants that have type @code{const char *}.
1464 @item -fno-elide-constructors
1465 @opindex fno-elide-constructors
1466 The C++ standard allows an implementation to omit creating a temporary
1467 which is only used to initialize another object of the same type.
1468 Specifying this option disables that optimization, and forces G++ to
1469 call the copy constructor in all cases.
1471 @item -fno-enforce-eh-specs
1472 @opindex fno-enforce-eh-specs
1473 Don't generate code to check for violation of exception specifications
1474 at runtime. This option violates the C++ standard, but may be useful
1475 for reducing code size in production builds, much like defining
1476 @samp{NDEBUG}. This does not give user code permission to throw
1477 exceptions in violation of the exception specifications; the compiler
1478 will still optimize based on the specifications, so throwing an
1479 unexpected exception will result in undefined behavior.
1482 @itemx -fno-for-scope
1484 @opindex fno-for-scope
1485 If @option{-ffor-scope} is specified, the scope of variables declared in
1486 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1487 as specified by the C++ standard.
1488 If @option{-fno-for-scope} is specified, the scope of variables declared in
1489 a @i{for-init-statement} extends to the end of the enclosing scope,
1490 as was the case in old versions of G++, and other (traditional)
1491 implementations of C++.
1493 The default if neither flag is given to follow the standard,
1494 but to allow and give a warning for old-style code that would
1495 otherwise be invalid, or have different behavior.
1497 @item -fno-gnu-keywords
1498 @opindex fno-gnu-keywords
1499 Do not recognize @code{typeof} as a keyword, so that code can use this
1500 word as an identifier. You can use the keyword @code{__typeof__} instead.
1501 @option{-ansi} implies @option{-fno-gnu-keywords}.
1503 @item -fno-implicit-templates
1504 @opindex fno-implicit-templates
1505 Never emit code for non-inline templates which are instantiated
1506 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1507 @xref{Template Instantiation}, for more information.
1509 @item -fno-implicit-inline-templates
1510 @opindex fno-implicit-inline-templates
1511 Don't emit code for implicit instantiations of inline templates, either.
1512 The default is to handle inlines differently so that compiles with and
1513 without optimization will need the same set of explicit instantiations.
1515 @item -fno-implement-inlines
1516 @opindex fno-implement-inlines
1517 To save space, do not emit out-of-line copies of inline functions
1518 controlled by @samp{#pragma implementation}. This will cause linker
1519 errors if these functions are not inlined everywhere they are called.
1521 @item -fms-extensions
1522 @opindex fms-extensions
1523 Disable pedantic warnings about constructs used in MFC, such as implicit
1524 int and getting a pointer to member function via non-standard syntax.
1526 @item -fno-nonansi-builtins
1527 @opindex fno-nonansi-builtins
1528 Disable built-in declarations of functions that are not mandated by
1529 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1530 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1532 @item -fno-operator-names
1533 @opindex fno-operator-names
1534 Do not treat the operator name keywords @code{and}, @code{bitand},
1535 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1536 synonyms as keywords.
1538 @item -fno-optional-diags
1539 @opindex fno-optional-diags
1540 Disable diagnostics that the standard says a compiler does not need to
1541 issue. Currently, the only such diagnostic issued by G++ is the one for
1542 a name having multiple meanings within a class.
1545 @opindex fpermissive
1546 Downgrade some diagnostics about nonconformant code from errors to
1547 warnings. Thus, using @option{-fpermissive} will allow some
1548 nonconforming code to compile.
1552 Enable automatic template instantiation at link time. This option also
1553 implies @option{-fno-implicit-templates}. @xref{Template
1554 Instantiation}, for more information.
1558 Disable generation of information about every class with virtual
1559 functions for use by the C++ runtime type identification features
1560 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1561 of the language, you can save some space by using this flag. Note that
1562 exception handling uses the same information, but it will generate it as
1567 Emit statistics about front-end processing at the end of the compilation.
1568 This information is generally only useful to the G++ development team.
1570 @item -ftemplate-depth-@var{n}
1571 @opindex ftemplate-depth
1572 Set the maximum instantiation depth for template classes to @var{n}.
1573 A limit on the template instantiation depth is needed to detect
1574 endless recursions during template class instantiation. ANSI/ISO C++
1575 conforming programs must not rely on a maximum depth greater than 17.
1577 @item -fno-threadsafe-statics
1578 @opindex fno-threadsafe-statics
1579 Do not emit the extra code to use the routines specified in the C++
1580 ABI for thread-safe initialization of local statics. You can use this
1581 option to reduce code size slightly in code that doesn't need to be
1584 @item -fuse-cxa-atexit
1585 @opindex fuse-cxa-atexit
1586 Register destructors for objects with static storage duration with the
1587 @code{__cxa_atexit} function rather than the @code{atexit} function.
1588 This option is required for fully standards-compliant handling of static
1589 destructors, but will only work if your C library supports
1590 @code{__cxa_atexit}.
1592 @item -fvisibility-inlines-hidden
1593 @opindex fvisibility-inlines-hidden
1594 Causes all inlined methods to be marked with
1595 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1596 appear in the export table of a DSO and do not require a PLT indirection
1597 when used within the DSO@. Enabling this option can have a dramatic effect
1598 on load and link times of a DSO as it massively reduces the size of the
1599 dynamic export table when the library makes heavy use of templates. While
1600 it can cause bloating through duplication of code within each DSO where
1601 it is used, often the wastage is less than the considerable space occupied
1602 by a long symbol name in the export table which is typical when using
1603 templates and namespaces. For even more savings, combine with the
1604 @option{-fvisibility=hidden} switch.
1608 Do not use weak symbol support, even if it is provided by the linker.
1609 By default, G++ will use weak symbols if they are available. This
1610 option exists only for testing, and should not be used by end-users;
1611 it will result in inferior code and has no benefits. This option may
1612 be removed in a future release of G++.
1616 Do not search for header files in the standard directories specific to
1617 C++, but do still search the other standard directories. (This option
1618 is used when building the C++ library.)
1621 In addition, these optimization, warning, and code generation options
1622 have meanings only for C++ programs:
1625 @item -fno-default-inline
1626 @opindex fno-default-inline
1627 Do not assume @samp{inline} for functions defined inside a class scope.
1628 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1629 functions will have linkage like inline functions; they just won't be
1632 @item -Wabi @r{(C++ only)}
1634 Warn when G++ generates code that is probably not compatible with the
1635 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1636 all such cases, there are probably some cases that are not warned about,
1637 even though G++ is generating incompatible code. There may also be
1638 cases where warnings are emitted even though the code that is generated
1641 You should rewrite your code to avoid these warnings if you are
1642 concerned about the fact that code generated by G++ may not be binary
1643 compatible with code generated by other compilers.
1645 The known incompatibilities at this point include:
1650 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1651 pack data into the same byte as a base class. For example:
1654 struct A @{ virtual void f(); int f1 : 1; @};
1655 struct B : public A @{ int f2 : 1; @};
1659 In this case, G++ will place @code{B::f2} into the same byte
1660 as@code{A::f1}; other compilers will not. You can avoid this problem
1661 by explicitly padding @code{A} so that its size is a multiple of the
1662 byte size on your platform; that will cause G++ and other compilers to
1663 layout @code{B} identically.
1666 Incorrect handling of tail-padding for virtual bases. G++ does not use
1667 tail padding when laying out virtual bases. For example:
1670 struct A @{ virtual void f(); char c1; @};
1671 struct B @{ B(); char c2; @};
1672 struct C : public A, public virtual B @{@};
1676 In this case, G++ will not place @code{B} into the tail-padding for
1677 @code{A}; other compilers will. You can avoid this problem by
1678 explicitly padding @code{A} so that its size is a multiple of its
1679 alignment (ignoring virtual base classes); that will cause G++ and other
1680 compilers to layout @code{C} identically.
1683 Incorrect handling of bit-fields with declared widths greater than that
1684 of their underlying types, when the bit-fields appear in a union. For
1688 union U @{ int i : 4096; @};
1692 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1693 union too small by the number of bits in an @code{int}.
1696 Empty classes can be placed at incorrect offsets. For example:
1706 struct C : public B, public A @{@};
1710 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1711 it should be placed at offset zero. G++ mistakenly believes that the
1712 @code{A} data member of @code{B} is already at offset zero.
1715 Names of template functions whose types involve @code{typename} or
1716 template template parameters can be mangled incorrectly.
1719 template <typename Q>
1720 void f(typename Q::X) @{@}
1722 template <template <typename> class Q>
1723 void f(typename Q<int>::X) @{@}
1727 Instantiations of these templates may be mangled incorrectly.
1731 @item -Wctor-dtor-privacy @r{(C++ only)}
1732 @opindex Wctor-dtor-privacy
1733 Warn when a class seems unusable because all the constructors or
1734 destructors in that class are private, and it has neither friends nor
1735 public static member functions.
1737 @item -Wnon-virtual-dtor @r{(C++ only)}
1738 @opindex Wnon-virtual-dtor
1739 Warn when a class appears to be polymorphic, thereby requiring a virtual
1740 destructor, yet it declares a non-virtual one. This warning is also
1741 enabled if -Weffc++ is specified.
1743 @item -Wreorder @r{(C++ only)}
1745 @cindex reordering, warning
1746 @cindex warning for reordering of member initializers
1747 Warn when the order of member initializers given in the code does not
1748 match the order in which they must be executed. For instance:
1754 A(): j (0), i (1) @{ @}
1758 The compiler will rearrange the member initializers for @samp{i}
1759 and @samp{j} to match the declaration order of the members, emitting
1760 a warning to that effect. This warning is enabled by @option{-Wall}.
1763 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1766 @item -Weffc++ @r{(C++ only)}
1768 Warn about violations of the following style guidelines from Scott Meyers'
1769 @cite{Effective C++} book:
1773 Item 11: Define a copy constructor and an assignment operator for classes
1774 with dynamically allocated memory.
1777 Item 12: Prefer initialization to assignment in constructors.
1780 Item 14: Make destructors virtual in base classes.
1783 Item 15: Have @code{operator=} return a reference to @code{*this}.
1786 Item 23: Don't try to return a reference when you must return an object.
1790 Also warn about violations of the following style guidelines from
1791 Scott Meyers' @cite{More Effective C++} book:
1795 Item 6: Distinguish between prefix and postfix forms of increment and
1796 decrement operators.
1799 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1803 When selecting this option, be aware that the standard library
1804 headers do not obey all of these guidelines; use @samp{grep -v}
1805 to filter out those warnings.
1807 @item -Wno-deprecated @r{(C++ only)}
1808 @opindex Wno-deprecated
1809 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1811 @item -Wstrict-null-sentinel @r{(C++ only)}
1812 @opindex Wstrict-null-sentinel
1813 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1814 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1815 to @code{__null}. Although it is a null pointer constant not a null pointer,
1816 it is guaranteed to of the same size as a pointer. But this use is
1817 not portable across different compilers.
1819 @item -Wno-non-template-friend @r{(C++ only)}
1820 @opindex Wno-non-template-friend
1821 Disable warnings when non-templatized friend functions are declared
1822 within a template. Since the advent of explicit template specification
1823 support in G++, if the name of the friend is an unqualified-id (i.e.,
1824 @samp{friend foo(int)}), the C++ language specification demands that the
1825 friend declare or define an ordinary, nontemplate function. (Section
1826 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1827 could be interpreted as a particular specialization of a templatized
1828 function. Because this non-conforming behavior is no longer the default
1829 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1830 check existing code for potential trouble spots and is on by default.
1831 This new compiler behavior can be turned off with
1832 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1833 but disables the helpful warning.
1835 @item -Wold-style-cast @r{(C++ only)}
1836 @opindex Wold-style-cast
1837 Warn if an old-style (C-style) cast to a non-void type is used within
1838 a C++ program. The new-style casts (@samp{dynamic_cast},
1839 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1840 less vulnerable to unintended effects and much easier to search for.
1842 @item -Woverloaded-virtual @r{(C++ only)}
1843 @opindex Woverloaded-virtual
1844 @cindex overloaded virtual fn, warning
1845 @cindex warning for overloaded virtual fn
1846 Warn when a function declaration hides virtual functions from a
1847 base class. For example, in:
1854 struct B: public A @{
1859 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1867 will fail to compile.
1869 @item -Wno-pmf-conversions @r{(C++ only)}
1870 @opindex Wno-pmf-conversions
1871 Disable the diagnostic for converting a bound pointer to member function
1874 @item -Wsign-promo @r{(C++ only)}
1875 @opindex Wsign-promo
1876 Warn when overload resolution chooses a promotion from unsigned or
1877 enumerated type to a signed type, over a conversion to an unsigned type of
1878 the same size. Previous versions of G++ would try to preserve
1879 unsignedness, but the standard mandates the current behavior.
1884 A& operator = (int);
1894 In this example, G++ will synthesize a default @samp{A& operator =
1895 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1898 @node Objective-C and Objective-C++ Dialect Options
1899 @section Options Controlling Objective-C and Objective-C++ Dialects
1901 @cindex compiler options, Objective-C and Objective-C++
1902 @cindex Objective-C and Objective-C++ options, command line
1903 @cindex options, Objective-C and Objective-C++
1904 (NOTE: This manual does not describe the Objective-C and Objective-C++
1905 languages themselves. See @xref{Standards,,Language Standards
1906 Supported by GCC}, for references.)
1908 This section describes the command-line options that are only meaningful
1909 for Objective-C and Objective-C++ programs, but you can also use most of
1910 the language-independent GNU compiler options.
1911 For example, you might compile a file @code{some_class.m} like this:
1914 gcc -g -fgnu-runtime -O -c some_class.m
1918 In this example, @option{-fgnu-runtime} is an option meant only for
1919 Objective-C and Objective-C++ programs; you can use the other options with
1920 any language supported by GCC@.
1922 Note that since Objective-C is an extension of the C language, Objective-C
1923 compilations may also use options specific to the C front-end (e.g.,
1924 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1925 C++-specific options (e.g., @option{-Wabi}).
1927 Here is a list of options that are @emph{only} for compiling Objective-C
1928 and Objective-C++ programs:
1931 @item -fconstant-string-class=@var{class-name}
1932 @opindex fconstant-string-class
1933 Use @var{class-name} as the name of the class to instantiate for each
1934 literal string specified with the syntax @code{@@"@dots{}"}. The default
1935 class name is @code{NXConstantString} if the GNU runtime is being used, and
1936 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1937 @option{-fconstant-cfstrings} option, if also present, will override the
1938 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1939 to be laid out as constant CoreFoundation strings.
1942 @opindex fgnu-runtime
1943 Generate object code compatible with the standard GNU Objective-C
1944 runtime. This is the default for most types of systems.
1946 @item -fnext-runtime
1947 @opindex fnext-runtime
1948 Generate output compatible with the NeXT runtime. This is the default
1949 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1950 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1953 @item -fno-nil-receivers
1954 @opindex fno-nil-receivers
1955 Assume that all Objective-C message dispatches (e.g.,
1956 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1957 is not @code{nil}. This allows for more efficient entry points in the runtime
1958 to be used. Currently, this option is only available in conjunction with
1959 the NeXT runtime on Mac OS X 10.3 and later.
1961 @item -fobjc-call-cxx-cdtors
1962 @opindex fobjc-call-cxx-cdtors
1963 For each Objective-C class, check if any of its instance variables is a
1964 C++ object with a non-trivial default constructor. If so, synthesize a
1965 special @code{- (id) .cxx_construct} instance method that will run
1966 non-trivial default constructors on any such instance variables, in order,
1967 and then return @code{self}. Similarly, check if any instance variable
1968 is a C++ object with a non-trivial destructor, and if so, synthesize a
1969 special @code{- (void) .cxx_destruct} method that will run
1970 all such default destructors, in reverse order.
1972 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
1973 thusly generated will only operate on instance variables declared in the
1974 current Objective-C class, and not those inherited from superclasses. It
1975 is the responsibility of the Objective-C runtime to invoke all such methods
1976 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
1977 will be invoked by the runtime immediately after a new object
1978 instance is allocated; the @code{- (void) .cxx_destruct} methods will
1979 be invoked immediately before the runtime deallocates an object instance.
1981 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
1982 support for invoking the @code{- (id) .cxx_construct} and
1983 @code{- (void) .cxx_destruct} methods.
1985 @item -fobjc-direct-dispatch
1986 @opindex fobjc-direct-dispatch
1987 Allow fast jumps to the message dispatcher. On Darwin this is
1988 accomplished via the comm page.
1990 @item -fobjc-exceptions
1991 @opindex fobjc-exceptions
1992 Enable syntactic support for structured exception handling in Objective-C,
1993 similar to what is offered by C++ and Java. This option is
1994 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2003 @@catch (AnObjCClass *exc) @{
2010 @@catch (AnotherClass *exc) @{
2013 @@catch (id allOthers) @{
2023 The @code{@@throw} statement may appear anywhere in an Objective-C or
2024 Objective-C++ program; when used inside of a @code{@@catch} block, the
2025 @code{@@throw} may appear without an argument (as shown above), in which case
2026 the object caught by the @code{@@catch} will be rethrown.
2028 Note that only (pointers to) Objective-C objects may be thrown and
2029 caught using this scheme. When an object is thrown, it will be caught
2030 by the nearest @code{@@catch} clause capable of handling objects of that type,
2031 analogously to how @code{catch} blocks work in C++ and Java. A
2032 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2033 any and all Objective-C exceptions not caught by previous @code{@@catch}
2036 The @code{@@finally} clause, if present, will be executed upon exit from the
2037 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2038 regardless of whether any exceptions are thrown, caught or rethrown
2039 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2040 of the @code{finally} clause in Java.
2042 There are several caveats to using the new exception mechanism:
2046 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2047 idioms provided by the @code{NSException} class, the new
2048 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2049 systems, due to additional functionality needed in the (NeXT) Objective-C
2053 As mentioned above, the new exceptions do not support handling
2054 types other than Objective-C objects. Furthermore, when used from
2055 Objective-C++, the Objective-C exception model does not interoperate with C++
2056 exceptions at this time. This means you cannot @code{@@throw} an exception
2057 from Objective-C and @code{catch} it in C++, or vice versa
2058 (i.e., @code{throw @dots{} @@catch}).
2061 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2062 blocks for thread-safe execution:
2065 @@synchronized (ObjCClass *guard) @{
2070 Upon entering the @code{@@synchronized} block, a thread of execution shall
2071 first check whether a lock has been placed on the corresponding @code{guard}
2072 object by another thread. If it has, the current thread shall wait until
2073 the other thread relinquishes its lock. Once @code{guard} becomes available,
2074 the current thread will place its own lock on it, execute the code contained in
2075 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2076 making @code{guard} available to other threads).
2078 Unlike Java, Objective-C does not allow for entire methods to be marked
2079 @code{@@synchronized}. Note that throwing exceptions out of
2080 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2081 to be unlocked properly.
2085 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2087 @item -freplace-objc-classes
2088 @opindex freplace-objc-classes
2089 Emit a special marker instructing @command{ld(1)} not to statically link in
2090 the resulting object file, and allow @command{dyld(1)} to load it in at
2091 run time instead. This is used in conjunction with the Fix-and-Continue
2092 debugging mode, where the object file in question may be recompiled and
2093 dynamically reloaded in the course of program execution, without the need
2094 to restart the program itself. Currently, Fix-and-Continue functionality
2095 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2100 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2101 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2102 compile time) with static class references that get initialized at load time,
2103 which improves run-time performance. Specifying the @option{-fzero-link} flag
2104 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2105 to be retained. This is useful in Zero-Link debugging mode, since it allows
2106 for individual class implementations to be modified during program execution.
2110 Dump interface declarations for all classes seen in the source file to a
2111 file named @file{@var{sourcename}.decl}.
2113 @item -Wassign-intercept
2114 @opindex Wassign-intercept
2115 Warn whenever an Objective-C assignment is being intercepted by the
2119 @opindex Wno-protocol
2120 If a class is declared to implement a protocol, a warning is issued for
2121 every method in the protocol that is not implemented by the class. The
2122 default behavior is to issue a warning for every method not explicitly
2123 implemented in the class, even if a method implementation is inherited
2124 from the superclass. If you use the @option{-Wno-protocol} option, then
2125 methods inherited from the superclass are considered to be implemented,
2126 and no warning is issued for them.
2130 Warn if multiple methods of different types for the same selector are
2131 found during compilation. The check is performed on the list of methods
2132 in the final stage of compilation. Additionally, a check is performed
2133 for each selector appearing in a @code{@@selector(@dots{})}
2134 expression, and a corresponding method for that selector has been found
2135 during compilation. Because these checks scan the method table only at
2136 the end of compilation, these warnings are not produced if the final
2137 stage of compilation is not reached, for example because an error is
2138 found during compilation, or because the @option{-fsyntax-only} option is
2141 @item -Wstrict-selector-match
2142 @opindex Wstrict-selector-match
2143 Warn if multiple methods with differing argument and/or return types are
2144 found for a given selector when attempting to send a message using this
2145 selector to a receiver of type @code{id} or @code{Class}. When this flag
2146 is off (which is the default behavior), the compiler will omit such warnings
2147 if any differences found are confined to types which share the same size
2150 @item -Wundeclared-selector
2151 @opindex Wundeclared-selector
2152 Warn if a @code{@@selector(@dots{})} expression referring to an
2153 undeclared selector is found. A selector is considered undeclared if no
2154 method with that name has been declared before the
2155 @code{@@selector(@dots{})} expression, either explicitly in an
2156 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2157 an @code{@@implementation} section. This option always performs its
2158 checks as soon as a @code{@@selector(@dots{})} expression is found,
2159 while @option{-Wselector} only performs its checks in the final stage of
2160 compilation. This also enforces the coding style convention
2161 that methods and selectors must be declared before being used.
2163 @item -print-objc-runtime-info
2164 @opindex print-objc-runtime-info
2165 Generate C header describing the largest structure that is passed by
2170 @node Language Independent Options
2171 @section Options to Control Diagnostic Messages Formatting
2172 @cindex options to control diagnostics formatting
2173 @cindex diagnostic messages
2174 @cindex message formatting
2176 Traditionally, diagnostic messages have been formatted irrespective of
2177 the output device's aspect (e.g.@: its width, @dots{}). The options described
2178 below can be used to control the diagnostic messages formatting
2179 algorithm, e.g.@: how many characters per line, how often source location
2180 information should be reported. Right now, only the C++ front end can
2181 honor these options. However it is expected, in the near future, that
2182 the remaining front ends would be able to digest them correctly.
2185 @item -fmessage-length=@var{n}
2186 @opindex fmessage-length
2187 Try to format error messages so that they fit on lines of about @var{n}
2188 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2189 the front ends supported by GCC@. If @var{n} is zero, then no
2190 line-wrapping will be done; each error message will appear on a single
2193 @opindex fdiagnostics-show-location
2194 @item -fdiagnostics-show-location=once
2195 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2196 reporter to emit @emph{once} source location information; that is, in
2197 case the message is too long to fit on a single physical line and has to
2198 be wrapped, the source location won't be emitted (as prefix) again,
2199 over and over, in subsequent continuation lines. This is the default
2202 @item -fdiagnostics-show-location=every-line
2203 Only meaningful in line-wrapping mode. Instructs the diagnostic
2204 messages reporter to emit the same source location information (as
2205 prefix) for physical lines that result from the process of breaking
2206 a message which is too long to fit on a single line.
2208 @item -fdiagnostics-show-options
2209 @opindex fdiagnostics-show-options
2210 This option instructs the diagnostic machinery to add text to each
2211 diagnostic emitted, which indicates which command line option directly
2212 controls that diagnostic, when such an option is known to the
2213 diagnostic machinery.
2217 @node Warning Options
2218 @section Options to Request or Suppress Warnings
2219 @cindex options to control warnings
2220 @cindex warning messages
2221 @cindex messages, warning
2222 @cindex suppressing warnings
2224 Warnings are diagnostic messages that report constructions which
2225 are not inherently erroneous but which are risky or suggest there
2226 may have been an error.
2228 You can request many specific warnings with options beginning @samp{-W},
2229 for example @option{-Wimplicit} to request warnings on implicit
2230 declarations. Each of these specific warning options also has a
2231 negative form beginning @samp{-Wno-} to turn off warnings;
2232 for example, @option{-Wno-implicit}. This manual lists only one of the
2233 two forms, whichever is not the default.
2235 The following options control the amount and kinds of warnings produced
2236 by GCC; for further, language-specific options also refer to
2237 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2241 @cindex syntax checking
2243 @opindex fsyntax-only
2244 Check the code for syntax errors, but don't do anything beyond that.
2248 Issue all the warnings demanded by strict ISO C and ISO C++;
2249 reject all programs that use forbidden extensions, and some other
2250 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2251 version of the ISO C standard specified by any @option{-std} option used.
2253 Valid ISO C and ISO C++ programs should compile properly with or without
2254 this option (though a rare few will require @option{-ansi} or a
2255 @option{-std} option specifying the required version of ISO C)@. However,
2256 without this option, certain GNU extensions and traditional C and C++
2257 features are supported as well. With this option, they are rejected.
2259 @option{-pedantic} does not cause warning messages for use of the
2260 alternate keywords whose names begin and end with @samp{__}. Pedantic
2261 warnings are also disabled in the expression that follows
2262 @code{__extension__}. However, only system header files should use
2263 these escape routes; application programs should avoid them.
2264 @xref{Alternate Keywords}.
2266 Some users try to use @option{-pedantic} to check programs for strict ISO
2267 C conformance. They soon find that it does not do quite what they want:
2268 it finds some non-ISO practices, but not all---only those for which
2269 ISO C @emph{requires} a diagnostic, and some others for which
2270 diagnostics have been added.
2272 A feature to report any failure to conform to ISO C might be useful in
2273 some instances, but would require considerable additional work and would
2274 be quite different from @option{-pedantic}. We don't have plans to
2275 support such a feature in the near future.
2277 Where the standard specified with @option{-std} represents a GNU
2278 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2279 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2280 extended dialect is based. Warnings from @option{-pedantic} are given
2281 where they are required by the base standard. (It would not make sense
2282 for such warnings to be given only for features not in the specified GNU
2283 C dialect, since by definition the GNU dialects of C include all
2284 features the compiler supports with the given option, and there would be
2285 nothing to warn about.)
2287 @item -pedantic-errors
2288 @opindex pedantic-errors
2289 Like @option{-pedantic}, except that errors are produced rather than
2294 Inhibit all warning messages.
2298 Inhibit warning messages about the use of @samp{#import}.
2300 @item -Wchar-subscripts
2301 @opindex Wchar-subscripts
2302 Warn if an array subscript has type @code{char}. This is a common cause
2303 of error, as programmers often forget that this type is signed on some
2305 This warning is enabled by @option{-Wall}.
2309 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2310 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2311 This warning is enabled by @option{-Wall}.
2313 @item -Wfatal-errors
2314 @opindex Wfatal-errors
2315 This option causes the compiler to abort compilation on the first error
2316 occurred rather than trying to keep going and printing further error
2321 @opindex ffreestanding
2322 @opindex fno-builtin
2323 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2324 the arguments supplied have types appropriate to the format string
2325 specified, and that the conversions specified in the format string make
2326 sense. This includes standard functions, and others specified by format
2327 attributes (@pxref{Function Attributes}), in the @code{printf},
2328 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2329 not in the C standard) families (or other target-specific families).
2330 Which functions are checked without format attributes having been
2331 specified depends on the standard version selected, and such checks of
2332 functions without the attribute specified are disabled by
2333 @option{-ffreestanding} or @option{-fno-builtin}.
2335 The formats are checked against the format features supported by GNU
2336 libc version 2.2. These include all ISO C90 and C99 features, as well
2337 as features from the Single Unix Specification and some BSD and GNU
2338 extensions. Other library implementations may not support all these
2339 features; GCC does not support warning about features that go beyond a
2340 particular library's limitations. However, if @option{-pedantic} is used
2341 with @option{-Wformat}, warnings will be given about format features not
2342 in the selected standard version (but not for @code{strfmon} formats,
2343 since those are not in any version of the C standard). @xref{C Dialect
2344 Options,,Options Controlling C Dialect}.
2346 Since @option{-Wformat} also checks for null format arguments for
2347 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2349 @option{-Wformat} is included in @option{-Wall}. For more control over some
2350 aspects of format checking, the options @option{-Wformat-y2k},
2351 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2352 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2353 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2356 @opindex Wformat-y2k
2357 If @option{-Wformat} is specified, also warn about @code{strftime}
2358 formats which may yield only a two-digit year.
2360 @item -Wno-format-extra-args
2361 @opindex Wno-format-extra-args
2362 If @option{-Wformat} is specified, do not warn about excess arguments to a
2363 @code{printf} or @code{scanf} format function. The C standard specifies
2364 that such arguments are ignored.
2366 Where the unused arguments lie between used arguments that are
2367 specified with @samp{$} operand number specifications, normally
2368 warnings are still given, since the implementation could not know what
2369 type to pass to @code{va_arg} to skip the unused arguments. However,
2370 in the case of @code{scanf} formats, this option will suppress the
2371 warning if the unused arguments are all pointers, since the Single
2372 Unix Specification says that such unused arguments are allowed.
2374 @item -Wno-format-zero-length
2375 @opindex Wno-format-zero-length
2376 If @option{-Wformat} is specified, do not warn about zero-length formats.
2377 The C standard specifies that zero-length formats are allowed.
2379 @item -Wformat-nonliteral
2380 @opindex Wformat-nonliteral
2381 If @option{-Wformat} is specified, also warn if the format string is not a
2382 string literal and so cannot be checked, unless the format function
2383 takes its format arguments as a @code{va_list}.
2385 @item -Wformat-security
2386 @opindex Wformat-security
2387 If @option{-Wformat} is specified, also warn about uses of format
2388 functions that represent possible security problems. At present, this
2389 warns about calls to @code{printf} and @code{scanf} functions where the
2390 format string is not a string literal and there are no format arguments,
2391 as in @code{printf (foo);}. This may be a security hole if the format
2392 string came from untrusted input and contains @samp{%n}. (This is
2393 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2394 in future warnings may be added to @option{-Wformat-security} that are not
2395 included in @option{-Wformat-nonliteral}.)
2399 Enable @option{-Wformat} plus format checks not included in
2400 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2401 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2405 Warn about passing a null pointer for arguments marked as
2406 requiring a non-null value by the @code{nonnull} function attribute.
2408 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2409 can be disabled with the @option{-Wno-nonnull} option.
2411 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2413 Warn about uninitialized variables which are initialized with themselves.
2414 Note this option can only be used with the @option{-Wuninitialized} option,
2415 which in turn only works with @option{-O1} and above.
2417 For example, GCC will warn about @code{i} being uninitialized in the
2418 following snippet only when @option{-Winit-self} has been specified:
2429 @item -Wimplicit-int
2430 @opindex Wimplicit-int
2431 Warn when a declaration does not specify a type.
2432 This warning is enabled by @option{-Wall}.
2434 @item -Wimplicit-function-declaration
2435 @itemx -Werror-implicit-function-declaration
2436 @opindex Wimplicit-function-declaration
2437 @opindex Werror-implicit-function-declaration
2438 Give a warning (or error) whenever a function is used before being
2439 declared. The form @option{-Wno-error-implicit-function-declaration}
2441 This warning is enabled by @option{-Wall} (as a warning, not an error).
2445 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2446 This warning is enabled by @option{-Wall}.
2450 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2451 function with external linkage, returning int, taking either zero
2452 arguments, two, or three arguments of appropriate types.
2453 This warning is enabled by @option{-Wall}.
2455 @item -Wmissing-braces
2456 @opindex Wmissing-braces
2457 Warn if an aggregate or union initializer is not fully bracketed. In
2458 the following example, the initializer for @samp{a} is not fully
2459 bracketed, but that for @samp{b} is fully bracketed.
2462 int a[2][2] = @{ 0, 1, 2, 3 @};
2463 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2466 This warning is enabled by @option{-Wall}.
2468 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2469 @opindex Wmissing-include-dirs
2470 Warn if a user-supplied include directory does not exist.
2473 @opindex Wparentheses
2474 Warn if parentheses are omitted in certain contexts, such
2475 as when there is an assignment in a context where a truth value
2476 is expected, or when operators are nested whose precedence people
2477 often get confused about. Only the warning for an assignment used as
2478 a truth value is supported when compiling C++; the other warnings are
2479 only supported when compiling C@.
2481 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2482 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2483 interpretation from that of ordinary mathematical notation.
2485 Also warn about constructions where there may be confusion to which
2486 @code{if} statement an @code{else} branch belongs. Here is an example of
2501 In C, every @code{else} branch belongs to the innermost possible @code{if}
2502 statement, which in this example is @code{if (b)}. This is often not
2503 what the programmer expected, as illustrated in the above example by
2504 indentation the programmer chose. When there is the potential for this
2505 confusion, GCC will issue a warning when this flag is specified.
2506 To eliminate the warning, add explicit braces around the innermost
2507 @code{if} statement so there is no way the @code{else} could belong to
2508 the enclosing @code{if}. The resulting code would look like this:
2524 This warning is enabled by @option{-Wall}.
2526 @item -Wsequence-point
2527 @opindex Wsequence-point
2528 Warn about code that may have undefined semantics because of violations
2529 of sequence point rules in the C standard.
2531 The C standard defines the order in which expressions in a C program are
2532 evaluated in terms of @dfn{sequence points}, which represent a partial
2533 ordering between the execution of parts of the program: those executed
2534 before the sequence point, and those executed after it. These occur
2535 after the evaluation of a full expression (one which is not part of a
2536 larger expression), after the evaluation of the first operand of a
2537 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2538 function is called (but after the evaluation of its arguments and the
2539 expression denoting the called function), and in certain other places.
2540 Other than as expressed by the sequence point rules, the order of
2541 evaluation of subexpressions of an expression is not specified. All
2542 these rules describe only a partial order rather than a total order,
2543 since, for example, if two functions are called within one expression
2544 with no sequence point between them, the order in which the functions
2545 are called is not specified. However, the standards committee have
2546 ruled that function calls do not overlap.
2548 It is not specified when between sequence points modifications to the
2549 values of objects take effect. Programs whose behavior depends on this
2550 have undefined behavior; the C standard specifies that ``Between the
2551 previous and next sequence point an object shall have its stored value
2552 modified at most once by the evaluation of an expression. Furthermore,
2553 the prior value shall be read only to determine the value to be
2554 stored.''. If a program breaks these rules, the results on any
2555 particular implementation are entirely unpredictable.
2557 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2558 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2559 diagnosed by this option, and it may give an occasional false positive
2560 result, but in general it has been found fairly effective at detecting
2561 this sort of problem in programs.
2563 The present implementation of this option only works for C programs. A
2564 future implementation may also work for C++ programs.
2566 The C standard is worded confusingly, therefore there is some debate
2567 over the precise meaning of the sequence point rules in subtle cases.
2568 Links to discussions of the problem, including proposed formal
2569 definitions, may be found on the GCC readings page, at
2570 @w{@uref{http://gcc.gnu.org/readings.html}}.
2572 This warning is enabled by @option{-Wall}.
2575 @opindex Wreturn-type
2576 Warn whenever a function is defined with a return-type that defaults to
2577 @code{int}. Also warn about any @code{return} statement with no
2578 return-value in a function whose return-type is not @code{void}.
2580 For C, also warn if the return type of a function has a type qualifier
2581 such as @code{const}. Such a type qualifier has no effect, since the
2582 value returned by a function is not an lvalue. ISO C prohibits
2583 qualified @code{void} return types on function definitions, so such
2584 return types always receive a warning even without this option.
2586 For C++, a function without return type always produces a diagnostic
2587 message, even when @option{-Wno-return-type} is specified. The only
2588 exceptions are @samp{main} and functions defined in system headers.
2590 This warning is enabled by @option{-Wall}.
2594 Warn whenever a @code{switch} statement has an index of enumerated type
2595 and lacks a @code{case} for one or more of the named codes of that
2596 enumeration. (The presence of a @code{default} label prevents this
2597 warning.) @code{case} labels outside the enumeration range also
2598 provoke warnings when this option is used.
2599 This warning is enabled by @option{-Wall}.
2601 @item -Wswitch-default
2602 @opindex Wswitch-switch
2603 Warn whenever a @code{switch} statement does not have a @code{default}
2607 @opindex Wswitch-enum
2608 Warn whenever a @code{switch} statement has an index of enumerated type
2609 and lacks a @code{case} for one or more of the named codes of that
2610 enumeration. @code{case} labels outside the enumeration range also
2611 provoke warnings when this option is used.
2615 Warn if any trigraphs are encountered that might change the meaning of
2616 the program (trigraphs within comments are not warned about).
2617 This warning is enabled by @option{-Wall}.
2619 @item -Wunused-function
2620 @opindex Wunused-function
2621 Warn whenever a static function is declared but not defined or a
2622 non-inline static function is unused.
2623 This warning is enabled by @option{-Wall}.
2625 @item -Wunused-label
2626 @opindex Wunused-label
2627 Warn whenever a label is declared but not used.
2628 This warning is enabled by @option{-Wall}.
2630 To suppress this warning use the @samp{unused} attribute
2631 (@pxref{Variable Attributes}).
2633 @item -Wunused-parameter
2634 @opindex Wunused-parameter
2635 Warn whenever a function parameter is unused aside from its declaration.
2637 To suppress this warning use the @samp{unused} attribute
2638 (@pxref{Variable Attributes}).
2640 @item -Wunused-variable
2641 @opindex Wunused-variable
2642 Warn whenever a local variable or non-constant static variable is unused
2643 aside from its declaration
2644 This warning is enabled by @option{-Wall}.
2646 To suppress this warning use the @samp{unused} attribute
2647 (@pxref{Variable Attributes}).
2649 @item -Wunused-value
2650 @opindex Wunused-value
2651 Warn whenever a statement computes a result that is explicitly not used.
2652 This warning is enabled by @option{-Wall}.
2654 To suppress this warning cast the expression to @samp{void}.
2658 All the above @option{-Wunused} options combined.
2660 In order to get a warning about an unused function parameter, you must
2661 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2662 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2664 @item -Wuninitialized
2665 @opindex Wuninitialized
2666 Warn if an automatic variable is used without first being initialized or
2667 if a variable may be clobbered by a @code{setjmp} call.
2669 These warnings are possible only in optimizing compilation,
2670 because they require data flow information that is computed only
2671 when optimizing. If you don't specify @option{-O}, you simply won't
2674 If you want to warn about code which uses the uninitialized value of the
2675 variable in its own initializer, use the @option{-Winit-self} option.
2677 These warnings occur for individual uninitialized or clobbered
2678 elements of structure, union or array variables as well as for
2679 variables which are uninitialized or clobbered as a whole. They do
2680 not occur for variables or elements declared @code{volatile}. Because
2681 these warnings depend on optimization, the exact variables or elements
2682 for which there are warnings will depend on the precise optimization
2683 options and version of GCC used.
2685 Note that there may be no warning about a variable that is used only
2686 to compute a value that itself is never used, because such
2687 computations may be deleted by data flow analysis before the warnings
2690 These warnings are made optional because GCC is not smart
2691 enough to see all the reasons why the code might be correct
2692 despite appearing to have an error. Here is one example of how
2713 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2714 always initialized, but GCC doesn't know this. Here is
2715 another common case:
2720 if (change_y) save_y = y, y = new_y;
2722 if (change_y) y = save_y;
2727 This has no bug because @code{save_y} is used only if it is set.
2729 @cindex @code{longjmp} warnings
2730 This option also warns when a non-volatile automatic variable might be
2731 changed by a call to @code{longjmp}. These warnings as well are possible
2732 only in optimizing compilation.
2734 The compiler sees only the calls to @code{setjmp}. It cannot know
2735 where @code{longjmp} will be called; in fact, a signal handler could
2736 call it at any point in the code. As a result, you may get a warning
2737 even when there is in fact no problem because @code{longjmp} cannot
2738 in fact be called at the place which would cause a problem.
2740 Some spurious warnings can be avoided if you declare all the functions
2741 you use that never return as @code{noreturn}. @xref{Function
2744 This warning is enabled by @option{-Wall}.
2746 @item -Wunknown-pragmas
2747 @opindex Wunknown-pragmas
2748 @cindex warning for unknown pragmas
2749 @cindex unknown pragmas, warning
2750 @cindex pragmas, warning of unknown
2751 Warn when a #pragma directive is encountered which is not understood by
2752 GCC@. If this command line option is used, warnings will even be issued
2753 for unknown pragmas in system header files. This is not the case if
2754 the warnings were only enabled by the @option{-Wall} command line option.
2757 @opindex Wno-pragmas
2759 Do not warn about misuses of pragmas, such as incorrect parameters,
2760 invalid syntax, or conflicts between pragmas. See also
2761 @samp{-Wunknown-pragmas}.
2763 @item -Wstrict-aliasing
2764 @opindex Wstrict-aliasing
2765 This option is only active when @option{-fstrict-aliasing} is active.
2766 It warns about code which might break the strict aliasing rules that the
2767 compiler is using for optimization. The warning does not catch all
2768 cases, but does attempt to catch the more common pitfalls. It is
2769 included in @option{-Wall}.
2771 @item -Wstrict-aliasing=2
2772 @opindex Wstrict-aliasing=2
2773 This option is only active when @option{-fstrict-aliasing} is active.
2774 It warns about code which might break the strict aliasing rules that the
2775 compiler is using for optimization. This warning catches more cases than
2776 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2777 cases that are safe.
2781 All of the above @samp{-W} options combined. This enables all the
2782 warnings about constructions that some users consider questionable, and
2783 that are easy to avoid (or modify to prevent the warning), even in
2784 conjunction with macros. This also enables some language-specific
2785 warnings described in @ref{C++ Dialect Options} and
2786 @ref{Objective-C and Objective-C++ Dialect Options}.
2789 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2790 Some of them warn about constructions that users generally do not
2791 consider questionable, but which occasionally you might wish to check
2792 for; others warn about constructions that are necessary or hard to avoid
2793 in some cases, and there is no simple way to modify the code to suppress
2800 (This option used to be called @option{-W}. The older name is still
2801 supported, but the newer name is more descriptive.) Print extra warning
2802 messages for these events:
2806 A function can return either with or without a value. (Falling
2807 off the end of the function body is considered returning without
2808 a value.) For example, this function would evoke such a
2822 An expression-statement or the left-hand side of a comma expression
2823 contains no side effects.
2824 To suppress the warning, cast the unused expression to void.
2825 For example, an expression such as @samp{x[i,j]} will cause a warning,
2826 but @samp{x[(void)i,j]} will not.
2829 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2832 Storage-class specifiers like @code{static} are not the first things in
2833 a declaration. According to the C Standard, this usage is obsolescent.
2836 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2840 A comparison between signed and unsigned values could produce an
2841 incorrect result when the signed value is converted to unsigned.
2842 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2845 An aggregate has an initializer which does not initialize all members.
2846 This warning can be independently controlled by
2847 @option{-Wmissing-field-initializers}.
2850 A function parameter is declared without a type specifier in K&R-style
2858 An empty body occurs in an @samp{if} or @samp{else} statement.
2861 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2862 @samp{>}, or @samp{>=}.
2865 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2868 Any of several floating-point events that often indicate errors, such as
2869 overflow, underflow, loss of precision, etc.
2871 @item @r{(C++ only)}
2872 An enumerator and a non-enumerator both appear in a conditional expression.
2874 @item @r{(C++ only)}
2875 A non-static reference or non-static @samp{const} member appears in a
2876 class without constructors.
2878 @item @r{(C++ only)}
2879 Ambiguous virtual bases.
2881 @item @r{(C++ only)}
2882 Subscripting an array which has been declared @samp{register}.
2884 @item @r{(C++ only)}
2885 Taking the address of a variable which has been declared @samp{register}.
2887 @item @r{(C++ only)}
2888 A base class is not initialized in a derived class' copy constructor.
2891 @item -Wno-div-by-zero
2892 @opindex Wno-div-by-zero
2893 @opindex Wdiv-by-zero
2894 Do not warn about compile-time integer division by zero. Floating point
2895 division by zero is not warned about, as it can be a legitimate way of
2896 obtaining infinities and NaNs.
2898 @item -Wsystem-headers
2899 @opindex Wsystem-headers
2900 @cindex warnings from system headers
2901 @cindex system headers, warnings from
2902 Print warning messages for constructs found in system header files.
2903 Warnings from system headers are normally suppressed, on the assumption
2904 that they usually do not indicate real problems and would only make the
2905 compiler output harder to read. Using this command line option tells
2906 GCC to emit warnings from system headers as if they occurred in user
2907 code. However, note that using @option{-Wall} in conjunction with this
2908 option will @emph{not} warn about unknown pragmas in system
2909 headers---for that, @option{-Wunknown-pragmas} must also be used.
2912 @opindex Wfloat-equal
2913 Warn if floating point values are used in equality comparisons.
2915 The idea behind this is that sometimes it is convenient (for the
2916 programmer) to consider floating-point values as approximations to
2917 infinitely precise real numbers. If you are doing this, then you need
2918 to compute (by analyzing the code, or in some other way) the maximum or
2919 likely maximum error that the computation introduces, and allow for it
2920 when performing comparisons (and when producing output, but that's a
2921 different problem). In particular, instead of testing for equality, you
2922 would check to see whether the two values have ranges that overlap; and
2923 this is done with the relational operators, so equality comparisons are
2926 @item -Wtraditional @r{(C only)}
2927 @opindex Wtraditional
2928 Warn about certain constructs that behave differently in traditional and
2929 ISO C@. Also warn about ISO C constructs that have no traditional C
2930 equivalent, and/or problematic constructs which should be avoided.
2934 Macro parameters that appear within string literals in the macro body.
2935 In traditional C macro replacement takes place within string literals,
2936 but does not in ISO C@.
2939 In traditional C, some preprocessor directives did not exist.
2940 Traditional preprocessors would only consider a line to be a directive
2941 if the @samp{#} appeared in column 1 on the line. Therefore
2942 @option{-Wtraditional} warns about directives that traditional C
2943 understands but would ignore because the @samp{#} does not appear as the
2944 first character on the line. It also suggests you hide directives like
2945 @samp{#pragma} not understood by traditional C by indenting them. Some
2946 traditional implementations would not recognize @samp{#elif}, so it
2947 suggests avoiding it altogether.
2950 A function-like macro that appears without arguments.
2953 The unary plus operator.
2956 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2957 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2958 constants.) Note, these suffixes appear in macros defined in the system
2959 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2960 Use of these macros in user code might normally lead to spurious
2961 warnings, however GCC's integrated preprocessor has enough context to
2962 avoid warning in these cases.
2965 A function declared external in one block and then used after the end of
2969 A @code{switch} statement has an operand of type @code{long}.
2972 A non-@code{static} function declaration follows a @code{static} one.
2973 This construct is not accepted by some traditional C compilers.
2976 The ISO type of an integer constant has a different width or
2977 signedness from its traditional type. This warning is only issued if
2978 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2979 typically represent bit patterns, are not warned about.
2982 Usage of ISO string concatenation is detected.
2985 Initialization of automatic aggregates.
2988 Identifier conflicts with labels. Traditional C lacks a separate
2989 namespace for labels.
2992 Initialization of unions. If the initializer is zero, the warning is
2993 omitted. This is done under the assumption that the zero initializer in
2994 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2995 initializer warnings and relies on default initialization to zero in the
2999 Conversions by prototypes between fixed/floating point values and vice
3000 versa. The absence of these prototypes when compiling with traditional
3001 C would cause serious problems. This is a subset of the possible
3002 conversion warnings, for the full set use @option{-Wconversion}.
3005 Use of ISO C style function definitions. This warning intentionally is
3006 @emph{not} issued for prototype declarations or variadic functions
3007 because these ISO C features will appear in your code when using
3008 libiberty's traditional C compatibility macros, @code{PARAMS} and
3009 @code{VPARAMS}. This warning is also bypassed for nested functions
3010 because that feature is already a GCC extension and thus not relevant to
3011 traditional C compatibility.
3014 @item -Wdeclaration-after-statement @r{(C only)}
3015 @opindex Wdeclaration-after-statement
3016 Warn when a declaration is found after a statement in a block. This
3017 construct, known from C++, was introduced with ISO C99 and is by default
3018 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3019 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3023 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3025 @item -Wno-endif-labels
3026 @opindex Wno-endif-labels
3027 @opindex Wendif-labels
3028 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3032 Warn whenever a local variable shadows another local variable, parameter or
3033 global variable or whenever a built-in function is shadowed.
3035 @item -Wlarger-than-@var{len}
3036 @opindex Wlarger-than
3037 Warn whenever an object of larger than @var{len} bytes is defined.
3039 @item -Wunsafe-loop-optimizations
3040 @opindex Wunsafe-loop-optimizations
3041 Warn if the loop cannot be optimized because the compiler could not
3042 assume anything on the bounds of the loop indices. With
3043 @option{-funsafe-loop-optimizations} warn if the compiler made
3046 @item -Wpointer-arith
3047 @opindex Wpointer-arith
3048 Warn about anything that depends on the ``size of'' a function type or
3049 of @code{void}. GNU C assigns these types a size of 1, for
3050 convenience in calculations with @code{void *} pointers and pointers
3053 @item -Wbad-function-cast @r{(C only)}
3054 @opindex Wbad-function-cast
3055 Warn whenever a function call is cast to a non-matching type.
3056 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3059 Warn about ISO C constructs that are outside of the common subset of
3060 ISO C and ISO C++, e.g.@: request for implicit conversion from
3061 @code{void *} to a pointer to non-@code{void} type.
3065 Warn whenever a pointer is cast so as to remove a type qualifier from
3066 the target type. For example, warn if a @code{const char *} is cast
3067 to an ordinary @code{char *}.
3070 @opindex Wcast-align
3071 Warn whenever a pointer is cast such that the required alignment of the
3072 target is increased. For example, warn if a @code{char *} is cast to
3073 an @code{int *} on machines where integers can only be accessed at
3074 two- or four-byte boundaries.
3076 @item -Wwrite-strings
3077 @opindex Wwrite-strings
3078 When compiling C, give string constants the type @code{const
3079 char[@var{length}]} so that
3080 copying the address of one into a non-@code{const} @code{char *}
3081 pointer will get a warning; when compiling C++, warn about the
3082 deprecated conversion from string constants to @code{char *}.
3083 These warnings will help you find at
3084 compile time code that can try to write into a string constant, but
3085 only if you have been very careful about using @code{const} in
3086 declarations and prototypes. Otherwise, it will just be a nuisance;
3087 this is why we did not make @option{-Wall} request these warnings.
3090 @opindex Wconversion
3091 Warn if a prototype causes a type conversion that is different from what
3092 would happen to the same argument in the absence of a prototype. This
3093 includes conversions of fixed point to floating and vice versa, and
3094 conversions changing the width or signedness of a fixed point argument
3095 except when the same as the default promotion.
3097 Also, warn if a negative integer constant expression is implicitly
3098 converted to an unsigned type. For example, warn about the assignment
3099 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3100 casts like @code{(unsigned) -1}.
3102 @item -Wsign-compare
3103 @opindex Wsign-compare
3104 @cindex warning for comparison of signed and unsigned values
3105 @cindex comparison of signed and unsigned values, warning
3106 @cindex signed and unsigned values, comparison warning
3107 Warn when a comparison between signed and unsigned values could produce
3108 an incorrect result when the signed value is converted to unsigned.
3109 This warning is also enabled by @option{-Wextra}; to get the other warnings
3110 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3112 @item -Waggregate-return
3113 @opindex Waggregate-return
3114 Warn if any functions that return structures or unions are defined or
3115 called. (In languages where you can return an array, this also elicits
3119 @opindex Walways-true
3120 Warn about comparisons which are always true such as testing if
3121 unsigned values are greater than or equal to zero. This warning is
3122 enabled by @option{-Wall}.
3124 @item -Wno-attributes
3125 @opindex Wno-attributes
3126 @opindex Wattributes
3127 Do not warn if an unexpected @code{__attribute__} is used, such as
3128 unrecognized attributes, function attributes applied to variables,
3129 etc. This will not stop errors for incorrect use of supported
3132 @item -Wstrict-prototypes @r{(C only)}
3133 @opindex Wstrict-prototypes
3134 Warn if a function is declared or defined without specifying the
3135 argument types. (An old-style function definition is permitted without
3136 a warning if preceded by a declaration which specifies the argument
3139 @item -Wold-style-definition @r{(C only)}
3140 @opindex Wold-style-definition
3141 Warn if an old-style function definition is used. A warning is given
3142 even if there is a previous prototype.
3144 @item -Wmissing-prototypes @r{(C only)}
3145 @opindex Wmissing-prototypes
3146 Warn if a global function is defined without a previous prototype
3147 declaration. This warning is issued even if the definition itself
3148 provides a prototype. The aim is to detect global functions that fail
3149 to be declared in header files.
3151 @item -Wmissing-declarations @r{(C only)}
3152 @opindex Wmissing-declarations
3153 Warn if a global function is defined without a previous declaration.
3154 Do so even if the definition itself provides a prototype.
3155 Use this option to detect global functions that are not declared in
3158 @item -Wmissing-field-initializers
3159 @opindex Wmissing-field-initializers
3162 Warn if a structure's initializer has some fields missing. For
3163 example, the following code would cause such a warning, because
3164 @code{x.h} is implicitly zero:
3167 struct s @{ int f, g, h; @};
3168 struct s x = @{ 3, 4 @};
3171 This option does not warn about designated initializers, so the following
3172 modification would not trigger a warning:
3175 struct s @{ int f, g, h; @};
3176 struct s x = @{ .f = 3, .g = 4 @};
3179 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3180 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3182 @item -Wmissing-noreturn
3183 @opindex Wmissing-noreturn
3184 Warn about functions which might be candidates for attribute @code{noreturn}.
3185 Note these are only possible candidates, not absolute ones. Care should
3186 be taken to manually verify functions actually do not ever return before
3187 adding the @code{noreturn} attribute, otherwise subtle code generation
3188 bugs could be introduced. You will not get a warning for @code{main} in
3189 hosted C environments.
3191 @item -Wmissing-format-attribute
3192 @opindex Wmissing-format-attribute
3194 Warn about function pointers which might be candidates for @code{format}
3195 attributes. Note these are only possible candidates, not absolute ones.
3196 GCC will guess that function pointers with @code{format} attributes that
3197 are used in assignment, initialization, parameter passing or return
3198 statements should have a corresponding @code{format} attribute in the
3199 resulting type. I.e.@: the left-hand side of the assignment or
3200 initialization, the type of the parameter variable, or the return type
3201 of the containing function respectively should also have a @code{format}
3202 attribute to avoid the warning.
3204 GCC will also warn about function definitions which might be
3205 candidates for @code{format} attributes. Again, these are only
3206 possible candidates. GCC will guess that @code{format} attributes
3207 might be appropriate for any function that calls a function like
3208 @code{vprintf} or @code{vscanf}, but this might not always be the
3209 case, and some functions for which @code{format} attributes are
3210 appropriate may not be detected.
3212 @item -Wno-multichar
3213 @opindex Wno-multichar
3215 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3216 Usually they indicate a typo in the user's code, as they have
3217 implementation-defined values, and should not be used in portable code.
3219 @item -Wnormalized=<none|id|nfc|nfkc>
3220 @opindex Wnormalized
3223 @cindex character set, input normalization
3224 In ISO C and ISO C++, two identifiers are different if they are
3225 different sequences of characters. However, sometimes when characters
3226 outside the basic ASCII character set are used, you can have two
3227 different character sequences that look the same. To avoid confusion,
3228 the ISO 10646 standard sets out some @dfn{normalization rules} which
3229 when applied ensure that two sequences that look the same are turned into
3230 the same sequence. GCC can warn you if you are using identifiers which
3231 have not been normalized; this option controls that warning.
3233 There are four levels of warning that GCC supports. The default is
3234 @option{-Wnormalized=nfc}, which warns about any identifier which is
3235 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3236 recommended form for most uses.
3238 Unfortunately, there are some characters which ISO C and ISO C++ allow
3239 in identifiers that when turned into NFC aren't allowable as
3240 identifiers. That is, there's no way to use these symbols in portable
3241 ISO C or C++ and have all your identifiers in NFC.
3242 @option{-Wnormalized=id} suppresses the warning for these characters.
3243 It is hoped that future versions of the standards involved will correct
3244 this, which is why this option is not the default.
3246 You can switch the warning off for all characters by writing
3247 @option{-Wnormalized=none}. You would only want to do this if you
3248 were using some other normalization scheme (like ``D''), because
3249 otherwise you can easily create bugs that are literally impossible to see.
3251 Some characters in ISO 10646 have distinct meanings but look identical
3252 in some fonts or display methodologies, especially once formatting has
3253 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3254 LETTER N'', will display just like a regular @code{n} which has been
3255 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3256 normalisation scheme to convert all these into a standard form as
3257 well, and GCC will warn if your code is not in NFKC if you use
3258 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3259 about every identifier that contains the letter O because it might be
3260 confused with the digit 0, and so is not the default, but may be
3261 useful as a local coding convention if the programming environment is
3262 unable to be fixed to display these characters distinctly.
3264 @item -Wno-deprecated-declarations
3265 @opindex Wno-deprecated-declarations
3266 Do not warn about uses of functions, variables, and types marked as
3267 deprecated by using the @code{deprecated} attribute.
3268 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3269 @pxref{Type Attributes}.)
3273 Warn if a structure is given the packed attribute, but the packed
3274 attribute has no effect on the layout or size of the structure.
3275 Such structures may be mis-aligned for little benefit. For
3276 instance, in this code, the variable @code{f.x} in @code{struct bar}
3277 will be misaligned even though @code{struct bar} does not itself
3278 have the packed attribute:
3285 @} __attribute__((packed));
3295 Warn if padding is included in a structure, either to align an element
3296 of the structure or to align the whole structure. Sometimes when this
3297 happens it is possible to rearrange the fields of the structure to
3298 reduce the padding and so make the structure smaller.
3300 @item -Wredundant-decls
3301 @opindex Wredundant-decls
3302 Warn if anything is declared more than once in the same scope, even in
3303 cases where multiple declaration is valid and changes nothing.
3305 @item -Wnested-externs @r{(C only)}
3306 @opindex Wnested-externs
3307 Warn if an @code{extern} declaration is encountered within a function.
3309 @item -Wunreachable-code
3310 @opindex Wunreachable-code
3311 Warn if the compiler detects that code will never be executed.
3313 This option is intended to warn when the compiler detects that at
3314 least a whole line of source code will never be executed, because
3315 some condition is never satisfied or because it is after a
3316 procedure that never returns.
3318 It is possible for this option to produce a warning even though there
3319 are circumstances under which part of the affected line can be executed,
3320 so care should be taken when removing apparently-unreachable code.
3322 For instance, when a function is inlined, a warning may mean that the
3323 line is unreachable in only one inlined copy of the function.
3325 This option is not made part of @option{-Wall} because in a debugging
3326 version of a program there is often substantial code which checks
3327 correct functioning of the program and is, hopefully, unreachable
3328 because the program does work. Another common use of unreachable
3329 code is to provide behavior which is selectable at compile-time.
3333 Warn if a function can not be inlined and it was declared as inline.
3334 Even with this option, the compiler will not warn about failures to
3335 inline functions declared in system headers.
3337 The compiler uses a variety of heuristics to determine whether or not
3338 to inline a function. For example, the compiler takes into account
3339 the size of the function being inlined and the amount of inlining
3340 that has already been done in the current function. Therefore,
3341 seemingly insignificant changes in the source program can cause the
3342 warnings produced by @option{-Winline} to appear or disappear.
3344 @item -Wno-invalid-offsetof @r{(C++ only)}
3345 @opindex Wno-invalid-offsetof
3346 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3347 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3348 to a non-POD type is undefined. In existing C++ implementations,
3349 however, @samp{offsetof} typically gives meaningful results even when
3350 applied to certain kinds of non-POD types. (Such as a simple
3351 @samp{struct} that fails to be a POD type only by virtue of having a
3352 constructor.) This flag is for users who are aware that they are
3353 writing nonportable code and who have deliberately chosen to ignore the
3356 The restrictions on @samp{offsetof} may be relaxed in a future version
3357 of the C++ standard.
3359 @item -Wno-int-to-pointer-cast @r{(C only)}
3360 @opindex Wno-int-to-pointer-cast
3361 Suppress warnings from casts to pointer type of an integer of a
3364 @item -Wno-pointer-to-int-cast @r{(C only)}
3365 @opindex Wno-pointer-to-int-cast
3366 Suppress warnings from casts from a pointer to an integer type of a
3370 @opindex Winvalid-pch
3371 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3372 the search path but can't be used.
3376 @opindex Wno-long-long
3377 Warn if @samp{long long} type is used. This is default. To inhibit
3378 the warning messages, use @option{-Wno-long-long}. Flags
3379 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3380 only when @option{-pedantic} flag is used.
3382 @item -Wvariadic-macros
3383 @opindex Wvariadic-macros
3384 @opindex Wno-variadic-macros
3385 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3386 alternate syntax when in pedantic ISO C99 mode. This is default.
3387 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3389 @item -Wvolatile-register-var
3390 @opindex Wvolatile-register-var
3391 @opindex Wno-volatile-register-var
3392 Warn if a register variable is declared volatile. The volatile
3393 modifier does not inhibit all optimizations that may eliminate reads
3394 and/or writes to register variables.
3396 @item -Wdisabled-optimization
3397 @opindex Wdisabled-optimization
3398 Warn if a requested optimization pass is disabled. This warning does
3399 not generally indicate that there is anything wrong with your code; it
3400 merely indicates that GCC's optimizers were unable to handle the code
3401 effectively. Often, the problem is that your code is too big or too
3402 complex; GCC will refuse to optimize programs when the optimization
3403 itself is likely to take inordinate amounts of time.
3405 @item -Wno-pointer-sign
3406 @opindex Wno-pointer-sign
3407 Don't warn for pointer argument passing or assignment with different signedness.
3408 Only useful in the negative form since this warning is enabled by default.
3409 This option is only supported for C and Objective-C@.
3413 Make all warnings into errors.
3417 Make the specified warning into an errors. The specifier for a
3418 warning is appended, for example @option{-Werror=switch} turns the
3419 warnings controlled by @option{-Wswitch} into errors. This switch
3420 takes a negative form, to be used to negate @option{-Werror} for
3421 specific warnings, for example @option{-Wno-error=switch} makes
3422 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3423 is in effect. You can use the @option{-fdiagnostics-show-option}
3424 option to have each controllable warning amended with the option which
3425 controls it, to determine what to use with this option.
3427 Note that specifying @option{-Werror=}@var{foo} automatically implies
3428 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3431 @item -Wstack-protector
3432 @opindex Wstack-protector
3433 This option is only active when @option{-fstack-protector} is active. It
3434 warns about functions that will not be protected against stack smashing.
3436 @item -Wstring-literal-comparison
3437 @opindex Wstring-literal-comparison
3438 Warn about suspicious comparisons to string literal constants. In C,
3439 direct comparisons against the memory address of a string literal, such
3440 as @code{if (x == "abc")}, typically indicate a programmer error, and
3441 even when intentional, result in unspecified behavior and are not portable.
3442 Usually these warnings alert that the programmer intended to use
3443 @code{strcmp}. This warning is enabled by @option{-Wall}.
3447 @node Debugging Options
3448 @section Options for Debugging Your Program or GCC
3449 @cindex options, debugging
3450 @cindex debugging information options
3452 GCC has various special options that are used for debugging
3453 either your program or GCC:
3458 Produce debugging information in the operating system's native format
3459 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3462 On most systems that use stabs format, @option{-g} enables use of extra
3463 debugging information that only GDB can use; this extra information
3464 makes debugging work better in GDB but will probably make other debuggers
3466 refuse to read the program. If you want to control for certain whether
3467 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3468 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3470 GCC allows you to use @option{-g} with
3471 @option{-O}. The shortcuts taken by optimized code may occasionally
3472 produce surprising results: some variables you declared may not exist
3473 at all; flow of control may briefly move where you did not expect it;
3474 some statements may not be executed because they compute constant
3475 results or their values were already at hand; some statements may
3476 execute in different places because they were moved out of loops.
3478 Nevertheless it proves possible to debug optimized output. This makes
3479 it reasonable to use the optimizer for programs that might have bugs.
3481 The following options are useful when GCC is generated with the
3482 capability for more than one debugging format.
3486 Produce debugging information for use by GDB@. This means to use the
3487 most expressive format available (DWARF 2, stabs, or the native format
3488 if neither of those are supported), including GDB extensions if at all
3493 Produce debugging information in stabs format (if that is supported),
3494 without GDB extensions. This is the format used by DBX on most BSD
3495 systems. On MIPS, Alpha and System V Release 4 systems this option
3496 produces stabs debugging output which is not understood by DBX or SDB@.
3497 On System V Release 4 systems this option requires the GNU assembler.
3499 @item -feliminate-unused-debug-symbols
3500 @opindex feliminate-unused-debug-symbols
3501 Produce debugging information in stabs format (if that is supported),
3502 for only symbols that are actually used.
3506 Produce debugging information in stabs format (if that is supported),
3507 using GNU extensions understood only by the GNU debugger (GDB)@. The
3508 use of these extensions is likely to make other debuggers crash or
3509 refuse to read the program.
3513 Produce debugging information in COFF format (if that is supported).
3514 This is the format used by SDB on most System V systems prior to
3519 Produce debugging information in XCOFF format (if that is supported).
3520 This is the format used by the DBX debugger on IBM RS/6000 systems.
3524 Produce debugging information in XCOFF format (if that is supported),
3525 using GNU extensions understood only by the GNU debugger (GDB)@. The
3526 use of these extensions is likely to make other debuggers crash or
3527 refuse to read the program, and may cause assemblers other than the GNU
3528 assembler (GAS) to fail with an error.
3532 Produce debugging information in DWARF version 2 format (if that is
3533 supported). This is the format used by DBX on IRIX 6. With this
3534 option, GCC uses features of DWARF version 3 when they are useful;
3535 version 3 is upward compatible with version 2, but may still cause
3536 problems for older debuggers.
3540 Produce debugging information in VMS debug format (if that is
3541 supported). This is the format used by DEBUG on VMS systems.
3544 @itemx -ggdb@var{level}
3545 @itemx -gstabs@var{level}
3546 @itemx -gcoff@var{level}
3547 @itemx -gxcoff@var{level}
3548 @itemx -gvms@var{level}
3549 Request debugging information and also use @var{level} to specify how
3550 much information. The default level is 2.
3552 Level 1 produces minimal information, enough for making backtraces in
3553 parts of the program that you don't plan to debug. This includes
3554 descriptions of functions and external variables, but no information
3555 about local variables and no line numbers.
3557 Level 3 includes extra information, such as all the macro definitions
3558 present in the program. Some debuggers support macro expansion when
3559 you use @option{-g3}.
3561 @option{-gdwarf-2} does not accept a concatenated debug level, because
3562 GCC used to support an option @option{-gdwarf} that meant to generate
3563 debug information in version 1 of the DWARF format (which is very
3564 different from version 2), and it would have been too confusing. That
3565 debug format is long obsolete, but the option cannot be changed now.
3566 Instead use an additional @option{-g@var{level}} option to change the
3567 debug level for DWARF2.
3569 @item -feliminate-dwarf2-dups
3570 @opindex feliminate-dwarf2-dups
3571 Compress DWARF2 debugging information by eliminating duplicated
3572 information about each symbol. This option only makes sense when
3573 generating DWARF2 debugging information with @option{-gdwarf-2}.
3575 @cindex @command{prof}
3578 Generate extra code to write profile information suitable for the
3579 analysis program @command{prof}. You must use this option when compiling
3580 the source files you want data about, and you must also use it when
3583 @cindex @command{gprof}
3586 Generate extra code to write profile information suitable for the
3587 analysis program @command{gprof}. You must use this option when compiling
3588 the source files you want data about, and you must also use it when
3593 Makes the compiler print out each function name as it is compiled, and
3594 print some statistics about each pass when it finishes.
3597 @opindex ftime-report
3598 Makes the compiler print some statistics about the time consumed by each
3599 pass when it finishes.
3602 @opindex fmem-report
3603 Makes the compiler print some statistics about permanent memory
3604 allocation when it finishes.
3606 @item -fprofile-arcs
3607 @opindex fprofile-arcs
3608 Add code so that program flow @dfn{arcs} are instrumented. During
3609 execution the program records how many times each branch and call is
3610 executed and how many times it is taken or returns. When the compiled
3611 program exits it saves this data to a file called
3612 @file{@var{auxname}.gcda} for each source file. The data may be used for
3613 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3614 test coverage analysis (@option{-ftest-coverage}). Each object file's
3615 @var{auxname} is generated from the name of the output file, if
3616 explicitly specified and it is not the final executable, otherwise it is
3617 the basename of the source file. In both cases any suffix is removed
3618 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3619 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3620 @xref{Cross-profiling}.
3622 @cindex @command{gcov}
3626 This option is used to compile and link code instrumented for coverage
3627 analysis. The option is a synonym for @option{-fprofile-arcs}
3628 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3629 linking). See the documentation for those options for more details.
3634 Compile the source files with @option{-fprofile-arcs} plus optimization
3635 and code generation options. For test coverage analysis, use the
3636 additional @option{-ftest-coverage} option. You do not need to profile
3637 every source file in a program.
3640 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3641 (the latter implies the former).
3644 Run the program on a representative workload to generate the arc profile
3645 information. This may be repeated any number of times. You can run
3646 concurrent instances of your program, and provided that the file system
3647 supports locking, the data files will be correctly updated. Also
3648 @code{fork} calls are detected and correctly handled (double counting
3652 For profile-directed optimizations, compile the source files again with
3653 the same optimization and code generation options plus
3654 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3655 Control Optimization}).
3658 For test coverage analysis, use @command{gcov} to produce human readable
3659 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3660 @command{gcov} documentation for further information.
3664 With @option{-fprofile-arcs}, for each function of your program GCC
3665 creates a program flow graph, then finds a spanning tree for the graph.
3666 Only arcs that are not on the spanning tree have to be instrumented: the
3667 compiler adds code to count the number of times that these arcs are
3668 executed. When an arc is the only exit or only entrance to a block, the
3669 instrumentation code can be added to the block; otherwise, a new basic
3670 block must be created to hold the instrumentation code.
3673 @item -ftest-coverage
3674 @opindex ftest-coverage
3675 Produce a notes file that the @command{gcov} code-coverage utility
3676 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3677 show program coverage. Each source file's note file is called
3678 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3679 above for a description of @var{auxname} and instructions on how to
3680 generate test coverage data. Coverage data will match the source files
3681 more closely, if you do not optimize.
3683 @item -d@var{letters}
3684 @item -fdump-rtl-@var{pass}
3686 Says to make debugging dumps during compilation at times specified by
3687 @var{letters}. This is used for debugging the RTL-based passes of the
3688 compiler. The file names for most of the dumps are made by appending a
3689 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3690 from the name of the output file, if explicitly specified and it is not
3691 an executable, otherwise it is the basename of the source file.
3693 Most debug dumps can be enabled either passing a letter to the @option{-d}
3694 option, or with a long @option{-fdump-rtl} switch; here are the possible
3695 letters for use in @var{letters} and @var{pass}, and their meanings:
3700 Annotate the assembler output with miscellaneous debugging information.
3703 @itemx -fdump-rtl-bp
3705 @opindex fdump-rtl-bp
3706 Dump after computing branch probabilities, to @file{@var{file}.09.bp}.
3709 @itemx -fdump-rtl-bbro
3711 @opindex fdump-rtl-bbro
3712 Dump after block reordering, to @file{@var{file}.30.bbro}.
3715 @itemx -fdump-rtl-combine
3717 @opindex fdump-rtl-combine
3718 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3721 @itemx -fdump-rtl-ce1
3722 @itemx -fdump-rtl-ce2
3724 @opindex fdump-rtl-ce1
3725 @opindex fdump-rtl-ce2
3726 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3727 first if conversion, to the file @file{@var{file}.11.ce1}. @option{-dC}
3728 and @option{-fdump-rtl-ce2} enable dumping after the second if
3729 conversion, to the file @file{@var{file}.18.ce2}.
3732 @itemx -fdump-rtl-btl
3733 @itemx -fdump-rtl-dbr
3735 @opindex fdump-rtl-btl
3736 @opindex fdump-rtl-dbr
3737 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3738 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3739 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3740 scheduling, to @file{@var{file}.36.dbr}.
3744 Dump all macro definitions, at the end of preprocessing, in addition to
3748 @itemx -fdump-rtl-ce3
3750 @opindex fdump-rtl-ce3
3751 Dump after the third if conversion, to @file{@var{file}.28.ce3}.
3754 @itemx -fdump-rtl-cfg
3755 @itemx -fdump-rtl-life
3757 @opindex fdump-rtl-cfg
3758 @opindex fdump-rtl-life
3759 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3760 and data flow analysis, to @file{@var{file}.08.cfg}. @option{-df}
3761 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3762 to @file{@var{file}.16.life}.
3765 @itemx -fdump-rtl-greg
3767 @opindex fdump-rtl-greg
3768 Dump after global register allocation, to @file{@var{file}.23.greg}.
3771 @itemx -fdump-rtl-gcse
3772 @itemx -fdump-rtl-bypass
3774 @opindex fdump-rtl-gcse
3775 @opindex fdump-rtl-bypass
3776 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3777 @file{@var{file}.05.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3778 enable dumping after jump bypassing and control flow optimizations, to
3779 @file{@var{file}.07.bypass}.
3782 @itemx -fdump-rtl-eh
3784 @opindex fdump-rtl-eh
3785 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3788 @itemx -fdump-rtl-sibling
3790 @opindex fdump-rtl-sibling
3791 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3794 @itemx -fdump-rtl-jump
3796 @opindex fdump-rtl-jump
3797 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3800 @itemx -fdump-rtl-stack
3802 @opindex fdump-rtl-stack
3803 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3806 @itemx -fdump-rtl-lreg
3808 @opindex fdump-rtl-lreg
3809 Dump after local register allocation, to @file{@var{file}.22.lreg}.
3812 @itemx -fdump-rtl-loop
3813 @itemx -fdump-rtl-loop2
3815 @opindex fdump-rtl-loop
3816 @opindex fdump-rtl-loop2
3817 @option{-dL} and @option{-fdump-rtl-loop} enable dumping after the first
3818 loop optimization pass, to @file{@var{file}.06.loop}. @option{-dL} and
3819 @option{-fdump-rtl-loop2} enable dumping after the second pass, to
3820 @file{@var{file}.13.loop2}.
3823 @itemx -fdump-rtl-sms
3825 @opindex fdump-rtl-sms
3826 Dump after modulo scheduling, to @file{@var{file}.20.sms}.
3829 @itemx -fdump-rtl-mach
3831 @opindex fdump-rtl-mach
3832 Dump after performing the machine dependent reorganization pass, to
3833 @file{@var{file}.35.mach}.
3836 @itemx -fdump-rtl-rnreg
3838 @opindex fdump-rtl-rnreg
3839 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3842 @itemx -fdump-rtl-regmove
3844 @opindex fdump-rtl-regmove
3845 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3848 @itemx -fdump-rtl-postreload
3850 @opindex fdump-rtl-postreload
3851 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3854 @itemx -fdump-rtl-expand
3856 @opindex fdump-rtl-expand
3857 Dump after RTL generation, to @file{@var{file}.00.expand}.
3860 @itemx -fdump-rtl-sched2
3862 @opindex fdump-rtl-sched2
3863 Dump after the second scheduling pass, to @file{@var{file}.32.sched2}.
3866 @itemx -fdump-rtl-cse
3868 @opindex fdump-rtl-cse
3869 Dump after CSE (including the jump optimization that sometimes follows
3870 CSE), to @file{@var{file}.04.cse}.
3873 @itemx -fdump-rtl-sched
3875 @opindex fdump-rtl-sched
3876 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3879 @itemx -fdump-rtl-cse2
3881 @opindex fdump-rtl-cse2
3882 Dump after the second CSE pass (including the jump optimization that
3883 sometimes follows CSE), to @file{@var{file}.15.cse2}.
3886 @itemx -fdump-rtl-tracer
3888 @opindex fdump-rtl-tracer
3889 Dump after running tracer, to @file{@var{file}.12.tracer}.
3892 @itemx -fdump-rtl-vpt
3893 @itemx -fdump-rtl-vartrack
3895 @opindex fdump-rtl-vpt
3896 @opindex fdump-rtl-vartrack
3897 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3898 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3899 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3900 to @file{@var{file}.34.vartrack}.
3903 @itemx -fdump-rtl-flow2
3905 @opindex fdump-rtl-flow2
3906 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3909 @itemx -fdump-rtl-peephole2
3911 @opindex fdump-rtl-peephole2
3912 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3915 @itemx -fdump-rtl-web
3917 @opindex fdump-rtl-web
3918 Dump after live range splitting, to @file{@var{file}.14.web}.
3921 @itemx -fdump-rtl-all
3923 @opindex fdump-rtl-all
3924 Produce all the dumps listed above.
3928 Produce a core dump whenever an error occurs.
3932 Print statistics on memory usage, at the end of the run, to
3937 Annotate the assembler output with a comment indicating which
3938 pattern and alternative was used. The length of each instruction is
3943 Dump the RTL in the assembler output as a comment before each instruction.
3944 Also turns on @option{-dp} annotation.
3948 For each of the other indicated dump files (either with @option{-d} or
3949 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3950 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3954 Just generate RTL for a function instead of compiling it. Usually used
3955 with @samp{r} (@option{-fdump-rtl-expand}).
3959 Dump debugging information during parsing, to standard error.
3962 @item -fdump-unnumbered
3963 @opindex fdump-unnumbered
3964 When doing debugging dumps (see @option{-d} option above), suppress instruction
3965 numbers and line number note output. This makes it more feasible to
3966 use diff on debugging dumps for compiler invocations with different
3967 options, in particular with and without @option{-g}.
3969 @item -fdump-translation-unit @r{(C++ only)}
3970 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
3971 @opindex fdump-translation-unit
3972 Dump a representation of the tree structure for the entire translation
3973 unit to a file. The file name is made by appending @file{.tu} to the
3974 source file name. If the @samp{-@var{options}} form is used, @var{options}
3975 controls the details of the dump as described for the
3976 @option{-fdump-tree} options.
3978 @item -fdump-class-hierarchy @r{(C++ only)}
3979 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3980 @opindex fdump-class-hierarchy
3981 Dump a representation of each class's hierarchy and virtual function
3982 table layout to a file. The file name is made by appending @file{.class}
3983 to the source file name. If the @samp{-@var{options}} form is used,
3984 @var{options} controls the details of the dump as described for the
3985 @option{-fdump-tree} options.
3987 @item -fdump-ipa-@var{switch}
3989 Control the dumping at various stages of inter-procedural analysis
3990 language tree to a file. The file name is generated by appending a switch
3991 specific suffix to the source file name. The following dumps are possible:
3995 Enables all inter-procedural analysis dumps; currently the only produced
3996 dump is the @samp{cgraph} dump.
3999 Dumps information about call-graph optimization, unused function removal,
4000 and inlining decisions.
4003 @item -fdump-tree-@var{switch}
4004 @itemx -fdump-tree-@var{switch}-@var{options}
4006 Control the dumping at various stages of processing the intermediate
4007 language tree to a file. The file name is generated by appending a switch
4008 specific suffix to the source file name. If the @samp{-@var{options}}
4009 form is used, @var{options} is a list of @samp{-} separated options that
4010 control the details of the dump. Not all options are applicable to all
4011 dumps, those which are not meaningful will be ignored. The following
4012 options are available
4016 Print the address of each node. Usually this is not meaningful as it
4017 changes according to the environment and source file. Its primary use
4018 is for tying up a dump file with a debug environment.
4020 Inhibit dumping of members of a scope or body of a function merely
4021 because that scope has been reached. Only dump such items when they
4022 are directly reachable by some other path. When dumping pretty-printed
4023 trees, this option inhibits dumping the bodies of control structures.
4025 Print a raw representation of the tree. By default, trees are
4026 pretty-printed into a C-like representation.
4028 Enable more detailed dumps (not honored by every dump option).
4030 Enable dumping various statistics about the pass (not honored by every dump
4033 Enable showing basic block boundaries (disabled in raw dumps).
4035 Enable showing virtual operands for every statement.
4037 Enable showing line numbers for statements.
4039 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4041 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4044 The following tree dumps are possible:
4048 Dump before any tree based optimization, to @file{@var{file}.original}.
4051 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4054 Dump after function inlining, to @file{@var{file}.inlined}.
4057 @opindex fdump-tree-gimple
4058 Dump each function before and after the gimplification pass to a file. The
4059 file name is made by appending @file{.gimple} to the source file name.
4062 @opindex fdump-tree-cfg
4063 Dump the control flow graph of each function to a file. The file name is
4064 made by appending @file{.cfg} to the source file name.
4067 @opindex fdump-tree-vcg
4068 Dump the control flow graph of each function to a file in VCG format. The
4069 file name is made by appending @file{.vcg} to the source file name. Note
4070 that if the file contains more than one function, the generated file cannot
4071 be used directly by VCG@. You will need to cut and paste each function's
4072 graph into its own separate file first.
4075 @opindex fdump-tree-ch
4076 Dump each function after copying loop headers. The file name is made by
4077 appending @file{.ch} to the source file name.
4080 @opindex fdump-tree-ssa
4081 Dump SSA related information to a file. The file name is made by appending
4082 @file{.ssa} to the source file name.
4085 @opindex fdump-tree-salias
4086 Dump structure aliasing variable information to a file. This file name
4087 is made by appending @file{.salias} to the source file name.
4090 @opindex fdump-tree-alias
4091 Dump aliasing information for each function. The file name is made by
4092 appending @file{.alias} to the source file name.
4095 @opindex fdump-tree-ccp
4096 Dump each function after CCP@. The file name is made by appending
4097 @file{.ccp} to the source file name.
4100 @opindex fdump-tree-storeccp
4101 Dump each function after STORE-CCP. The file name is made by appending
4102 @file{.storeccp} to the source file name.
4105 @opindex fdump-tree-pre
4106 Dump trees after partial redundancy elimination. The file name is made
4107 by appending @file{.pre} to the source file name.
4110 @opindex fdump-tree-fre
4111 Dump trees after full redundancy elimination. The file name is made
4112 by appending @file{.fre} to the source file name.
4115 @opindex fdump-tree-copyprop
4116 Dump trees after copy propagation. The file name is made
4117 by appending @file{.copyprop} to the source file name.
4119 @item store_copyprop
4120 @opindex fdump-tree-store_copyprop
4121 Dump trees after store copy-propagation. The file name is made
4122 by appending @file{.store_copyprop} to the source file name.
4125 @opindex fdump-tree-dce
4126 Dump each function after dead code elimination. The file name is made by
4127 appending @file{.dce} to the source file name.
4130 @opindex fdump-tree-mudflap
4131 Dump each function after adding mudflap instrumentation. The file name is
4132 made by appending @file{.mudflap} to the source file name.
4135 @opindex fdump-tree-sra
4136 Dump each function after performing scalar replacement of aggregates. The
4137 file name is made by appending @file{.sra} to the source file name.
4140 @opindex fdump-tree-sink
4141 Dump each function after performing code sinking. The file name is made
4142 by appending @file{.sink} to the source file name.
4145 @opindex fdump-tree-dom
4146 Dump each function after applying dominator tree optimizations. The file
4147 name is made by appending @file{.dom} to the source file name.
4150 @opindex fdump-tree-dse
4151 Dump each function after applying dead store elimination. The file
4152 name is made by appending @file{.dse} to the source file name.
4155 @opindex fdump-tree-phiopt
4156 Dump each function after optimizing PHI nodes into straightline code. The file
4157 name is made by appending @file{.phiopt} to the source file name.
4160 @opindex fdump-tree-forwprop
4161 Dump each function after forward propagating single use variables. The file
4162 name is made by appending @file{.forwprop} to the source file name.
4165 @opindex fdump-tree-copyrename
4166 Dump each function after applying the copy rename optimization. The file
4167 name is made by appending @file{.copyrename} to the source file name.
4170 @opindex fdump-tree-nrv
4171 Dump each function after applying the named return value optimization on
4172 generic trees. The file name is made by appending @file{.nrv} to the source
4176 @opindex fdump-tree-vect
4177 Dump each function after applying vectorization of loops. The file name is
4178 made by appending @file{.vect} to the source file name.
4181 @opindex fdump-tree-vrp
4182 Dump each function after Value Range Propagation (VRP). The file name
4183 is made by appending @file{.vrp} to the source file name.
4186 @opindex fdump-tree-all
4187 Enable all the available tree dumps with the flags provided in this option.
4190 @item -ftree-vectorizer-verbose=@var{n}
4191 @opindex ftree-vectorizer-verbose
4192 This option controls the amount of debugging output the vectorizer prints.
4193 This information is written to standard error, unless @option{-fdump-tree-all}
4194 or @option{-fdump-tree-vect} is specified, in which case it is output to the
4195 usual dump listing file, @file{.vect}.
4197 @item -frandom-seed=@var{string}
4198 @opindex frandom-string
4199 This option provides a seed that GCC uses when it would otherwise use
4200 random numbers. It is used to generate certain symbol names
4201 that have to be different in every compiled file. It is also used to
4202 place unique stamps in coverage data files and the object files that
4203 produce them. You can use the @option{-frandom-seed} option to produce
4204 reproducibly identical object files.
4206 The @var{string} should be different for every file you compile.
4208 @item -fsched-verbose=@var{n}
4209 @opindex fsched-verbose
4210 On targets that use instruction scheduling, this option controls the
4211 amount of debugging output the scheduler prints. This information is
4212 written to standard error, unless @option{-dS} or @option{-dR} is
4213 specified, in which case it is output to the usual dump
4214 listing file, @file{.sched} or @file{.sched2} respectively. However
4215 for @var{n} greater than nine, the output is always printed to standard
4218 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4219 same information as @option{-dRS}. For @var{n} greater than one, it
4220 also output basic block probabilities, detailed ready list information
4221 and unit/insn info. For @var{n} greater than two, it includes RTL
4222 at abort point, control-flow and regions info. And for @var{n} over
4223 four, @option{-fsched-verbose} also includes dependence info.
4227 Store the usual ``temporary'' intermediate files permanently; place them
4228 in the current directory and name them based on the source file. Thus,
4229 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4230 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4231 preprocessed @file{foo.i} output file even though the compiler now
4232 normally uses an integrated preprocessor.
4234 When used in combination with the @option{-x} command line option,
4235 @option{-save-temps} is sensible enough to avoid over writing an
4236 input source file with the same extension as an intermediate file.
4237 The corresponding intermediate file may be obtained by renaming the
4238 source file before using @option{-save-temps}.
4242 Report the CPU time taken by each subprocess in the compilation
4243 sequence. For C source files, this is the compiler proper and assembler
4244 (plus the linker if linking is done). The output looks like this:
4251 The first number on each line is the ``user time'', that is time spent
4252 executing the program itself. The second number is ``system time'',
4253 time spent executing operating system routines on behalf of the program.
4254 Both numbers are in seconds.
4256 @item -fvar-tracking
4257 @opindex fvar-tracking
4258 Run variable tracking pass. It computes where variables are stored at each
4259 position in code. Better debugging information is then generated
4260 (if the debugging information format supports this information).
4262 It is enabled by default when compiling with optimization (@option{-Os},
4263 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4264 the debug info format supports it.
4266 @item -print-file-name=@var{library}
4267 @opindex print-file-name
4268 Print the full absolute name of the library file @var{library} that
4269 would be used when linking---and don't do anything else. With this
4270 option, GCC does not compile or link anything; it just prints the
4273 @item -print-multi-directory
4274 @opindex print-multi-directory
4275 Print the directory name corresponding to the multilib selected by any
4276 other switches present in the command line. This directory is supposed
4277 to exist in @env{GCC_EXEC_PREFIX}.
4279 @item -print-multi-lib
4280 @opindex print-multi-lib
4281 Print the mapping from multilib directory names to compiler switches
4282 that enable them. The directory name is separated from the switches by
4283 @samp{;}, and each switch starts with an @samp{@@} instead of the
4284 @samp{-}, without spaces between multiple switches. This is supposed to
4285 ease shell-processing.
4287 @item -print-prog-name=@var{program}
4288 @opindex print-prog-name
4289 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4291 @item -print-libgcc-file-name
4292 @opindex print-libgcc-file-name
4293 Same as @option{-print-file-name=libgcc.a}.
4295 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4296 but you do want to link with @file{libgcc.a}. You can do
4299 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4302 @item -print-search-dirs
4303 @opindex print-search-dirs
4304 Print the name of the configured installation directory and a list of
4305 program and library directories @command{gcc} will search---and don't do anything else.
4307 This is useful when @command{gcc} prints the error message
4308 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4309 To resolve this you either need to put @file{cpp0} and the other compiler
4310 components where @command{gcc} expects to find them, or you can set the environment
4311 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4312 Don't forget the trailing @samp{/}.
4313 @xref{Environment Variables}.
4316 @opindex dumpmachine
4317 Print the compiler's target machine (for example,
4318 @samp{i686-pc-linux-gnu})---and don't do anything else.
4321 @opindex dumpversion
4322 Print the compiler version (for example, @samp{3.0})---and don't do
4327 Print the compiler's built-in specs---and don't do anything else. (This
4328 is used when GCC itself is being built.) @xref{Spec Files}.
4330 @item -feliminate-unused-debug-types
4331 @opindex feliminate-unused-debug-types
4332 Normally, when producing DWARF2 output, GCC will emit debugging
4333 information for all types declared in a compilation
4334 unit, regardless of whether or not they are actually used
4335 in that compilation unit. Sometimes this is useful, such as
4336 if, in the debugger, you want to cast a value to a type that is
4337 not actually used in your program (but is declared). More often,
4338 however, this results in a significant amount of wasted space.
4339 With this option, GCC will avoid producing debug symbol output
4340 for types that are nowhere used in the source file being compiled.
4343 @node Optimize Options
4344 @section Options That Control Optimization
4345 @cindex optimize options
4346 @cindex options, optimization
4348 These options control various sorts of optimizations.
4350 Without any optimization option, the compiler's goal is to reduce the
4351 cost of compilation and to make debugging produce the expected
4352 results. Statements are independent: if you stop the program with a
4353 breakpoint between statements, you can then assign a new value to any
4354 variable or change the program counter to any other statement in the
4355 function and get exactly the results you would expect from the source
4358 Turning on optimization flags makes the compiler attempt to improve
4359 the performance and/or code size at the expense of compilation time
4360 and possibly the ability to debug the program.
4362 The compiler performs optimization based on the knowledge it has of
4363 the program. Optimization levels @option{-O2} and above, in
4364 particular, enable @emph{unit-at-a-time} mode, which allows the
4365 compiler to consider information gained from later functions in
4366 the file when compiling a function. Compiling multiple files at
4367 once to a single output file in @emph{unit-at-a-time} mode allows
4368 the compiler to use information gained from all of the files when
4369 compiling each of them.
4371 Not all optimizations are controlled directly by a flag. Only
4372 optimizations that have a flag are listed.
4379 Optimize. Optimizing compilation takes somewhat more time, and a lot
4380 more memory for a large function.
4382 With @option{-O}, the compiler tries to reduce code size and execution
4383 time, without performing any optimizations that take a great deal of
4386 @option{-O} turns on the following optimization flags:
4387 @gccoptlist{-fdefer-pop @gol
4388 -fdelayed-branch @gol
4389 -fguess-branch-probability @gol
4390 -fcprop-registers @gol
4391 -floop-optimize @gol
4392 -fif-conversion @gol
4393 -fif-conversion2 @gol
4396 -ftree-dominator-opts @gol
4401 -ftree-copyrename @gol
4406 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4407 where doing so does not interfere with debugging.
4411 Optimize even more. GCC performs nearly all supported optimizations
4412 that do not involve a space-speed tradeoff. The compiler does not
4413 perform loop unrolling or function inlining when you specify @option{-O2}.
4414 As compared to @option{-O}, this option increases both compilation time
4415 and the performance of the generated code.
4417 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4418 also turns on the following optimization flags:
4419 @gccoptlist{-fthread-jumps @gol
4421 -foptimize-sibling-calls @gol
4422 -fcse-follow-jumps -fcse-skip-blocks @gol
4423 -fgcse -fgcse-lm @gol
4424 -fexpensive-optimizations @gol
4425 -fstrength-reduce @gol
4426 -frerun-cse-after-loop -frerun-loop-opt @gol
4429 -fschedule-insns -fschedule-insns2 @gol
4430 -fsched-interblock -fsched-spec @gol
4432 -fstrict-aliasing @gol
4433 -fdelete-null-pointer-checks @gol
4434 -freorder-blocks -freorder-functions @gol
4435 -funit-at-a-time @gol
4436 -falign-functions -falign-jumps @gol
4437 -falign-loops -falign-labels @gol
4441 Please note the warning under @option{-fgcse} about
4442 invoking @option{-O2} on programs that use computed gotos.
4446 Optimize yet more. @option{-O3} turns on all optimizations specified by
4447 @option{-O2} and also turns on the @option{-finline-functions},
4448 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4452 Do not optimize. This is the default.
4456 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4457 do not typically increase code size. It also performs further
4458 optimizations designed to reduce code size.
4460 @option{-Os} disables the following optimization flags:
4461 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4462 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4463 -fprefetch-loop-arrays -ftree-vect-loop-version}
4465 If you use multiple @option{-O} options, with or without level numbers,
4466 the last such option is the one that is effective.
4469 Options of the form @option{-f@var{flag}} specify machine-independent
4470 flags. Most flags have both positive and negative forms; the negative
4471 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4472 below, only one of the forms is listed---the one you typically will
4473 use. You can figure out the other form by either removing @samp{no-}
4476 The following options control specific optimizations. They are either
4477 activated by @option{-O} options or are related to ones that are. You
4478 can use the following flags in the rare cases when ``fine-tuning'' of
4479 optimizations to be performed is desired.
4482 @item -fno-default-inline
4483 @opindex fno-default-inline
4484 Do not make member functions inline by default merely because they are
4485 defined inside the class scope (C++ only). Otherwise, when you specify
4486 @w{@option{-O}}, member functions defined inside class scope are compiled
4487 inline by default; i.e., you don't need to add @samp{inline} in front of
4488 the member function name.
4490 @item -fno-defer-pop
4491 @opindex fno-defer-pop
4492 Always pop the arguments to each function call as soon as that function
4493 returns. For machines which must pop arguments after a function call,
4494 the compiler normally lets arguments accumulate on the stack for several
4495 function calls and pops them all at once.
4497 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4501 Force memory operands to be copied into registers before doing
4502 arithmetic on them. This produces better code by making all memory
4503 references potential common subexpressions. When they are not common
4504 subexpressions, instruction combination should eliminate the separate
4505 register-load. This option is now a nop and will be removed in 4.2.
4508 @opindex fforce-addr
4509 Force memory address constants to be copied into registers before
4510 doing arithmetic on them.
4512 @item -fomit-frame-pointer
4513 @opindex fomit-frame-pointer
4514 Don't keep the frame pointer in a register for functions that
4515 don't need one. This avoids the instructions to save, set up and
4516 restore frame pointers; it also makes an extra register available
4517 in many functions. @strong{It also makes debugging impossible on
4520 On some machines, such as the VAX, this flag has no effect, because
4521 the standard calling sequence automatically handles the frame pointer
4522 and nothing is saved by pretending it doesn't exist. The
4523 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4524 whether a target machine supports this flag. @xref{Registers,,Register
4525 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4527 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4529 @item -foptimize-sibling-calls
4530 @opindex foptimize-sibling-calls
4531 Optimize sibling and tail recursive calls.
4533 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4537 Don't pay attention to the @code{inline} keyword. Normally this option
4538 is used to keep the compiler from expanding any functions inline.
4539 Note that if you are not optimizing, no functions can be expanded inline.
4541 @item -finline-functions
4542 @opindex finline-functions
4543 Integrate all simple functions into their callers. The compiler
4544 heuristically decides which functions are simple enough to be worth
4545 integrating in this way.
4547 If all calls to a given function are integrated, and the function is
4548 declared @code{static}, then the function is normally not output as
4549 assembler code in its own right.
4551 Enabled at level @option{-O3}.
4553 @item -finline-functions-called-once
4554 @opindex finline-functions-called-once
4555 Consider all @code{static} functions called once for inlining into their
4556 caller even if they are not marked @code{inline}. If a call to a given
4557 function is integrated, then the function is not output as assembler code
4560 Enabled if @option{-funit-at-a-time} is enabled.
4562 @item -fearly-inlining
4563 @opindex fearly-inlining
4564 Inline functions marked by @code{always_inline} and functions whose body seems
4565 smaller than the function call overhead early before doing
4566 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4567 makes profiling significantly cheaper and usually inlining faster on programs
4568 having large chains of nested wrapper functions.
4572 @item -finline-limit=@var{n}
4573 @opindex finline-limit
4574 By default, GCC limits the size of functions that can be inlined. This flag
4575 allows the control of this limit for functions that are explicitly marked as
4576 inline (i.e., marked with the inline keyword or defined within the class
4577 definition in c++). @var{n} is the size of functions that can be inlined in
4578 number of pseudo instructions (not counting parameter handling). The default
4579 value of @var{n} is 600.
4580 Increasing this value can result in more inlined code at
4581 the cost of compilation time and memory consumption. Decreasing usually makes
4582 the compilation faster and less code will be inlined (which presumably
4583 means slower programs). This option is particularly useful for programs that
4584 use inlining heavily such as those based on recursive templates with C++.
4586 Inlining is actually controlled by a number of parameters, which may be
4587 specified individually by using @option{--param @var{name}=@var{value}}.
4588 The @option{-finline-limit=@var{n}} option sets some of these parameters
4592 @item max-inline-insns-single
4593 is set to @var{n}/2.
4594 @item max-inline-insns-auto
4595 is set to @var{n}/2.
4596 @item min-inline-insns
4597 is set to 130 or @var{n}/4, whichever is smaller.
4598 @item max-inline-insns-rtl
4602 See below for a documentation of the individual
4603 parameters controlling inlining.
4605 @emph{Note:} pseudo instruction represents, in this particular context, an
4606 abstract measurement of function's size. In no way does it represent a count
4607 of assembly instructions and as such its exact meaning might change from one
4608 release to an another.
4610 @item -fkeep-inline-functions
4611 @opindex fkeep-inline-functions
4612 In C, emit @code{static} functions that are declared @code{inline}
4613 into the object file, even if the function has been inlined into all
4614 of its callers. This switch does not affect functions using the
4615 @code{extern inline} extension in GNU C@. In C++, emit any and all
4616 inline functions into the object file.
4618 @item -fkeep-static-consts
4619 @opindex fkeep-static-consts
4620 Emit variables declared @code{static const} when optimization isn't turned
4621 on, even if the variables aren't referenced.
4623 GCC enables this option by default. If you want to force the compiler to
4624 check if the variable was referenced, regardless of whether or not
4625 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4627 @item -fmerge-constants
4628 Attempt to merge identical constants (string constants and floating point
4629 constants) across compilation units.
4631 This option is the default for optimized compilation if the assembler and
4632 linker support it. Use @option{-fno-merge-constants} to inhibit this
4635 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4637 @item -fmerge-all-constants
4638 Attempt to merge identical constants and identical variables.
4640 This option implies @option{-fmerge-constants}. In addition to
4641 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4642 arrays or initialized constant variables with integral or floating point
4643 types. Languages like C or C++ require each non-automatic variable to
4644 have distinct location, so using this option will result in non-conforming
4647 @item -fmodulo-sched
4648 @opindex fmodulo-sched
4649 Perform swing modulo scheduling immediately before the first scheduling
4650 pass. This pass looks at innermost loops and reorders their
4651 instructions by overlapping different iterations.
4653 @item -fno-branch-count-reg
4654 @opindex fno-branch-count-reg
4655 Do not use ``decrement and branch'' instructions on a count register,
4656 but instead generate a sequence of instructions that decrement a
4657 register, compare it against zero, then branch based upon the result.
4658 This option is only meaningful on architectures that support such
4659 instructions, which include x86, PowerPC, IA-64 and S/390.
4661 The default is @option{-fbranch-count-reg}, enabled when
4662 @option{-fstrength-reduce} is enabled.
4664 @item -fno-function-cse
4665 @opindex fno-function-cse
4666 Do not put function addresses in registers; make each instruction that
4667 calls a constant function contain the function's address explicitly.
4669 This option results in less efficient code, but some strange hacks
4670 that alter the assembler output may be confused by the optimizations
4671 performed when this option is not used.
4673 The default is @option{-ffunction-cse}
4675 @item -fno-zero-initialized-in-bss
4676 @opindex fno-zero-initialized-in-bss
4677 If the target supports a BSS section, GCC by default puts variables that
4678 are initialized to zero into BSS@. This can save space in the resulting
4681 This option turns off this behavior because some programs explicitly
4682 rely on variables going to the data section. E.g., so that the
4683 resulting executable can find the beginning of that section and/or make
4684 assumptions based on that.
4686 The default is @option{-fzero-initialized-in-bss}.
4688 @item -fbounds-check
4689 @opindex fbounds-check
4690 For front-ends that support it, generate additional code to check that
4691 indices used to access arrays are within the declared range. This is
4692 currently only supported by the Java and Fortran front-ends, where
4693 this option defaults to true and false respectively.
4695 @item -fmudflap -fmudflapth -fmudflapir
4699 @cindex bounds checking
4701 For front-ends that support it (C and C++), instrument all risky
4702 pointer/array dereferencing operations, some standard library
4703 string/heap functions, and some other associated constructs with
4704 range/validity tests. Modules so instrumented should be immune to
4705 buffer overflows, invalid heap use, and some other classes of C/C++
4706 programming errors. The instrumentation relies on a separate runtime
4707 library (@file{libmudflap}), which will be linked into a program if
4708 @option{-fmudflap} is given at link time. Run-time behavior of the
4709 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4710 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4713 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4714 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4715 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4716 instrumentation should ignore pointer reads. This produces less
4717 instrumentation (and therefore faster execution) and still provides
4718 some protection against outright memory corrupting writes, but allows
4719 erroneously read data to propagate within a program.
4723 @cindex openmp parallel
4724 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
4725 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
4726 compiler generates parallel code according to the OpenMP Application
4727 Program Interface v2.5. To generate the final exectuable, the runtime
4728 library @code{libgomp} must be linked in using @option{-lgomp}.
4730 @item -fstrength-reduce
4731 @opindex fstrength-reduce
4732 Perform the optimizations of loop strength reduction and
4733 elimination of iteration variables.
4735 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4737 @item -fthread-jumps
4738 @opindex fthread-jumps
4739 Perform optimizations where we check to see if a jump branches to a
4740 location where another comparison subsumed by the first is found. If
4741 so, the first branch is redirected to either the destination of the
4742 second branch or a point immediately following it, depending on whether
4743 the condition is known to be true or false.
4745 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4747 @item -fcse-follow-jumps
4748 @opindex fcse-follow-jumps
4749 In common subexpression elimination, scan through jump instructions
4750 when the target of the jump is not reached by any other path. For
4751 example, when CSE encounters an @code{if} statement with an
4752 @code{else} clause, CSE will follow the jump when the condition
4755 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4757 @item -fcse-skip-blocks
4758 @opindex fcse-skip-blocks
4759 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4760 follow jumps which conditionally skip over blocks. When CSE
4761 encounters a simple @code{if} statement with no else clause,
4762 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4763 body of the @code{if}.
4765 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4767 @item -frerun-cse-after-loop
4768 @opindex frerun-cse-after-loop
4769 Re-run common subexpression elimination after loop optimizations has been
4772 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4774 @item -frerun-loop-opt
4775 @opindex frerun-loop-opt
4776 Run the loop optimizer twice.
4778 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4782 Perform a global common subexpression elimination pass.
4783 This pass also performs global constant and copy propagation.
4785 @emph{Note:} When compiling a program using computed gotos, a GCC
4786 extension, you may get better runtime performance if you disable
4787 the global common subexpression elimination pass by adding
4788 @option{-fno-gcse} to the command line.
4790 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4794 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4795 attempt to move loads which are only killed by stores into themselves. This
4796 allows a loop containing a load/store sequence to be changed to a load outside
4797 the loop, and a copy/store within the loop.
4799 Enabled by default when gcse is enabled.
4803 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4804 global common subexpression elimination. This pass will attempt to move
4805 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4806 loops containing a load/store sequence can be changed to a load before
4807 the loop and a store after the loop.
4809 Not enabled at any optimization level.
4813 When @option{-fgcse-las} is enabled, the global common subexpression
4814 elimination pass eliminates redundant loads that come after stores to the
4815 same memory location (both partial and full redundancies).
4817 Not enabled at any optimization level.
4819 @item -fgcse-after-reload
4820 @opindex fgcse-after-reload
4821 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4822 pass is performed after reload. The purpose of this pass is to cleanup
4825 @item -floop-optimize
4826 @opindex floop-optimize
4827 Perform loop optimizations: move constant expressions out of loops, simplify
4828 exit test conditions and optionally do strength-reduction as well.
4830 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4832 @item -floop-optimize2
4833 @opindex floop-optimize2
4834 Perform loop optimizations using the new loop optimizer. The optimizations
4835 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4838 @item -funsafe-loop-optimizations
4839 @opindex funsafe-loop-optimizations
4840 If given, the loop optimizer will assume that loop indices do not
4841 overflow, and that the loops with nontrivial exit condition are not
4842 infinite. This enables a wider range of loop optimizations even if
4843 the loop optimizer itself cannot prove that these assumptions are valid.
4844 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4845 if it finds this kind of loop.
4847 @item -fcrossjumping
4848 @opindex crossjumping
4849 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4850 resulting code may or may not perform better than without cross-jumping.
4852 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4854 @item -fif-conversion
4855 @opindex if-conversion
4856 Attempt to transform conditional jumps into branch-less equivalents. This
4857 include use of conditional moves, min, max, set flags and abs instructions, and
4858 some tricks doable by standard arithmetics. The use of conditional execution
4859 on chips where it is available is controlled by @code{if-conversion2}.
4861 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4863 @item -fif-conversion2
4864 @opindex if-conversion2
4865 Use conditional execution (where available) to transform conditional jumps into
4866 branch-less equivalents.
4868 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4870 @item -fdelete-null-pointer-checks
4871 @opindex fdelete-null-pointer-checks
4872 Use global dataflow analysis to identify and eliminate useless checks
4873 for null pointers. The compiler assumes that dereferencing a null
4874 pointer would have halted the program. If a pointer is checked after
4875 it has already been dereferenced, it cannot be null.
4877 In some environments, this assumption is not true, and programs can
4878 safely dereference null pointers. Use
4879 @option{-fno-delete-null-pointer-checks} to disable this optimization
4880 for programs which depend on that behavior.
4882 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4884 @item -fexpensive-optimizations
4885 @opindex fexpensive-optimizations
4886 Perform a number of minor optimizations that are relatively expensive.
4888 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4890 @item -foptimize-register-move
4892 @opindex foptimize-register-move
4894 Attempt to reassign register numbers in move instructions and as
4895 operands of other simple instructions in order to maximize the amount of
4896 register tying. This is especially helpful on machines with two-operand
4899 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4902 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4904 @item -fdelayed-branch
4905 @opindex fdelayed-branch
4906 If supported for the target machine, attempt to reorder instructions
4907 to exploit instruction slots available after delayed branch
4910 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4912 @item -fschedule-insns
4913 @opindex fschedule-insns
4914 If supported for the target machine, attempt to reorder instructions to
4915 eliminate execution stalls due to required data being unavailable. This
4916 helps machines that have slow floating point or memory load instructions
4917 by allowing other instructions to be issued until the result of the load
4918 or floating point instruction is required.
4920 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4922 @item -fschedule-insns2
4923 @opindex fschedule-insns2
4924 Similar to @option{-fschedule-insns}, but requests an additional pass of
4925 instruction scheduling after register allocation has been done. This is
4926 especially useful on machines with a relatively small number of
4927 registers and where memory load instructions take more than one cycle.
4929 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4931 @item -fno-sched-interblock
4932 @opindex fno-sched-interblock
4933 Don't schedule instructions across basic blocks. This is normally
4934 enabled by default when scheduling before register allocation, i.e.@:
4935 with @option{-fschedule-insns} or at @option{-O2} or higher.
4937 @item -fno-sched-spec
4938 @opindex fno-sched-spec
4939 Don't allow speculative motion of non-load instructions. This is normally
4940 enabled by default when scheduling before register allocation, i.e.@:
4941 with @option{-fschedule-insns} or at @option{-O2} or higher.
4943 @item -fsched-spec-load
4944 @opindex fsched-spec-load
4945 Allow speculative motion of some load instructions. This only makes
4946 sense when scheduling before register allocation, i.e.@: with
4947 @option{-fschedule-insns} or at @option{-O2} or higher.
4949 @item -fsched-spec-load-dangerous
4950 @opindex fsched-spec-load-dangerous
4951 Allow speculative motion of more load instructions. This only makes
4952 sense when scheduling before register allocation, i.e.@: with
4953 @option{-fschedule-insns} or at @option{-O2} or higher.
4955 @item -fsched-stalled-insns=@var{n}
4956 @opindex fsched-stalled-insns
4957 Define how many insns (if any) can be moved prematurely from the queue
4958 of stalled insns into the ready list, during the second scheduling pass.
4960 @item -fsched-stalled-insns-dep=@var{n}
4961 @opindex fsched-stalled-insns-dep
4962 Define how many insn groups (cycles) will be examined for a dependency
4963 on a stalled insn that is candidate for premature removal from the queue
4964 of stalled insns. Has an effect only during the second scheduling pass,
4965 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4967 @item -fsched2-use-superblocks
4968 @opindex fsched2-use-superblocks
4969 When scheduling after register allocation, do use superblock scheduling
4970 algorithm. Superblock scheduling allows motion across basic block boundaries
4971 resulting on faster schedules. This option is experimental, as not all machine
4972 descriptions used by GCC model the CPU closely enough to avoid unreliable
4973 results from the algorithm.
4975 This only makes sense when scheduling after register allocation, i.e.@: with
4976 @option{-fschedule-insns2} or at @option{-O2} or higher.
4978 @item -fsched2-use-traces
4979 @opindex fsched2-use-traces
4980 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4981 allocation and additionally perform code duplication in order to increase the
4982 size of superblocks using tracer pass. See @option{-ftracer} for details on
4985 This mode should produce faster but significantly longer programs. Also
4986 without @option{-fbranch-probabilities} the traces constructed may not
4987 match the reality and hurt the performance. This only makes
4988 sense when scheduling after register allocation, i.e.@: with
4989 @option{-fschedule-insns2} or at @option{-O2} or higher.
4991 @item -freschedule-modulo-scheduled-loops
4992 @opindex fscheduling-in-modulo-scheduled-loops
4993 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4994 we may want to prevent the later scheduling passes from changing its schedule, we use this
4995 option to control that.
4997 @item -fcaller-saves
4998 @opindex fcaller-saves
4999 Enable values to be allocated in registers that will be clobbered by
5000 function calls, by emitting extra instructions to save and restore the
5001 registers around such calls. Such allocation is done only when it
5002 seems to result in better code than would otherwise be produced.
5004 This option is always enabled by default on certain machines, usually
5005 those which have no call-preserved registers to use instead.
5007 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5010 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5011 enabled by default at @option{-O2} and @option{-O3}.
5014 Perform Full Redundancy Elimination (FRE) on trees. The difference
5015 between FRE and PRE is that FRE only considers expressions
5016 that are computed on all paths leading to the redundant computation.
5017 This analysis faster than PRE, though it exposes fewer redundancies.
5018 This flag is enabled by default at @option{-O} and higher.
5020 @item -ftree-copy-prop
5021 Perform copy propagation on trees. This pass eliminates unnecessary
5022 copy operations. This flag is enabled by default at @option{-O} and
5025 @item -ftree-store-copy-prop
5026 Perform copy propagation of memory loads and stores. This pass
5027 eliminates unnecessary copy operations in memory references
5028 (structures, global variables, arrays, etc). This flag is enabled by
5029 default at @option{-O2} and higher.
5032 Perform structural alias analysis on trees. This flag
5033 is enabled by default at @option{-O} and higher.
5036 Perform forward store motion on trees. This flag is
5037 enabled by default at @option{-O} and higher.
5040 Perform sparse conditional constant propagation (CCP) on trees. This
5041 pass only operates on local scalar variables and is enabled by default
5042 at @option{-O} and higher.
5044 @item -ftree-store-ccp
5045 Perform sparse conditional constant propagation (CCP) on trees. This
5046 pass operates on both local scalar variables and memory stores and
5047 loads (global variables, structures, arrays, etc). This flag is
5048 enabled by default at @option{-O2} and higher.
5051 Perform dead code elimination (DCE) on trees. This flag is enabled by
5052 default at @option{-O} and higher.
5054 @item -ftree-dominator-opts
5055 Perform a variety of simple scalar cleanups (constant/copy
5056 propagation, redundancy elimination, range propagation and expression
5057 simplification) based on a dominator tree traversal. This also
5058 performs jump threading (to reduce jumps to jumps). This flag is
5059 enabled by default at @option{-O} and higher.
5062 Perform loop header copying on trees. This is beneficial since it increases
5063 effectiveness of code motion optimizations. It also saves one jump. This flag
5064 is enabled by default at @option{-O} and higher. It is not enabled
5065 for @option{-Os}, since it usually increases code size.
5067 @item -ftree-loop-optimize
5068 Perform loop optimizations on trees. This flag is enabled by default
5069 at @option{-O} and higher.
5071 @item -ftree-loop-linear
5072 Perform linear loop transformations on tree. This flag can improve cache
5073 performance and allow further loop optimizations to take place.
5075 @item -ftree-loop-im
5076 Perform loop invariant motion on trees. This pass moves only invariants that
5077 would be hard to handle at RTL level (function calls, operations that expand to
5078 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5079 operands of conditions that are invariant out of the loop, so that we can use
5080 just trivial invariantness analysis in loop unswitching. The pass also includes
5083 @item -ftree-loop-ivcanon
5084 Create a canonical counter for number of iterations in the loop for that
5085 determining number of iterations requires complicated analysis. Later
5086 optimizations then may determine the number easily. Useful especially
5087 in connection with unrolling.
5090 Perform induction variable optimizations (strength reduction, induction
5091 variable merging and induction variable elimination) on trees.
5094 Perform scalar replacement of aggregates. This pass replaces structure
5095 references with scalars to prevent committing structures to memory too
5096 early. This flag is enabled by default at @option{-O} and higher.
5098 @item -ftree-copyrename
5099 Perform copy renaming on trees. This pass attempts to rename compiler
5100 temporaries to other variables at copy locations, usually resulting in
5101 variable names which more closely resemble the original variables. This flag
5102 is enabled by default at @option{-O} and higher.
5105 Perform temporary expression replacement during the SSA->normal phase. Single
5106 use/single def temporaries are replaced at their use location with their
5107 defining expression. This results in non-GIMPLE code, but gives the expanders
5108 much more complex trees to work on resulting in better RTL generation. This is
5109 enabled by default at @option{-O} and higher.
5112 Perform live range splitting during the SSA->normal phase. Distinct live
5113 ranges of a variable are split into unique variables, allowing for better
5114 optimization later. This is enabled by default at @option{-O} and higher.
5116 @item -ftree-vectorize
5117 Perform loop vectorization on trees.
5119 @item -ftree-vect-loop-version
5120 @opindex ftree-vect-loop-version
5121 Perform loop versioning when doing loop vectorization on trees. When a loop
5122 appears to be vectorizable except that data alignment or data dependence cannot
5123 be determined at compile time then vectorized and non-vectorized versions of
5124 the loop are generated along with runtime checks for alignment or dependence
5125 to control which version is executed. This option is enabled by default
5126 except at level @option{-Os} where it is disabled.
5129 Perform Value Range Propagation on trees. This is similar to the
5130 constant propagation pass, but instead of values, ranges of values are
5131 propagated. This allows the optimizers to remove unnecessary range
5132 checks like array bound checks and null pointer checks. This is
5133 enabled by default at @option{-O2} and higher. Null pointer check
5134 elimination is only done if @option{-fdelete-null-pointer-checks} is
5139 Perform tail duplication to enlarge superblock size. This transformation
5140 simplifies the control flow of the function allowing other optimizations to do
5143 @item -funroll-loops
5144 @opindex funroll-loops
5145 Unroll loops whose number of iterations can be determined at compile
5146 time or upon entry to the loop. @option{-funroll-loops} implies both
5147 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
5148 option makes code larger, and may or may not make it run faster.
5150 @item -funroll-all-loops
5151 @opindex funroll-all-loops
5152 Unroll all loops, even if their number of iterations is uncertain when
5153 the loop is entered. This usually makes programs run more slowly.
5154 @option{-funroll-all-loops} implies the same options as
5155 @option{-funroll-loops},
5157 @item -fsplit-ivs-in-unroller
5158 @opindex -fsplit-ivs-in-unroller
5159 Enables expressing of values of induction variables in later iterations
5160 of the unrolled loop using the value in the first iteration. This breaks
5161 long dependency chains, thus improving efficiency of the scheduling passes.
5163 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5164 same effect. However in cases the loop body is more complicated than
5165 a single basic block, this is not reliable. It also does not work at all
5166 on some of the architectures due to restrictions in the CSE pass.
5168 This optimization is enabled by default.
5170 @item -fvariable-expansion-in-unroller
5171 @opindex -fvariable-expansion-in-unroller
5172 With this option, the compiler will create multiple copies of some
5173 local variables when unrolling a loop which can result in superior code.
5175 @item -fprefetch-loop-arrays
5176 @opindex fprefetch-loop-arrays
5177 If supported by the target machine, generate instructions to prefetch
5178 memory to improve the performance of loops that access large arrays.
5180 These options may generate better or worse code; results are highly
5181 dependent on the structure of loops within the source code.
5184 @itemx -fno-peephole2
5185 @opindex fno-peephole
5186 @opindex fno-peephole2
5187 Disable any machine-specific peephole optimizations. The difference
5188 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5189 are implemented in the compiler; some targets use one, some use the
5190 other, a few use both.
5192 @option{-fpeephole} is enabled by default.
5193 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5195 @item -fno-guess-branch-probability
5196 @opindex fno-guess-branch-probability
5197 Do not guess branch probabilities using heuristics.
5199 GCC will use heuristics to guess branch probabilities if they are
5200 not provided by profiling feedback (@option{-fprofile-arcs}). These
5201 heuristics are based on the control flow graph. If some branch probabilities
5202 are specified by @samp{__builtin_expect}, then the heuristics will be
5203 used to guess branch probabilities for the rest of the control flow graph,
5204 taking the @samp{__builtin_expect} info into account. The interactions
5205 between the heuristics and @samp{__builtin_expect} can be complex, and in
5206 some cases, it may be useful to disable the heuristics so that the effects
5207 of @samp{__builtin_expect} are easier to understand.
5209 The default is @option{-fguess-branch-probability} at levels
5210 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5212 @item -freorder-blocks
5213 @opindex freorder-blocks
5214 Reorder basic blocks in the compiled function in order to reduce number of
5215 taken branches and improve code locality.
5217 Enabled at levels @option{-O2}, @option{-O3}.
5219 @item -freorder-blocks-and-partition
5220 @opindex freorder-blocks-and-partition
5221 In addition to reordering basic blocks in the compiled function, in order
5222 to reduce number of taken branches, partitions hot and cold basic blocks
5223 into separate sections of the assembly and .o files, to improve
5224 paging and cache locality performance.
5226 This optimization is automatically turned off in the presence of
5227 exception handling, for linkonce sections, for functions with a user-defined
5228 section attribute and on any architecture that does not support named
5231 @item -freorder-functions
5232 @opindex freorder-functions
5233 Reorder functions in the object file in order to
5234 improve code locality. This is implemented by using special
5235 subsections @code{.text.hot} for most frequently executed functions and
5236 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5237 the linker so object file format must support named sections and linker must
5238 place them in a reasonable way.
5240 Also profile feedback must be available in to make this option effective. See
5241 @option{-fprofile-arcs} for details.
5243 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5245 @item -fstrict-aliasing
5246 @opindex fstrict-aliasing
5247 Allows the compiler to assume the strictest aliasing rules applicable to
5248 the language being compiled. For C (and C++), this activates
5249 optimizations based on the type of expressions. In particular, an
5250 object of one type is assumed never to reside at the same address as an
5251 object of a different type, unless the types are almost the same. For
5252 example, an @code{unsigned int} can alias an @code{int}, but not a
5253 @code{void*} or a @code{double}. A character type may alias any other
5256 Pay special attention to code like this:
5269 The practice of reading from a different union member than the one most
5270 recently written to (called ``type-punning'') is common. Even with
5271 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5272 is accessed through the union type. So, the code above will work as
5273 expected. However, this code might not:
5284 Every language that wishes to perform language-specific alias analysis
5285 should define a function that computes, given an @code{tree}
5286 node, an alias set for the node. Nodes in different alias sets are not
5287 allowed to alias. For an example, see the C front-end function
5288 @code{c_get_alias_set}.
5290 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5292 @item -falign-functions
5293 @itemx -falign-functions=@var{n}
5294 @opindex falign-functions
5295 Align the start of functions to the next power-of-two greater than
5296 @var{n}, skipping up to @var{n} bytes. For instance,
5297 @option{-falign-functions=32} aligns functions to the next 32-byte
5298 boundary, but @option{-falign-functions=24} would align to the next
5299 32-byte boundary only if this can be done by skipping 23 bytes or less.
5301 @option{-fno-align-functions} and @option{-falign-functions=1} are
5302 equivalent and mean that functions will not be aligned.
5304 Some assemblers only support this flag when @var{n} is a power of two;
5305 in that case, it is rounded up.
5307 If @var{n} is not specified or is zero, use a machine-dependent default.
5309 Enabled at levels @option{-O2}, @option{-O3}.
5311 @item -falign-labels
5312 @itemx -falign-labels=@var{n}
5313 @opindex falign-labels
5314 Align all branch targets to a power-of-two boundary, skipping up to
5315 @var{n} bytes like @option{-falign-functions}. This option can easily
5316 make code slower, because it must insert dummy operations for when the
5317 branch target is reached in the usual flow of the code.
5319 @option{-fno-align-labels} and @option{-falign-labels=1} are
5320 equivalent and mean that labels will not be aligned.
5322 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5323 are greater than this value, then their values are used instead.
5325 If @var{n} is not specified or is zero, use a machine-dependent default
5326 which is very likely to be @samp{1}, meaning no alignment.
5328 Enabled at levels @option{-O2}, @option{-O3}.
5331 @itemx -falign-loops=@var{n}
5332 @opindex falign-loops
5333 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5334 like @option{-falign-functions}. The hope is that the loop will be
5335 executed many times, which will make up for any execution of the dummy
5338 @option{-fno-align-loops} and @option{-falign-loops=1} are
5339 equivalent and mean that loops will not be aligned.
5341 If @var{n} is not specified or is zero, use a machine-dependent default.
5343 Enabled at levels @option{-O2}, @option{-O3}.
5346 @itemx -falign-jumps=@var{n}
5347 @opindex falign-jumps
5348 Align branch targets to a power-of-two boundary, for branch targets
5349 where the targets can only be reached by jumping, skipping up to @var{n}
5350 bytes like @option{-falign-functions}. In this case, no dummy operations
5353 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5354 equivalent and mean that loops will not be aligned.
5356 If @var{n} is not specified or is zero, use a machine-dependent default.
5358 Enabled at levels @option{-O2}, @option{-O3}.
5360 @item -funit-at-a-time
5361 @opindex funit-at-a-time
5362 Parse the whole compilation unit before starting to produce code.
5363 This allows some extra optimizations to take place but consumes
5364 more memory (in general). There are some compatibility issues
5365 with @emph{unit-at-a-time} mode:
5368 enabling @emph{unit-at-a-time} mode may change the order
5369 in which functions, variables, and top-level @code{asm} statements
5370 are emitted, and will likely break code relying on some particular
5371 ordering. The majority of such top-level @code{asm} statements,
5372 though, can be replaced by @code{section} attributes. The
5373 @option{fno-toplevel-reorder} option may be used to keep the ordering
5374 used in the input file, at the cost of some optimizations.
5377 @emph{unit-at-a-time} mode removes unreferenced static variables
5378 and functions. This may result in undefined references
5379 when an @code{asm} statement refers directly to variables or functions
5380 that are otherwise unused. In that case either the variable/function
5381 shall be listed as an operand of the @code{asm} statement operand or,
5382 in the case of top-level @code{asm} statements the attribute @code{used}
5383 shall be used on the declaration.
5386 Static functions now can use non-standard passing conventions that
5387 may break @code{asm} statements calling functions directly. Again,
5388 attribute @code{used} will prevent this behavior.
5391 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5392 but this scheme may not be supported by future releases of GCC@.
5394 Enabled at levels @option{-O2}, @option{-O3}.
5396 @item -fno-toplevel-reorder
5397 Do not reorder top-level functions, variables, and @code{asm}
5398 statements. Output them in the same order that they appear in the
5399 input file. When this option is used, unreferenced static variables
5400 will not be removed. This option is intended to support existing code
5401 which relies on a particular ordering. For new code, it is better to
5406 Constructs webs as commonly used for register allocation purposes and assign
5407 each web individual pseudo register. This allows the register allocation pass
5408 to operate on pseudos directly, but also strengthens several other optimization
5409 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5410 however, make debugging impossible, since variables will no longer stay in a
5413 Enabled by default with @option{-funroll-loops}.
5415 @item -fwhole-program
5416 @opindex fwhole-program
5417 Assume that the current compilation unit represents whole program being
5418 compiled. All public functions and variables with the exception of @code{main}
5419 and those merged by attribute @code{externally_visible} become static functions
5420 and in a affect gets more aggressively optimized by interprocedural optimizers.
5421 While this option is equivalent to proper use of @code{static} keyword for
5422 programs consisting of single file, in combination with option
5423 @option{--combine} this flag can be used to compile most of smaller scale C
5424 programs since the functions and variables become local for the whole combined
5425 compilation unit, not for the single source file itself.
5428 @item -fno-cprop-registers
5429 @opindex fno-cprop-registers
5430 After register allocation and post-register allocation instruction splitting,
5431 we perform a copy-propagation pass to try to reduce scheduling dependencies
5432 and occasionally eliminate the copy.
5434 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5436 @item -fprofile-generate
5437 @opindex fprofile-generate
5439 Enable options usually used for instrumenting application to produce
5440 profile useful for later recompilation with profile feedback based
5441 optimization. You must use @option{-fprofile-generate} both when
5442 compiling and when linking your program.
5444 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5447 @opindex fprofile-use
5448 Enable profile feedback directed optimizations, and optimizations
5449 generally profitable only with profile feedback available.
5451 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5452 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer},
5453 @code{-fno-loop-optimize}.
5457 The following options control compiler behavior regarding floating
5458 point arithmetic. These options trade off between speed and
5459 correctness. All must be specifically enabled.
5463 @opindex ffloat-store
5464 Do not store floating point variables in registers, and inhibit other
5465 options that might change whether a floating point value is taken from a
5468 @cindex floating point precision
5469 This option prevents undesirable excess precision on machines such as
5470 the 68000 where the floating registers (of the 68881) keep more
5471 precision than a @code{double} is supposed to have. Similarly for the
5472 x86 architecture. For most programs, the excess precision does only
5473 good, but a few programs rely on the precise definition of IEEE floating
5474 point. Use @option{-ffloat-store} for such programs, after modifying
5475 them to store all pertinent intermediate computations into variables.
5479 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5480 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5481 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5482 and @option{fcx-limited-range}.
5484 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5486 This option should never be turned on by any @option{-O} option since
5487 it can result in incorrect output for programs which depend on
5488 an exact implementation of IEEE or ISO rules/specifications for
5491 @item -fno-math-errno
5492 @opindex fno-math-errno
5493 Do not set ERRNO after calling math functions that are executed
5494 with a single instruction, e.g., sqrt. A program that relies on
5495 IEEE exceptions for math error handling may want to use this flag
5496 for speed while maintaining IEEE arithmetic compatibility.
5498 This option should never be turned on by any @option{-O} option since
5499 it can result in incorrect output for programs which depend on
5500 an exact implementation of IEEE or ISO rules/specifications for
5503 The default is @option{-fmath-errno}.
5505 On Darwin systems, the math library never sets @code{errno}. There is therefore
5506 no reason for the compiler to consider the possibility that it might,
5507 and @option{-fno-math-errno} is the default.
5509 @item -funsafe-math-optimizations
5510 @opindex funsafe-math-optimizations
5511 Allow optimizations for floating-point arithmetic that (a) assume
5512 that arguments and results are valid and (b) may violate IEEE or
5513 ANSI standards. When used at link-time, it may include libraries
5514 or startup files that change the default FPU control word or other
5515 similar optimizations.
5517 This option should never be turned on by any @option{-O} option since
5518 it can result in incorrect output for programs which depend on
5519 an exact implementation of IEEE or ISO rules/specifications for
5522 The default is @option{-fno-unsafe-math-optimizations}.
5524 @item -ffinite-math-only
5525 @opindex ffinite-math-only
5526 Allow optimizations for floating-point arithmetic that assume
5527 that arguments and results are not NaNs or +-Infs.
5529 This option should never be turned on by any @option{-O} option since
5530 it can result in incorrect output for programs which depend on
5531 an exact implementation of IEEE or ISO rules/specifications.
5533 The default is @option{-fno-finite-math-only}.
5535 @item -fno-trapping-math
5536 @opindex fno-trapping-math
5537 Compile code assuming that floating-point operations cannot generate
5538 user-visible traps. These traps include division by zero, overflow,
5539 underflow, inexact result and invalid operation. This option implies
5540 @option{-fno-signaling-nans}. Setting this option may allow faster
5541 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5543 This option should never be turned on by any @option{-O} option since
5544 it can result in incorrect output for programs which depend on
5545 an exact implementation of IEEE or ISO rules/specifications for
5548 The default is @option{-ftrapping-math}.
5550 @item -frounding-math
5551 @opindex frounding-math
5552 Disable transformations and optimizations that assume default floating
5553 point rounding behavior. This is round-to-zero for all floating point
5554 to integer conversions, and round-to-nearest for all other arithmetic
5555 truncations. This option should be specified for programs that change
5556 the FP rounding mode dynamically, or that may be executed with a
5557 non-default rounding mode. This option disables constant folding of
5558 floating point expressions at compile-time (which may be affected by
5559 rounding mode) and arithmetic transformations that are unsafe in the
5560 presence of sign-dependent rounding modes.
5562 The default is @option{-fno-rounding-math}.
5564 This option is experimental and does not currently guarantee to
5565 disable all GCC optimizations that are affected by rounding mode.
5566 Future versions of GCC may provide finer control of this setting
5567 using C99's @code{FENV_ACCESS} pragma. This command line option
5568 will be used to specify the default state for @code{FENV_ACCESS}.
5570 @item -frtl-abstract-sequences
5571 @opindex frtl-abstract-sequences
5572 It is a size optimization method. This option is to find identical
5573 sequences of code, which can be turned into pseudo-procedures and
5574 then replace all occurrences with calls to the newly created
5575 subroutine. It is kind of an opposite of @option{-finline-functions}.
5576 This optimization runs at RTL level.
5578 @item -fsignaling-nans
5579 @opindex fsignaling-nans
5580 Compile code assuming that IEEE signaling NaNs may generate user-visible
5581 traps during floating-point operations. Setting this option disables
5582 optimizations that may change the number of exceptions visible with
5583 signaling NaNs. This option implies @option{-ftrapping-math}.
5585 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5588 The default is @option{-fno-signaling-nans}.
5590 This option is experimental and does not currently guarantee to
5591 disable all GCC optimizations that affect signaling NaN behavior.
5593 @item -fsingle-precision-constant
5594 @opindex fsingle-precision-constant
5595 Treat floating point constant as single precision constant instead of
5596 implicitly converting it to double precision constant.
5598 @item -fcx-limited-range
5599 @itemx -fno-cx-limited-range
5600 @opindex fcx-limited-range
5601 @opindex fno-cx-limited-range
5602 When enabled, this option states that a range reduction step is not
5603 needed when performing complex division. The default is
5604 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5606 This option controls the default setting of the ISO C99
5607 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5612 The following options control optimizations that may improve
5613 performance, but are not enabled by any @option{-O} options. This
5614 section includes experimental options that may produce broken code.
5617 @item -fbranch-probabilities
5618 @opindex fbranch-probabilities
5619 After running a program compiled with @option{-fprofile-arcs}
5620 (@pxref{Debugging Options,, Options for Debugging Your Program or
5621 @command{gcc}}), you can compile it a second time using
5622 @option{-fbranch-probabilities}, to improve optimizations based on
5623 the number of times each branch was taken. When the program
5624 compiled with @option{-fprofile-arcs} exits it saves arc execution
5625 counts to a file called @file{@var{sourcename}.gcda} for each source
5626 file The information in this data file is very dependent on the
5627 structure of the generated code, so you must use the same source code
5628 and the same optimization options for both compilations.
5630 With @option{-fbranch-probabilities}, GCC puts a
5631 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5632 These can be used to improve optimization. Currently, they are only
5633 used in one place: in @file{reorg.c}, instead of guessing which path a
5634 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5635 exactly determine which path is taken more often.
5637 @item -fprofile-values
5638 @opindex fprofile-values
5639 If combined with @option{-fprofile-arcs}, it adds code so that some
5640 data about values of expressions in the program is gathered.
5642 With @option{-fbranch-probabilities}, it reads back the data gathered
5643 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5644 notes to instructions for their later usage in optimizations.
5646 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5650 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5651 a code to gather information about values of expressions.
5653 With @option{-fbranch-probabilities}, it reads back the data gathered
5654 and actually performs the optimizations based on them.
5655 Currently the optimizations include specialization of division operation
5656 using the knowledge about the value of the denominator.
5658 @item -frename-registers
5659 @opindex frename-registers
5660 Attempt to avoid false dependencies in scheduled code by making use
5661 of registers left over after register allocation. This optimization
5662 will most benefit processors with lots of registers. Depending on the
5663 debug information format adopted by the target, however, it can
5664 make debugging impossible, since variables will no longer stay in
5665 a ``home register''.
5667 Enabled by default with @option{-funroll-loops}.
5671 Perform tail duplication to enlarge superblock size. This transformation
5672 simplifies the control flow of the function allowing other optimizations to do
5675 Enabled with @option{-fprofile-use}.
5677 @item -funroll-loops
5678 @opindex funroll-loops
5679 Unroll loops whose number of iterations can be determined at compile time or
5680 upon entry to the loop. @option{-funroll-loops} implies
5681 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5682 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5683 small constant number of iterations). This option makes code larger, and may
5684 or may not make it run faster.
5686 Enabled with @option{-fprofile-use}.
5688 @item -funroll-all-loops
5689 @opindex funroll-all-loops
5690 Unroll all loops, even if their number of iterations is uncertain when
5691 the loop is entered. This usually makes programs run more slowly.
5692 @option{-funroll-all-loops} implies the same options as
5693 @option{-funroll-loops}.
5696 @opindex fpeel-loops
5697 Peels the loops for that there is enough information that they do not
5698 roll much (from profile feedback). It also turns on complete loop peeling
5699 (i.e.@: complete removal of loops with small constant number of iterations).
5701 Enabled with @option{-fprofile-use}.
5703 @item -fmove-loop-invariants
5704 @opindex fmove-loop-invariants
5705 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5706 at level @option{-O1}
5708 @item -funswitch-loops
5709 @opindex funswitch-loops
5710 Move branches with loop invariant conditions out of the loop, with duplicates
5711 of the loop on both branches (modified according to result of the condition).
5713 @item -fprefetch-loop-arrays
5714 @opindex fprefetch-loop-arrays
5715 If supported by the target machine, generate instructions to prefetch
5716 memory to improve the performance of loops that access large arrays.
5718 Disabled at level @option{-Os}.
5720 @item -ffunction-sections
5721 @itemx -fdata-sections
5722 @opindex ffunction-sections
5723 @opindex fdata-sections
5724 Place each function or data item into its own section in the output
5725 file if the target supports arbitrary sections. The name of the
5726 function or the name of the data item determines the section's name
5729 Use these options on systems where the linker can perform optimizations
5730 to improve locality of reference in the instruction space. Most systems
5731 using the ELF object format and SPARC processors running Solaris 2 have
5732 linkers with such optimizations. AIX may have these optimizations in
5735 Only use these options when there are significant benefits from doing
5736 so. When you specify these options, the assembler and linker will
5737 create larger object and executable files and will also be slower.
5738 You will not be able to use @code{gprof} on all systems if you
5739 specify this option and you may have problems with debugging if
5740 you specify both this option and @option{-g}.
5742 @item -fbranch-target-load-optimize
5743 @opindex fbranch-target-load-optimize
5744 Perform branch target register load optimization before prologue / epilogue
5746 The use of target registers can typically be exposed only during reload,
5747 thus hoisting loads out of loops and doing inter-block scheduling needs
5748 a separate optimization pass.
5750 @item -fbranch-target-load-optimize2
5751 @opindex fbranch-target-load-optimize2
5752 Perform branch target register load optimization after prologue / epilogue
5755 @item -fbtr-bb-exclusive
5756 @opindex fbtr-bb-exclusive
5757 When performing branch target register load optimization, don't reuse
5758 branch target registers in within any basic block.
5760 @item -fstack-protector
5761 Emit extra code to check for buffer overflows, such as stack smashing
5762 attacks. This is done by adding a guard variable to functions with
5763 vulnerable objects. This includes functions that call alloca, and
5764 functions with buffers larger than 8 bytes. The guards are initialized
5765 when a function is entered and then checked when the function exits.
5766 If a guard check fails, an error message is printed and the program exits.
5768 @item -fstack-protector-all
5769 Like @option{-fstack-protector} except that all functions are protected.
5771 @item --param @var{name}=@var{value}
5773 In some places, GCC uses various constants to control the amount of
5774 optimization that is done. For example, GCC will not inline functions
5775 that contain more that a certain number of instructions. You can
5776 control some of these constants on the command-line using the
5777 @option{--param} option.
5779 The names of specific parameters, and the meaning of the values, are
5780 tied to the internals of the compiler, and are subject to change
5781 without notice in future releases.
5783 In each case, the @var{value} is an integer. The allowable choices for
5784 @var{name} are given in the following table:
5787 @item salias-max-implicit-fields
5788 The maximum number of fields in a variable without direct
5789 structure accesses for which structure aliasing will consider trying
5790 to track each field. The default is 5
5792 @item salias-max-array-elements
5793 The maximum number of elements an array can have and its elements
5794 still be tracked individually by structure aliasing. The default is 4
5796 @item sra-max-structure-size
5797 The maximum structure size, in bytes, at which the scalar replacement
5798 of aggregates (SRA) optimization will perform block copies. The
5799 default value, 0, implies that GCC will select the most appropriate
5802 @item sra-field-structure-ratio
5803 The threshold ratio (as a percentage) between instantiated fields and
5804 the complete structure size. We say that if the ratio of the number
5805 of bytes in instantiated fields to the number of bytes in the complete
5806 structure exceeds this parameter, then block copies are not used. The
5809 @item max-crossjump-edges
5810 The maximum number of incoming edges to consider for crossjumping.
5811 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5812 the number of edges incoming to each block. Increasing values mean
5813 more aggressive optimization, making the compile time increase with
5814 probably small improvement in executable size.
5816 @item min-crossjump-insns
5817 The minimum number of instructions which must be matched at the end
5818 of two blocks before crossjumping will be performed on them. This
5819 value is ignored in the case where all instructions in the block being
5820 crossjumped from are matched. The default value is 5.
5822 @item max-grow-copy-bb-insns
5823 The maximum code size expansion factor when copying basic blocks
5824 instead of jumping. The expansion is relative to a jump instruction.
5825 The default value is 8.
5827 @item max-goto-duplication-insns
5828 The maximum number of instructions to duplicate to a block that jumps
5829 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5830 passes, GCC factors computed gotos early in the compilation process,
5831 and unfactors them as late as possible. Only computed jumps at the
5832 end of a basic blocks with no more than max-goto-duplication-insns are
5833 unfactored. The default value is 8.
5835 @item max-delay-slot-insn-search
5836 The maximum number of instructions to consider when looking for an
5837 instruction to fill a delay slot. If more than this arbitrary number of
5838 instructions is searched, the time savings from filling the delay slot
5839 will be minimal so stop searching. Increasing values mean more
5840 aggressive optimization, making the compile time increase with probably
5841 small improvement in executable run time.
5843 @item max-delay-slot-live-search
5844 When trying to fill delay slots, the maximum number of instructions to
5845 consider when searching for a block with valid live register
5846 information. Increasing this arbitrarily chosen value means more
5847 aggressive optimization, increasing the compile time. This parameter
5848 should be removed when the delay slot code is rewritten to maintain the
5851 @item max-gcse-memory
5852 The approximate maximum amount of memory that will be allocated in
5853 order to perform the global common subexpression elimination
5854 optimization. If more memory than specified is required, the
5855 optimization will not be done.
5857 @item max-gcse-passes
5858 The maximum number of passes of GCSE to run. The default is 1.
5860 @item max-pending-list-length
5861 The maximum number of pending dependencies scheduling will allow
5862 before flushing the current state and starting over. Large functions
5863 with few branches or calls can create excessively large lists which
5864 needlessly consume memory and resources.
5866 @item max-inline-insns-single
5867 Several parameters control the tree inliner used in gcc.
5868 This number sets the maximum number of instructions (counted in GCC's
5869 internal representation) in a single function that the tree inliner
5870 will consider for inlining. This only affects functions declared
5871 inline and methods implemented in a class declaration (C++).
5872 The default value is 450.
5874 @item max-inline-insns-auto
5875 When you use @option{-finline-functions} (included in @option{-O3}),
5876 a lot of functions that would otherwise not be considered for inlining
5877 by the compiler will be investigated. To those functions, a different
5878 (more restrictive) limit compared to functions declared inline can
5880 The default value is 90.
5882 @item large-function-insns
5883 The limit specifying really large functions. For functions larger than this
5884 limit after inlining inlining is constrained by
5885 @option{--param large-function-growth}. This parameter is useful primarily
5886 to avoid extreme compilation time caused by non-linear algorithms used by the
5888 This parameter is ignored when @option{-funit-at-a-time} is not used.
5889 The default value is 2700.
5891 @item large-function-growth
5892 Specifies maximal growth of large function caused by inlining in percents.
5893 This parameter is ignored when @option{-funit-at-a-time} is not used.
5894 The default value is 100 which limits large function growth to 2.0 times
5897 @item large-unit-insns
5898 The limit specifying large translation unit. Growth caused by inlining of
5899 units larger than this limit is limited by @option{--param inline-unit-growth}.
5900 For small units this might be too tight (consider unit consisting of function A
5901 that is inline and B that just calls A three time. If B is small relative to
5902 A, the growth of unit is 300\% and yet such inlining is very sane. For very
5903 large units consisting of small inlininable functions however the overall unit
5904 growth limit is needed to avoid exponential explosion of code size. Thus for
5905 smaller units, the size is increased to @option{--param large-unit-insns}
5906 before applying @option{--param inline-unit-growth}. The default is 10000
5908 @item inline-unit-growth
5909 Specifies maximal overall growth of the compilation unit caused by inlining.
5910 This parameter is ignored when @option{-funit-at-a-time} is not used.
5911 The default value is 50 which limits unit growth to 1.5 times the original
5914 @item max-inline-insns-recursive
5915 @itemx max-inline-insns-recursive-auto
5916 Specifies maximum number of instructions out-of-line copy of self recursive inline
5917 function can grow into by performing recursive inlining.
5919 For functions declared inline @option{--param max-inline-insns-recursive} is
5920 taken into acount. For function not declared inline, recursive inlining
5921 happens only when @option{-finline-functions} (included in @option{-O3}) is
5922 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5923 default value is 450.
5925 @item max-inline-recursive-depth
5926 @itemx max-inline-recursive-depth-auto
5927 Specifies maximum recursion depth used by the recursive inlining.
5929 For functions declared inline @option{--param max-inline-recursive-depth} is
5930 taken into acount. For function not declared inline, recursive inlining
5931 happens only when @option{-finline-functions} (included in @option{-O3}) is
5932 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5933 default value is 450.
5935 @item min-inline-recursive-probability
5936 Recursive inlining is profitable only for function having deep recursion
5937 in average and can hurt for function having little recursion depth by
5938 increasing the prologue size or complexity of function body to other
5941 When profile feedback is available (see @option{-fprofile-generate}) the actual
5942 recursion depth can be guessed from probability that function will recurse via
5943 given call expression. This parameter limits inlining only to call expression
5944 whose probability exceeds given threshold (in percents). The default value is
5947 @item inline-call-cost
5948 Specify cost of call instruction relative to simple arithmetics operations
5949 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
5950 functions and at the same time increases size of leaf function that is believed to
5951 reduce function size by being inlined. In effect it increases amount of
5952 inlining for code having large abstraction penalty (many functions that just
5953 pass the arguments to other functions) and decrease inlining for code with low
5954 abstraction penalty. The default value is 16.
5956 @item max-unrolled-insns
5957 The maximum number of instructions that a loop should have if that loop
5958 is unrolled, and if the loop is unrolled, it determines how many times
5959 the loop code is unrolled.
5961 @item max-average-unrolled-insns
5962 The maximum number of instructions biased by probabilities of their execution
5963 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5964 it determines how many times the loop code is unrolled.
5966 @item max-unroll-times
5967 The maximum number of unrollings of a single loop.
5969 @item max-peeled-insns
5970 The maximum number of instructions that a loop should have if that loop
5971 is peeled, and if the loop is peeled, it determines how many times
5972 the loop code is peeled.
5974 @item max-peel-times
5975 The maximum number of peelings of a single loop.
5977 @item max-completely-peeled-insns
5978 The maximum number of insns of a completely peeled loop.
5980 @item max-completely-peel-times
5981 The maximum number of iterations of a loop to be suitable for complete peeling.
5983 @item max-unswitch-insns
5984 The maximum number of insns of an unswitched loop.
5986 @item max-unswitch-level
5987 The maximum number of branches unswitched in a single loop.
5990 The minimum cost of an expensive expression in the loop invariant motion.
5992 @item iv-consider-all-candidates-bound
5993 Bound on number of candidates for induction variables below that
5994 all candidates are considered for each use in induction variable
5995 optimizations. Only the most relevant candidates are considered
5996 if there are more candidates, to avoid quadratic time complexity.
5998 @item iv-max-considered-uses
5999 The induction variable optimizations give up on loops that contain more
6000 induction variable uses.
6002 @item iv-always-prune-cand-set-bound
6003 If number of candidates in the set is smaller than this value,
6004 we always try to remove unnecessary ivs from the set during its
6005 optimization when a new iv is added to the set.
6007 @item scev-max-expr-size
6008 Bound on size of expressions used in the scalar evolutions analyzer.
6009 Large expressions slow the analyzer.
6011 @item vect-max-version-checks
6012 The maximum number of runtime checks that can be performed when doing
6013 loop versioning in the vectorizer. See option ftree-vect-loop-version
6014 for more information.
6016 @item max-iterations-to-track
6018 The maximum number of iterations of a loop the brute force algorithm
6019 for analysis of # of iterations of the loop tries to evaluate.
6021 @item hot-bb-count-fraction
6022 Select fraction of the maximal count of repetitions of basic block in program
6023 given basic block needs to have to be considered hot.
6025 @item hot-bb-frequency-fraction
6026 Select fraction of the maximal frequency of executions of basic block in
6027 function given basic block needs to have to be considered hot
6029 @item max-predicted-iterations
6030 The maximum number of loop iterations we predict statically. This is useful
6031 in cases where function contain single loop with known bound and other loop
6032 with unknown. We predict the known number of iterations correctly, while
6033 the unknown number of iterations average to roughly 10. This means that the
6034 loop without bounds would appear artificially cold relative to the other one.
6036 @item tracer-dynamic-coverage
6037 @itemx tracer-dynamic-coverage-feedback
6039 This value is used to limit superblock formation once the given percentage of
6040 executed instructions is covered. This limits unnecessary code size
6043 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6044 feedback is available. The real profiles (as opposed to statically estimated
6045 ones) are much less balanced allowing the threshold to be larger value.
6047 @item tracer-max-code-growth
6048 Stop tail duplication once code growth has reached given percentage. This is
6049 rather hokey argument, as most of the duplicates will be eliminated later in
6050 cross jumping, so it may be set to much higher values than is the desired code
6053 @item tracer-min-branch-ratio
6055 Stop reverse growth when the reverse probability of best edge is less than this
6056 threshold (in percent).
6058 @item tracer-min-branch-ratio
6059 @itemx tracer-min-branch-ratio-feedback
6061 Stop forward growth if the best edge do have probability lower than this
6064 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6065 compilation for profile feedback and one for compilation without. The value
6066 for compilation with profile feedback needs to be more conservative (higher) in
6067 order to make tracer effective.
6069 @item max-cse-path-length
6071 Maximum number of basic blocks on path that cse considers. The default is 10.
6074 The maximum instructions CSE process before flushing. The default is 1000.
6076 @item global-var-threshold
6078 Counts the number of function calls (@var{n}) and the number of
6079 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
6080 single artificial variable will be created to represent all the
6081 call-clobbered variables at function call sites. This artificial
6082 variable will then be made to alias every call-clobbered variable.
6083 (done as @code{int * size_t} on the host machine; beware overflow).
6085 @item max-aliased-vops
6087 Maximum number of virtual operands allowed to represent aliases
6088 before triggering the alias grouping heuristic. Alias grouping
6089 reduces compile times and memory consumption needed for aliasing at
6090 the expense of precision loss in alias information.
6092 @item ggc-min-expand
6094 GCC uses a garbage collector to manage its own memory allocation. This
6095 parameter specifies the minimum percentage by which the garbage
6096 collector's heap should be allowed to expand between collections.
6097 Tuning this may improve compilation speed; it has no effect on code
6100 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6101 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6102 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6103 GCC is not able to calculate RAM on a particular platform, the lower
6104 bound of 30% is used. Setting this parameter and
6105 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6106 every opportunity. This is extremely slow, but can be useful for
6109 @item ggc-min-heapsize
6111 Minimum size of the garbage collector's heap before it begins bothering
6112 to collect garbage. The first collection occurs after the heap expands
6113 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6114 tuning this may improve compilation speed, and has no effect on code
6117 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6118 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6119 with a lower bound of 4096 (four megabytes) and an upper bound of
6120 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6121 particular platform, the lower bound is used. Setting this parameter
6122 very large effectively disables garbage collection. Setting this
6123 parameter and @option{ggc-min-expand} to zero causes a full collection
6124 to occur at every opportunity.
6126 @item max-reload-search-insns
6127 The maximum number of instruction reload should look backward for equivalent
6128 register. Increasing values mean more aggressive optimization, making the
6129 compile time increase with probably slightly better performance. The default
6132 @item max-cselib-memory-location
6133 The maximum number of memory locations cselib should take into acount.
6134 Increasing values mean more aggressive optimization, making the compile time
6135 increase with probably slightly better performance. The default value is 500.
6137 @item max-flow-memory-location
6138 Similar as @option{max-cselib-memory-location} but for dataflow liveness.
6139 The default value is 100.
6141 @item reorder-blocks-duplicate
6142 @itemx reorder-blocks-duplicate-feedback
6144 Used by basic block reordering pass to decide whether to use unconditional
6145 branch or duplicate the code on its destination. Code is duplicated when its
6146 estimated size is smaller than this value multiplied by the estimated size of
6147 unconditional jump in the hot spots of the program.
6149 The @option{reorder-block-duplicate-feedback} is used only when profile
6150 feedback is available and may be set to higher values than
6151 @option{reorder-block-duplicate} since information about the hot spots is more
6154 @item max-sched-region-blocks
6155 The maximum number of blocks in a region to be considered for
6156 interblock scheduling. The default value is 10.
6158 @item max-sched-region-insns
6159 The maximum number of insns in a region to be considered for
6160 interblock scheduling. The default value is 100.
6162 @item min-sched-prob
6163 The minimum probability of reaching a source block for interblock
6164 speculative scheduling. The default value is 40.
6166 @item max-last-value-rtl
6168 The maximum size measured as number of RTLs that can be recorded in an expression
6169 in combiner for a pseudo register as last known value of that register. The default
6172 @item integer-share-limit
6173 Small integer constants can use a shared data structure, reducing the
6174 compiler's memory usage and increasing its speed. This sets the maximum
6175 value of a shared integer constant's. The default value is 256.
6177 @item min-virtual-mappings
6178 Specifies the minimum number of virtual mappings in the incremental
6179 SSA updater that should be registered to trigger the virtual mappings
6180 heuristic defined by virtual-mappings-ratio. The default value is
6183 @item virtual-mappings-ratio
6184 If the number of virtual mappings is virtual-mappings-ratio bigger
6185 than the number of virtual symbols to be updated, then the incremental
6186 SSA updater switches to a full update for those symbols. The default
6189 @item ssp-buffer-size
6190 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6191 protection when @option{-fstack-protection} is used.
6193 @item max-jump-thread-duplication-stmts
6194 Maximum number of statements allowed in a block that needs to be
6195 duplicated when threading jumps.
6199 @node Preprocessor Options
6200 @section Options Controlling the Preprocessor
6201 @cindex preprocessor options
6202 @cindex options, preprocessor
6204 These options control the C preprocessor, which is run on each C source
6205 file before actual compilation.
6207 If you use the @option{-E} option, nothing is done except preprocessing.
6208 Some of these options make sense only together with @option{-E} because
6209 they cause the preprocessor output to be unsuitable for actual
6214 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6215 and pass @var{option} directly through to the preprocessor. If
6216 @var{option} contains commas, it is split into multiple options at the
6217 commas. However, many options are modified, translated or interpreted
6218 by the compiler driver before being passed to the preprocessor, and
6219 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6220 interface is undocumented and subject to change, so whenever possible
6221 you should avoid using @option{-Wp} and let the driver handle the
6224 @item -Xpreprocessor @var{option}
6225 @opindex preprocessor
6226 Pass @var{option} as an option to the preprocessor. You can use this to
6227 supply system-specific preprocessor options which GCC does not know how to
6230 If you want to pass an option that takes an argument, you must use
6231 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6234 @include cppopts.texi
6236 @node Assembler Options
6237 @section Passing Options to the Assembler
6239 @c prevent bad page break with this line
6240 You can pass options to the assembler.
6243 @item -Wa,@var{option}
6245 Pass @var{option} as an option to the assembler. If @var{option}
6246 contains commas, it is split into multiple options at the commas.
6248 @item -Xassembler @var{option}
6250 Pass @var{option} as an option to the assembler. You can use this to
6251 supply system-specific assembler options which GCC does not know how to
6254 If you want to pass an option that takes an argument, you must use
6255 @option{-Xassembler} twice, once for the option and once for the argument.
6260 @section Options for Linking
6261 @cindex link options
6262 @cindex options, linking
6264 These options come into play when the compiler links object files into
6265 an executable output file. They are meaningless if the compiler is
6266 not doing a link step.
6270 @item @var{object-file-name}
6271 A file name that does not end in a special recognized suffix is
6272 considered to name an object file or library. (Object files are
6273 distinguished from libraries by the linker according to the file
6274 contents.) If linking is done, these object files are used as input
6283 If any of these options is used, then the linker is not run, and
6284 object file names should not be used as arguments. @xref{Overall
6288 @item -l@var{library}
6289 @itemx -l @var{library}
6291 Search the library named @var{library} when linking. (The second
6292 alternative with the library as a separate argument is only for
6293 POSIX compliance and is not recommended.)
6295 It makes a difference where in the command you write this option; the
6296 linker searches and processes libraries and object files in the order they
6297 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6298 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6299 to functions in @samp{z}, those functions may not be loaded.
6301 The linker searches a standard list of directories for the library,
6302 which is actually a file named @file{lib@var{library}.a}. The linker
6303 then uses this file as if it had been specified precisely by name.
6305 The directories searched include several standard system directories
6306 plus any that you specify with @option{-L}.
6308 Normally the files found this way are library files---archive files
6309 whose members are object files. The linker handles an archive file by
6310 scanning through it for members which define symbols that have so far
6311 been referenced but not defined. But if the file that is found is an
6312 ordinary object file, it is linked in the usual fashion. The only
6313 difference between using an @option{-l} option and specifying a file name
6314 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6315 and searches several directories.
6319 You need this special case of the @option{-l} option in order to
6320 link an Objective-C or Objective-C++ program.
6323 @opindex nostartfiles
6324 Do not use the standard system startup files when linking.
6325 The standard system libraries are used normally, unless @option{-nostdlib}
6326 or @option{-nodefaultlibs} is used.
6328 @item -nodefaultlibs
6329 @opindex nodefaultlibs
6330 Do not use the standard system libraries when linking.
6331 Only the libraries you specify will be passed to the linker.
6332 The standard startup files are used normally, unless @option{-nostartfiles}
6333 is used. The compiler may generate calls to @code{memcmp},
6334 @code{memset}, @code{memcpy} and @code{memmove}.
6335 These entries are usually resolved by entries in
6336 libc. These entry points should be supplied through some other
6337 mechanism when this option is specified.
6341 Do not use the standard system startup files or libraries when linking.
6342 No startup files and only the libraries you specify will be passed to
6343 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6344 @code{memcpy} and @code{memmove}.
6345 These entries are usually resolved by entries in
6346 libc. These entry points should be supplied through some other
6347 mechanism when this option is specified.
6349 @cindex @option{-lgcc}, use with @option{-nostdlib}
6350 @cindex @option{-nostdlib} and unresolved references
6351 @cindex unresolved references and @option{-nostdlib}
6352 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6353 @cindex @option{-nodefaultlibs} and unresolved references
6354 @cindex unresolved references and @option{-nodefaultlibs}
6355 One of the standard libraries bypassed by @option{-nostdlib} and
6356 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6357 that GCC uses to overcome shortcomings of particular machines, or special
6358 needs for some languages.
6359 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6360 Collection (GCC) Internals},
6361 for more discussion of @file{libgcc.a}.)
6362 In most cases, you need @file{libgcc.a} even when you want to avoid
6363 other standard libraries. In other words, when you specify @option{-nostdlib}
6364 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6365 This ensures that you have no unresolved references to internal GCC
6366 library subroutines. (For example, @samp{__main}, used to ensure C++
6367 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6368 GNU Compiler Collection (GCC) Internals}.)
6372 Produce a position independent executable on targets which support it.
6373 For predictable results, you must also specify the same set of options
6374 that were used to generate code (@option{-fpie}, @option{-fPIE},
6375 or model suboptions) when you specify this option.
6379 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6380 that support it. This instructs the linker to add all symbols, not
6381 only used ones, to the dynamic symbol table. This option is needed
6382 for some uses of @code{dlopen} or to allow obtaining backtraces
6383 from within a program.
6387 Remove all symbol table and relocation information from the executable.
6391 On systems that support dynamic linking, this prevents linking with the shared
6392 libraries. On other systems, this option has no effect.
6396 Produce a shared object which can then be linked with other objects to
6397 form an executable. Not all systems support this option. For predictable
6398 results, you must also specify the same set of options that were used to
6399 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6400 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6401 needs to build supplementary stub code for constructors to work. On
6402 multi-libbed systems, @samp{gcc -shared} must select the correct support
6403 libraries to link against. Failing to supply the correct flags may lead
6404 to subtle defects. Supplying them in cases where they are not necessary
6407 @item -shared-libgcc
6408 @itemx -static-libgcc
6409 @opindex shared-libgcc
6410 @opindex static-libgcc
6411 On systems that provide @file{libgcc} as a shared library, these options
6412 force the use of either the shared or static version respectively.
6413 If no shared version of @file{libgcc} was built when the compiler was
6414 configured, these options have no effect.
6416 There are several situations in which an application should use the
6417 shared @file{libgcc} instead of the static version. The most common
6418 of these is when the application wishes to throw and catch exceptions
6419 across different shared libraries. In that case, each of the libraries
6420 as well as the application itself should use the shared @file{libgcc}.
6422 Therefore, the G++ and GCJ drivers automatically add
6423 @option{-shared-libgcc} whenever you build a shared library or a main
6424 executable, because C++ and Java programs typically use exceptions, so
6425 this is the right thing to do.
6427 If, instead, you use the GCC driver to create shared libraries, you may
6428 find that they will not always be linked with the shared @file{libgcc}.
6429 If GCC finds, at its configuration time, that you have a non-GNU linker
6430 or a GNU linker that does not support option @option{--eh-frame-hdr},
6431 it will link the shared version of @file{libgcc} into shared libraries
6432 by default. Otherwise, it will take advantage of the linker and optimize
6433 away the linking with the shared version of @file{libgcc}, linking with
6434 the static version of libgcc by default. This allows exceptions to
6435 propagate through such shared libraries, without incurring relocation
6436 costs at library load time.
6438 However, if a library or main executable is supposed to throw or catch
6439 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6440 for the languages used in the program, or using the option
6441 @option{-shared-libgcc}, such that it is linked with the shared
6446 Bind references to global symbols when building a shared object. Warn
6447 about any unresolved references (unless overridden by the link editor
6448 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6451 @item -Xlinker @var{option}
6453 Pass @var{option} as an option to the linker. You can use this to
6454 supply system-specific linker options which GCC does not know how to
6457 If you want to pass an option that takes an argument, you must use
6458 @option{-Xlinker} twice, once for the option and once for the argument.
6459 For example, to pass @option{-assert definitions}, you must write
6460 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6461 @option{-Xlinker "-assert definitions"}, because this passes the entire
6462 string as a single argument, which is not what the linker expects.
6464 @item -Wl,@var{option}
6466 Pass @var{option} as an option to the linker. If @var{option} contains
6467 commas, it is split into multiple options at the commas.
6469 @item -u @var{symbol}
6471 Pretend the symbol @var{symbol} is undefined, to force linking of
6472 library modules to define it. You can use @option{-u} multiple times with
6473 different symbols to force loading of additional library modules.
6476 @node Directory Options
6477 @section Options for Directory Search
6478 @cindex directory options
6479 @cindex options, directory search
6482 These options specify directories to search for header files, for
6483 libraries and for parts of the compiler:
6488 Add the directory @var{dir} to the head of the list of directories to be
6489 searched for header files. This can be used to override a system header
6490 file, substituting your own version, since these directories are
6491 searched before the system header file directories. However, you should
6492 not use this option to add directories that contain vendor-supplied
6493 system header files (use @option{-isystem} for that). If you use more than
6494 one @option{-I} option, the directories are scanned in left-to-right
6495 order; the standard system directories come after.
6497 If a standard system include directory, or a directory specified with
6498 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6499 option will be ignored. The directory will still be searched but as a
6500 system directory at its normal position in the system include chain.
6501 This is to ensure that GCC's procedure to fix buggy system headers and
6502 the ordering for the include_next directive are not inadvertently changed.
6503 If you really need to change the search order for system directories,
6504 use the @option{-nostdinc} and/or @option{-isystem} options.
6506 @item -iquote@var{dir}
6508 Add the directory @var{dir} to the head of the list of directories to
6509 be searched for header files only for the case of @samp{#include
6510 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6511 otherwise just like @option{-I}.
6515 Add directory @var{dir} to the list of directories to be searched
6518 @item -B@var{prefix}
6520 This option specifies where to find the executables, libraries,
6521 include files, and data files of the compiler itself.
6523 The compiler driver program runs one or more of the subprograms
6524 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6525 @var{prefix} as a prefix for each program it tries to run, both with and
6526 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6528 For each subprogram to be run, the compiler driver first tries the
6529 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6530 was not specified, the driver tries two standard prefixes, which are
6531 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6532 those results in a file name that is found, the unmodified program
6533 name is searched for using the directories specified in your
6534 @env{PATH} environment variable.
6536 The compiler will check to see if the path provided by the @option{-B}
6537 refers to a directory, and if necessary it will add a directory
6538 separator character at the end of the path.
6540 @option{-B} prefixes that effectively specify directory names also apply
6541 to libraries in the linker, because the compiler translates these
6542 options into @option{-L} options for the linker. They also apply to
6543 includes files in the preprocessor, because the compiler translates these
6544 options into @option{-isystem} options for the preprocessor. In this case,
6545 the compiler appends @samp{include} to the prefix.
6547 The run-time support file @file{libgcc.a} can also be searched for using
6548 the @option{-B} prefix, if needed. If it is not found there, the two
6549 standard prefixes above are tried, and that is all. The file is left
6550 out of the link if it is not found by those means.
6552 Another way to specify a prefix much like the @option{-B} prefix is to use
6553 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6556 As a special kludge, if the path provided by @option{-B} is
6557 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6558 9, then it will be replaced by @file{[dir/]include}. This is to help
6559 with boot-strapping the compiler.
6561 @item -specs=@var{file}
6563 Process @var{file} after the compiler reads in the standard @file{specs}
6564 file, in order to override the defaults that the @file{gcc} driver
6565 program uses when determining what switches to pass to @file{cc1},
6566 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6567 @option{-specs=@var{file}} can be specified on the command line, and they
6568 are processed in order, from left to right.
6570 @item --sysroot=@var{dir}
6572 Use @var{dir} as the logical root directory for headers and libraries.
6573 For example, if the compiler would normally search for headers in
6574 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6575 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6577 If you use both this option and the @option{-isysroot} option, then
6578 the @option{--sysroot} option will apply to libraries, but the
6579 @option{-isysroot} option will apply to header files.
6581 The GNU linker (beginning with version 2.16) has the necessary support
6582 for this option. If your linker does not support this option, the
6583 header file aspect of @option{--sysroot} will still work, but the
6584 library aspect will not.
6588 This option has been deprecated. Please use @option{-iquote} instead for
6589 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6590 Any directories you specify with @option{-I} options before the @option{-I-}
6591 option are searched only for the case of @samp{#include "@var{file}"};
6592 they are not searched for @samp{#include <@var{file}>}.
6594 If additional directories are specified with @option{-I} options after
6595 the @option{-I-}, these directories are searched for all @samp{#include}
6596 directives. (Ordinarily @emph{all} @option{-I} directories are used
6599 In addition, the @option{-I-} option inhibits the use of the current
6600 directory (where the current input file came from) as the first search
6601 directory for @samp{#include "@var{file}"}. There is no way to
6602 override this effect of @option{-I-}. With @option{-I.} you can specify
6603 searching the directory which was current when the compiler was
6604 invoked. That is not exactly the same as what the preprocessor does
6605 by default, but it is often satisfactory.
6607 @option{-I-} does not inhibit the use of the standard system directories
6608 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6615 @section Specifying subprocesses and the switches to pass to them
6618 @command{gcc} is a driver program. It performs its job by invoking a
6619 sequence of other programs to do the work of compiling, assembling and
6620 linking. GCC interprets its command-line parameters and uses these to
6621 deduce which programs it should invoke, and which command-line options
6622 it ought to place on their command lines. This behavior is controlled
6623 by @dfn{spec strings}. In most cases there is one spec string for each
6624 program that GCC can invoke, but a few programs have multiple spec
6625 strings to control their behavior. The spec strings built into GCC can
6626 be overridden by using the @option{-specs=} command-line switch to specify
6629 @dfn{Spec files} are plaintext files that are used to construct spec
6630 strings. They consist of a sequence of directives separated by blank
6631 lines. The type of directive is determined by the first non-whitespace
6632 character on the line and it can be one of the following:
6635 @item %@var{command}
6636 Issues a @var{command} to the spec file processor. The commands that can
6640 @item %include <@var{file}>
6642 Search for @var{file} and insert its text at the current point in the
6645 @item %include_noerr <@var{file}>
6646 @cindex %include_noerr
6647 Just like @samp{%include}, but do not generate an error message if the include
6648 file cannot be found.
6650 @item %rename @var{old_name} @var{new_name}
6652 Rename the spec string @var{old_name} to @var{new_name}.
6656 @item *[@var{spec_name}]:
6657 This tells the compiler to create, override or delete the named spec
6658 string. All lines after this directive up to the next directive or
6659 blank line are considered to be the text for the spec string. If this
6660 results in an empty string then the spec will be deleted. (Or, if the
6661 spec did not exist, then nothing will happened.) Otherwise, if the spec
6662 does not currently exist a new spec will be created. If the spec does
6663 exist then its contents will be overridden by the text of this
6664 directive, unless the first character of that text is the @samp{+}
6665 character, in which case the text will be appended to the spec.
6667 @item [@var{suffix}]:
6668 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6669 and up to the next directive or blank line are considered to make up the
6670 spec string for the indicated suffix. When the compiler encounters an
6671 input file with the named suffix, it will processes the spec string in
6672 order to work out how to compile that file. For example:
6679 This says that any input file whose name ends in @samp{.ZZ} should be
6680 passed to the program @samp{z-compile}, which should be invoked with the
6681 command-line switch @option{-input} and with the result of performing the
6682 @samp{%i} substitution. (See below.)
6684 As an alternative to providing a spec string, the text that follows a
6685 suffix directive can be one of the following:
6688 @item @@@var{language}
6689 This says that the suffix is an alias for a known @var{language}. This is
6690 similar to using the @option{-x} command-line switch to GCC to specify a
6691 language explicitly. For example:
6698 Says that .ZZ files are, in fact, C++ source files.
6701 This causes an error messages saying:
6704 @var{name} compiler not installed on this system.
6708 GCC already has an extensive list of suffixes built into it.
6709 This directive will add an entry to the end of the list of suffixes, but
6710 since the list is searched from the end backwards, it is effectively
6711 possible to override earlier entries using this technique.
6715 GCC has the following spec strings built into it. Spec files can
6716 override these strings or create their own. Note that individual
6717 targets can also add their own spec strings to this list.
6720 asm Options to pass to the assembler
6721 asm_final Options to pass to the assembler post-processor
6722 cpp Options to pass to the C preprocessor
6723 cc1 Options to pass to the C compiler
6724 cc1plus Options to pass to the C++ compiler
6725 endfile Object files to include at the end of the link
6726 link Options to pass to the linker
6727 lib Libraries to include on the command line to the linker
6728 libgcc Decides which GCC support library to pass to the linker
6729 linker Sets the name of the linker
6730 predefines Defines to be passed to the C preprocessor
6731 signed_char Defines to pass to CPP to say whether @code{char} is signed
6733 startfile Object files to include at the start of the link
6736 Here is a small example of a spec file:
6742 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6745 This example renames the spec called @samp{lib} to @samp{old_lib} and
6746 then overrides the previous definition of @samp{lib} with a new one.
6747 The new definition adds in some extra command-line options before
6748 including the text of the old definition.
6750 @dfn{Spec strings} are a list of command-line options to be passed to their
6751 corresponding program. In addition, the spec strings can contain
6752 @samp{%}-prefixed sequences to substitute variable text or to
6753 conditionally insert text into the command line. Using these constructs
6754 it is possible to generate quite complex command lines.
6756 Here is a table of all defined @samp{%}-sequences for spec
6757 strings. Note that spaces are not generated automatically around the
6758 results of expanding these sequences. Therefore you can concatenate them
6759 together or combine them with constant text in a single argument.
6763 Substitute one @samp{%} into the program name or argument.
6766 Substitute the name of the input file being processed.
6769 Substitute the basename of the input file being processed.
6770 This is the substring up to (and not including) the last period
6771 and not including the directory.
6774 This is the same as @samp{%b}, but include the file suffix (text after
6778 Marks the argument containing or following the @samp{%d} as a
6779 temporary file name, so that that file will be deleted if GCC exits
6780 successfully. Unlike @samp{%g}, this contributes no text to the
6783 @item %g@var{suffix}
6784 Substitute a file name that has suffix @var{suffix} and is chosen
6785 once per compilation, and mark the argument in the same way as
6786 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6787 name is now chosen in a way that is hard to predict even when previously
6788 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6789 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6790 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6791 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6792 was simply substituted with a file name chosen once per compilation,
6793 without regard to any appended suffix (which was therefore treated
6794 just like ordinary text), making such attacks more likely to succeed.
6796 @item %u@var{suffix}
6797 Like @samp{%g}, but generates a new temporary file name even if
6798 @samp{%u@var{suffix}} was already seen.
6800 @item %U@var{suffix}
6801 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6802 new one if there is no such last file name. In the absence of any
6803 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6804 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6805 would involve the generation of two distinct file names, one
6806 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6807 simply substituted with a file name chosen for the previous @samp{%u},
6808 without regard to any appended suffix.
6810 @item %j@var{suffix}
6811 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6812 writable, and if save-temps is off; otherwise, substitute the name
6813 of a temporary file, just like @samp{%u}. This temporary file is not
6814 meant for communication between processes, but rather as a junk
6817 @item %|@var{suffix}
6818 @itemx %m@var{suffix}
6819 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6820 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6821 all. These are the two most common ways to instruct a program that it
6822 should read from standard input or write to standard output. If you
6823 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6824 construct: see for example @file{f/lang-specs.h}.
6826 @item %.@var{SUFFIX}
6827 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6828 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6829 terminated by the next space or %.
6832 Marks the argument containing or following the @samp{%w} as the
6833 designated output file of this compilation. This puts the argument
6834 into the sequence of arguments that @samp{%o} will substitute later.
6837 Substitutes the names of all the output files, with spaces
6838 automatically placed around them. You should write spaces
6839 around the @samp{%o} as well or the results are undefined.
6840 @samp{%o} is for use in the specs for running the linker.
6841 Input files whose names have no recognized suffix are not compiled
6842 at all, but they are included among the output files, so they will
6846 Substitutes the suffix for object files. Note that this is
6847 handled specially when it immediately follows @samp{%g, %u, or %U},
6848 because of the need for those to form complete file names. The
6849 handling is such that @samp{%O} is treated exactly as if it had already
6850 been substituted, except that @samp{%g, %u, and %U} do not currently
6851 support additional @var{suffix} characters following @samp{%O} as they would
6852 following, for example, @samp{.o}.
6855 Substitutes the standard macro predefinitions for the
6856 current target machine. Use this when running @code{cpp}.
6859 Like @samp{%p}, but puts @samp{__} before and after the name of each
6860 predefined macro, except for macros that start with @samp{__} or with
6861 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6865 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6866 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
6867 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6868 and @option{-imultilib} as necessary.
6871 Current argument is the name of a library or startup file of some sort.
6872 Search for that file in a standard list of directories and substitute
6873 the full name found.
6876 Print @var{str} as an error message. @var{str} is terminated by a newline.
6877 Use this when inconsistent options are detected.
6880 Substitute the contents of spec string @var{name} at this point.
6883 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6885 @item %x@{@var{option}@}
6886 Accumulate an option for @samp{%X}.
6889 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6893 Output the accumulated assembler options specified by @option{-Wa}.
6896 Output the accumulated preprocessor options specified by @option{-Wp}.
6899 Process the @code{asm} spec. This is used to compute the
6900 switches to be passed to the assembler.
6903 Process the @code{asm_final} spec. This is a spec string for
6904 passing switches to an assembler post-processor, if such a program is
6908 Process the @code{link} spec. This is the spec for computing the
6909 command line passed to the linker. Typically it will make use of the
6910 @samp{%L %G %S %D and %E} sequences.
6913 Dump out a @option{-L} option for each directory that GCC believes might
6914 contain startup files. If the target supports multilibs then the
6915 current multilib directory will be prepended to each of these paths.
6918 Process the @code{lib} spec. This is a spec string for deciding which
6919 libraries should be included on the command line to the linker.
6922 Process the @code{libgcc} spec. This is a spec string for deciding
6923 which GCC support library should be included on the command line to the linker.
6926 Process the @code{startfile} spec. This is a spec for deciding which
6927 object files should be the first ones passed to the linker. Typically
6928 this might be a file named @file{crt0.o}.
6931 Process the @code{endfile} spec. This is a spec string that specifies
6932 the last object files that will be passed to the linker.
6935 Process the @code{cpp} spec. This is used to construct the arguments
6936 to be passed to the C preprocessor.
6939 Process the @code{cc1} spec. This is used to construct the options to be
6940 passed to the actual C compiler (@samp{cc1}).
6943 Process the @code{cc1plus} spec. This is used to construct the options to be
6944 passed to the actual C++ compiler (@samp{cc1plus}).
6947 Substitute the variable part of a matched option. See below.
6948 Note that each comma in the substituted string is replaced by
6952 Remove all occurrences of @code{-S} from the command line. Note---this
6953 command is position dependent. @samp{%} commands in the spec string
6954 before this one will see @code{-S}, @samp{%} commands in the spec string
6955 after this one will not.
6957 @item %:@var{function}(@var{args})
6958 Call the named function @var{function}, passing it @var{args}.
6959 @var{args} is first processed as a nested spec string, then split
6960 into an argument vector in the usual fashion. The function returns
6961 a string which is processed as if it had appeared literally as part
6962 of the current spec.
6964 The following built-in spec functions are provided:
6967 @item @code{if-exists}
6968 The @code{if-exists} spec function takes one argument, an absolute
6969 pathname to a file. If the file exists, @code{if-exists} returns the
6970 pathname. Here is a small example of its usage:
6974 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6977 @item @code{if-exists-else}
6978 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6979 spec function, except that it takes two arguments. The first argument is
6980 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6981 returns the pathname. If it does not exist, it returns the second argument.
6982 This way, @code{if-exists-else} can be used to select one file or another,
6983 based on the existence of the first. Here is a small example of its usage:
6987 crt0%O%s %:if-exists(crti%O%s) \
6988 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6991 @item @code{replace-outfile}
6992 The @code{replace-outfile} spec function takes two arguments. It looks for the
6993 first argument in the outfiles array and replaces it with the second argument. Here
6994 is a small example of its usage:
6997 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7003 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7004 If that switch was not specified, this substitutes nothing. Note that
7005 the leading dash is omitted when specifying this option, and it is
7006 automatically inserted if the substitution is performed. Thus the spec
7007 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7008 and would output the command line option @option{-foo}.
7010 @item %W@{@code{S}@}
7011 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7014 @item %@{@code{S}*@}
7015 Substitutes all the switches specified to GCC whose names start
7016 with @code{-S}, but which also take an argument. This is used for
7017 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7018 GCC considers @option{-o foo} as being
7019 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7020 text, including the space. Thus two arguments would be generated.
7022 @item %@{@code{S}*&@code{T}*@}
7023 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7024 (the order of @code{S} and @code{T} in the spec is not significant).
7025 There can be any number of ampersand-separated variables; for each the
7026 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7028 @item %@{@code{S}:@code{X}@}
7029 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7031 @item %@{!@code{S}:@code{X}@}
7032 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7034 @item %@{@code{S}*:@code{X}@}
7035 Substitutes @code{X} if one or more switches whose names start with
7036 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7037 once, no matter how many such switches appeared. However, if @code{%*}
7038 appears somewhere in @code{X}, then @code{X} will be substituted once
7039 for each matching switch, with the @code{%*} replaced by the part of
7040 that switch that matched the @code{*}.
7042 @item %@{.@code{S}:@code{X}@}
7043 Substitutes @code{X}, if processing a file with suffix @code{S}.
7045 @item %@{!.@code{S}:@code{X}@}
7046 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7048 @item %@{@code{S}|@code{P}:@code{X}@}
7049 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7050 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7051 although they have a stronger binding than the @samp{|}. If @code{%*}
7052 appears in @code{X}, all of the alternatives must be starred, and only
7053 the first matching alternative is substituted.
7055 For example, a spec string like this:
7058 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7061 will output the following command-line options from the following input
7062 command-line options:
7067 -d fred.c -foo -baz -boggle
7068 -d jim.d -bar -baz -boggle
7071 @item %@{S:X; T:Y; :D@}
7073 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7074 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7075 be as many clauses as you need. This may be combined with @code{.},
7076 @code{!}, @code{|}, and @code{*} as needed.
7081 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7082 construct may contain other nested @samp{%} constructs or spaces, or
7083 even newlines. They are processed as usual, as described above.
7084 Trailing white space in @code{X} is ignored. White space may also
7085 appear anywhere on the left side of the colon in these constructs,
7086 except between @code{.} or @code{*} and the corresponding word.
7088 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7089 handled specifically in these constructs. If another value of
7090 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7091 @option{-W} switch is found later in the command line, the earlier
7092 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7093 just one letter, which passes all matching options.
7095 The character @samp{|} at the beginning of the predicate text is used to
7096 indicate that a command should be piped to the following command, but
7097 only if @option{-pipe} is specified.
7099 It is built into GCC which switches take arguments and which do not.
7100 (You might think it would be useful to generalize this to allow each
7101 compiler's spec to say which switches take arguments. But this cannot
7102 be done in a consistent fashion. GCC cannot even decide which input
7103 files have been specified without knowing which switches take arguments,
7104 and it must know which input files to compile in order to tell which
7107 GCC also knows implicitly that arguments starting in @option{-l} are to be
7108 treated as compiler output files, and passed to the linker in their
7109 proper position among the other output files.
7111 @c man begin OPTIONS
7113 @node Target Options
7114 @section Specifying Target Machine and Compiler Version
7115 @cindex target options
7116 @cindex cross compiling
7117 @cindex specifying machine version
7118 @cindex specifying compiler version and target machine
7119 @cindex compiler version, specifying
7120 @cindex target machine, specifying
7122 The usual way to run GCC is to run the executable called @file{gcc}, or
7123 @file{<machine>-gcc} when cross-compiling, or
7124 @file{<machine>-gcc-<version>} to run a version other than the one that
7125 was installed last. Sometimes this is inconvenient, so GCC provides
7126 options that will switch to another cross-compiler or version.
7129 @item -b @var{machine}
7131 The argument @var{machine} specifies the target machine for compilation.
7133 The value to use for @var{machine} is the same as was specified as the
7134 machine type when configuring GCC as a cross-compiler. For
7135 example, if a cross-compiler was configured with @samp{configure
7136 arm-elf}, meaning to compile for an arm processor with elf binaries,
7137 then you would specify @option{-b arm-elf} to run that cross compiler.
7138 Because there are other options beginning with @option{-b}, the
7139 configuration must contain a hyphen.
7141 @item -V @var{version}
7143 The argument @var{version} specifies which version of GCC to run.
7144 This is useful when multiple versions are installed. For example,
7145 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7148 The @option{-V} and @option{-b} options work by running the
7149 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7150 use them if you can just run that directly.
7152 @node Submodel Options
7153 @section Hardware Models and Configurations
7154 @cindex submodel options
7155 @cindex specifying hardware config
7156 @cindex hardware models and configurations, specifying
7157 @cindex machine dependent options
7159 Earlier we discussed the standard option @option{-b} which chooses among
7160 different installed compilers for completely different target
7161 machines, such as VAX vs.@: 68000 vs.@: 80386.
7163 In addition, each of these target machine types can have its own
7164 special options, starting with @samp{-m}, to choose among various
7165 hardware models or configurations---for example, 68010 vs 68020,
7166 floating coprocessor or none. A single installed version of the
7167 compiler can compile for any model or configuration, according to the
7170 Some configurations of the compiler also support additional special
7171 options, usually for compatibility with other compilers on the same
7174 @c This list is ordered alphanumerically by subsection name.
7175 @c It should be the same order and spelling as these options are listed
7176 @c in Machine Dependent Options
7182 * Blackfin Options::
7186 * DEC Alpha Options::
7187 * DEC Alpha/VMS Options::
7191 * i386 and x86-64 Options::
7204 * RS/6000 and PowerPC Options::
7205 * S/390 and zSeries Options::
7208 * System V Options::
7209 * TMS320C3x/C4x Options::
7213 * Xstormy16 Options::
7219 @subsection ARC Options
7222 These options are defined for ARC implementations:
7227 Compile code for little endian mode. This is the default.
7231 Compile code for big endian mode.
7234 @opindex mmangle-cpu
7235 Prepend the name of the cpu to all public symbol names.
7236 In multiple-processor systems, there are many ARC variants with different
7237 instruction and register set characteristics. This flag prevents code
7238 compiled for one cpu to be linked with code compiled for another.
7239 No facility exists for handling variants that are ``almost identical''.
7240 This is an all or nothing option.
7242 @item -mcpu=@var{cpu}
7244 Compile code for ARC variant @var{cpu}.
7245 Which variants are supported depend on the configuration.
7246 All variants support @option{-mcpu=base}, this is the default.
7248 @item -mtext=@var{text-section}
7249 @itemx -mdata=@var{data-section}
7250 @itemx -mrodata=@var{readonly-data-section}
7254 Put functions, data, and readonly data in @var{text-section},
7255 @var{data-section}, and @var{readonly-data-section} respectively
7256 by default. This can be overridden with the @code{section} attribute.
7257 @xref{Variable Attributes}.
7262 @subsection ARM Options
7265 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7269 @item -mabi=@var{name}
7271 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7272 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7275 @opindex mapcs-frame
7276 Generate a stack frame that is compliant with the ARM Procedure Call
7277 Standard for all functions, even if this is not strictly necessary for
7278 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7279 with this option will cause the stack frames not to be generated for
7280 leaf functions. The default is @option{-mno-apcs-frame}.
7284 This is a synonym for @option{-mapcs-frame}.
7287 @c not currently implemented
7288 @item -mapcs-stack-check
7289 @opindex mapcs-stack-check
7290 Generate code to check the amount of stack space available upon entry to
7291 every function (that actually uses some stack space). If there is
7292 insufficient space available then either the function
7293 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7294 called, depending upon the amount of stack space required. The run time
7295 system is required to provide these functions. The default is
7296 @option{-mno-apcs-stack-check}, since this produces smaller code.
7298 @c not currently implemented
7300 @opindex mapcs-float
7301 Pass floating point arguments using the float point registers. This is
7302 one of the variants of the APCS@. This option is recommended if the
7303 target hardware has a floating point unit or if a lot of floating point
7304 arithmetic is going to be performed by the code. The default is
7305 @option{-mno-apcs-float}, since integer only code is slightly increased in
7306 size if @option{-mapcs-float} is used.
7308 @c not currently implemented
7309 @item -mapcs-reentrant
7310 @opindex mapcs-reentrant
7311 Generate reentrant, position independent code. The default is
7312 @option{-mno-apcs-reentrant}.
7315 @item -mthumb-interwork
7316 @opindex mthumb-interwork
7317 Generate code which supports calling between the ARM and Thumb
7318 instruction sets. Without this option the two instruction sets cannot
7319 be reliably used inside one program. The default is
7320 @option{-mno-thumb-interwork}, since slightly larger code is generated
7321 when @option{-mthumb-interwork} is specified.
7323 @item -mno-sched-prolog
7324 @opindex mno-sched-prolog
7325 Prevent the reordering of instructions in the function prolog, or the
7326 merging of those instruction with the instructions in the function's
7327 body. This means that all functions will start with a recognizable set
7328 of instructions (or in fact one of a choice from a small set of
7329 different function prologues), and this information can be used to
7330 locate the start if functions inside an executable piece of code. The
7331 default is @option{-msched-prolog}.
7334 @opindex mhard-float
7335 Generate output containing floating point instructions. This is the
7339 @opindex msoft-float
7340 Generate output containing library calls for floating point.
7341 @strong{Warning:} the requisite libraries are not available for all ARM
7342 targets. Normally the facilities of the machine's usual C compiler are
7343 used, but this cannot be done directly in cross-compilation. You must make
7344 your own arrangements to provide suitable library functions for
7347 @option{-msoft-float} changes the calling convention in the output file;
7348 therefore, it is only useful if you compile @emph{all} of a program with
7349 this option. In particular, you need to compile @file{libgcc.a}, the
7350 library that comes with GCC, with @option{-msoft-float} in order for
7353 @item -mfloat-abi=@var{name}
7355 Specifies which ABI to use for floating point values. Permissible values
7356 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7358 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7359 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7360 of floating point instructions, but still uses the soft-float calling
7363 @item -mlittle-endian
7364 @opindex mlittle-endian
7365 Generate code for a processor running in little-endian mode. This is
7366 the default for all standard configurations.
7369 @opindex mbig-endian
7370 Generate code for a processor running in big-endian mode; the default is
7371 to compile code for a little-endian processor.
7373 @item -mwords-little-endian
7374 @opindex mwords-little-endian
7375 This option only applies when generating code for big-endian processors.
7376 Generate code for a little-endian word order but a big-endian byte
7377 order. That is, a byte order of the form @samp{32107654}. Note: this
7378 option should only be used if you require compatibility with code for
7379 big-endian ARM processors generated by versions of the compiler prior to
7382 @item -mcpu=@var{name}
7384 This specifies the name of the target ARM processor. GCC uses this name
7385 to determine what kind of instructions it can emit when generating
7386 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7387 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7388 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7389 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7390 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7391 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7392 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7393 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7394 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7395 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7396 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7397 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7398 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7399 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7402 @itemx -mtune=@var{name}
7404 This option is very similar to the @option{-mcpu=} option, except that
7405 instead of specifying the actual target processor type, and hence
7406 restricting which instructions can be used, it specifies that GCC should
7407 tune the performance of the code as if the target were of the type
7408 specified in this option, but still choosing the instructions that it
7409 will generate based on the cpu specified by a @option{-mcpu=} option.
7410 For some ARM implementations better performance can be obtained by using
7413 @item -march=@var{name}
7415 This specifies the name of the target ARM architecture. GCC uses this
7416 name to determine what kind of instructions it can emit when generating
7417 assembly code. This option can be used in conjunction with or instead
7418 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7419 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7420 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7421 @samp{iwmmxt}, @samp{ep9312}.
7423 @item -mfpu=@var{name}
7424 @itemx -mfpe=@var{number}
7425 @itemx -mfp=@var{number}
7429 This specifies what floating point hardware (or hardware emulation) is
7430 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7431 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7432 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7433 with older versions of GCC@.
7435 If @option{-msoft-float} is specified this specifies the format of
7436 floating point values.
7438 @item -mstructure-size-boundary=@var{n}
7439 @opindex mstructure-size-boundary
7440 The size of all structures and unions will be rounded up to a multiple
7441 of the number of bits set by this option. Permissible values are 8, 32
7442 and 64. The default value varies for different toolchains. For the COFF
7443 targeted toolchain the default value is 8. A value of 64 is only allowed
7444 if the underlying ABI supports it.
7446 Specifying the larger number can produce faster, more efficient code, but
7447 can also increase the size of the program. Different values are potentially
7448 incompatible. Code compiled with one value cannot necessarily expect to
7449 work with code or libraries compiled with another value, if they exchange
7450 information using structures or unions.
7452 @item -mabort-on-noreturn
7453 @opindex mabort-on-noreturn
7454 Generate a call to the function @code{abort} at the end of a
7455 @code{noreturn} function. It will be executed if the function tries to
7459 @itemx -mno-long-calls
7460 @opindex mlong-calls
7461 @opindex mno-long-calls
7462 Tells the compiler to perform function calls by first loading the
7463 address of the function into a register and then performing a subroutine
7464 call on this register. This switch is needed if the target function
7465 will lie outside of the 64 megabyte addressing range of the offset based
7466 version of subroutine call instruction.
7468 Even if this switch is enabled, not all function calls will be turned
7469 into long calls. The heuristic is that static functions, functions
7470 which have the @samp{short-call} attribute, functions that are inside
7471 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7472 definitions have already been compiled within the current compilation
7473 unit, will not be turned into long calls. The exception to this rule is
7474 that weak function definitions, functions with the @samp{long-call}
7475 attribute or the @samp{section} attribute, and functions that are within
7476 the scope of a @samp{#pragma long_calls} directive, will always be
7477 turned into long calls.
7479 This feature is not enabled by default. Specifying
7480 @option{-mno-long-calls} will restore the default behavior, as will
7481 placing the function calls within the scope of a @samp{#pragma
7482 long_calls_off} directive. Note these switches have no effect on how
7483 the compiler generates code to handle function calls via function
7486 @item -mnop-fun-dllimport
7487 @opindex mnop-fun-dllimport
7488 Disable support for the @code{dllimport} attribute.
7490 @item -msingle-pic-base
7491 @opindex msingle-pic-base
7492 Treat the register used for PIC addressing as read-only, rather than
7493 loading it in the prologue for each function. The run-time system is
7494 responsible for initializing this register with an appropriate value
7495 before execution begins.
7497 @item -mpic-register=@var{reg}
7498 @opindex mpic-register
7499 Specify the register to be used for PIC addressing. The default is R10
7500 unless stack-checking is enabled, when R9 is used.
7502 @item -mcirrus-fix-invalid-insns
7503 @opindex mcirrus-fix-invalid-insns
7504 @opindex mno-cirrus-fix-invalid-insns
7505 Insert NOPs into the instruction stream to in order to work around
7506 problems with invalid Maverick instruction combinations. This option
7507 is only valid if the @option{-mcpu=ep9312} option has been used to
7508 enable generation of instructions for the Cirrus Maverick floating
7509 point co-processor. This option is not enabled by default, since the
7510 problem is only present in older Maverick implementations. The default
7511 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7514 @item -mpoke-function-name
7515 @opindex mpoke-function-name
7516 Write the name of each function into the text section, directly
7517 preceding the function prologue. The generated code is similar to this:
7521 .ascii "arm_poke_function_name", 0
7524 .word 0xff000000 + (t1 - t0)
7525 arm_poke_function_name
7527 stmfd sp!, @{fp, ip, lr, pc@}
7531 When performing a stack backtrace, code can inspect the value of
7532 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7533 location @code{pc - 12} and the top 8 bits are set, then we know that
7534 there is a function name embedded immediately preceding this location
7535 and has length @code{((pc[-3]) & 0xff000000)}.
7539 Generate code for the 16-bit Thumb instruction set. The default is to
7540 use the 32-bit ARM instruction set.
7543 @opindex mtpcs-frame
7544 Generate a stack frame that is compliant with the Thumb Procedure Call
7545 Standard for all non-leaf functions. (A leaf function is one that does
7546 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7548 @item -mtpcs-leaf-frame
7549 @opindex mtpcs-leaf-frame
7550 Generate a stack frame that is compliant with the Thumb Procedure Call
7551 Standard for all leaf functions. (A leaf function is one that does
7552 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7554 @item -mcallee-super-interworking
7555 @opindex mcallee-super-interworking
7556 Gives all externally visible functions in the file being compiled an ARM
7557 instruction set header which switches to Thumb mode before executing the
7558 rest of the function. This allows these functions to be called from
7559 non-interworking code.
7561 @item -mcaller-super-interworking
7562 @opindex mcaller-super-interworking
7563 Allows calls via function pointers (including virtual functions) to
7564 execute correctly regardless of whether the target code has been
7565 compiled for interworking or not. There is a small overhead in the cost
7566 of executing a function pointer if this option is enabled.
7568 @item -mtp=@var{name}
7570 Specify the access model for the thread local storage pointer. The valid
7571 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7572 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7573 (supported in the arm6k architecture), and @option{auto}, which uses the
7574 best available method for the selected processor. The default setting is
7580 @subsection AVR Options
7583 These options are defined for AVR implementations:
7586 @item -mmcu=@var{mcu}
7588 Specify ATMEL AVR instruction set or MCU type.
7590 Instruction set avr1 is for the minimal AVR core, not supported by the C
7591 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7592 attiny11, attiny12, attiny15, attiny28).
7594 Instruction set avr2 (default) is for the classic AVR core with up to
7595 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7596 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7597 at90c8534, at90s8535).
7599 Instruction set avr3 is for the classic AVR core with up to 128K program
7600 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7602 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7603 memory space (MCU types: atmega8, atmega83, atmega85).
7605 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7606 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7607 atmega64, atmega128, at43usb355, at94k).
7611 Output instruction sizes to the asm file.
7613 @item -minit-stack=@var{N}
7614 @opindex minit-stack
7615 Specify the initial stack address, which may be a symbol or numeric value,
7616 @samp{__stack} is the default.
7618 @item -mno-interrupts
7619 @opindex mno-interrupts
7620 Generated code is not compatible with hardware interrupts.
7621 Code size will be smaller.
7623 @item -mcall-prologues
7624 @opindex mcall-prologues
7625 Functions prologues/epilogues expanded as call to appropriate
7626 subroutines. Code size will be smaller.
7628 @item -mno-tablejump
7629 @opindex mno-tablejump
7630 Do not generate tablejump insns which sometimes increase code size.
7633 @opindex mtiny-stack
7634 Change only the low 8 bits of the stack pointer.
7638 Assume int to be 8 bit integer. This affects the sizes of all types: A
7639 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7640 and long long will be 4 bytes. Please note that this option does not
7641 comply to the C standards, but it will provide you with smaller code
7645 @node Blackfin Options
7646 @subsection Blackfin Options
7647 @cindex Blackfin Options
7650 @item -momit-leaf-frame-pointer
7651 @opindex momit-leaf-frame-pointer
7652 Don't keep the frame pointer in a register for leaf functions. This
7653 avoids the instructions to save, set up and restore frame pointers and
7654 makes an extra register available in leaf functions. The option
7655 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7656 which might make debugging harder.
7658 @item -mspecld-anomaly
7659 @opindex mspecld-anomaly
7660 When enabled, the compiler will ensure that the generated code does not
7661 contain speculative loads after jump instructions. This option is enabled
7664 @item -mno-specld-anomaly
7665 @opindex mno-specld-anomaly
7666 Don't generate extra code to prevent speculative loads from occurring.
7668 @item -mcsync-anomaly
7669 @opindex mcsync-anomaly
7670 When enabled, the compiler will ensure that the generated code does not
7671 contain CSYNC or SSYNC instructions too soon after conditional branches.
7672 This option is enabled by default.
7674 @item -mno-csync-anomaly
7675 @opindex mno-csync-anomaly
7676 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7677 occurring too soon after a conditional branch.
7681 When enabled, the compiler is free to take advantage of the knowledge that
7682 the entire program fits into the low 64k of memory.
7685 @opindex mno-low-64k
7686 Assume that the program is arbitrarily large. This is the default.
7688 @item -mid-shared-library
7689 @opindex mid-shared-library
7690 Generate code that supports shared libraries via the library ID method.
7691 This allows for execute in place and shared libraries in an environment
7692 without virtual memory management. This option implies @option{-fPIC}.
7694 @item -mno-id-shared-library
7695 @opindex mno-id-shared-library
7696 Generate code that doesn't assume ID based shared libraries are being used.
7697 This is the default.
7699 @item -mshared-library-id=n
7700 @opindex mshared-library-id
7701 Specified the identification number of the ID based shared library being
7702 compiled. Specifying a value of 0 will generate more compact code, specifying
7703 other values will force the allocation of that number to the current
7704 library but is no more space or time efficient than omitting this option.
7707 @itemx -mno-long-calls
7708 @opindex mlong-calls
7709 @opindex mno-long-calls
7710 Tells the compiler to perform function calls by first loading the
7711 address of the function into a register and then performing a subroutine
7712 call on this register. This switch is needed if the target function
7713 will lie outside of the 24 bit addressing range of the offset based
7714 version of subroutine call instruction.
7716 This feature is not enabled by default. Specifying
7717 @option{-mno-long-calls} will restore the default behavior. Note these
7718 switches have no effect on how the compiler generates code to handle
7719 function calls via function pointers.
7723 @subsection CRIS Options
7724 @cindex CRIS Options
7726 These options are defined specifically for the CRIS ports.
7729 @item -march=@var{architecture-type}
7730 @itemx -mcpu=@var{architecture-type}
7733 Generate code for the specified architecture. The choices for
7734 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7735 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7736 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7739 @item -mtune=@var{architecture-type}
7741 Tune to @var{architecture-type} everything applicable about the generated
7742 code, except for the ABI and the set of available instructions. The
7743 choices for @var{architecture-type} are the same as for
7744 @option{-march=@var{architecture-type}}.
7746 @item -mmax-stack-frame=@var{n}
7747 @opindex mmax-stack-frame
7748 Warn when the stack frame of a function exceeds @var{n} bytes.
7750 @item -melinux-stacksize=@var{n}
7751 @opindex melinux-stacksize
7752 Only available with the @samp{cris-axis-aout} target. Arranges for
7753 indications in the program to the kernel loader that the stack of the
7754 program should be set to @var{n} bytes.
7760 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7761 @option{-march=v3} and @option{-march=v8} respectively.
7763 @item -mmul-bug-workaround
7764 @itemx -mno-mul-bug-workaround
7765 @opindex mmul-bug-workaround
7766 @opindex mno-mul-bug-workaround
7767 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7768 models where it applies. This option is active by default.
7772 Enable CRIS-specific verbose debug-related information in the assembly
7773 code. This option also has the effect to turn off the @samp{#NO_APP}
7774 formatted-code indicator to the assembler at the beginning of the
7779 Do not use condition-code results from previous instruction; always emit
7780 compare and test instructions before use of condition codes.
7782 @item -mno-side-effects
7783 @opindex mno-side-effects
7784 Do not emit instructions with side-effects in addressing modes other than
7788 @itemx -mno-stack-align
7790 @itemx -mno-data-align
7791 @itemx -mconst-align
7792 @itemx -mno-const-align
7793 @opindex mstack-align
7794 @opindex mno-stack-align
7795 @opindex mdata-align
7796 @opindex mno-data-align
7797 @opindex mconst-align
7798 @opindex mno-const-align
7799 These options (no-options) arranges (eliminate arrangements) for the
7800 stack-frame, individual data and constants to be aligned for the maximum
7801 single data access size for the chosen CPU model. The default is to
7802 arrange for 32-bit alignment. ABI details such as structure layout are
7803 not affected by these options.
7811 Similar to the stack- data- and const-align options above, these options
7812 arrange for stack-frame, writable data and constants to all be 32-bit,
7813 16-bit or 8-bit aligned. The default is 32-bit alignment.
7815 @item -mno-prologue-epilogue
7816 @itemx -mprologue-epilogue
7817 @opindex mno-prologue-epilogue
7818 @opindex mprologue-epilogue
7819 With @option{-mno-prologue-epilogue}, the normal function prologue and
7820 epilogue that sets up the stack-frame are omitted and no return
7821 instructions or return sequences are generated in the code. Use this
7822 option only together with visual inspection of the compiled code: no
7823 warnings or errors are generated when call-saved registers must be saved,
7824 or storage for local variable needs to be allocated.
7830 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7831 instruction sequences that load addresses for functions from the PLT part
7832 of the GOT rather than (traditional on other architectures) calls to the
7833 PLT@. The default is @option{-mgotplt}.
7837 Legacy no-op option only recognized with the cris-axis-aout target.
7841 Legacy no-op option only recognized with the cris-axis-elf and
7842 cris-axis-linux-gnu targets.
7846 Only recognized with the cris-axis-aout target, where it selects a
7847 GNU/linux-like multilib, include files and instruction set for
7852 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7856 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7857 to link with input-output functions from a simulator library. Code,
7858 initialized data and zero-initialized data are allocated consecutively.
7862 Like @option{-sim}, but pass linker options to locate initialized data at
7863 0x40000000 and zero-initialized data at 0x80000000.
7867 @subsection CRX Options
7870 These options are defined specifically for the CRX ports.
7876 Enable the use of multiply-accumulate instructions. Disabled by default.
7880 Push instructions will be used to pass outgoing arguments when functions
7881 are called. Enabled by default.
7884 @node Darwin Options
7885 @subsection Darwin Options
7886 @cindex Darwin options
7888 These options are defined for all architectures running the Darwin operating
7891 FSF GCC on Darwin does not create ``fat'' object files; it will create
7892 an object file for the single architecture that it was built to
7893 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7894 @option{-arch} options are used; it does so by running the compiler or
7895 linker multiple times and joining the results together with
7898 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7899 @samp{i686}) is determined by the flags that specify the ISA
7900 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7901 @option{-force_cpusubtype_ALL} option can be used to override this.
7903 The Darwin tools vary in their behavior when presented with an ISA
7904 mismatch. The assembler, @file{as}, will only permit instructions to
7905 be used that are valid for the subtype of the file it is generating,
7906 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7907 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7908 and print an error if asked to create a shared library with a less
7909 restrictive subtype than its input files (for instance, trying to put
7910 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7911 for executables, @file{ld}, will quietly give the executable the most
7912 restrictive subtype of any of its input files.
7917 Add the framework directory @var{dir} to the head of the list of
7918 directories to be searched for header files. These directories are
7919 interleaved with those specified by @option{-I} options and are
7920 scanned in a left-to-right order.
7922 A framework directory is a directory with frameworks in it. A
7923 framework is a directory with a @samp{"Headers"} and/or
7924 @samp{"PrivateHeaders"} directory contained directly in it that ends
7925 in @samp{".framework"}. The name of a framework is the name of this
7926 directory excluding the @samp{".framework"}. Headers associated with
7927 the framework are found in one of those two directories, with
7928 @samp{"Headers"} being searched first. A subframework is a framework
7929 directory that is in a framework's @samp{"Frameworks"} directory.
7930 Includes of subframework headers can only appear in a header of a
7931 framework that contains the subframework, or in a sibling subframework
7932 header. Two subframeworks are siblings if they occur in the same
7933 framework. A subframework should not have the same name as a
7934 framework, a warning will be issued if this is violated. Currently a
7935 subframework cannot have subframeworks, in the future, the mechanism
7936 may be extended to support this. The standard frameworks can be found
7937 in @samp{"/System/Library/Frameworks"} and
7938 @samp{"/Library/Frameworks"}. An example include looks like
7939 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7940 the name of the framework and header.h is found in the
7941 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7945 Emit debugging information for symbols that are used. For STABS
7946 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7947 This is by default ON@.
7951 Emit debugging information for all symbols and types.
7953 @item -mmacosx-version-min=@var{version}
7954 The earliest version of MacOS X that this executable will run on
7955 is @var{version}. Typical values of @var{version} include @code{10.1},
7956 @code{10.2}, and @code{10.3.9}.
7958 The default for this option is to make choices that seem to be most
7961 @item -mone-byte-bool
7962 @opindex -mone-byte-bool
7963 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7964 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7965 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7966 option has no effect on x86.
7968 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7969 to generate code that is not binary compatible with code generated
7970 without that switch. Using this switch may require recompiling all
7971 other modules in a program, including system libraries. Use this
7972 switch to conform to a non-default data model.
7974 @item -mfix-and-continue
7975 @itemx -ffix-and-continue
7976 @itemx -findirect-data
7977 @opindex mfix-and-continue
7978 @opindex ffix-and-continue
7979 @opindex findirect-data
7980 Generate code suitable for fast turn around development. Needed to
7981 enable gdb to dynamically load @code{.o} files into already running
7982 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7983 are provided for backwards compatibility.
7987 Loads all members of static archive libraries.
7988 See man ld(1) for more information.
7990 @item -arch_errors_fatal
7991 @opindex arch_errors_fatal
7992 Cause the errors having to do with files that have the wrong architecture
7996 @opindex bind_at_load
7997 Causes the output file to be marked such that the dynamic linker will
7998 bind all undefined references when the file is loaded or launched.
8002 Produce a Mach-o bundle format file.
8003 See man ld(1) for more information.
8005 @item -bundle_loader @var{executable}
8006 @opindex bundle_loader
8007 This option specifies the @var{executable} that will be loading the build
8008 output file being linked. See man ld(1) for more information.
8011 @opindex -dynamiclib
8012 When passed this option, GCC will produce a dynamic library instead of
8013 an executable when linking, using the Darwin @file{libtool} command.
8015 @item -force_cpusubtype_ALL
8016 @opindex -force_cpusubtype_ALL
8017 This causes GCC's output file to have the @var{ALL} subtype, instead of
8018 one controlled by the @option{-mcpu} or @option{-march} option.
8020 @item -allowable_client @var{client_name}
8022 @itemx -compatibility_version
8023 @itemx -current_version
8025 @itemx -dependency-file
8027 @itemx -dylinker_install_name
8029 @itemx -exported_symbols_list
8031 @itemx -flat_namespace
8032 @itemx -force_flat_namespace
8033 @itemx -headerpad_max_install_names
8036 @itemx -install_name
8037 @itemx -keep_private_externs
8038 @itemx -multi_module
8039 @itemx -multiply_defined
8040 @itemx -multiply_defined_unused
8042 @itemx -no_dead_strip_inits_and_terms
8043 @itemx -nofixprebinding
8046 @itemx -noseglinkedit
8047 @itemx -pagezero_size
8049 @itemx -prebind_all_twolevel_modules
8050 @itemx -private_bundle
8051 @itemx -read_only_relocs
8053 @itemx -sectobjectsymbols
8057 @itemx -sectobjectsymbols
8060 @itemx -segs_read_only_addr
8061 @itemx -segs_read_write_addr
8062 @itemx -seg_addr_table
8063 @itemx -seg_addr_table_filename
8066 @itemx -segs_read_only_addr
8067 @itemx -segs_read_write_addr
8068 @itemx -single_module
8071 @itemx -sub_umbrella
8072 @itemx -twolevel_namespace
8075 @itemx -unexported_symbols_list
8076 @itemx -weak_reference_mismatches
8079 @opindex allowable_client
8080 @opindex client_name
8081 @opindex compatibility_version
8082 @opindex current_version
8084 @opindex dependency-file
8086 @opindex dylinker_install_name
8088 @opindex exported_symbols_list
8090 @opindex flat_namespace
8091 @opindex force_flat_namespace
8092 @opindex headerpad_max_install_names
8095 @opindex install_name
8096 @opindex keep_private_externs
8097 @opindex multi_module
8098 @opindex multiply_defined
8099 @opindex multiply_defined_unused
8101 @opindex no_dead_strip_inits_and_terms
8102 @opindex nofixprebinding
8103 @opindex nomultidefs
8105 @opindex noseglinkedit
8106 @opindex pagezero_size
8108 @opindex prebind_all_twolevel_modules
8109 @opindex private_bundle
8110 @opindex read_only_relocs
8112 @opindex sectobjectsymbols
8116 @opindex sectobjectsymbols
8119 @opindex segs_read_only_addr
8120 @opindex segs_read_write_addr
8121 @opindex seg_addr_table
8122 @opindex seg_addr_table_filename
8123 @opindex seglinkedit
8125 @opindex segs_read_only_addr
8126 @opindex segs_read_write_addr
8127 @opindex single_module
8129 @opindex sub_library
8130 @opindex sub_umbrella
8131 @opindex twolevel_namespace
8134 @opindex unexported_symbols_list
8135 @opindex weak_reference_mismatches
8136 @opindex whatsloaded
8138 These options are passed to the Darwin linker. The Darwin linker man page
8139 describes them in detail.
8142 @node DEC Alpha Options
8143 @subsection DEC Alpha Options
8145 These @samp{-m} options are defined for the DEC Alpha implementations:
8148 @item -mno-soft-float
8150 @opindex mno-soft-float
8151 @opindex msoft-float
8152 Use (do not use) the hardware floating-point instructions for
8153 floating-point operations. When @option{-msoft-float} is specified,
8154 functions in @file{libgcc.a} will be used to perform floating-point
8155 operations. Unless they are replaced by routines that emulate the
8156 floating-point operations, or compiled in such a way as to call such
8157 emulations routines, these routines will issue floating-point
8158 operations. If you are compiling for an Alpha without floating-point
8159 operations, you must ensure that the library is built so as not to call
8162 Note that Alpha implementations without floating-point operations are
8163 required to have floating-point registers.
8168 @opindex mno-fp-regs
8169 Generate code that uses (does not use) the floating-point register set.
8170 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8171 register set is not used, floating point operands are passed in integer
8172 registers as if they were integers and floating-point results are passed
8173 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8174 so any function with a floating-point argument or return value called by code
8175 compiled with @option{-mno-fp-regs} must also be compiled with that
8178 A typical use of this option is building a kernel that does not use,
8179 and hence need not save and restore, any floating-point registers.
8183 The Alpha architecture implements floating-point hardware optimized for
8184 maximum performance. It is mostly compliant with the IEEE floating
8185 point standard. However, for full compliance, software assistance is
8186 required. This option generates code fully IEEE compliant code
8187 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8188 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8189 defined during compilation. The resulting code is less efficient but is
8190 able to correctly support denormalized numbers and exceptional IEEE
8191 values such as not-a-number and plus/minus infinity. Other Alpha
8192 compilers call this option @option{-ieee_with_no_inexact}.
8194 @item -mieee-with-inexact
8195 @opindex mieee-with-inexact
8196 This is like @option{-mieee} except the generated code also maintains
8197 the IEEE @var{inexact-flag}. Turning on this option causes the
8198 generated code to implement fully-compliant IEEE math. In addition to
8199 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8200 macro. On some Alpha implementations the resulting code may execute
8201 significantly slower than the code generated by default. Since there is
8202 very little code that depends on the @var{inexact-flag}, you should
8203 normally not specify this option. Other Alpha compilers call this
8204 option @option{-ieee_with_inexact}.
8206 @item -mfp-trap-mode=@var{trap-mode}
8207 @opindex mfp-trap-mode
8208 This option controls what floating-point related traps are enabled.
8209 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8210 The trap mode can be set to one of four values:
8214 This is the default (normal) setting. The only traps that are enabled
8215 are the ones that cannot be disabled in software (e.g., division by zero
8219 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8223 Like @samp{su}, but the instructions are marked to be safe for software
8224 completion (see Alpha architecture manual for details).
8227 Like @samp{su}, but inexact traps are enabled as well.
8230 @item -mfp-rounding-mode=@var{rounding-mode}
8231 @opindex mfp-rounding-mode
8232 Selects the IEEE rounding mode. Other Alpha compilers call this option
8233 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8238 Normal IEEE rounding mode. Floating point numbers are rounded towards
8239 the nearest machine number or towards the even machine number in case
8243 Round towards minus infinity.
8246 Chopped rounding mode. Floating point numbers are rounded towards zero.
8249 Dynamic rounding mode. A field in the floating point control register
8250 (@var{fpcr}, see Alpha architecture reference manual) controls the
8251 rounding mode in effect. The C library initializes this register for
8252 rounding towards plus infinity. Thus, unless your program modifies the
8253 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8256 @item -mtrap-precision=@var{trap-precision}
8257 @opindex mtrap-precision
8258 In the Alpha architecture, floating point traps are imprecise. This
8259 means without software assistance it is impossible to recover from a
8260 floating trap and program execution normally needs to be terminated.
8261 GCC can generate code that can assist operating system trap handlers
8262 in determining the exact location that caused a floating point trap.
8263 Depending on the requirements of an application, different levels of
8264 precisions can be selected:
8268 Program precision. This option is the default and means a trap handler
8269 can only identify which program caused a floating point exception.
8272 Function precision. The trap handler can determine the function that
8273 caused a floating point exception.
8276 Instruction precision. The trap handler can determine the exact
8277 instruction that caused a floating point exception.
8280 Other Alpha compilers provide the equivalent options called
8281 @option{-scope_safe} and @option{-resumption_safe}.
8283 @item -mieee-conformant
8284 @opindex mieee-conformant
8285 This option marks the generated code as IEEE conformant. You must not
8286 use this option unless you also specify @option{-mtrap-precision=i} and either
8287 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8288 is to emit the line @samp{.eflag 48} in the function prologue of the
8289 generated assembly file. Under DEC Unix, this has the effect that
8290 IEEE-conformant math library routines will be linked in.
8292 @item -mbuild-constants
8293 @opindex mbuild-constants
8294 Normally GCC examines a 32- or 64-bit integer constant to
8295 see if it can construct it from smaller constants in two or three
8296 instructions. If it cannot, it will output the constant as a literal and
8297 generate code to load it from the data segment at runtime.
8299 Use this option to require GCC to construct @emph{all} integer constants
8300 using code, even if it takes more instructions (the maximum is six).
8302 You would typically use this option to build a shared library dynamic
8303 loader. Itself a shared library, it must relocate itself in memory
8304 before it can find the variables and constants in its own data segment.
8310 Select whether to generate code to be assembled by the vendor-supplied
8311 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8329 Indicate whether GCC should generate code to use the optional BWX,
8330 CIX, FIX and MAX instruction sets. The default is to use the instruction
8331 sets supported by the CPU type specified via @option{-mcpu=} option or that
8332 of the CPU on which GCC was built if none was specified.
8337 @opindex mfloat-ieee
8338 Generate code that uses (does not use) VAX F and G floating point
8339 arithmetic instead of IEEE single and double precision.
8341 @item -mexplicit-relocs
8342 @itemx -mno-explicit-relocs
8343 @opindex mexplicit-relocs
8344 @opindex mno-explicit-relocs
8345 Older Alpha assemblers provided no way to generate symbol relocations
8346 except via assembler macros. Use of these macros does not allow
8347 optimal instruction scheduling. GNU binutils as of version 2.12
8348 supports a new syntax that allows the compiler to explicitly mark
8349 which relocations should apply to which instructions. This option
8350 is mostly useful for debugging, as GCC detects the capabilities of
8351 the assembler when it is built and sets the default accordingly.
8355 @opindex msmall-data
8356 @opindex mlarge-data
8357 When @option{-mexplicit-relocs} is in effect, static data is
8358 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8359 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8360 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8361 16-bit relocations off of the @code{$gp} register. This limits the
8362 size of the small data area to 64KB, but allows the variables to be
8363 directly accessed via a single instruction.
8365 The default is @option{-mlarge-data}. With this option the data area
8366 is limited to just below 2GB@. Programs that require more than 2GB of
8367 data must use @code{malloc} or @code{mmap} to allocate the data in the
8368 heap instead of in the program's data segment.
8370 When generating code for shared libraries, @option{-fpic} implies
8371 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8375 @opindex msmall-text
8376 @opindex mlarge-text
8377 When @option{-msmall-text} is used, the compiler assumes that the
8378 code of the entire program (or shared library) fits in 4MB, and is
8379 thus reachable with a branch instruction. When @option{-msmall-data}
8380 is used, the compiler can assume that all local symbols share the
8381 same @code{$gp} value, and thus reduce the number of instructions
8382 required for a function call from 4 to 1.
8384 The default is @option{-mlarge-text}.
8386 @item -mcpu=@var{cpu_type}
8388 Set the instruction set and instruction scheduling parameters for
8389 machine type @var{cpu_type}. You can specify either the @samp{EV}
8390 style name or the corresponding chip number. GCC supports scheduling
8391 parameters for the EV4, EV5 and EV6 family of processors and will
8392 choose the default values for the instruction set from the processor
8393 you specify. If you do not specify a processor type, GCC will default
8394 to the processor on which the compiler was built.
8396 Supported values for @var{cpu_type} are
8402 Schedules as an EV4 and has no instruction set extensions.
8406 Schedules as an EV5 and has no instruction set extensions.
8410 Schedules as an EV5 and supports the BWX extension.
8415 Schedules as an EV5 and supports the BWX and MAX extensions.
8419 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8423 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8426 @item -mtune=@var{cpu_type}
8428 Set only the instruction scheduling parameters for machine type
8429 @var{cpu_type}. The instruction set is not changed.
8431 @item -mmemory-latency=@var{time}
8432 @opindex mmemory-latency
8433 Sets the latency the scheduler should assume for typical memory
8434 references as seen by the application. This number is highly
8435 dependent on the memory access patterns used by the application
8436 and the size of the external cache on the machine.
8438 Valid options for @var{time} are
8442 A decimal number representing clock cycles.
8448 The compiler contains estimates of the number of clock cycles for
8449 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8450 (also called Dcache, Scache, and Bcache), as well as to main memory.
8451 Note that L3 is only valid for EV5.
8456 @node DEC Alpha/VMS Options
8457 @subsection DEC Alpha/VMS Options
8459 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8462 @item -mvms-return-codes
8463 @opindex mvms-return-codes
8464 Return VMS condition codes from main. The default is to return POSIX
8465 style condition (e.g.@ error) codes.
8469 @subsection FRV Options
8476 Only use the first 32 general purpose registers.
8481 Use all 64 general purpose registers.
8486 Use only the first 32 floating point registers.
8491 Use all 64 floating point registers
8494 @opindex mhard-float
8496 Use hardware instructions for floating point operations.
8499 @opindex msoft-float
8501 Use library routines for floating point operations.
8506 Dynamically allocate condition code registers.
8511 Do not try to dynamically allocate condition code registers, only
8512 use @code{icc0} and @code{fcc0}.
8517 Change ABI to use double word insns.
8522 Do not use double word instructions.
8527 Use floating point double instructions.
8532 Do not use floating point double instructions.
8537 Use media instructions.
8542 Do not use media instructions.
8547 Use multiply and add/subtract instructions.
8552 Do not use multiply and add/subtract instructions.
8557 Select the FDPIC ABI, that uses function descriptors to represent
8558 pointers to functions. Without any PIC/PIE-related options, it
8559 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8560 assumes GOT entries and small data are within a 12-bit range from the
8561 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8562 are computed with 32 bits.
8565 @opindex minline-plt
8567 Enable inlining of PLT entries in function calls to functions that are
8568 not known to bind locally. It has no effect without @option{-mfdpic}.
8569 It's enabled by default if optimizing for speed and compiling for
8570 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8571 optimization option such as @option{-O3} or above is present in the
8577 Assume a large TLS segment when generating thread-local code.
8582 Do not assume a large TLS segment when generating thread-local code.
8587 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8588 that is known to be in read-only sections. It's enabled by default,
8589 except for @option{-fpic} or @option{-fpie}: even though it may help
8590 make the global offset table smaller, it trades 1 instruction for 4.
8591 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8592 one of which may be shared by multiple symbols, and it avoids the need
8593 for a GOT entry for the referenced symbol, so it's more likely to be a
8594 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8596 @item -multilib-library-pic
8597 @opindex multilib-library-pic
8599 Link with the (library, not FD) pic libraries. It's implied by
8600 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8601 @option{-fpic} without @option{-mfdpic}. You should never have to use
8607 Follow the EABI requirement of always creating a frame pointer whenever
8608 a stack frame is allocated. This option is enabled by default and can
8609 be disabled with @option{-mno-linked-fp}.
8612 @opindex mlong-calls
8614 Use indirect addressing to call functions outside the current
8615 compilation unit. This allows the functions to be placed anywhere
8616 within the 32-bit address space.
8618 @item -malign-labels
8619 @opindex malign-labels
8621 Try to align labels to an 8-byte boundary by inserting nops into the
8622 previous packet. This option only has an effect when VLIW packing
8623 is enabled. It doesn't create new packets; it merely adds nops to
8627 @opindex mlibrary-pic
8629 Generate position-independent EABI code.
8634 Use only the first four media accumulator registers.
8639 Use all eight media accumulator registers.
8644 Pack VLIW instructions.
8649 Do not pack VLIW instructions.
8654 Do not mark ABI switches in e_flags.
8659 Enable the use of conditional-move instructions (default).
8661 This switch is mainly for debugging the compiler and will likely be removed
8662 in a future version.
8664 @item -mno-cond-move
8665 @opindex mno-cond-move
8667 Disable the use of conditional-move instructions.
8669 This switch is mainly for debugging the compiler and will likely be removed
8670 in a future version.
8675 Enable the use of conditional set instructions (default).
8677 This switch is mainly for debugging the compiler and will likely be removed
8678 in a future version.
8683 Disable the use of conditional set instructions.
8685 This switch is mainly for debugging the compiler and will likely be removed
8686 in a future version.
8691 Enable the use of conditional execution (default).
8693 This switch is mainly for debugging the compiler and will likely be removed
8694 in a future version.
8696 @item -mno-cond-exec
8697 @opindex mno-cond-exec
8699 Disable the use of conditional execution.
8701 This switch is mainly for debugging the compiler and will likely be removed
8702 in a future version.
8705 @opindex mvliw-branch
8707 Run a pass to pack branches into VLIW instructions (default).
8709 This switch is mainly for debugging the compiler and will likely be removed
8710 in a future version.
8712 @item -mno-vliw-branch
8713 @opindex mno-vliw-branch
8715 Do not run a pass to pack branches into VLIW instructions.
8717 This switch is mainly for debugging the compiler and will likely be removed
8718 in a future version.
8720 @item -mmulti-cond-exec
8721 @opindex mmulti-cond-exec
8723 Enable optimization of @code{&&} and @code{||} in conditional execution
8726 This switch is mainly for debugging the compiler and will likely be removed
8727 in a future version.
8729 @item -mno-multi-cond-exec
8730 @opindex mno-multi-cond-exec
8732 Disable optimization of @code{&&} and @code{||} in conditional execution.
8734 This switch is mainly for debugging the compiler and will likely be removed
8735 in a future version.
8737 @item -mnested-cond-exec
8738 @opindex mnested-cond-exec
8740 Enable nested conditional execution optimizations (default).
8742 This switch is mainly for debugging the compiler and will likely be removed
8743 in a future version.
8745 @item -mno-nested-cond-exec
8746 @opindex mno-nested-cond-exec
8748 Disable nested conditional execution optimizations.
8750 This switch is mainly for debugging the compiler and will likely be removed
8751 in a future version.
8753 @item -moptimize-membar
8754 @opindex moptimize-membar
8756 This switch removes redundant @code{membar} instructions from the
8757 compiler generated code. It is enabled by default.
8759 @item -mno-optimize-membar
8760 @opindex mno-optimize-membar
8762 This switch disables the automatic removal of redundant @code{membar}
8763 instructions from the generated code.
8765 @item -mtomcat-stats
8766 @opindex mtomcat-stats
8768 Cause gas to print out tomcat statistics.
8770 @item -mcpu=@var{cpu}
8773 Select the processor type for which to generate code. Possible values are
8774 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8775 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8779 @node H8/300 Options
8780 @subsection H8/300 Options
8782 These @samp{-m} options are defined for the H8/300 implementations:
8787 Shorten some address references at link time, when possible; uses the
8788 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8789 ld, Using ld}, for a fuller description.
8793 Generate code for the H8/300H@.
8797 Generate code for the H8S@.
8801 Generate code for the H8S and H8/300H in the normal mode. This switch
8802 must be used either with @option{-mh} or @option{-ms}.
8806 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8810 Make @code{int} data 32 bits by default.
8814 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8815 The default for the H8/300H and H8S is to align longs and floats on 4
8817 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8818 This option has no effect on the H8/300.
8822 @subsection HPPA Options
8823 @cindex HPPA Options
8825 These @samp{-m} options are defined for the HPPA family of computers:
8828 @item -march=@var{architecture-type}
8830 Generate code for the specified architecture. The choices for
8831 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8832 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8833 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8834 architecture option for your machine. Code compiled for lower numbered
8835 architectures will run on higher numbered architectures, but not the
8839 @itemx -mpa-risc-1-1
8840 @itemx -mpa-risc-2-0
8841 @opindex mpa-risc-1-0
8842 @opindex mpa-risc-1-1
8843 @opindex mpa-risc-2-0
8844 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8847 @opindex mbig-switch
8848 Generate code suitable for big switch tables. Use this option only if
8849 the assembler/linker complain about out of range branches within a switch
8852 @item -mjump-in-delay
8853 @opindex mjump-in-delay
8854 Fill delay slots of function calls with unconditional jump instructions
8855 by modifying the return pointer for the function call to be the target
8856 of the conditional jump.
8858 @item -mdisable-fpregs
8859 @opindex mdisable-fpregs
8860 Prevent floating point registers from being used in any manner. This is
8861 necessary for compiling kernels which perform lazy context switching of
8862 floating point registers. If you use this option and attempt to perform
8863 floating point operations, the compiler will abort.
8865 @item -mdisable-indexing
8866 @opindex mdisable-indexing
8867 Prevent the compiler from using indexing address modes. This avoids some
8868 rather obscure problems when compiling MIG generated code under MACH@.
8870 @item -mno-space-regs
8871 @opindex mno-space-regs
8872 Generate code that assumes the target has no space registers. This allows
8873 GCC to generate faster indirect calls and use unscaled index address modes.
8875 Such code is suitable for level 0 PA systems and kernels.
8877 @item -mfast-indirect-calls
8878 @opindex mfast-indirect-calls
8879 Generate code that assumes calls never cross space boundaries. This
8880 allows GCC to emit code which performs faster indirect calls.
8882 This option will not work in the presence of shared libraries or nested
8885 @item -mfixed-range=@var{register-range}
8886 @opindex mfixed-range
8887 Generate code treating the given register range as fixed registers.
8888 A fixed register is one that the register allocator can not use. This is
8889 useful when compiling kernel code. A register range is specified as
8890 two registers separated by a dash. Multiple register ranges can be
8891 specified separated by a comma.
8893 @item -mlong-load-store
8894 @opindex mlong-load-store
8895 Generate 3-instruction load and store sequences as sometimes required by
8896 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8899 @item -mportable-runtime
8900 @opindex mportable-runtime
8901 Use the portable calling conventions proposed by HP for ELF systems.
8905 Enable the use of assembler directives only GAS understands.
8907 @item -mschedule=@var{cpu-type}
8909 Schedule code according to the constraints for the machine type
8910 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8911 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8912 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8913 proper scheduling option for your machine. The default scheduling is
8917 @opindex mlinker-opt
8918 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8919 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8920 linkers in which they give bogus error messages when linking some programs.
8923 @opindex msoft-float
8924 Generate output containing library calls for floating point.
8925 @strong{Warning:} the requisite libraries are not available for all HPPA
8926 targets. Normally the facilities of the machine's usual C compiler are
8927 used, but this cannot be done directly in cross-compilation. You must make
8928 your own arrangements to provide suitable library functions for
8929 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8930 does provide software floating point support.
8932 @option{-msoft-float} changes the calling convention in the output file;
8933 therefore, it is only useful if you compile @emph{all} of a program with
8934 this option. In particular, you need to compile @file{libgcc.a}, the
8935 library that comes with GCC, with @option{-msoft-float} in order for
8940 Generate the predefine, @code{_SIO}, for server IO@. The default is
8941 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8942 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
8943 options are available under HP-UX and HI-UX@.
8947 Use GNU ld specific options. This passes @option{-shared} to ld when
8948 building a shared library. It is the default when GCC is configured,
8949 explicitly or implicitly, with the GNU linker. This option does not
8950 have any affect on which ld is called, it only changes what parameters
8951 are passed to that ld. The ld that is called is determined by the
8952 @option{--with-ld} configure option, GCC's program search path, and
8953 finally by the user's @env{PATH}. The linker used by GCC can be printed
8954 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
8955 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
8959 Use HP ld specific options. This passes @option{-b} to ld when building
8960 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8961 links. It is the default when GCC is configured, explicitly or
8962 implicitly, with the HP linker. This option does not have any affect on
8963 which ld is called, it only changes what parameters are passed to that
8964 ld. The ld that is called is determined by the @option{--with-ld}
8965 configure option, GCC's program search path, and finally by the user's
8966 @env{PATH}. The linker used by GCC can be printed using @samp{which
8967 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
8968 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
8971 @opindex mno-long-calls
8972 Generate code that uses long call sequences. This ensures that a call
8973 is always able to reach linker generated stubs. The default is to generate
8974 long calls only when the distance from the call site to the beginning
8975 of the function or translation unit, as the case may be, exceeds a
8976 predefined limit set by the branch type being used. The limits for
8977 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8978 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8981 Distances are measured from the beginning of functions when using the
8982 @option{-ffunction-sections} option, or when using the @option{-mgas}
8983 and @option{-mno-portable-runtime} options together under HP-UX with
8986 It is normally not desirable to use this option as it will degrade
8987 performance. However, it may be useful in large applications,
8988 particularly when partial linking is used to build the application.
8990 The types of long calls used depends on the capabilities of the
8991 assembler and linker, and the type of code being generated. The
8992 impact on systems that support long absolute calls, and long pic
8993 symbol-difference or pc-relative calls should be relatively small.
8994 However, an indirect call is used on 32-bit ELF systems in pic code
8995 and it is quite long.
8997 @item -munix=@var{unix-std}
8999 Generate compiler predefines and select a startfile for the specified
9000 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9001 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9002 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9003 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9004 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9007 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9008 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9009 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9010 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9011 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9012 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9014 It is @emph{important} to note that this option changes the interfaces
9015 for various library routines. It also affects the operational behavior
9016 of the C library. Thus, @emph{extreme} care is needed in using this
9019 Library code that is intended to operate with more than one UNIX
9020 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9021 as appropriate. Most GNU software doesn't provide this capability.
9025 Suppress the generation of link options to search libdld.sl when the
9026 @option{-static} option is specified on HP-UX 10 and later.
9030 The HP-UX implementation of setlocale in libc has a dependency on
9031 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9032 when the @option{-static} option is specified, special link options
9033 are needed to resolve this dependency.
9035 On HP-UX 10 and later, the GCC driver adds the necessary options to
9036 link with libdld.sl when the @option{-static} option is specified.
9037 This causes the resulting binary to be dynamic. On the 64-bit port,
9038 the linkers generate dynamic binaries by default in any case. The
9039 @option{-nolibdld} option can be used to prevent the GCC driver from
9040 adding these link options.
9044 Add support for multithreading with the @dfn{dce thread} library
9045 under HP-UX@. This option sets flags for both the preprocessor and
9049 @node i386 and x86-64 Options
9050 @subsection Intel 386 and AMD x86-64 Options
9051 @cindex i386 Options
9052 @cindex x86-64 Options
9053 @cindex Intel 386 Options
9054 @cindex AMD x86-64 Options
9056 These @samp{-m} options are defined for the i386 and x86-64 family of
9060 @item -mtune=@var{cpu-type}
9062 Tune to @var{cpu-type} everything applicable about the generated code, except
9063 for the ABI and the set of available instructions. The choices for
9067 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9068 If you know the CPU on which your code will run, then you should use
9069 the corresponding @option{-mtune} option instead of
9070 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9071 of your application will have, then you should use this option.
9073 As new processors are deployed in the marketplace, the behavior of this
9074 option will change. Therefore, if you upgrade to a newer version of
9075 GCC, the code generated option will change to reflect the processors
9076 that were most common when that version of GCC was released.
9078 There is no @option{-march=generic} option because @option{-march}
9079 indicates the instruction set the compiler can use, and there is no
9080 generic instruction set applicable to all processors. In contrast,
9081 @option{-mtune} indicates the processor (or, in this case, collection of
9082 processors) for which the code is optimized.
9084 Original Intel's i386 CPU@.
9086 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9088 Intel Pentium CPU with no MMX support.
9090 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9092 Intel PentiumPro CPU@.
9094 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9095 instruction set will be used, so the code will run on all i686 familly chips.
9097 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9098 @item pentium3, pentium3m
9099 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9102 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9103 support. Used by Centrino notebooks.
9104 @item pentium4, pentium4m
9105 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9107 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9110 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9111 SSE2 and SSE3 instruction set support.
9113 AMD K6 CPU with MMX instruction set support.
9115 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9116 @item athlon, athlon-tbird
9117 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9119 @item athlon-4, athlon-xp, athlon-mp
9120 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9121 instruction set support.
9122 @item k8, opteron, athlon64, athlon-fx
9123 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9124 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9126 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9129 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9130 instruction set support.
9132 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9133 implemented for this chip.)
9135 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9136 implemented for this chip.)
9139 While picking a specific @var{cpu-type} will schedule things appropriately
9140 for that particular chip, the compiler will not generate any code that
9141 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9144 @item -march=@var{cpu-type}
9146 Generate instructions for the machine type @var{cpu-type}. The choices
9147 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9148 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9150 @item -mcpu=@var{cpu-type}
9152 A deprecated synonym for @option{-mtune}.
9161 @opindex mpentiumpro
9162 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9163 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9164 These synonyms are deprecated.
9166 @item -mfpmath=@var{unit}
9168 Generate floating point arithmetics for selected unit @var{unit}. The choices
9173 Use the standard 387 floating point coprocessor present majority of chips and
9174 emulated otherwise. Code compiled with this option will run almost everywhere.
9175 The temporary results are computed in 80bit precision instead of precision
9176 specified by the type resulting in slightly different results compared to most
9177 of other chips. See @option{-ffloat-store} for more detailed description.
9179 This is the default choice for i386 compiler.
9182 Use scalar floating point instructions present in the SSE instruction set.
9183 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9184 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9185 instruction set supports only single precision arithmetics, thus the double and
9186 extended precision arithmetics is still done using 387. Later version, present
9187 only in Pentium4 and the future AMD x86-64 chips supports double precision
9190 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9191 or @option{-msse2} switches to enable SSE extensions and make this option
9192 effective. For the x86-64 compiler, these extensions are enabled by default.
9194 The resulting code should be considerably faster in the majority of cases and avoid
9195 the numerical instability problems of 387 code, but may break some existing
9196 code that expects temporaries to be 80bit.
9198 This is the default choice for the x86-64 compiler.
9201 Attempt to utilize both instruction sets at once. This effectively double the
9202 amount of available registers and on chips with separate execution units for
9203 387 and SSE the execution resources too. Use this option with care, as it is
9204 still experimental, because the GCC register allocator does not model separate
9205 functional units well resulting in instable performance.
9208 @item -masm=@var{dialect}
9209 @opindex masm=@var{dialect}
9210 Output asm instructions using selected @var{dialect}. Supported
9211 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9212 not support @samp{intel}.
9217 @opindex mno-ieee-fp
9218 Control whether or not the compiler uses IEEE floating point
9219 comparisons. These handle correctly the case where the result of a
9220 comparison is unordered.
9223 @opindex msoft-float
9224 Generate output containing library calls for floating point.
9225 @strong{Warning:} the requisite libraries are not part of GCC@.
9226 Normally the facilities of the machine's usual C compiler are used, but
9227 this can't be done directly in cross-compilation. You must make your
9228 own arrangements to provide suitable library functions for
9231 On machines where a function returns floating point results in the 80387
9232 register stack, some floating point opcodes may be emitted even if
9233 @option{-msoft-float} is used.
9235 @item -mno-fp-ret-in-387
9236 @opindex mno-fp-ret-in-387
9237 Do not use the FPU registers for return values of functions.
9239 The usual calling convention has functions return values of types
9240 @code{float} and @code{double} in an FPU register, even if there
9241 is no FPU@. The idea is that the operating system should emulate
9244 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9245 in ordinary CPU registers instead.
9247 @item -mno-fancy-math-387
9248 @opindex mno-fancy-math-387
9249 Some 387 emulators do not support the @code{sin}, @code{cos} and
9250 @code{sqrt} instructions for the 387. Specify this option to avoid
9251 generating those instructions. This option is the default on FreeBSD,
9252 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9253 indicates that the target cpu will always have an FPU and so the
9254 instruction will not need emulation. As of revision 2.6.1, these
9255 instructions are not generated unless you also use the
9256 @option{-funsafe-math-optimizations} switch.
9258 @item -malign-double
9259 @itemx -mno-align-double
9260 @opindex malign-double
9261 @opindex mno-align-double
9262 Control whether GCC aligns @code{double}, @code{long double}, and
9263 @code{long long} variables on a two word boundary or a one word
9264 boundary. Aligning @code{double} variables on a two word boundary will
9265 produce code that runs somewhat faster on a @samp{Pentium} at the
9266 expense of more memory.
9268 @strong{Warning:} if you use the @option{-malign-double} switch,
9269 structures containing the above types will be aligned differently than
9270 the published application binary interface specifications for the 386
9271 and will not be binary compatible with structures in code compiled
9272 without that switch.
9274 @item -m96bit-long-double
9275 @itemx -m128bit-long-double
9276 @opindex m96bit-long-double
9277 @opindex m128bit-long-double
9278 These switches control the size of @code{long double} type. The i386
9279 application binary interface specifies the size to be 96 bits,
9280 so @option{-m96bit-long-double} is the default in 32 bit mode.
9282 Modern architectures (Pentium and newer) would prefer @code{long double}
9283 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9284 conforming to the ABI, this would not be possible. So specifying a
9285 @option{-m128bit-long-double} will align @code{long double}
9286 to a 16 byte boundary by padding the @code{long double} with an additional
9289 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9290 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9292 Notice that neither of these options enable any extra precision over the x87
9293 standard of 80 bits for a @code{long double}.
9295 @strong{Warning:} if you override the default value for your target ABI, the
9296 structures and arrays containing @code{long double} variables will change
9297 their size as well as function calling convention for function taking
9298 @code{long double} will be modified. Hence they will not be binary
9299 compatible with arrays or structures in code compiled without that switch.
9301 @item -mmlarge-data-threshold=@var{number}
9302 @opindex mlarge-data-threshold=@var{number}
9303 When @option{-mcmodel=medium} is specified, the data greater than
9304 @var{threshold} are placed in large data section. This value must be the
9305 same across all object linked into the binary and defaults to 65535.
9308 @itemx -mno-svr3-shlib
9309 @opindex msvr3-shlib
9310 @opindex mno-svr3-shlib
9311 Control whether GCC places uninitialized local variables into the
9312 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9313 into @code{bss}. These options are meaningful only on System V Release 3.
9317 Use a different function-calling convention, in which functions that
9318 take a fixed number of arguments return with the @code{ret} @var{num}
9319 instruction, which pops their arguments while returning. This saves one
9320 instruction in the caller since there is no need to pop the arguments
9323 You can specify that an individual function is called with this calling
9324 sequence with the function attribute @samp{stdcall}. You can also
9325 override the @option{-mrtd} option by using the function attribute
9326 @samp{cdecl}. @xref{Function Attributes}.
9328 @strong{Warning:} this calling convention is incompatible with the one
9329 normally used on Unix, so you cannot use it if you need to call
9330 libraries compiled with the Unix compiler.
9332 Also, you must provide function prototypes for all functions that
9333 take variable numbers of arguments (including @code{printf});
9334 otherwise incorrect code will be generated for calls to those
9337 In addition, seriously incorrect code will result if you call a
9338 function with too many arguments. (Normally, extra arguments are
9339 harmlessly ignored.)
9341 @item -mregparm=@var{num}
9343 Control how many registers are used to pass integer arguments. By
9344 default, no registers are used to pass arguments, and at most 3
9345 registers can be used. You can control this behavior for a specific
9346 function by using the function attribute @samp{regparm}.
9347 @xref{Function Attributes}.
9349 @strong{Warning:} if you use this switch, and
9350 @var{num} is nonzero, then you must build all modules with the same
9351 value, including any libraries. This includes the system libraries and
9355 @opindex msseregparm
9356 Use SSE register passing conventions for float and double arguments
9357 and return values. You can control this behavior for a specific
9358 function by using the function attribute @samp{sseregparm}.
9359 @xref{Function Attributes}.
9361 @strong{Warning:} if you use this switch then you must build all
9362 modules with the same value, including any libraries. This includes
9363 the system libraries and startup modules.
9365 @item -mpreferred-stack-boundary=@var{num}
9366 @opindex mpreferred-stack-boundary
9367 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9368 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9369 the default is 4 (16 bytes or 128 bits), except when optimizing for code
9370 size (@option{-Os}), in which case the default is the minimum correct
9371 alignment (4 bytes for x86, and 8 bytes for x86-64).
9373 On Pentium and PentiumPro, @code{double} and @code{long double} values
9374 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9375 suffer significant run time performance penalties. On Pentium III, the
9376 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
9377 penalties if it is not 16 byte aligned.
9379 To ensure proper alignment of this values on the stack, the stack boundary
9380 must be as aligned as that required by any value stored on the stack.
9381 Further, every function must be generated such that it keeps the stack
9382 aligned. Thus calling a function compiled with a higher preferred
9383 stack boundary from a function compiled with a lower preferred stack
9384 boundary will most likely misalign the stack. It is recommended that
9385 libraries that use callbacks always use the default setting.
9387 This extra alignment does consume extra stack space, and generally
9388 increases code size. Code that is sensitive to stack space usage, such
9389 as embedded systems and operating system kernels, may want to reduce the
9390 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9408 These switches enable or disable the use of instructions in the MMX,
9409 SSE, SSE2 or 3DNow! extended instruction sets. These extensions are
9410 also available as built-in functions: see @ref{X86 Built-in Functions},
9411 for details of the functions enabled and disabled by these switches.
9413 To have SSE/SSE2 instructions generated automatically from floating-point
9414 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9416 These options will enable GCC to use these extended instructions in
9417 generated code, even without @option{-mfpmath=sse}. Applications which
9418 perform runtime CPU detection must compile separate files for each
9419 supported architecture, using the appropriate flags. In particular,
9420 the file containing the CPU detection code should be compiled without
9424 @itemx -mno-push-args
9426 @opindex mno-push-args
9427 Use PUSH operations to store outgoing parameters. This method is shorter
9428 and usually equally fast as method using SUB/MOV operations and is enabled
9429 by default. In some cases disabling it may improve performance because of
9430 improved scheduling and reduced dependencies.
9432 @item -maccumulate-outgoing-args
9433 @opindex maccumulate-outgoing-args
9434 If enabled, the maximum amount of space required for outgoing arguments will be
9435 computed in the function prologue. This is faster on most modern CPUs
9436 because of reduced dependencies, improved scheduling and reduced stack usage
9437 when preferred stack boundary is not equal to 2. The drawback is a notable
9438 increase in code size. This switch implies @option{-mno-push-args}.
9442 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9443 on thread-safe exception handling must compile and link all code with the
9444 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9445 @option{-D_MT}; when linking, it links in a special thread helper library
9446 @option{-lmingwthrd} which cleans up per thread exception handling data.
9448 @item -mno-align-stringops
9449 @opindex mno-align-stringops
9450 Do not align destination of inlined string operations. This switch reduces
9451 code size and improves performance in case the destination is already aligned,
9452 but GCC doesn't know about it.
9454 @item -minline-all-stringops
9455 @opindex minline-all-stringops
9456 By default GCC inlines string operations only when destination is known to be
9457 aligned at least to 4 byte boundary. This enables more inlining, increase code
9458 size, but may improve performance of code that depends on fast memcpy, strlen
9459 and memset for short lengths.
9461 @item -momit-leaf-frame-pointer
9462 @opindex momit-leaf-frame-pointer
9463 Don't keep the frame pointer in a register for leaf functions. This
9464 avoids the instructions to save, set up and restore frame pointers and
9465 makes an extra register available in leaf functions. The option
9466 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9467 which might make debugging harder.
9469 @item -mtls-direct-seg-refs
9470 @itemx -mno-tls-direct-seg-refs
9471 @opindex mtls-direct-seg-refs
9472 Controls whether TLS variables may be accessed with offsets from the
9473 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9474 or whether the thread base pointer must be added. Whether or not this
9475 is legal depends on the operating system, and whether it maps the
9476 segment to cover the entire TLS area.
9478 For systems that use GNU libc, the default is on.
9481 These @samp{-m} switches are supported in addition to the above
9482 on AMD x86-64 processors in 64-bit environments.
9489 Generate code for a 32-bit or 64-bit environment.
9490 The 32-bit environment sets int, long and pointer to 32 bits and
9491 generates code that runs on any i386 system.
9492 The 64-bit environment sets int to 32 bits and long and pointer
9493 to 64 bits and generates code for AMD's x86-64 architecture.
9496 @opindex no-red-zone
9497 Do not use a so called red zone for x86-64 code. The red zone is mandated
9498 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9499 stack pointer that will not be modified by signal or interrupt handlers
9500 and therefore can be used for temporary data without adjusting the stack
9501 pointer. The flag @option{-mno-red-zone} disables this red zone.
9503 @item -mcmodel=small
9504 @opindex mcmodel=small
9505 Generate code for the small code model: the program and its symbols must
9506 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9507 Programs can be statically or dynamically linked. This is the default
9510 @item -mcmodel=kernel
9511 @opindex mcmodel=kernel
9512 Generate code for the kernel code model. The kernel runs in the
9513 negative 2 GB of the address space.
9514 This model has to be used for Linux kernel code.
9516 @item -mcmodel=medium
9517 @opindex mcmodel=medium
9518 Generate code for the medium model: The program is linked in the lower 2
9519 GB of the address space but symbols can be located anywhere in the
9520 address space. Programs can be statically or dynamically linked, but
9521 building of shared libraries are not supported with the medium model.
9523 @item -mcmodel=large
9524 @opindex mcmodel=large
9525 Generate code for the large model: This model makes no assumptions
9526 about addresses and sizes of sections. Currently GCC does not implement
9531 @subsection IA-64 Options
9532 @cindex IA-64 Options
9534 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9538 @opindex mbig-endian
9539 Generate code for a big endian target. This is the default for HP-UX@.
9541 @item -mlittle-endian
9542 @opindex mlittle-endian
9543 Generate code for a little endian target. This is the default for AIX5
9550 Generate (or don't) code for the GNU assembler. This is the default.
9551 @c Also, this is the default if the configure option @option{--with-gnu-as}
9558 Generate (or don't) code for the GNU linker. This is the default.
9559 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9564 Generate code that does not use a global pointer register. The result
9565 is not position independent code, and violates the IA-64 ABI@.
9567 @item -mvolatile-asm-stop
9568 @itemx -mno-volatile-asm-stop
9569 @opindex mvolatile-asm-stop
9570 @opindex mno-volatile-asm-stop
9571 Generate (or don't) a stop bit immediately before and after volatile asm
9574 @item -mregister-names
9575 @itemx -mno-register-names
9576 @opindex mregister-names
9577 @opindex mno-register-names
9578 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9579 the stacked registers. This may make assembler output more readable.
9585 Disable (or enable) optimizations that use the small data section. This may
9586 be useful for working around optimizer bugs.
9589 @opindex mconstant-gp
9590 Generate code that uses a single constant global pointer value. This is
9591 useful when compiling kernel code.
9595 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9596 This is useful when compiling firmware code.
9598 @item -minline-float-divide-min-latency
9599 @opindex minline-float-divide-min-latency
9600 Generate code for inline divides of floating point values
9601 using the minimum latency algorithm.
9603 @item -minline-float-divide-max-throughput
9604 @opindex minline-float-divide-max-throughput
9605 Generate code for inline divides of floating point values
9606 using the maximum throughput algorithm.
9608 @item -minline-int-divide-min-latency
9609 @opindex minline-int-divide-min-latency
9610 Generate code for inline divides of integer values
9611 using the minimum latency algorithm.
9613 @item -minline-int-divide-max-throughput
9614 @opindex minline-int-divide-max-throughput
9615 Generate code for inline divides of integer values
9616 using the maximum throughput algorithm.
9618 @item -minline-sqrt-min-latency
9619 @opindex minline-sqrt-min-latency
9620 Generate code for inline square roots
9621 using the minimum latency algorithm.
9623 @item -minline-sqrt-max-throughput
9624 @opindex minline-sqrt-max-throughput
9625 Generate code for inline square roots
9626 using the maximum throughput algorithm.
9628 @item -mno-dwarf2-asm
9630 @opindex mno-dwarf2-asm
9631 @opindex mdwarf2-asm
9632 Don't (or do) generate assembler code for the DWARF2 line number debugging
9633 info. This may be useful when not using the GNU assembler.
9635 @item -mearly-stop-bits
9636 @itemx -mno-early-stop-bits
9637 @opindex mearly-stop-bits
9638 @opindex mno-early-stop-bits
9639 Allow stop bits to be placed earlier than immediately preceding the
9640 instruction that triggered the stop bit. This can improve instruction
9641 scheduling, but does not always do so.
9643 @item -mfixed-range=@var{register-range}
9644 @opindex mfixed-range
9645 Generate code treating the given register range as fixed registers.
9646 A fixed register is one that the register allocator can not use. This is
9647 useful when compiling kernel code. A register range is specified as
9648 two registers separated by a dash. Multiple register ranges can be
9649 specified separated by a comma.
9651 @item -mtls-size=@var{tls-size}
9653 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9656 @item -mtune=@var{cpu-type}
9658 Tune the instruction scheduling for a particular CPU, Valid values are
9659 itanium, itanium1, merced, itanium2, and mckinley.
9665 Add support for multithreading using the POSIX threads library. This
9666 option sets flags for both the preprocessor and linker. It does
9667 not affect the thread safety of object code produced by the compiler or
9668 that of libraries supplied with it. These are HP-UX specific flags.
9674 Generate code for a 32-bit or 64-bit environment.
9675 The 32-bit environment sets int, long and pointer to 32 bits.
9676 The 64-bit environment sets int to 32 bits and long and pointer
9677 to 64 bits. These are HP-UX specific flags.
9682 @subsection M32C Options
9683 @cindex M32C options
9686 @item -mcpu=@var{name}
9688 Select the CPU for which code is generated. @var{name} may be one of
9689 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9690 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9695 Specifies that the program will be run on the simulator. This causes
9696 an alternate runtime library to be linked in which supports, for
9697 example, file I/O. You must not use this option when generating
9698 programs that will run on real hardware; you must provide your own
9699 runtime library for whatever I/O functions are needed.
9701 @item -memregs=@var{number}
9703 Specifies the number of memory-based pseudo-registers GCC will use
9704 during code generation. These pseudo-registers will be used like real
9705 registers, so there is a tradeoff between GCC's ability to fit the
9706 code into available registers, and the performance penalty of using
9707 memory instead of registers. Note that all modules in a program must
9708 be compiled with the same value for this option. Because of that, you
9709 must not use this option with the default runtime libraries gcc
9714 @node M32R/D Options
9715 @subsection M32R/D Options
9716 @cindex M32R/D options
9718 These @option{-m} options are defined for Renesas M32R/D architectures:
9723 Generate code for the M32R/2@.
9727 Generate code for the M32R/X@.
9731 Generate code for the M32R@. This is the default.
9734 @opindex mmodel=small
9735 Assume all objects live in the lower 16MB of memory (so that their addresses
9736 can be loaded with the @code{ld24} instruction), and assume all subroutines
9737 are reachable with the @code{bl} instruction.
9738 This is the default.
9740 The addressability of a particular object can be set with the
9741 @code{model} attribute.
9743 @item -mmodel=medium
9744 @opindex mmodel=medium
9745 Assume objects may be anywhere in the 32-bit address space (the compiler
9746 will generate @code{seth/add3} instructions to load their addresses), and
9747 assume all subroutines are reachable with the @code{bl} instruction.
9750 @opindex mmodel=large
9751 Assume objects may be anywhere in the 32-bit address space (the compiler
9752 will generate @code{seth/add3} instructions to load their addresses), and
9753 assume subroutines may not be reachable with the @code{bl} instruction
9754 (the compiler will generate the much slower @code{seth/add3/jl}
9755 instruction sequence).
9758 @opindex msdata=none
9759 Disable use of the small data area. Variables will be put into
9760 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
9761 @code{section} attribute has been specified).
9762 This is the default.
9764 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
9765 Objects may be explicitly put in the small data area with the
9766 @code{section} attribute using one of these sections.
9769 @opindex msdata=sdata
9770 Put small global and static data in the small data area, but do not
9771 generate special code to reference them.
9775 Put small global and static data in the small data area, and generate
9776 special instructions to reference them.
9780 @cindex smaller data references
9781 Put global and static objects less than or equal to @var{num} bytes
9782 into the small data or bss sections instead of the normal data or bss
9783 sections. The default value of @var{num} is 8.
9784 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
9785 for this option to have any effect.
9787 All modules should be compiled with the same @option{-G @var{num}} value.
9788 Compiling with different values of @var{num} may or may not work; if it
9789 doesn't the linker will give an error message---incorrect code will not be
9794 Makes the M32R specific code in the compiler display some statistics
9795 that might help in debugging programs.
9798 @opindex malign-loops
9799 Align all loops to a 32-byte boundary.
9801 @item -mno-align-loops
9802 @opindex mno-align-loops
9803 Do not enforce a 32-byte alignment for loops. This is the default.
9805 @item -missue-rate=@var{number}
9806 @opindex missue-rate=@var{number}
9807 Issue @var{number} instructions per cycle. @var{number} can only be 1
9810 @item -mbranch-cost=@var{number}
9811 @opindex mbranch-cost=@var{number}
9812 @var{number} can only be 1 or 2. If it is 1 then branches will be
9813 preferred over conditional code, if it is 2, then the opposite will
9816 @item -mflush-trap=@var{number}
9817 @opindex mflush-trap=@var{number}
9818 Specifies the trap number to use to flush the cache. The default is
9819 12. Valid numbers are between 0 and 15 inclusive.
9821 @item -mno-flush-trap
9822 @opindex mno-flush-trap
9823 Specifies that the cache cannot be flushed by using a trap.
9825 @item -mflush-func=@var{name}
9826 @opindex mflush-func=@var{name}
9827 Specifies the name of the operating system function to call to flush
9828 the cache. The default is @emph{_flush_cache}, but a function call
9829 will only be used if a trap is not available.
9831 @item -mno-flush-func
9832 @opindex mno-flush-func
9833 Indicates that there is no OS function for flushing the cache.
9837 @node M680x0 Options
9838 @subsection M680x0 Options
9839 @cindex M680x0 options
9841 These are the @samp{-m} options defined for the 68000 series. The default
9842 values for these options depends on which style of 68000 was selected when
9843 the compiler was configured; the defaults for the most common choices are
9851 Generate output for a 68000. This is the default
9852 when the compiler is configured for 68000-based systems.
9854 Use this option for microcontrollers with a 68000 or EC000 core,
9855 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
9861 Generate output for a 68020. This is the default
9862 when the compiler is configured for 68020-based systems.
9866 Generate output containing 68881 instructions for floating point.
9867 This is the default for most 68020 systems unless @option{--nfp} was
9868 specified when the compiler was configured.
9872 Generate output for a 68030. This is the default when the compiler is
9873 configured for 68030-based systems.
9877 Generate output for a 68040. This is the default when the compiler is
9878 configured for 68040-based systems.
9880 This option inhibits the use of 68881/68882 instructions that have to be
9881 emulated by software on the 68040. Use this option if your 68040 does not
9882 have code to emulate those instructions.
9886 Generate output for a 68060. This is the default when the compiler is
9887 configured for 68060-based systems.
9889 This option inhibits the use of 68020 and 68881/68882 instructions that
9890 have to be emulated by software on the 68060. Use this option if your 68060
9891 does not have code to emulate those instructions.
9895 Generate output for a CPU32. This is the default
9896 when the compiler is configured for CPU32-based systems.
9898 Use this option for microcontrollers with a
9899 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
9900 68336, 68340, 68341, 68349 and 68360.
9904 Generate output for a 520X ``coldfire'' family cpu. This is the default
9905 when the compiler is configured for 520X-based systems.
9907 Use this option for microcontroller with a 5200 core, including
9908 the MCF5202, MCF5203, MCF5204 and MCF5202.
9912 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
9913 This includes use of hardware floating point instructions.
9917 Generate output for a 68040, without using any of the new instructions.
9918 This results in code which can run relatively efficiently on either a
9919 68020/68881 or a 68030 or a 68040. The generated code does use the
9920 68881 instructions that are emulated on the 68040.
9924 Generate output for a 68060, without using any of the new instructions.
9925 This results in code which can run relatively efficiently on either a
9926 68020/68881 or a 68030 or a 68040. The generated code does use the
9927 68881 instructions that are emulated on the 68060.
9930 @opindex msoft-float
9931 Generate output containing library calls for floating point.
9932 @strong{Warning:} the requisite libraries are not available for all m68k
9933 targets. Normally the facilities of the machine's usual C compiler are
9934 used, but this can't be done directly in cross-compilation. You must
9935 make your own arrangements to provide suitable library functions for
9936 cross-compilation. The embedded targets @samp{m68k-*-aout} and
9937 @samp{m68k-*-coff} do provide software floating point support.
9941 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9942 Additionally, parameters passed on the stack are also aligned to a
9943 16-bit boundary even on targets whose API mandates promotion to 32-bit.
9946 @opindex mnobitfield
9947 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
9948 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
9952 Do use the bit-field instructions. The @option{-m68020} option implies
9953 @option{-mbitfield}. This is the default if you use a configuration
9954 designed for a 68020.
9958 Use a different function-calling convention, in which functions
9959 that take a fixed number of arguments return with the @code{rtd}
9960 instruction, which pops their arguments while returning. This
9961 saves one instruction in the caller since there is no need to pop
9962 the arguments there.
9964 This calling convention is incompatible with the one normally
9965 used on Unix, so you cannot use it if you need to call libraries
9966 compiled with the Unix compiler.
9968 Also, you must provide function prototypes for all functions that
9969 take variable numbers of arguments (including @code{printf});
9970 otherwise incorrect code will be generated for calls to those
9973 In addition, seriously incorrect code will result if you call a
9974 function with too many arguments. (Normally, extra arguments are
9975 harmlessly ignored.)
9977 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
9978 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
9981 @itemx -mno-align-int
9983 @opindex mno-align-int
9984 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
9985 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
9986 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
9987 Aligning variables on 32-bit boundaries produces code that runs somewhat
9988 faster on processors with 32-bit busses at the expense of more memory.
9990 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
9991 align structures containing the above types differently than
9992 most published application binary interface specifications for the m68k.
9996 Use the pc-relative addressing mode of the 68000 directly, instead of
9997 using a global offset table. At present, this option implies @option{-fpic},
9998 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
9999 not presently supported with @option{-mpcrel}, though this could be supported for
10000 68020 and higher processors.
10002 @item -mno-strict-align
10003 @itemx -mstrict-align
10004 @opindex mno-strict-align
10005 @opindex mstrict-align
10006 Do not (do) assume that unaligned memory references will be handled by
10010 Generate code that allows the data segment to be located in a different
10011 area of memory from the text segment. This allows for execute in place in
10012 an environment without virtual memory management. This option implies
10015 @item -mno-sep-data
10016 Generate code that assumes that the data segment follows the text segment.
10017 This is the default.
10019 @item -mid-shared-library
10020 Generate code that supports shared libraries via the library ID method.
10021 This allows for execute in place and shared libraries in an environment
10022 without virtual memory management. This option implies @option{-fPIC}.
10024 @item -mno-id-shared-library
10025 Generate code that doesn't assume ID based shared libraries are being used.
10026 This is the default.
10028 @item -mshared-library-id=n
10029 Specified the identification number of the ID based shared library being
10030 compiled. Specifying a value of 0 will generate more compact code, specifying
10031 other values will force the allocation of that number to the current
10032 library but is no more space or time efficient than omitting this option.
10036 @node M68hc1x Options
10037 @subsection M68hc1x Options
10038 @cindex M68hc1x options
10040 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10041 microcontrollers. The default values for these options depends on
10042 which style of microcontroller was selected when the compiler was configured;
10043 the defaults for the most common choices are given below.
10050 Generate output for a 68HC11. This is the default
10051 when the compiler is configured for 68HC11-based systems.
10057 Generate output for a 68HC12. This is the default
10058 when the compiler is configured for 68HC12-based systems.
10064 Generate output for a 68HCS12.
10066 @item -mauto-incdec
10067 @opindex mauto-incdec
10068 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10075 Enable the use of 68HC12 min and max instructions.
10078 @itemx -mno-long-calls
10079 @opindex mlong-calls
10080 @opindex mno-long-calls
10081 Treat all calls as being far away (near). If calls are assumed to be
10082 far away, the compiler will use the @code{call} instruction to
10083 call a function and the @code{rtc} instruction for returning.
10087 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10089 @item -msoft-reg-count=@var{count}
10090 @opindex msoft-reg-count
10091 Specify the number of pseudo-soft registers which are used for the
10092 code generation. The maximum number is 32. Using more pseudo-soft
10093 register may or may not result in better code depending on the program.
10094 The default is 4 for 68HC11 and 2 for 68HC12.
10098 @node MCore Options
10099 @subsection MCore Options
10100 @cindex MCore options
10102 These are the @samp{-m} options defined for the Motorola M*Core
10108 @itemx -mno-hardlit
10110 @opindex mno-hardlit
10111 Inline constants into the code stream if it can be done in two
10112 instructions or less.
10118 Use the divide instruction. (Enabled by default).
10120 @item -mrelax-immediate
10121 @itemx -mno-relax-immediate
10122 @opindex mrelax-immediate
10123 @opindex mno-relax-immediate
10124 Allow arbitrary sized immediates in bit operations.
10126 @item -mwide-bitfields
10127 @itemx -mno-wide-bitfields
10128 @opindex mwide-bitfields
10129 @opindex mno-wide-bitfields
10130 Always treat bit-fields as int-sized.
10132 @item -m4byte-functions
10133 @itemx -mno-4byte-functions
10134 @opindex m4byte-functions
10135 @opindex mno-4byte-functions
10136 Force all functions to be aligned to a four byte boundary.
10138 @item -mcallgraph-data
10139 @itemx -mno-callgraph-data
10140 @opindex mcallgraph-data
10141 @opindex mno-callgraph-data
10142 Emit callgraph information.
10145 @itemx -mno-slow-bytes
10146 @opindex mslow-bytes
10147 @opindex mno-slow-bytes
10148 Prefer word access when reading byte quantities.
10150 @item -mlittle-endian
10151 @itemx -mbig-endian
10152 @opindex mlittle-endian
10153 @opindex mbig-endian
10154 Generate code for a little endian target.
10160 Generate code for the 210 processor.
10164 @subsection MIPS Options
10165 @cindex MIPS options
10171 Generate big-endian code.
10175 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10178 @item -march=@var{arch}
10180 Generate code that will run on @var{arch}, which can be the name of a
10181 generic MIPS ISA, or the name of a particular processor.
10183 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10184 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10185 The processor names are:
10186 @samp{4kc}, @samp{4km}, @samp{4kp},
10187 @samp{5kc}, @samp{5kf},
10189 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10192 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10193 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10194 @samp{rm7000}, @samp{rm9000},
10197 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10198 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10199 The special value @samp{from-abi} selects the
10200 most compatible architecture for the selected ABI (that is,
10201 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10203 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10204 (for example, @samp{-march=r2k}). Prefixes are optional, and
10205 @samp{vr} may be written @samp{r}.
10207 GCC defines two macros based on the value of this option. The first
10208 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10209 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10210 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10211 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10212 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10214 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10215 above. In other words, it will have the full prefix and will not
10216 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10217 the macro names the resolved architecture (either @samp{"mips1"} or
10218 @samp{"mips3"}). It names the default architecture when no
10219 @option{-march} option is given.
10221 @item -mtune=@var{arch}
10223 Optimize for @var{arch}. Among other things, this option controls
10224 the way instructions are scheduled, and the perceived cost of arithmetic
10225 operations. The list of @var{arch} values is the same as for
10228 When this option is not used, GCC will optimize for the processor
10229 specified by @option{-march}. By using @option{-march} and
10230 @option{-mtune} together, it is possible to generate code that will
10231 run on a family of processors, but optimize the code for one
10232 particular member of that family.
10234 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10235 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10236 @samp{-march} ones described above.
10240 Equivalent to @samp{-march=mips1}.
10244 Equivalent to @samp{-march=mips2}.
10248 Equivalent to @samp{-march=mips3}.
10252 Equivalent to @samp{-march=mips4}.
10256 Equivalent to @samp{-march=mips32}.
10260 Equivalent to @samp{-march=mips32r2}.
10264 Equivalent to @samp{-march=mips64}.
10269 @opindex mno-mips16
10270 Generate (do not generate) MIPS16 code. If GCC is targetting a
10271 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10283 Generate code for the given ABI@.
10285 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10286 generates 64-bit code when you select a 64-bit architecture, but you
10287 can use @option{-mgp32} to get 32-bit code instead.
10289 For information about the O64 ABI, see
10290 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10293 @itemx -mno-abicalls
10295 @opindex mno-abicalls
10296 Generate (do not generate) SVR4-style position-independent code.
10297 @option{-mabicalls} is the default for SVR4-based systems.
10303 Lift (do not lift) the usual restrictions on the size of the global
10306 GCC normally uses a single instruction to load values from the GOT@.
10307 While this is relatively efficient, it will only work if the GOT
10308 is smaller than about 64k. Anything larger will cause the linker
10309 to report an error such as:
10311 @cindex relocation truncated to fit (MIPS)
10313 relocation truncated to fit: R_MIPS_GOT16 foobar
10316 If this happens, you should recompile your code with @option{-mxgot}.
10317 It should then work with very large GOTs, although it will also be
10318 less efficient, since it will take three instructions to fetch the
10319 value of a global symbol.
10321 Note that some linkers can create multiple GOTs. If you have such a
10322 linker, you should only need to use @option{-mxgot} when a single object
10323 file accesses more than 64k's worth of GOT entries. Very few do.
10325 These options have no effect unless GCC is generating position
10330 Assume that general-purpose registers are 32 bits wide.
10334 Assume that general-purpose registers are 64 bits wide.
10338 Assume that floating-point registers are 32 bits wide.
10342 Assume that floating-point registers are 64 bits wide.
10345 @opindex mhard-float
10346 Use floating-point coprocessor instructions.
10349 @opindex msoft-float
10350 Do not use floating-point coprocessor instructions. Implement
10351 floating-point calculations using library calls instead.
10353 @item -msingle-float
10354 @opindex msingle-float
10355 Assume that the floating-point coprocessor only supports single-precision
10358 @itemx -mdouble-float
10359 @opindex mdouble-float
10360 Assume that the floating-point coprocessor supports double-precision
10361 operations. This is the default.
10367 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10369 @itemx -mpaired-single
10370 @itemx -mno-paired-single
10371 @opindex mpaired-single
10372 @opindex mno-paired-single
10373 Use (do not use) paired-single floating-point instructions.
10374 @xref{MIPS Paired-Single Support}. This option can only be used
10375 when generating 64-bit code and requires hardware floating-point
10376 support to be enabled.
10381 @opindex mno-mips3d
10382 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10383 The option @option{-mips3d} implies @option{-mpaired-single}.
10387 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10388 an explanation of the default and the way that the pointer size is
10393 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10395 The default size of @code{int}s, @code{long}s and pointers depends on
10396 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10397 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10398 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10399 or the same size as integer registers, whichever is smaller.
10405 Assume (do not assume) that all symbols have 32-bit values, regardless
10406 of the selected ABI@. This option is useful in combination with
10407 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10408 to generate shorter and faster references to symbolic addresses.
10412 @cindex smaller data references (MIPS)
10413 @cindex gp-relative references (MIPS)
10414 Put global and static items less than or equal to @var{num} bytes into
10415 the small data or bss section instead of the normal data or bss section.
10416 This allows the data to be accessed using a single instruction.
10418 All modules should be compiled with the same @option{-G @var{num}}
10421 @item -membedded-data
10422 @itemx -mno-embedded-data
10423 @opindex membedded-data
10424 @opindex mno-embedded-data
10425 Allocate variables to the read-only data section first if possible, then
10426 next in the small data section if possible, otherwise in data. This gives
10427 slightly slower code than the default, but reduces the amount of RAM required
10428 when executing, and thus may be preferred for some embedded systems.
10430 @item -muninit-const-in-rodata
10431 @itemx -mno-uninit-const-in-rodata
10432 @opindex muninit-const-in-rodata
10433 @opindex mno-uninit-const-in-rodata
10434 Put uninitialized @code{const} variables in the read-only data section.
10435 This option is only meaningful in conjunction with @option{-membedded-data}.
10437 @item -msplit-addresses
10438 @itemx -mno-split-addresses
10439 @opindex msplit-addresses
10440 @opindex mno-split-addresses
10441 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10442 relocation operators. This option has been superseded by
10443 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10445 @item -mexplicit-relocs
10446 @itemx -mno-explicit-relocs
10447 @opindex mexplicit-relocs
10448 @opindex mno-explicit-relocs
10449 Use (do not use) assembler relocation operators when dealing with symbolic
10450 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10451 is to use assembler macros instead.
10453 @option{-mexplicit-relocs} is the default if GCC was configured
10454 to use an assembler that supports relocation operators.
10456 @item -mcheck-zero-division
10457 @itemx -mno-check-zero-division
10458 @opindex mcheck-zero-division
10459 @opindex mno-check-zero-division
10460 Trap (do not trap) on integer division by zero. The default is
10461 @option{-mcheck-zero-division}.
10463 @item -mdivide-traps
10464 @itemx -mdivide-breaks
10465 @opindex mdivide-traps
10466 @opindex mdivide-breaks
10467 MIPS systems check for division by zero by generating either a
10468 conditional trap or a break instruction. Using traps results in
10469 smaller code, but is only supported on MIPS II and later. Also, some
10470 versions of the Linux kernel have a bug that prevents trap from
10471 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10472 allow conditional traps on architectures that support them and
10473 @option{-mdivide-breaks} to force the use of breaks.
10475 The default is usually @option{-mdivide-traps}, but this can be
10476 overridden at configure time using @option{--with-divide=breaks}.
10477 Divide-by-zero checks can be completely disabled using
10478 @option{-mno-check-zero-division}.
10483 @opindex mno-memcpy
10484 Force (do not force) the use of @code{memcpy()} for non-trivial block
10485 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10486 most constant-sized copies.
10489 @itemx -mno-long-calls
10490 @opindex mlong-calls
10491 @opindex mno-long-calls
10492 Disable (do not disable) use of the @code{jal} instruction. Calling
10493 functions using @code{jal} is more efficient but requires the caller
10494 and callee to be in the same 256 megabyte segment.
10496 This option has no effect on abicalls code. The default is
10497 @option{-mno-long-calls}.
10503 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10504 instructions, as provided by the R4650 ISA@.
10507 @itemx -mno-fused-madd
10508 @opindex mfused-madd
10509 @opindex mno-fused-madd
10510 Enable (disable) use of the floating point multiply-accumulate
10511 instructions, when they are available. The default is
10512 @option{-mfused-madd}.
10514 When multiply-accumulate instructions are used, the intermediate
10515 product is calculated to infinite precision and is not subject to
10516 the FCSR Flush to Zero bit. This may be undesirable in some
10521 Tell the MIPS assembler to not run its preprocessor over user
10522 assembler files (with a @samp{.s} suffix) when assembling them.
10525 @itemx -mno-fix-r4000
10526 @opindex mfix-r4000
10527 @opindex mno-fix-r4000
10528 Work around certain R4000 CPU errata:
10531 A double-word or a variable shift may give an incorrect result if executed
10532 immediately after starting an integer division.
10534 A double-word or a variable shift may give an incorrect result if executed
10535 while an integer multiplication is in progress.
10537 An integer division may give an incorrect result if started in a delay slot
10538 of a taken branch or a jump.
10542 @itemx -mno-fix-r4400
10543 @opindex mfix-r4400
10544 @opindex mno-fix-r4400
10545 Work around certain R4400 CPU errata:
10548 A double-word or a variable shift may give an incorrect result if executed
10549 immediately after starting an integer division.
10553 @itemx -mno-fix-vr4120
10554 @opindex mfix-vr4120
10555 Work around certain VR4120 errata:
10558 @code{dmultu} does not always produce the correct result.
10560 @code{div} and @code{ddiv} do not always produce the correct result if one
10561 of the operands is negative.
10563 The workarounds for the division errata rely on special functions in
10564 @file{libgcc.a}. At present, these functions are only provided by
10565 the @code{mips64vr*-elf} configurations.
10567 Other VR4120 errata require a nop to be inserted between certain pairs of
10568 instructions. These errata are handled by the assembler, not by GCC itself.
10571 @opindex mfix-vr4130
10572 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10573 workarounds are implemented by the assembler rather than by GCC,
10574 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10575 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10576 instructions are available instead.
10579 @itemx -mno-fix-sb1
10581 Work around certain SB-1 CPU core errata.
10582 (This flag currently works around the SB-1 revision 2
10583 ``F1'' and ``F2'' floating point errata.)
10585 @item -mflush-func=@var{func}
10586 @itemx -mno-flush-func
10587 @opindex mflush-func
10588 Specifies the function to call to flush the I and D caches, or to not
10589 call any such function. If called, the function must take the same
10590 arguments as the common @code{_flush_func()}, that is, the address of the
10591 memory range for which the cache is being flushed, the size of the
10592 memory range, and the number 3 (to flush both caches). The default
10593 depends on the target GCC was configured for, but commonly is either
10594 @samp{_flush_func} or @samp{__cpu_flush}.
10596 @item -mbranch-likely
10597 @itemx -mno-branch-likely
10598 @opindex mbranch-likely
10599 @opindex mno-branch-likely
10600 Enable or disable use of Branch Likely instructions, regardless of the
10601 default for the selected architecture. By default, Branch Likely
10602 instructions may be generated if they are supported by the selected
10603 architecture. An exception is for the MIPS32 and MIPS64 architectures
10604 and processors which implement those architectures; for those, Branch
10605 Likely instructions will not be generated by default because the MIPS32
10606 and MIPS64 architectures specifically deprecate their use.
10608 @item -mfp-exceptions
10609 @itemx -mno-fp-exceptions
10610 @opindex mfp-exceptions
10611 Specifies whether FP exceptions are enabled. This affects how we schedule
10612 FP instructions for some processors. The default is that FP exceptions are
10615 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10616 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10619 @item -mvr4130-align
10620 @itemx -mno-vr4130-align
10621 @opindex mvr4130-align
10622 The VR4130 pipeline is two-way superscalar, but can only issue two
10623 instructions together if the first one is 8-byte aligned. When this
10624 option is enabled, GCC will align pairs of instructions that it
10625 thinks should execute in parallel.
10627 This option only has an effect when optimizing for the VR4130.
10628 It normally makes code faster, but at the expense of making it bigger.
10629 It is enabled by default at optimization level @option{-O3}.
10633 @subsection MMIX Options
10634 @cindex MMIX Options
10636 These options are defined for the MMIX:
10640 @itemx -mno-libfuncs
10642 @opindex mno-libfuncs
10643 Specify that intrinsic library functions are being compiled, passing all
10644 values in registers, no matter the size.
10647 @itemx -mno-epsilon
10649 @opindex mno-epsilon
10650 Generate floating-point comparison instructions that compare with respect
10651 to the @code{rE} epsilon register.
10653 @item -mabi=mmixware
10655 @opindex mabi-mmixware
10657 Generate code that passes function parameters and return values that (in
10658 the called function) are seen as registers @code{$0} and up, as opposed to
10659 the GNU ABI which uses global registers @code{$231} and up.
10661 @item -mzero-extend
10662 @itemx -mno-zero-extend
10663 @opindex mzero-extend
10664 @opindex mno-zero-extend
10665 When reading data from memory in sizes shorter than 64 bits, use (do not
10666 use) zero-extending load instructions by default, rather than
10667 sign-extending ones.
10670 @itemx -mno-knuthdiv
10672 @opindex mno-knuthdiv
10673 Make the result of a division yielding a remainder have the same sign as
10674 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10675 remainder follows the sign of the dividend. Both methods are
10676 arithmetically valid, the latter being almost exclusively used.
10678 @item -mtoplevel-symbols
10679 @itemx -mno-toplevel-symbols
10680 @opindex mtoplevel-symbols
10681 @opindex mno-toplevel-symbols
10682 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10683 code can be used with the @code{PREFIX} assembly directive.
10687 Generate an executable in the ELF format, rather than the default
10688 @samp{mmo} format used by the @command{mmix} simulator.
10690 @item -mbranch-predict
10691 @itemx -mno-branch-predict
10692 @opindex mbranch-predict
10693 @opindex mno-branch-predict
10694 Use (do not use) the probable-branch instructions, when static branch
10695 prediction indicates a probable branch.
10697 @item -mbase-addresses
10698 @itemx -mno-base-addresses
10699 @opindex mbase-addresses
10700 @opindex mno-base-addresses
10701 Generate (do not generate) code that uses @emph{base addresses}. Using a
10702 base address automatically generates a request (handled by the assembler
10703 and the linker) for a constant to be set up in a global register. The
10704 register is used for one or more base address requests within the range 0
10705 to 255 from the value held in the register. The generally leads to short
10706 and fast code, but the number of different data items that can be
10707 addressed is limited. This means that a program that uses lots of static
10708 data may require @option{-mno-base-addresses}.
10710 @item -msingle-exit
10711 @itemx -mno-single-exit
10712 @opindex msingle-exit
10713 @opindex mno-single-exit
10714 Force (do not force) generated code to have a single exit point in each
10718 @node MN10300 Options
10719 @subsection MN10300 Options
10720 @cindex MN10300 options
10722 These @option{-m} options are defined for Matsushita MN10300 architectures:
10727 Generate code to avoid bugs in the multiply instructions for the MN10300
10728 processors. This is the default.
10730 @item -mno-mult-bug
10731 @opindex mno-mult-bug
10732 Do not generate code to avoid bugs in the multiply instructions for the
10733 MN10300 processors.
10737 Generate code which uses features specific to the AM33 processor.
10741 Do not generate code which uses features specific to the AM33 processor. This
10744 @item -mreturn-pointer-on-d0
10745 @opindex mreturn-pointer-on-d0
10746 When generating a function which returns a pointer, return the pointer
10747 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
10748 only in a0, and attempts to call such functions without a prototype
10749 would result in errors. Note that this option is on by default; use
10750 @option{-mno-return-pointer-on-d0} to disable it.
10754 Do not link in the C run-time initialization object file.
10758 Indicate to the linker that it should perform a relaxation optimization pass
10759 to shorten branches, calls and absolute memory addresses. This option only
10760 has an effect when used on the command line for the final link step.
10762 This option makes symbolic debugging impossible.
10766 @subsection MT Options
10769 These @option{-m} options are defined for Morpho MT architectures:
10773 @item -march=@var{cpu-type}
10775 Generate code that will run on @var{cpu-type}, which is the name of a system
10776 representing a certain processor type. Possible values for
10777 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
10778 @samp{ms1-16-003} and @samp{ms2}.
10780 When this option is not used, the default is @option{-march=ms1-16-002}.
10784 Use byte loads and stores when generating code.
10788 Do not use byte loads and stores when generating code.
10792 Use simulator runtime
10796 Do not link in the C run-time initialization object file
10797 @file{crti.o}. Other run-time initialization and termination files
10798 such as @file{startup.o} and @file{exit.o} are still included on the
10799 linker command line.
10803 @node PDP-11 Options
10804 @subsection PDP-11 Options
10805 @cindex PDP-11 Options
10807 These options are defined for the PDP-11:
10812 Use hardware FPP floating point. This is the default. (FIS floating
10813 point on the PDP-11/40 is not supported.)
10816 @opindex msoft-float
10817 Do not use hardware floating point.
10821 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10825 Return floating-point results in memory. This is the default.
10829 Generate code for a PDP-11/40.
10833 Generate code for a PDP-11/45. This is the default.
10837 Generate code for a PDP-11/10.
10839 @item -mbcopy-builtin
10840 @opindex bcopy-builtin
10841 Use inline @code{movmemhi} patterns for copying memory. This is the
10846 Do not use inline @code{movmemhi} patterns for copying memory.
10852 Use 16-bit @code{int}. This is the default.
10858 Use 32-bit @code{int}.
10861 @itemx -mno-float32
10863 @opindex mno-float32
10864 Use 64-bit @code{float}. This is the default.
10867 @itemx -mno-float64
10869 @opindex mno-float64
10870 Use 32-bit @code{float}.
10874 Use @code{abshi2} pattern. This is the default.
10878 Do not use @code{abshi2} pattern.
10880 @item -mbranch-expensive
10881 @opindex mbranch-expensive
10882 Pretend that branches are expensive. This is for experimenting with
10883 code generation only.
10885 @item -mbranch-cheap
10886 @opindex mbranch-cheap
10887 Do not pretend that branches are expensive. This is the default.
10891 Generate code for a system with split I&D@.
10895 Generate code for a system without split I&D@. This is the default.
10899 Use Unix assembler syntax. This is the default when configured for
10900 @samp{pdp11-*-bsd}.
10904 Use DEC assembler syntax. This is the default when configured for any
10905 PDP-11 target other than @samp{pdp11-*-bsd}.
10908 @node PowerPC Options
10909 @subsection PowerPC Options
10910 @cindex PowerPC options
10912 These are listed under @xref{RS/6000 and PowerPC Options}.
10914 @node RS/6000 and PowerPC Options
10915 @subsection IBM RS/6000 and PowerPC Options
10916 @cindex RS/6000 and PowerPC Options
10917 @cindex IBM RS/6000 and PowerPC Options
10919 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
10926 @itemx -mno-powerpc
10927 @itemx -mpowerpc-gpopt
10928 @itemx -mno-powerpc-gpopt
10929 @itemx -mpowerpc-gfxopt
10930 @itemx -mno-powerpc-gfxopt
10932 @itemx -mno-powerpc64
10936 @itemx -mno-popcntb
10942 @opindex mno-power2
10944 @opindex mno-powerpc
10945 @opindex mpowerpc-gpopt
10946 @opindex mno-powerpc-gpopt
10947 @opindex mpowerpc-gfxopt
10948 @opindex mno-powerpc-gfxopt
10949 @opindex mpowerpc64
10950 @opindex mno-powerpc64
10954 @opindex mno-popcntb
10957 GCC supports two related instruction set architectures for the
10958 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
10959 instructions supported by the @samp{rios} chip set used in the original
10960 RS/6000 systems and the @dfn{PowerPC} instruction set is the
10961 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
10962 the IBM 4xx, 6xx, and follow-on microprocessors.
10964 Neither architecture is a subset of the other. However there is a
10965 large common subset of instructions supported by both. An MQ
10966 register is included in processors supporting the POWER architecture.
10968 You use these options to specify which instructions are available on the
10969 processor you are using. The default value of these options is
10970 determined when configuring GCC@. Specifying the
10971 @option{-mcpu=@var{cpu_type}} overrides the specification of these
10972 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
10973 rather than the options listed above.
10975 The @option{-mpower} option allows GCC to generate instructions that
10976 are found only in the POWER architecture and to use the MQ register.
10977 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
10978 to generate instructions that are present in the POWER2 architecture but
10979 not the original POWER architecture.
10981 The @option{-mpowerpc} option allows GCC to generate instructions that
10982 are found only in the 32-bit subset of the PowerPC architecture.
10983 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
10984 GCC to use the optional PowerPC architecture instructions in the
10985 General Purpose group, including floating-point square root. Specifying
10986 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
10987 use the optional PowerPC architecture instructions in the Graphics
10988 group, including floating-point select.
10990 The @option{-mmfcrf} option allows GCC to generate the move from
10991 condition register field instruction implemented on the POWER4
10992 processor and other processors that support the PowerPC V2.01
10994 The @option{-mpopcntb} option allows GCC to generate the popcount and
10995 double precision FP reciprocal estimate instruction implemented on the
10996 POWER5 processor and other processors that support the PowerPC V2.02
10998 The @option{-mfprnd} option allows GCC to generate the FP round to
10999 integer instructions implemented on the POWER5+ processor and other
11000 processors that support the PowerPC V2.03 architecture.
11002 The @option{-mpowerpc64} option allows GCC to generate the additional
11003 64-bit instructions that are found in the full PowerPC64 architecture
11004 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11005 @option{-mno-powerpc64}.
11007 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11008 will use only the instructions in the common subset of both
11009 architectures plus some special AIX common-mode calls, and will not use
11010 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11011 permits GCC to use any instruction from either architecture and to
11012 allow use of the MQ register; specify this for the Motorola MPC601.
11014 @item -mnew-mnemonics
11015 @itemx -mold-mnemonics
11016 @opindex mnew-mnemonics
11017 @opindex mold-mnemonics
11018 Select which mnemonics to use in the generated assembler code. With
11019 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11020 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11021 assembler mnemonics defined for the POWER architecture. Instructions
11022 defined in only one architecture have only one mnemonic; GCC uses that
11023 mnemonic irrespective of which of these options is specified.
11025 GCC defaults to the mnemonics appropriate for the architecture in
11026 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11027 value of these option. Unless you are building a cross-compiler, you
11028 should normally not specify either @option{-mnew-mnemonics} or
11029 @option{-mold-mnemonics}, but should instead accept the default.
11031 @item -mcpu=@var{cpu_type}
11033 Set architecture type, register usage, choice of mnemonics, and
11034 instruction scheduling parameters for machine type @var{cpu_type}.
11035 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11036 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11037 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11038 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11039 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11040 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11041 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11042 @samp{power4}, @samp{power5}, @samp{power5+},
11043 @samp{common}, @samp{powerpc}, @samp{powerpc64},
11044 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11046 @option{-mcpu=common} selects a completely generic processor. Code
11047 generated under this option will run on any POWER or PowerPC processor.
11048 GCC will use only the instructions in the common subset of both
11049 architectures, and will not use the MQ register. GCC assumes a generic
11050 processor model for scheduling purposes.
11052 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11053 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11054 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11055 types, with an appropriate, generic processor model assumed for
11056 scheduling purposes.
11058 The other options specify a specific processor. Code generated under
11059 those options will run best on that processor, and may not run at all on
11062 The @option{-mcpu} options automatically enable or disable the
11063 following options: @option{-maltivec}, @option{-mfprnd},
11064 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11065 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11066 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11067 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw}.
11068 The particular options
11069 set for any particular CPU will vary between compiler versions,
11070 depending on what setting seems to produce optimal code for that CPU;
11071 it doesn't necessarily reflect the actual hardware's capabilities. If
11072 you wish to set an individual option to a particular value, you may
11073 specify it after the @option{-mcpu} option, like @samp{-mcpu=970
11076 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11077 not enabled or disabled by the @option{-mcpu} option at present because
11078 AIX does not have full support for these options. You may still
11079 enable or disable them individually if you're sure it'll work in your
11082 @item -mtune=@var{cpu_type}
11084 Set the instruction scheduling parameters for machine type
11085 @var{cpu_type}, but do not set the architecture type, register usage, or
11086 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11087 values for @var{cpu_type} are used for @option{-mtune} as for
11088 @option{-mcpu}. If both are specified, the code generated will use the
11089 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11090 scheduling parameters set by @option{-mtune}.
11096 Generate code to compute division as reciprocal estimate and iterative
11097 refinement, creating opportunities for increased throughput. This
11098 feature requires: optional PowerPC Graphics instruction set for single
11099 precision and FRE instruction for double precision, assuming divides
11100 cannot generate user-visible traps, and the domain values not include
11101 Infinities, denormals or zero denominator.
11104 @itemx -mno-altivec
11106 @opindex mno-altivec
11107 Generate code that uses (does not use) AltiVec instructions, and also
11108 enable the use of built-in functions that allow more direct access to
11109 the AltiVec instruction set. You may also need to set
11110 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11116 @opindex mno-vrsave
11117 Generate VRSAVE instructions when generating AltiVec code.
11121 Extend the current ABI with SPE ABI extensions. This does not change
11122 the default ABI, instead it adds the SPE ABI extensions to the current
11126 @opindex mabi=no-spe
11127 Disable Booke SPE ABI extensions for the current ABI@.
11130 @opindex msecure-plt
11131 Generate code that allows ld and ld.so to build executables and shared
11132 libraries with non-exec .plt and .got sections. This is a PowerPC
11133 32-bit SYSV ABI option.
11137 Generate code that uses a BSS .plt section that ld.so fills in, and
11138 requires .plt and .got sections that are both writable and executable.
11139 This is a PowerPC 32-bit SYSV ABI option.
11145 This switch enables or disables the generation of ISEL instructions.
11147 @item -misel=@var{yes/no}
11148 This switch has been deprecated. Use @option{-misel} and
11149 @option{-mno-isel} instead.
11155 This switch enables or disables the generation of SPE simd
11158 @item -mspe=@var{yes/no}
11159 This option has been deprecated. Use @option{-mspe} and
11160 @option{-mno-spe} instead.
11162 @item -mfloat-gprs=@var{yes/single/double/no}
11163 @itemx -mfloat-gprs
11164 @opindex mfloat-gprs
11165 This switch enables or disables the generation of floating point
11166 operations on the general purpose registers for architectures that
11169 The argument @var{yes} or @var{single} enables the use of
11170 single-precision floating point operations.
11172 The argument @var{double} enables the use of single and
11173 double-precision floating point operations.
11175 The argument @var{no} disables floating point operations on the
11176 general purpose registers.
11178 This option is currently only available on the MPC854x.
11184 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11185 targets (including GNU/Linux). The 32-bit environment sets int, long
11186 and pointer to 32 bits and generates code that runs on any PowerPC
11187 variant. The 64-bit environment sets int to 32 bits and long and
11188 pointer to 64 bits, and generates code for PowerPC64, as for
11189 @option{-mpowerpc64}.
11192 @itemx -mno-fp-in-toc
11193 @itemx -mno-sum-in-toc
11194 @itemx -mminimal-toc
11196 @opindex mno-fp-in-toc
11197 @opindex mno-sum-in-toc
11198 @opindex mminimal-toc
11199 Modify generation of the TOC (Table Of Contents), which is created for
11200 every executable file. The @option{-mfull-toc} option is selected by
11201 default. In that case, GCC will allocate at least one TOC entry for
11202 each unique non-automatic variable reference in your program. GCC
11203 will also place floating-point constants in the TOC@. However, only
11204 16,384 entries are available in the TOC@.
11206 If you receive a linker error message that saying you have overflowed
11207 the available TOC space, you can reduce the amount of TOC space used
11208 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11209 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11210 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11211 generate code to calculate the sum of an address and a constant at
11212 run-time instead of putting that sum into the TOC@. You may specify one
11213 or both of these options. Each causes GCC to produce very slightly
11214 slower and larger code at the expense of conserving TOC space.
11216 If you still run out of space in the TOC even when you specify both of
11217 these options, specify @option{-mminimal-toc} instead. This option causes
11218 GCC to make only one TOC entry for every file. When you specify this
11219 option, GCC will produce code that is slower and larger but which
11220 uses extremely little TOC space. You may wish to use this option
11221 only on files that contain less frequently executed code.
11227 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11228 @code{long} type, and the infrastructure needed to support them.
11229 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11230 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11231 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11234 @itemx -mno-xl-compat
11235 @opindex mxl-compat
11236 @opindex mno-xl-compat
11237 Produce code that conforms more closely to IBM XLC semantics when using
11238 AIX-compatible ABI. Pass floating-point arguments to prototyped
11239 functions beyond the register save area (RSA) on the stack in addition
11240 to argument FPRs. Do not assume that most significant double in 128
11241 bit long double value is properly rounded when comparing values.
11243 The AIX calling convention was extended but not initially documented to
11244 handle an obscure K&R C case of calling a function that takes the
11245 address of its arguments with fewer arguments than declared. AIX XL
11246 compilers access floating point arguments which do not fit in the
11247 RSA from the stack when a subroutine is compiled without
11248 optimization. Because always storing floating-point arguments on the
11249 stack is inefficient and rarely needed, this option is not enabled by
11250 default and only is necessary when calling subroutines compiled by AIX
11251 XL compilers without optimization.
11255 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11256 application written to use message passing with special startup code to
11257 enable the application to run. The system must have PE installed in the
11258 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11259 must be overridden with the @option{-specs=} option to specify the
11260 appropriate directory location. The Parallel Environment does not
11261 support threads, so the @option{-mpe} option and the @option{-pthread}
11262 option are incompatible.
11264 @item -malign-natural
11265 @itemx -malign-power
11266 @opindex malign-natural
11267 @opindex malign-power
11268 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11269 @option{-malign-natural} overrides the ABI-defined alignment of larger
11270 types, such as floating-point doubles, on their natural size-based boundary.
11271 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11272 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11274 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11278 @itemx -mhard-float
11279 @opindex msoft-float
11280 @opindex mhard-float
11281 Generate code that does not use (uses) the floating-point register set.
11282 Software floating point emulation is provided if you use the
11283 @option{-msoft-float} option, and pass the option to GCC when linking.
11286 @itemx -mno-multiple
11288 @opindex mno-multiple
11289 Generate code that uses (does not use) the load multiple word
11290 instructions and the store multiple word instructions. These
11291 instructions are generated by default on POWER systems, and not
11292 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11293 endian PowerPC systems, since those instructions do not work when the
11294 processor is in little endian mode. The exceptions are PPC740 and
11295 PPC750 which permit the instructions usage in little endian mode.
11300 @opindex mno-string
11301 Generate code that uses (does not use) the load string instructions
11302 and the store string word instructions to save multiple registers and
11303 do small block moves. These instructions are generated by default on
11304 POWER systems, and not generated on PowerPC systems. Do not use
11305 @option{-mstring} on little endian PowerPC systems, since those
11306 instructions do not work when the processor is in little endian mode.
11307 The exceptions are PPC740 and PPC750 which permit the instructions
11308 usage in little endian mode.
11313 @opindex mno-update
11314 Generate code that uses (does not use) the load or store instructions
11315 that update the base register to the address of the calculated memory
11316 location. These instructions are generated by default. If you use
11317 @option{-mno-update}, there is a small window between the time that the
11318 stack pointer is updated and the address of the previous frame is
11319 stored, which means code that walks the stack frame across interrupts or
11320 signals may get corrupted data.
11323 @itemx -mno-fused-madd
11324 @opindex mfused-madd
11325 @opindex mno-fused-madd
11326 Generate code that uses (does not use) the floating point multiply and
11327 accumulate instructions. These instructions are generated by default if
11328 hardware floating is used.
11334 Generate code that uses (does not use) the half-word multiply and
11335 multiply-accumulate instructions on the IBM 405 and 440 processors.
11336 These instructions are generated by default when targetting those
11339 @item -mno-bit-align
11341 @opindex mno-bit-align
11342 @opindex mbit-align
11343 On System V.4 and embedded PowerPC systems do not (do) force structures
11344 and unions that contain bit-fields to be aligned to the base type of the
11347 For example, by default a structure containing nothing but 8
11348 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11349 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11350 the structure would be aligned to a 1 byte boundary and be one byte in
11353 @item -mno-strict-align
11354 @itemx -mstrict-align
11355 @opindex mno-strict-align
11356 @opindex mstrict-align
11357 On System V.4 and embedded PowerPC systems do not (do) assume that
11358 unaligned memory references will be handled by the system.
11360 @item -mrelocatable
11361 @itemx -mno-relocatable
11362 @opindex mrelocatable
11363 @opindex mno-relocatable
11364 On embedded PowerPC systems generate code that allows (does not allow)
11365 the program to be relocated to a different address at runtime. If you
11366 use @option{-mrelocatable} on any module, all objects linked together must
11367 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11369 @item -mrelocatable-lib
11370 @itemx -mno-relocatable-lib
11371 @opindex mrelocatable-lib
11372 @opindex mno-relocatable-lib
11373 On embedded PowerPC systems generate code that allows (does not allow)
11374 the program to be relocated to a different address at runtime. Modules
11375 compiled with @option{-mrelocatable-lib} can be linked with either modules
11376 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11377 with modules compiled with the @option{-mrelocatable} options.
11383 On System V.4 and embedded PowerPC systems do not (do) assume that
11384 register 2 contains a pointer to a global area pointing to the addresses
11385 used in the program.
11388 @itemx -mlittle-endian
11390 @opindex mlittle-endian
11391 On System V.4 and embedded PowerPC systems compile code for the
11392 processor in little endian mode. The @option{-mlittle-endian} option is
11393 the same as @option{-mlittle}.
11396 @itemx -mbig-endian
11398 @opindex mbig-endian
11399 On System V.4 and embedded PowerPC systems compile code for the
11400 processor in big endian mode. The @option{-mbig-endian} option is
11401 the same as @option{-mbig}.
11403 @item -mdynamic-no-pic
11404 @opindex mdynamic-no-pic
11405 On Darwin and Mac OS X systems, compile code so that it is not
11406 relocatable, but that its external references are relocatable. The
11407 resulting code is suitable for applications, but not shared
11410 @item -mprioritize-restricted-insns=@var{priority}
11411 @opindex mprioritize-restricted-insns
11412 This option controls the priority that is assigned to
11413 dispatch-slot restricted instructions during the second scheduling
11414 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11415 @var{no/highest/second-highest} priority to dispatch slot restricted
11418 @item -msched-costly-dep=@var{dependence_type}
11419 @opindex msched-costly-dep
11420 This option controls which dependences are considered costly
11421 by the target during instruction scheduling. The argument
11422 @var{dependence_type} takes one of the following values:
11423 @var{no}: no dependence is costly,
11424 @var{all}: all dependences are costly,
11425 @var{true_store_to_load}: a true dependence from store to load is costly,
11426 @var{store_to_load}: any dependence from store to load is costly,
11427 @var{number}: any dependence which latency >= @var{number} is costly.
11429 @item -minsert-sched-nops=@var{scheme}
11430 @opindex minsert-sched-nops
11431 This option controls which nop insertion scheme will be used during
11432 the second scheduling pass. The argument @var{scheme} takes one of the
11434 @var{no}: Don't insert nops.
11435 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11436 according to the scheduler's grouping.
11437 @var{regroup_exact}: Insert nops to force costly dependent insns into
11438 separate groups. Insert exactly as many nops as needed to force an insn
11439 to a new group, according to the estimated processor grouping.
11440 @var{number}: Insert nops to force costly dependent insns into
11441 separate groups. Insert @var{number} nops to force an insn to a new group.
11444 @opindex mcall-sysv
11445 On System V.4 and embedded PowerPC systems compile code using calling
11446 conventions that adheres to the March 1995 draft of the System V
11447 Application Binary Interface, PowerPC processor supplement. This is the
11448 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11450 @item -mcall-sysv-eabi
11451 @opindex mcall-sysv-eabi
11452 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11454 @item -mcall-sysv-noeabi
11455 @opindex mcall-sysv-noeabi
11456 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11458 @item -mcall-solaris
11459 @opindex mcall-solaris
11460 On System V.4 and embedded PowerPC systems compile code for the Solaris
11464 @opindex mcall-linux
11465 On System V.4 and embedded PowerPC systems compile code for the
11466 Linux-based GNU system.
11470 On System V.4 and embedded PowerPC systems compile code for the
11471 Hurd-based GNU system.
11473 @item -mcall-netbsd
11474 @opindex mcall-netbsd
11475 On System V.4 and embedded PowerPC systems compile code for the
11476 NetBSD operating system.
11478 @item -maix-struct-return
11479 @opindex maix-struct-return
11480 Return all structures in memory (as specified by the AIX ABI)@.
11482 @item -msvr4-struct-return
11483 @opindex msvr4-struct-return
11484 Return structures smaller than 8 bytes in registers (as specified by the
11487 @item -mabi=@var{abi-type}
11489 Extend the current ABI with a particular extension, or remove such extension.
11490 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11494 @itemx -mno-prototype
11495 @opindex mprototype
11496 @opindex mno-prototype
11497 On System V.4 and embedded PowerPC systems assume that all calls to
11498 variable argument functions are properly prototyped. Otherwise, the
11499 compiler must insert an instruction before every non prototyped call to
11500 set or clear bit 6 of the condition code register (@var{CR}) to
11501 indicate whether floating point values were passed in the floating point
11502 registers in case the function takes a variable arguments. With
11503 @option{-mprototype}, only calls to prototyped variable argument functions
11504 will set or clear the bit.
11508 On embedded PowerPC systems, assume that the startup module is called
11509 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11510 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11515 On embedded PowerPC systems, assume that the startup module is called
11516 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11521 On embedded PowerPC systems, assume that the startup module is called
11522 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11525 @item -myellowknife
11526 @opindex myellowknife
11527 On embedded PowerPC systems, assume that the startup module is called
11528 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11533 On System V.4 and embedded PowerPC systems, specify that you are
11534 compiling for a VxWorks system.
11538 Specify that you are compiling for the WindISS simulation environment.
11542 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11543 header to indicate that @samp{eabi} extended relocations are used.
11549 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11550 Embedded Applications Binary Interface (eabi) which is a set of
11551 modifications to the System V.4 specifications. Selecting @option{-meabi}
11552 means that the stack is aligned to an 8 byte boundary, a function
11553 @code{__eabi} is called to from @code{main} to set up the eabi
11554 environment, and the @option{-msdata} option can use both @code{r2} and
11555 @code{r13} to point to two separate small data areas. Selecting
11556 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11557 do not call an initialization function from @code{main}, and the
11558 @option{-msdata} option will only use @code{r13} to point to a single
11559 small data area. The @option{-meabi} option is on by default if you
11560 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11563 @opindex msdata=eabi
11564 On System V.4 and embedded PowerPC systems, put small initialized
11565 @code{const} global and static data in the @samp{.sdata2} section, which
11566 is pointed to by register @code{r2}. Put small initialized
11567 non-@code{const} global and static data in the @samp{.sdata} section,
11568 which is pointed to by register @code{r13}. Put small uninitialized
11569 global and static data in the @samp{.sbss} section, which is adjacent to
11570 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11571 incompatible with the @option{-mrelocatable} option. The
11572 @option{-msdata=eabi} option also sets the @option{-memb} option.
11575 @opindex msdata=sysv
11576 On System V.4 and embedded PowerPC systems, put small global and static
11577 data in the @samp{.sdata} section, which is pointed to by register
11578 @code{r13}. Put small uninitialized global and static data in the
11579 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11580 The @option{-msdata=sysv} option is incompatible with the
11581 @option{-mrelocatable} option.
11583 @item -msdata=default
11585 @opindex msdata=default
11587 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11588 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11589 same as @option{-msdata=sysv}.
11592 @opindex msdata-data
11593 On System V.4 and embedded PowerPC systems, put small global
11594 data in the @samp{.sdata} section. Put small uninitialized global
11595 data in the @samp{.sbss} section. Do not use register @code{r13}
11596 to address small data however. This is the default behavior unless
11597 other @option{-msdata} options are used.
11601 @opindex msdata=none
11603 On embedded PowerPC systems, put all initialized global and static data
11604 in the @samp{.data} section, and all uninitialized data in the
11605 @samp{.bss} section.
11609 @cindex smaller data references (PowerPC)
11610 @cindex .sdata/.sdata2 references (PowerPC)
11611 On embedded PowerPC systems, put global and static items less than or
11612 equal to @var{num} bytes into the small data or bss sections instead of
11613 the normal data or bss section. By default, @var{num} is 8. The
11614 @option{-G @var{num}} switch is also passed to the linker.
11615 All modules should be compiled with the same @option{-G @var{num}} value.
11618 @itemx -mno-regnames
11620 @opindex mno-regnames
11621 On System V.4 and embedded PowerPC systems do (do not) emit register
11622 names in the assembly language output using symbolic forms.
11625 @itemx -mno-longcall
11627 @opindex mno-longcall
11628 Default to making all function calls indirectly, using a register, so
11629 that functions which reside further than 32 megabytes (33,554,432
11630 bytes) from the current location can be called. This setting can be
11631 overridden by the @code{shortcall} function attribute, or by
11632 @code{#pragma longcall(0)}.
11634 Some linkers are capable of detecting out-of-range calls and generating
11635 glue code on the fly. On these systems, long calls are unnecessary and
11636 generate slower code. As of this writing, the AIX linker can do this,
11637 as can the GNU linker for PowerPC/64. It is planned to add this feature
11638 to the GNU linker for 32-bit PowerPC systems as well.
11640 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11641 callee, L42'', plus a ``branch island'' (glue code). The two target
11642 addresses represent the callee and the ``branch island''. The
11643 Darwin/PPC linker will prefer the first address and generate a ``bl
11644 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11645 otherwise, the linker will generate ``bl L42'' to call the ``branch
11646 island''. The ``branch island'' is appended to the body of the
11647 calling function; it computes the full 32-bit address of the callee
11650 On Mach-O (Darwin) systems, this option directs the compiler emit to
11651 the glue for every direct call, and the Darwin linker decides whether
11652 to use or discard it.
11654 In the future, we may cause GCC to ignore all longcall specifications
11655 when the linker is known to generate glue.
11659 Adds support for multithreading with the @dfn{pthreads} library.
11660 This option sets flags for both the preprocessor and linker.
11664 @node S/390 and zSeries Options
11665 @subsection S/390 and zSeries Options
11666 @cindex S/390 and zSeries Options
11668 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
11672 @itemx -msoft-float
11673 @opindex mhard-float
11674 @opindex msoft-float
11675 Use (do not use) the hardware floating-point instructions and registers
11676 for floating-point operations. When @option{-msoft-float} is specified,
11677 functions in @file{libgcc.a} will be used to perform floating-point
11678 operations. When @option{-mhard-float} is specified, the compiler
11679 generates IEEE floating-point instructions. This is the default.
11682 @itemx -mno-backchain
11683 @opindex mbackchain
11684 @opindex mno-backchain
11685 Store (do not store) the address of the caller's frame as backchain pointer
11686 into the callee's stack frame.
11687 A backchain may be needed to allow debugging using tools that do not understand
11688 DWARF-2 call frame information.
11689 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
11690 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
11691 the backchain is placed into the topmost word of the 96/160 byte register
11694 In general, code compiled with @option{-mbackchain} is call-compatible with
11695 code compiled with @option{-mmo-backchain}; however, use of the backchain
11696 for debugging purposes usually requires that the whole binary is built with
11697 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
11698 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11699 to build a linux kernel use @option{-msoft-float}.
11701 The default is to not maintain the backchain.
11703 @item -mpacked-stack
11704 @item -mno-packed-stack
11705 @opindex mpacked-stack
11706 @opindex mno-packed-stack
11707 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
11708 specified, the compiler uses the all fields of the 96/160 byte register save
11709 area only for their default purpose; unused fields still take up stack space.
11710 When @option{-mpacked-stack} is specified, register save slots are densely
11711 packed at the top of the register save area; unused space is reused for other
11712 purposes, allowing for more efficient use of the available stack space.
11713 However, when @option{-mbackchain} is also in effect, the topmost word of
11714 the save area is always used to store the backchain, and the return address
11715 register is always saved two words below the backchain.
11717 As long as the stack frame backchain is not used, code generated with
11718 @option{-mpacked-stack} is call-compatible with code generated with
11719 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
11720 S/390 or zSeries generated code that uses the stack frame backchain at run
11721 time, not just for debugging purposes. Such code is not call-compatible
11722 with code compiled with @option{-mpacked-stack}. Also, note that the
11723 combination of @option{-mbackchain},
11724 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11725 to build a linux kernel use @option{-msoft-float}.
11727 The default is to not use the packed stack layout.
11730 @itemx -mno-small-exec
11731 @opindex msmall-exec
11732 @opindex mno-small-exec
11733 Generate (or do not generate) code using the @code{bras} instruction
11734 to do subroutine calls.
11735 This only works reliably if the total executable size does not
11736 exceed 64k. The default is to use the @code{basr} instruction instead,
11737 which does not have this limitation.
11743 When @option{-m31} is specified, generate code compliant to the
11744 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
11745 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
11746 particular to generate 64-bit instructions. For the @samp{s390}
11747 targets, the default is @option{-m31}, while the @samp{s390x}
11748 targets default to @option{-m64}.
11754 When @option{-mzarch} is specified, generate code using the
11755 instructions available on z/Architecture.
11756 When @option{-mesa} is specified, generate code using the
11757 instructions available on ESA/390. Note that @option{-mesa} is
11758 not possible with @option{-m64}.
11759 When generating code compliant to the GNU/Linux for S/390 ABI,
11760 the default is @option{-mesa}. When generating code compliant
11761 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
11767 Generate (or do not generate) code using the @code{mvcle} instruction
11768 to perform block moves. When @option{-mno-mvcle} is specified,
11769 use a @code{mvc} loop instead. This is the default unless optimizing for
11776 Print (or do not print) additional debug information when compiling.
11777 The default is to not print debug information.
11779 @item -march=@var{cpu-type}
11781 Generate code that will run on @var{cpu-type}, which is the name of a system
11782 representing a certain processor type. Possible values for
11783 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
11784 When generating code using the instructions available on z/Architecture,
11785 the default is @option{-march=z900}. Otherwise, the default is
11786 @option{-march=g5}.
11788 @item -mtune=@var{cpu-type}
11790 Tune to @var{cpu-type} everything applicable about the generated code,
11791 except for the ABI and the set of available instructions.
11792 The list of @var{cpu-type} values is the same as for @option{-march}.
11793 The default is the value used for @option{-march}.
11796 @itemx -mno-tpf-trace
11797 @opindex mtpf-trace
11798 @opindex mno-tpf-trace
11799 Generate code that adds (does not add) in TPF OS specific branches to trace
11800 routines in the operating system. This option is off by default, even
11801 when compiling for the TPF OS@.
11804 @itemx -mno-fused-madd
11805 @opindex mfused-madd
11806 @opindex mno-fused-madd
11807 Generate code that uses (does not use) the floating point multiply and
11808 accumulate instructions. These instructions are generated by default if
11809 hardware floating point is used.
11811 @item -mwarn-framesize=@var{framesize}
11812 @opindex mwarn-framesize
11813 Emit a warning if the current function exceeds the given frame size. Because
11814 this is a compile time check it doesn't need to be a real problem when the program
11815 runs. It is intended to identify functions which most probably cause
11816 a stack overflow. It is useful to be used in an environment with limited stack
11817 size e.g.@: the linux kernel.
11819 @item -mwarn-dynamicstack
11820 @opindex mwarn-dynamicstack
11821 Emit a warning if the function calls alloca or uses dynamically
11822 sized arrays. This is generally a bad idea with a limited stack size.
11824 @item -mstack-guard=@var{stack-guard}
11825 @item -mstack-size=@var{stack-size}
11826 @opindex mstack-guard
11827 @opindex mstack-size
11828 These arguments always have to be used in conjunction. If they are present the s390
11829 back end emits additional instructions in the function prologue which trigger a trap
11830 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
11831 (remember that the stack on s390 grows downward). These options are intended to
11832 be used to help debugging stack overflow problems. The additionally emitted code
11833 causes only little overhead and hence can also be used in production like systems
11834 without greater performance degradation. The given values have to be exact
11835 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
11837 In order to be efficient the extra code makes the assumption that the stack starts
11838 at an address aligned to the value given by @var{stack-size}.
11842 @subsection SH Options
11844 These @samp{-m} options are defined for the SH implementations:
11849 Generate code for the SH1.
11853 Generate code for the SH2.
11856 Generate code for the SH2e.
11860 Generate code for the SH3.
11864 Generate code for the SH3e.
11868 Generate code for the SH4 without a floating-point unit.
11870 @item -m4-single-only
11871 @opindex m4-single-only
11872 Generate code for the SH4 with a floating-point unit that only
11873 supports single-precision arithmetic.
11877 Generate code for the SH4 assuming the floating-point unit is in
11878 single-precision mode by default.
11882 Generate code for the SH4.
11886 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
11887 floating-point unit is not used.
11889 @item -m4a-single-only
11890 @opindex m4a-single-only
11891 Generate code for the SH4a, in such a way that no double-precision
11892 floating point operations are used.
11895 @opindex m4a-single
11896 Generate code for the SH4a assuming the floating-point unit is in
11897 single-precision mode by default.
11901 Generate code for the SH4a.
11905 Same as @option{-m4a-nofpu}, except that it implicitly passes
11906 @option{-dsp} to the assembler. GCC doesn't generate any DSP
11907 instructions at the moment.
11911 Compile code for the processor in big endian mode.
11915 Compile code for the processor in little endian mode.
11919 Align doubles at 64-bit boundaries. Note that this changes the calling
11920 conventions, and thus some functions from the standard C library will
11921 not work unless you recompile it first with @option{-mdalign}.
11925 Shorten some address references at link time, when possible; uses the
11926 linker option @option{-relax}.
11930 Use 32-bit offsets in @code{switch} tables. The default is to use
11935 Enable the use of the instruction @code{fmovd}.
11939 Comply with the calling conventions defined by Renesas.
11943 Comply with the calling conventions defined by Renesas.
11947 Comply with the calling conventions defined for GCC before the Renesas
11948 conventions were available. This option is the default for all
11949 targets of the SH toolchain except for @samp{sh-symbianelf}.
11952 @opindex mnomacsave
11953 Mark the @code{MAC} register as call-clobbered, even if
11954 @option{-mhitachi} is given.
11958 Increase IEEE-compliance of floating-point code.
11959 At the moment, this is equivalent to @option{-fno-finite-math-only}.
11960 When generating 16 bit SH opcodes, getting IEEE-conforming results for
11961 comparisons of NANs / infinities incurs extra overhead in every
11962 floating point comparison, therefore the default is set to
11963 @option{-ffinite-math-only}.
11967 Dump instruction size and location in the assembly code.
11970 @opindex mpadstruct
11971 This option is deprecated. It pads structures to multiple of 4 bytes,
11972 which is incompatible with the SH ABI@.
11976 Optimize for space instead of speed. Implied by @option{-Os}.
11979 @opindex mprefergot
11980 When generating position-independent code, emit function calls using
11981 the Global Offset Table instead of the Procedure Linkage Table.
11985 Generate a library function call to invalidate instruction cache
11986 entries, after fixing up a trampoline. This library function call
11987 doesn't assume it can write to the whole memory address space. This
11988 is the default when the target is @code{sh-*-linux*}.
11990 @item -multcost=@var{number}
11991 @opindex multcost=@var{number}
11992 Set the cost to assume for a multiply insn.
11994 @item -mdiv=@var{strategy}
11995 @opindex mdiv=@var{strategy}
11996 Set the division strategy to use for SHmedia code. @var{strategy} must be
11997 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
11998 inv:call2, inv:fp .
11999 "fp" performs the operation in floating point. This has a very high latency,
12000 but needs only a few instructions, so it might be a good choice if
12001 your code has enough easily exploitable ILP to allow the compiler to
12002 schedule the floating point instructions together with other instructions.
12003 Division by zero causes a floating point exception.
12004 "inv" uses integer operations to calculate the inverse of the divisor,
12005 and then multiplies the dividend with the inverse. This strategy allows
12006 cse and hoisting of the inverse calculation. Division by zero calculates
12007 an unspecified result, but does not trap.
12008 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12009 have been found, or if the entire operation has been hoisted to the same
12010 place, the last stages of the inverse calculation are intertwined with the
12011 final multiply to reduce the overall latency, at the expense of using a few
12012 more instructions, and thus offering fewer scheduling opportunities with
12014 "call" calls a library function that usually implements the inv:minlat
12016 This gives high code density for m5-*media-nofpu compilations.
12017 "call2" uses a different entry point of the same library function, where it
12018 assumes that a pointer to a lookup table has already been set up, which
12019 exposes the pointer load to cse / code hoisting optimizations.
12020 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12021 code generation, but if the code stays unoptimized, revert to the "call",
12022 "call2", or "fp" strategies, respectively. Note that the
12023 potentially-trapping side effect of division by zero is carried by a
12024 separate instruction, so it is possible that all the integer instructions
12025 are hoisted out, but the marker for the side effect stays where it is.
12026 A recombination to fp operations or a call is not possible in that case.
12027 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12028 that the inverse calculation was nor separated from the multiply, they speed
12029 up division where the dividend fits into 20 bits (plus sign where applicable),
12030 by inserting a test to skip a number of operations in this case; this test
12031 slows down the case of larger dividends. inv20u assumes the case of a such
12032 a small dividend to be unlikely, and inv20l assumes it to be likely.
12034 @item -mdivsi3_libfunc=@var{name}
12035 @opindex mdivsi3_libfunc=@var{name}
12036 Set the name of the library function used for 32 bit signed division to
12037 @var{name}. This only affect the name used in the call and inv:call
12038 division strategies, and the compiler will still expect the same
12039 sets of input/output/clobbered registers as if this option was not present.
12041 @item -madjust-unroll
12042 @opindex madjust-unroll
12043 Throttle unrolling to avoid thrashing target registers.
12044 This option only has an effect if the gcc code base supports the
12045 TARGET_ADJUST_UNROLL_MAX target hook.
12047 @item -mindexed-addressing
12048 @opindex mindexed-addressing
12049 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12050 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12051 semantics for the indexed addressing mode. The architecture allows the
12052 implementation of processors with 64 bit MMU, which the OS could use to
12053 get 32 bit addressing, but since no current hardware implementation supports
12054 this or any other way to make the indexed addressing mode safe to use in
12055 the 32 bit ABI, the default is -mno-indexed-addressing.
12057 @item -mgettrcost=@var{number}
12058 @opindex mgettrcost=@var{number}
12059 Set the cost assumed for the gettr instruction to @var{number}.
12060 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12064 Assume pt* instructions won't trap. This will generally generate better
12065 scheduled code, but is unsafe on current hardware. The current architecture
12066 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12067 This has the unintentional effect of making it unsafe to schedule ptabs /
12068 ptrel before a branch, or hoist it out of a loop. For example,
12069 __do_global_ctors, a part of libgcc that runs constructors at program
12070 startup, calls functions in a list which is delimited by -1. With the
12071 -mpt-fixed option, the ptabs will be done before testing against -1.
12072 That means that all the constructors will be run a bit quicker, but when
12073 the loop comes to the end of the list, the program crashes because ptabs
12074 loads -1 into a target register. Since this option is unsafe for any
12075 hardware implementing the current architecture specification, the default
12076 is -mno-pt-fixed. Unless the user specifies a specific cost with
12077 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12078 this deters register allocation using target registers for storing
12081 @item -minvalid-symbols
12082 @opindex minvalid-symbols
12083 Assume symbols might be invalid. Ordinary function symbols generated by
12084 the compiler will always be valid to load with movi/shori/ptabs or
12085 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12086 to generate symbols that will cause ptabs / ptrel to trap.
12087 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12088 It will then prevent cross-basic-block cse, hoisting and most scheduling
12089 of symbol loads. The default is @option{-mno-invalid-symbols}.
12092 @node SPARC Options
12093 @subsection SPARC Options
12094 @cindex SPARC options
12096 These @samp{-m} options are supported on the SPARC:
12099 @item -mno-app-regs
12101 @opindex mno-app-regs
12103 Specify @option{-mapp-regs} to generate output using the global registers
12104 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12107 To be fully SVR4 ABI compliant at the cost of some performance loss,
12108 specify @option{-mno-app-regs}. You should compile libraries and system
12109 software with this option.
12112 @itemx -mhard-float
12114 @opindex mhard-float
12115 Generate output containing floating point instructions. This is the
12119 @itemx -msoft-float
12121 @opindex msoft-float
12122 Generate output containing library calls for floating point.
12123 @strong{Warning:} the requisite libraries are not available for all SPARC
12124 targets. Normally the facilities of the machine's usual C compiler are
12125 used, but this cannot be done directly in cross-compilation. You must make
12126 your own arrangements to provide suitable library functions for
12127 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12128 @samp{sparclite-*-*} do provide software floating point support.
12130 @option{-msoft-float} changes the calling convention in the output file;
12131 therefore, it is only useful if you compile @emph{all} of a program with
12132 this option. In particular, you need to compile @file{libgcc.a}, the
12133 library that comes with GCC, with @option{-msoft-float} in order for
12136 @item -mhard-quad-float
12137 @opindex mhard-quad-float
12138 Generate output containing quad-word (long double) floating point
12141 @item -msoft-quad-float
12142 @opindex msoft-quad-float
12143 Generate output containing library calls for quad-word (long double)
12144 floating point instructions. The functions called are those specified
12145 in the SPARC ABI@. This is the default.
12147 As of this writing, there are no SPARC implementations that have hardware
12148 support for the quad-word floating point instructions. They all invoke
12149 a trap handler for one of these instructions, and then the trap handler
12150 emulates the effect of the instruction. Because of the trap handler overhead,
12151 this is much slower than calling the ABI library routines. Thus the
12152 @option{-msoft-quad-float} option is the default.
12154 @item -mno-unaligned-doubles
12155 @itemx -munaligned-doubles
12156 @opindex mno-unaligned-doubles
12157 @opindex munaligned-doubles
12158 Assume that doubles have 8 byte alignment. This is the default.
12160 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12161 alignment only if they are contained in another type, or if they have an
12162 absolute address. Otherwise, it assumes they have 4 byte alignment.
12163 Specifying this option avoids some rare compatibility problems with code
12164 generated by other compilers. It is not the default because it results
12165 in a performance loss, especially for floating point code.
12167 @item -mno-faster-structs
12168 @itemx -mfaster-structs
12169 @opindex mno-faster-structs
12170 @opindex mfaster-structs
12171 With @option{-mfaster-structs}, the compiler assumes that structures
12172 should have 8 byte alignment. This enables the use of pairs of
12173 @code{ldd} and @code{std} instructions for copies in structure
12174 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12175 However, the use of this changed alignment directly violates the SPARC
12176 ABI@. Thus, it's intended only for use on targets where the developer
12177 acknowledges that their resulting code will not be directly in line with
12178 the rules of the ABI@.
12180 @item -mimpure-text
12181 @opindex mimpure-text
12182 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12183 the compiler to not pass @option{-z text} to the linker when linking a
12184 shared object. Using this option, you can link position-dependent
12185 code into a shared object.
12187 @option{-mimpure-text} suppresses the ``relocations remain against
12188 allocatable but non-writable sections'' linker error message.
12189 However, the necessary relocations will trigger copy-on-write, and the
12190 shared object is not actually shared across processes. Instead of
12191 using @option{-mimpure-text}, you should compile all source code with
12192 @option{-fpic} or @option{-fPIC}.
12194 This option is only available on SunOS and Solaris.
12196 @item -mcpu=@var{cpu_type}
12198 Set the instruction set, register set, and instruction scheduling parameters
12199 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12200 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12201 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12202 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
12203 @samp{ultrasparc3}.
12205 Default instruction scheduling parameters are used for values that select
12206 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12207 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12209 Here is a list of each supported architecture and their supported
12214 v8: supersparc, hypersparc
12215 sparclite: f930, f934, sparclite86x
12217 v9: ultrasparc, ultrasparc3
12220 By default (unless configured otherwise), GCC generates code for the V7
12221 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12222 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12223 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12224 SPARCStation 1, 2, IPX etc.
12226 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12227 architecture. The only difference from V7 code is that the compiler emits
12228 the integer multiply and integer divide instructions which exist in SPARC-V8
12229 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12230 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12233 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12234 the SPARC architecture. This adds the integer multiply, integer divide step
12235 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12236 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12237 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12238 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12239 MB86934 chip, which is the more recent SPARClite with FPU@.
12241 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12242 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12243 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12244 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12245 optimizes it for the TEMIC SPARClet chip.
12247 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12248 architecture. This adds 64-bit integer and floating-point move instructions,
12249 3 additional floating-point condition code registers and conditional move
12250 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12251 optimizes it for the Sun UltraSPARC I/II chips. With
12252 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12253 Sun UltraSPARC III chip.
12255 @item -mtune=@var{cpu_type}
12257 Set the instruction scheduling parameters for machine type
12258 @var{cpu_type}, but do not set the instruction set or register set that the
12259 option @option{-mcpu=@var{cpu_type}} would.
12261 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12262 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12263 that select a particular cpu implementation. Those are @samp{cypress},
12264 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12265 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
12266 @samp{ultrasparc3}.
12271 @opindex mno-v8plus
12272 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12273 difference from the V8 ABI is that the global and out registers are
12274 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12275 mode for all SPARC-V9 processors.
12281 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12282 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12285 These @samp{-m} options are supported in addition to the above
12286 on SPARC-V9 processors in 64-bit environments:
12289 @item -mlittle-endian
12290 @opindex mlittle-endian
12291 Generate code for a processor running in little-endian mode. It is only
12292 available for a few configurations and most notably not on Solaris and Linux.
12298 Generate code for a 32-bit or 64-bit environment.
12299 The 32-bit environment sets int, long and pointer to 32 bits.
12300 The 64-bit environment sets int to 32 bits and long and pointer
12303 @item -mcmodel=medlow
12304 @opindex mcmodel=medlow
12305 Generate code for the Medium/Low code model: 64-bit addresses, programs
12306 must be linked in the low 32 bits of memory. Programs can be statically
12307 or dynamically linked.
12309 @item -mcmodel=medmid
12310 @opindex mcmodel=medmid
12311 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12312 must be linked in the low 44 bits of memory, the text and data segments must
12313 be less than 2GB in size and the data segment must be located within 2GB of
12316 @item -mcmodel=medany
12317 @opindex mcmodel=medany
12318 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12319 may be linked anywhere in memory, the text and data segments must be less
12320 than 2GB in size and the data segment must be located within 2GB of the
12323 @item -mcmodel=embmedany
12324 @opindex mcmodel=embmedany
12325 Generate code for the Medium/Anywhere code model for embedded systems:
12326 64-bit addresses, the text and data segments must be less than 2GB in
12327 size, both starting anywhere in memory (determined at link time). The
12328 global register %g4 points to the base of the data segment. Programs
12329 are statically linked and PIC is not supported.
12332 @itemx -mno-stack-bias
12333 @opindex mstack-bias
12334 @opindex mno-stack-bias
12335 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12336 frame pointer if present, are offset by @minus{}2047 which must be added back
12337 when making stack frame references. This is the default in 64-bit mode.
12338 Otherwise, assume no such offset is present.
12341 These switches are supported in addition to the above on Solaris:
12346 Add support for multithreading using the Solaris threads library. This
12347 option sets flags for both the preprocessor and linker. This option does
12348 not affect the thread safety of object code produced by the compiler or
12349 that of libraries supplied with it.
12353 Add support for multithreading using the POSIX threads library. This
12354 option sets flags for both the preprocessor and linker. This option does
12355 not affect the thread safety of object code produced by the compiler or
12356 that of libraries supplied with it.
12359 @node System V Options
12360 @subsection Options for System V
12362 These additional options are available on System V Release 4 for
12363 compatibility with other compilers on those systems:
12368 Create a shared object.
12369 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12373 Identify the versions of each tool used by the compiler, in a
12374 @code{.ident} assembler directive in the output.
12378 Refrain from adding @code{.ident} directives to the output file (this is
12381 @item -YP,@var{dirs}
12383 Search the directories @var{dirs}, and no others, for libraries
12384 specified with @option{-l}.
12386 @item -Ym,@var{dir}
12388 Look in the directory @var{dir} to find the M4 preprocessor.
12389 The assembler uses this option.
12390 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12391 @c the generic assembler that comes with Solaris takes just -Ym.
12394 @node TMS320C3x/C4x Options
12395 @subsection TMS320C3x/C4x Options
12396 @cindex TMS320C3x/C4x Options
12398 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12402 @item -mcpu=@var{cpu_type}
12404 Set the instruction set, register set, and instruction scheduling
12405 parameters for machine type @var{cpu_type}. Supported values for
12406 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12407 @samp{c44}. The default is @samp{c40} to generate code for the
12412 @itemx -msmall-memory
12414 @opindex mbig-memory
12416 @opindex msmall-memory
12418 Generates code for the big or small memory model. The small memory
12419 model assumed that all data fits into one 64K word page. At run-time
12420 the data page (DP) register must be set to point to the 64K page
12421 containing the .bss and .data program sections. The big memory model is
12422 the default and requires reloading of the DP register for every direct
12429 Allow (disallow) allocation of general integer operands into the block
12430 count register BK@.
12436 Enable (disable) generation of code using decrement and branch,
12437 DBcond(D), instructions. This is enabled by default for the C4x. To be
12438 on the safe side, this is disabled for the C3x, since the maximum
12439 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12440 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12441 that it can utilize the decrement and branch instruction, but will give
12442 up if there is more than one memory reference in the loop. Thus a loop
12443 where the loop counter is decremented can generate slightly more
12444 efficient code, in cases where the RPTB instruction cannot be utilized.
12446 @item -mdp-isr-reload
12448 @opindex mdp-isr-reload
12450 Force the DP register to be saved on entry to an interrupt service
12451 routine (ISR), reloaded to point to the data section, and restored on
12452 exit from the ISR@. This should not be required unless someone has
12453 violated the small memory model by modifying the DP register, say within
12460 For the C3x use the 24-bit MPYI instruction for integer multiplies
12461 instead of a library call to guarantee 32-bit results. Note that if one
12462 of the operands is a constant, then the multiplication will be performed
12463 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12464 then squaring operations are performed inline instead of a library call.
12467 @itemx -mno-fast-fix
12469 @opindex mno-fast-fix
12470 The C3x/C4x FIX instruction to convert a floating point value to an
12471 integer value chooses the nearest integer less than or equal to the
12472 floating point value rather than to the nearest integer. Thus if the
12473 floating point number is negative, the result will be incorrectly
12474 truncated an additional code is necessary to detect and correct this
12475 case. This option can be used to disable generation of the additional
12476 code required to correct the result.
12482 Enable (disable) generation of repeat block sequences using the RPTB
12483 instruction for zero overhead looping. The RPTB construct is only used
12484 for innermost loops that do not call functions or jump across the loop
12485 boundaries. There is no advantage having nested RPTB loops due to the
12486 overhead required to save and restore the RC, RS, and RE registers.
12487 This is enabled by default with @option{-O2}.
12489 @item -mrpts=@var{count}
12493 Enable (disable) the use of the single instruction repeat instruction
12494 RPTS@. If a repeat block contains a single instruction, and the loop
12495 count can be guaranteed to be less than the value @var{count}, GCC will
12496 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12497 then a RPTS will be emitted even if the loop count cannot be determined
12498 at compile time. Note that the repeated instruction following RPTS does
12499 not have to be reloaded from memory each iteration, thus freeing up the
12500 CPU buses for operands. However, since interrupts are blocked by this
12501 instruction, it is disabled by default.
12503 @item -mloop-unsigned
12504 @itemx -mno-loop-unsigned
12505 @opindex mloop-unsigned
12506 @opindex mno-loop-unsigned
12507 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12508 is @math{2^{31} + 1} since these instructions test if the iteration count is
12509 negative to terminate the loop. If the iteration count is unsigned
12510 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12511 exceeded. This switch allows an unsigned iteration count.
12515 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12516 with. This also enforces compatibility with the API employed by the TI
12517 C3x C compiler. For example, long doubles are passed as structures
12518 rather than in floating point registers.
12524 Generate code that uses registers (stack) for passing arguments to functions.
12525 By default, arguments are passed in registers where possible rather
12526 than by pushing arguments on to the stack.
12528 @item -mparallel-insns
12529 @itemx -mno-parallel-insns
12530 @opindex mparallel-insns
12531 @opindex mno-parallel-insns
12532 Allow the generation of parallel instructions. This is enabled by
12533 default with @option{-O2}.
12535 @item -mparallel-mpy
12536 @itemx -mno-parallel-mpy
12537 @opindex mparallel-mpy
12538 @opindex mno-parallel-mpy
12539 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12540 provided @option{-mparallel-insns} is also specified. These instructions have
12541 tight register constraints which can pessimize the code generation
12542 of large functions.
12547 @subsection V850 Options
12548 @cindex V850 Options
12550 These @samp{-m} options are defined for V850 implementations:
12554 @itemx -mno-long-calls
12555 @opindex mlong-calls
12556 @opindex mno-long-calls
12557 Treat all calls as being far away (near). If calls are assumed to be
12558 far away, the compiler will always load the functions address up into a
12559 register, and call indirect through the pointer.
12565 Do not optimize (do optimize) basic blocks that use the same index
12566 pointer 4 or more times to copy pointer into the @code{ep} register, and
12567 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12568 option is on by default if you optimize.
12570 @item -mno-prolog-function
12571 @itemx -mprolog-function
12572 @opindex mno-prolog-function
12573 @opindex mprolog-function
12574 Do not use (do use) external functions to save and restore registers
12575 at the prologue and epilogue of a function. The external functions
12576 are slower, but use less code space if more than one function saves
12577 the same number of registers. The @option{-mprolog-function} option
12578 is on by default if you optimize.
12582 Try to make the code as small as possible. At present, this just turns
12583 on the @option{-mep} and @option{-mprolog-function} options.
12585 @item -mtda=@var{n}
12587 Put static or global variables whose size is @var{n} bytes or less into
12588 the tiny data area that register @code{ep} points to. The tiny data
12589 area can hold up to 256 bytes in total (128 bytes for byte references).
12591 @item -msda=@var{n}
12593 Put static or global variables whose size is @var{n} bytes or less into
12594 the small data area that register @code{gp} points to. The small data
12595 area can hold up to 64 kilobytes.
12597 @item -mzda=@var{n}
12599 Put static or global variables whose size is @var{n} bytes or less into
12600 the first 32 kilobytes of memory.
12604 Specify that the target processor is the V850.
12607 @opindex mbig-switch
12608 Generate code suitable for big switch tables. Use this option only if
12609 the assembler/linker complain about out of range branches within a switch
12614 This option will cause r2 and r5 to be used in the code generated by
12615 the compiler. This setting is the default.
12617 @item -mno-app-regs
12618 @opindex mno-app-regs
12619 This option will cause r2 and r5 to be treated as fixed registers.
12623 Specify that the target processor is the V850E1. The preprocessor
12624 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
12625 this option is used.
12629 Specify that the target processor is the V850E@. The preprocessor
12630 constant @samp{__v850e__} will be defined if this option is used.
12632 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
12633 are defined then a default target processor will be chosen and the
12634 relevant @samp{__v850*__} preprocessor constant will be defined.
12636 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
12637 defined, regardless of which processor variant is the target.
12639 @item -mdisable-callt
12640 @opindex mdisable-callt
12641 This option will suppress generation of the CALLT instruction for the
12642 v850e and v850e1 flavors of the v850 architecture. The default is
12643 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
12648 @subsection VAX Options
12649 @cindex VAX options
12651 These @samp{-m} options are defined for the VAX:
12656 Do not output certain jump instructions (@code{aobleq} and so on)
12657 that the Unix assembler for the VAX cannot handle across long
12662 Do output those jump instructions, on the assumption that you
12663 will assemble with the GNU assembler.
12667 Output code for g-format floating point numbers instead of d-format.
12670 @node x86-64 Options
12671 @subsection x86-64 Options
12672 @cindex x86-64 options
12674 These are listed under @xref{i386 and x86-64 Options}.
12676 @node Xstormy16 Options
12677 @subsection Xstormy16 Options
12678 @cindex Xstormy16 Options
12680 These options are defined for Xstormy16:
12685 Choose startup files and linker script suitable for the simulator.
12688 @node Xtensa Options
12689 @subsection Xtensa Options
12690 @cindex Xtensa Options
12692 These options are supported for Xtensa targets:
12696 @itemx -mno-const16
12698 @opindex mno-const16
12699 Enable or disable use of @code{CONST16} instructions for loading
12700 constant values. The @code{CONST16} instruction is currently not a
12701 standard option from Tensilica. When enabled, @code{CONST16}
12702 instructions are always used in place of the standard @code{L32R}
12703 instructions. The use of @code{CONST16} is enabled by default only if
12704 the @code{L32R} instruction is not available.
12707 @itemx -mno-fused-madd
12708 @opindex mfused-madd
12709 @opindex mno-fused-madd
12710 Enable or disable use of fused multiply/add and multiply/subtract
12711 instructions in the floating-point option. This has no effect if the
12712 floating-point option is not also enabled. Disabling fused multiply/add
12713 and multiply/subtract instructions forces the compiler to use separate
12714 instructions for the multiply and add/subtract operations. This may be
12715 desirable in some cases where strict IEEE 754-compliant results are
12716 required: the fused multiply add/subtract instructions do not round the
12717 intermediate result, thereby producing results with @emph{more} bits of
12718 precision than specified by the IEEE standard. Disabling fused multiply
12719 add/subtract instructions also ensures that the program output is not
12720 sensitive to the compiler's ability to combine multiply and add/subtract
12723 @item -mtext-section-literals
12724 @itemx -mno-text-section-literals
12725 @opindex mtext-section-literals
12726 @opindex mno-text-section-literals
12727 Control the treatment of literal pools. The default is
12728 @option{-mno-text-section-literals}, which places literals in a separate
12729 section in the output file. This allows the literal pool to be placed
12730 in a data RAM/ROM, and it also allows the linker to combine literal
12731 pools from separate object files to remove redundant literals and
12732 improve code size. With @option{-mtext-section-literals}, the literals
12733 are interspersed in the text section in order to keep them as close as
12734 possible to their references. This may be necessary for large assembly
12737 @item -mtarget-align
12738 @itemx -mno-target-align
12739 @opindex mtarget-align
12740 @opindex mno-target-align
12741 When this option is enabled, GCC instructs the assembler to
12742 automatically align instructions to reduce branch penalties at the
12743 expense of some code density. The assembler attempts to widen density
12744 instructions to align branch targets and the instructions following call
12745 instructions. If there are not enough preceding safe density
12746 instructions to align a target, no widening will be performed. The
12747 default is @option{-mtarget-align}. These options do not affect the
12748 treatment of auto-aligned instructions like @code{LOOP}, which the
12749 assembler will always align, either by widening density instructions or
12750 by inserting no-op instructions.
12753 @itemx -mno-longcalls
12754 @opindex mlongcalls
12755 @opindex mno-longcalls
12756 When this option is enabled, GCC instructs the assembler to translate
12757 direct calls to indirect calls unless it can determine that the target
12758 of a direct call is in the range allowed by the call instruction. This
12759 translation typically occurs for calls to functions in other source
12760 files. Specifically, the assembler translates a direct @code{CALL}
12761 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
12762 The default is @option{-mno-longcalls}. This option should be used in
12763 programs where the call target can potentially be out of range. This
12764 option is implemented in the assembler, not the compiler, so the
12765 assembly code generated by GCC will still show direct call
12766 instructions---look at the disassembled object code to see the actual
12767 instructions. Note that the assembler will use an indirect call for
12768 every cross-file call, not just those that really will be out of range.
12771 @node zSeries Options
12772 @subsection zSeries Options
12773 @cindex zSeries options
12775 These are listed under @xref{S/390 and zSeries Options}.
12777 @node Code Gen Options
12778 @section Options for Code Generation Conventions
12779 @cindex code generation conventions
12780 @cindex options, code generation
12781 @cindex run-time options
12783 These machine-independent options control the interface conventions
12784 used in code generation.
12786 Most of them have both positive and negative forms; the negative form
12787 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
12788 one of the forms is listed---the one which is not the default. You
12789 can figure out the other form by either removing @samp{no-} or adding
12793 @item -fbounds-check
12794 @opindex fbounds-check
12795 For front-ends that support it, generate additional code to check that
12796 indices used to access arrays are within the declared range. This is
12797 currently only supported by the Java and Fortran 77 front-ends, where
12798 this option defaults to true and false respectively.
12802 This option generates traps for signed overflow on addition, subtraction,
12803 multiplication operations.
12807 This option instructs the compiler to assume that signed arithmetic
12808 overflow of addition, subtraction and multiplication wraps around
12809 using twos-complement representation. This flag enables some optimizations
12810 and disables others. This option is enabled by default for the Java
12811 front-end, as required by the Java language specification.
12814 @opindex fexceptions
12815 Enable exception handling. Generates extra code needed to propagate
12816 exceptions. For some targets, this implies GCC will generate frame
12817 unwind information for all functions, which can produce significant data
12818 size overhead, although it does not affect execution. If you do not
12819 specify this option, GCC will enable it by default for languages like
12820 C++ which normally require exception handling, and disable it for
12821 languages like C that do not normally require it. However, you may need
12822 to enable this option when compiling C code that needs to interoperate
12823 properly with exception handlers written in C++. You may also wish to
12824 disable this option if you are compiling older C++ programs that don't
12825 use exception handling.
12827 @item -fnon-call-exceptions
12828 @opindex fnon-call-exceptions
12829 Generate code that allows trapping instructions to throw exceptions.
12830 Note that this requires platform-specific runtime support that does
12831 not exist everywhere. Moreover, it only allows @emph{trapping}
12832 instructions to throw exceptions, i.e.@: memory references or floating
12833 point instructions. It does not allow exceptions to be thrown from
12834 arbitrary signal handlers such as @code{SIGALRM}.
12836 @item -funwind-tables
12837 @opindex funwind-tables
12838 Similar to @option{-fexceptions}, except that it will just generate any needed
12839 static data, but will not affect the generated code in any other way.
12840 You will normally not enable this option; instead, a language processor
12841 that needs this handling would enable it on your behalf.
12843 @item -fasynchronous-unwind-tables
12844 @opindex fasynchronous-unwind-tables
12845 Generate unwind table in dwarf2 format, if supported by target machine. The
12846 table is exact at each instruction boundary, so it can be used for stack
12847 unwinding from asynchronous events (such as debugger or garbage collector).
12849 @item -fpcc-struct-return
12850 @opindex fpcc-struct-return
12851 Return ``short'' @code{struct} and @code{union} values in memory like
12852 longer ones, rather than in registers. This convention is less
12853 efficient, but it has the advantage of allowing intercallability between
12854 GCC-compiled files and files compiled with other compilers, particularly
12855 the Portable C Compiler (pcc).
12857 The precise convention for returning structures in memory depends
12858 on the target configuration macros.
12860 Short structures and unions are those whose size and alignment match
12861 that of some integer type.
12863 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
12864 switch is not binary compatible with code compiled with the
12865 @option{-freg-struct-return} switch.
12866 Use it to conform to a non-default application binary interface.
12868 @item -freg-struct-return
12869 @opindex freg-struct-return
12870 Return @code{struct} and @code{union} values in registers when possible.
12871 This is more efficient for small structures than
12872 @option{-fpcc-struct-return}.
12874 If you specify neither @option{-fpcc-struct-return} nor
12875 @option{-freg-struct-return}, GCC defaults to whichever convention is
12876 standard for the target. If there is no standard convention, GCC
12877 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
12878 the principal compiler. In those cases, we can choose the standard, and
12879 we chose the more efficient register return alternative.
12881 @strong{Warning:} code compiled with the @option{-freg-struct-return}
12882 switch is not binary compatible with code compiled with the
12883 @option{-fpcc-struct-return} switch.
12884 Use it to conform to a non-default application binary interface.
12886 @item -fshort-enums
12887 @opindex fshort-enums
12888 Allocate to an @code{enum} type only as many bytes as it needs for the
12889 declared range of possible values. Specifically, the @code{enum} type
12890 will be equivalent to the smallest integer type which has enough room.
12892 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
12893 code that is not binary compatible with code generated without that switch.
12894 Use it to conform to a non-default application binary interface.
12896 @item -fshort-double
12897 @opindex fshort-double
12898 Use the same size for @code{double} as for @code{float}.
12900 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
12901 code that is not binary compatible with code generated without that switch.
12902 Use it to conform to a non-default application binary interface.
12904 @item -fshort-wchar
12905 @opindex fshort-wchar
12906 Override the underlying type for @samp{wchar_t} to be @samp{short
12907 unsigned int} instead of the default for the target. This option is
12908 useful for building programs to run under WINE@.
12910 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
12911 code that is not binary compatible with code generated without that switch.
12912 Use it to conform to a non-default application binary interface.
12914 @item -fshared-data
12915 @opindex fshared-data
12916 Requests that the data and non-@code{const} variables of this
12917 compilation be shared data rather than private data. The distinction
12918 makes sense only on certain operating systems, where shared data is
12919 shared between processes running the same program, while private data
12920 exists in one copy per process.
12923 @opindex fno-common
12924 In C, allocate even uninitialized global variables in the data section of the
12925 object file, rather than generating them as common blocks. This has the
12926 effect that if the same variable is declared (without @code{extern}) in
12927 two different compilations, you will get an error when you link them.
12928 The only reason this might be useful is if you wish to verify that the
12929 program will work on other systems which always work this way.
12933 Ignore the @samp{#ident} directive.
12935 @item -finhibit-size-directive
12936 @opindex finhibit-size-directive
12937 Don't output a @code{.size} assembler directive, or anything else that
12938 would cause trouble if the function is split in the middle, and the
12939 two halves are placed at locations far apart in memory. This option is
12940 used when compiling @file{crtstuff.c}; you should not need to use it
12943 @item -fverbose-asm
12944 @opindex fverbose-asm
12945 Put extra commentary information in the generated assembly code to
12946 make it more readable. This option is generally only of use to those
12947 who actually need to read the generated assembly code (perhaps while
12948 debugging the compiler itself).
12950 @option{-fno-verbose-asm}, the default, causes the
12951 extra information to be omitted and is useful when comparing two assembler
12956 @cindex global offset table
12958 Generate position-independent code (PIC) suitable for use in a shared
12959 library, if supported for the target machine. Such code accesses all
12960 constant addresses through a global offset table (GOT)@. The dynamic
12961 loader resolves the GOT entries when the program starts (the dynamic
12962 loader is not part of GCC; it is part of the operating system). If
12963 the GOT size for the linked executable exceeds a machine-specific
12964 maximum size, you get an error message from the linker indicating that
12965 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
12966 instead. (These maximums are 8k on the SPARC and 32k
12967 on the m68k and RS/6000. The 386 has no such limit.)
12969 Position-independent code requires special support, and therefore works
12970 only on certain machines. For the 386, GCC supports PIC for System V
12971 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
12972 position-independent.
12974 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
12979 If supported for the target machine, emit position-independent code,
12980 suitable for dynamic linking and avoiding any limit on the size of the
12981 global offset table. This option makes a difference on the m68k,
12982 PowerPC and SPARC@.
12984 Position-independent code requires special support, and therefore works
12985 only on certain machines.
12987 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
12994 These options are similar to @option{-fpic} and @option{-fPIC}, but
12995 generated position independent code can be only linked into executables.
12996 Usually these options are used when @option{-pie} GCC option will be
12997 used during linking.
12999 @item -fno-jump-tables
13000 @opindex fno-jump-tables
13001 Do not use jump tables for switch statements even where it would be
13002 more efficient than other code generation strategies. This option is
13003 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13004 building code which forms part of a dynamic linker and cannot
13005 reference the address of a jump table. On some targets, jump tables
13006 do not require a GOT and this option is not needed.
13008 @item -ffixed-@var{reg}
13010 Treat the register named @var{reg} as a fixed register; generated code
13011 should never refer to it (except perhaps as a stack pointer, frame
13012 pointer or in some other fixed role).
13014 @var{reg} must be the name of a register. The register names accepted
13015 are machine-specific and are defined in the @code{REGISTER_NAMES}
13016 macro in the machine description macro file.
13018 This flag does not have a negative form, because it specifies a
13021 @item -fcall-used-@var{reg}
13022 @opindex fcall-used
13023 Treat the register named @var{reg} as an allocable register that is
13024 clobbered by function calls. It may be allocated for temporaries or
13025 variables that do not live across a call. Functions compiled this way
13026 will not save and restore the register @var{reg}.
13028 It is an error to used this flag with the frame pointer or stack pointer.
13029 Use of this flag for other registers that have fixed pervasive roles in
13030 the machine's execution model will produce disastrous results.
13032 This flag does not have a negative form, because it specifies a
13035 @item -fcall-saved-@var{reg}
13036 @opindex fcall-saved
13037 Treat the register named @var{reg} as an allocable register saved by
13038 functions. It may be allocated even for temporaries or variables that
13039 live across a call. Functions compiled this way will save and restore
13040 the register @var{reg} if they use it.
13042 It is an error to used this flag with the frame pointer or stack pointer.
13043 Use of this flag for other registers that have fixed pervasive roles in
13044 the machine's execution model will produce disastrous results.
13046 A different sort of disaster will result from the use of this flag for
13047 a register in which function values may be returned.
13049 This flag does not have a negative form, because it specifies a
13052 @item -fpack-struct[=@var{n}]
13053 @opindex fpack-struct
13054 Without a value specified, pack all structure members together without
13055 holes. When a value is specified (which must be a small power of two), pack
13056 structure members according to this value, representing the maximum
13057 alignment (that is, objects with default alignment requirements larger than
13058 this will be output potentially unaligned at the next fitting location.
13060 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13061 code that is not binary compatible with code generated without that switch.
13062 Additionally, it makes the code suboptimal.
13063 Use it to conform to a non-default application binary interface.
13065 @item -finstrument-functions
13066 @opindex finstrument-functions
13067 Generate instrumentation calls for entry and exit to functions. Just
13068 after function entry and just before function exit, the following
13069 profiling functions will be called with the address of the current
13070 function and its call site. (On some platforms,
13071 @code{__builtin_return_address} does not work beyond the current
13072 function, so the call site information may not be available to the
13073 profiling functions otherwise.)
13076 void __cyg_profile_func_enter (void *this_fn,
13078 void __cyg_profile_func_exit (void *this_fn,
13082 The first argument is the address of the start of the current function,
13083 which may be looked up exactly in the symbol table.
13085 This instrumentation is also done for functions expanded inline in other
13086 functions. The profiling calls will indicate where, conceptually, the
13087 inline function is entered and exited. This means that addressable
13088 versions of such functions must be available. If all your uses of a
13089 function are expanded inline, this may mean an additional expansion of
13090 code size. If you use @samp{extern inline} in your C code, an
13091 addressable version of such functions must be provided. (This is
13092 normally the case anyways, but if you get lucky and the optimizer always
13093 expands the functions inline, you might have gotten away without
13094 providing static copies.)
13096 A function may be given the attribute @code{no_instrument_function}, in
13097 which case this instrumentation will not be done. This can be used, for
13098 example, for the profiling functions listed above, high-priority
13099 interrupt routines, and any functions from which the profiling functions
13100 cannot safely be called (perhaps signal handlers, if the profiling
13101 routines generate output or allocate memory).
13103 @item -fstack-check
13104 @opindex fstack-check
13105 Generate code to verify that you do not go beyond the boundary of the
13106 stack. You should specify this flag if you are running in an
13107 environment with multiple threads, but only rarely need to specify it in
13108 a single-threaded environment since stack overflow is automatically
13109 detected on nearly all systems if there is only one stack.
13111 Note that this switch does not actually cause checking to be done; the
13112 operating system must do that. The switch causes generation of code
13113 to ensure that the operating system sees the stack being extended.
13115 @item -fstack-limit-register=@var{reg}
13116 @itemx -fstack-limit-symbol=@var{sym}
13117 @itemx -fno-stack-limit
13118 @opindex fstack-limit-register
13119 @opindex fstack-limit-symbol
13120 @opindex fno-stack-limit
13121 Generate code to ensure that the stack does not grow beyond a certain value,
13122 either the value of a register or the address of a symbol. If the stack
13123 would grow beyond the value, a signal is raised. For most targets,
13124 the signal is raised before the stack overruns the boundary, so
13125 it is possible to catch the signal without taking special precautions.
13127 For instance, if the stack starts at absolute address @samp{0x80000000}
13128 and grows downwards, you can use the flags
13129 @option{-fstack-limit-symbol=__stack_limit} and
13130 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13131 of 128KB@. Note that this may only work with the GNU linker.
13133 @cindex aliasing of parameters
13134 @cindex parameters, aliased
13135 @item -fargument-alias
13136 @itemx -fargument-noalias
13137 @itemx -fargument-noalias-global
13138 @opindex fargument-alias
13139 @opindex fargument-noalias
13140 @opindex fargument-noalias-global
13141 Specify the possible relationships among parameters and between
13142 parameters and global data.
13144 @option{-fargument-alias} specifies that arguments (parameters) may
13145 alias each other and may alias global storage.@*
13146 @option{-fargument-noalias} specifies that arguments do not alias
13147 each other, but may alias global storage.@*
13148 @option{-fargument-noalias-global} specifies that arguments do not
13149 alias each other and do not alias global storage.
13151 Each language will automatically use whatever option is required by
13152 the language standard. You should not need to use these options yourself.
13154 @item -fleading-underscore
13155 @opindex fleading-underscore
13156 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13157 change the way C symbols are represented in the object file. One use
13158 is to help link with legacy assembly code.
13160 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13161 generate code that is not binary compatible with code generated without that
13162 switch. Use it to conform to a non-default application binary interface.
13163 Not all targets provide complete support for this switch.
13165 @item -ftls-model=@var{model}
13166 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13167 The @var{model} argument should be one of @code{global-dynamic},
13168 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13170 The default without @option{-fpic} is @code{initial-exec}; with
13171 @option{-fpic} the default is @code{global-dynamic}.
13173 @item -fvisibility=@var{default|internal|hidden|protected}
13174 @opindex fvisibility
13175 Set the default ELF image symbol visibility to the specified option---all
13176 symbols will be marked with this unless overridden within the code.
13177 Using this feature can very substantially improve linking and
13178 load times of shared object libraries, produce more optimized
13179 code, provide near-perfect API export and prevent symbol clashes.
13180 It is @strong{strongly} recommended that you use this in any shared objects
13183 Despite the nomenclature, @code{default} always means public ie;
13184 available to be linked against from outside the shared object.
13185 @code{protected} and @code{internal} are pretty useless in real-world
13186 usage so the only other commonly used option will be @code{hidden}.
13187 The default if @option{-fvisibility} isn't specified is
13188 @code{default}, i.e., make every
13189 symbol public---this causes the same behavior as previous versions of
13192 A good explanation of the benefits offered by ensuring ELF
13193 symbols have the correct visibility is given by ``How To Write
13194 Shared Libraries'' by Ulrich Drepper (which can be found at
13195 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13196 solution made possible by this option to marking things hidden when
13197 the default is public is to make the default hidden and mark things
13198 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13199 and @code{__attribute__ ((visibility("default")))} instead of
13200 @code{__declspec(dllexport)} you get almost identical semantics with
13201 identical syntax. This is a great boon to those working with
13202 cross-platform projects.
13204 For those adding visibility support to existing code, you may find
13205 @samp{#pragma GCC visibility} of use. This works by you enclosing
13206 the declarations you wish to set visibility for with (for example)
13207 @samp{#pragma GCC visibility push(hidden)} and
13208 @samp{#pragma GCC visibility pop}.
13209 Bear in mind that symbol visibility should be viewed @strong{as
13210 part of the API interface contract} and thus all new code should
13211 always specify visibility when it is not the default ie; declarations
13212 only for use within the local DSO should @strong{always} be marked explicitly
13213 as hidden as so to avoid PLT indirection overheads---making this
13214 abundantly clear also aids readability and self-documentation of the code.
13215 Note that due to ISO C++ specification requirements, operator new and
13216 operator delete must always be of default visibility.
13218 An overview of these techniques, their benefits and how to use them
13219 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13225 @node Environment Variables
13226 @section Environment Variables Affecting GCC
13227 @cindex environment variables
13229 @c man begin ENVIRONMENT
13230 This section describes several environment variables that affect how GCC
13231 operates. Some of them work by specifying directories or prefixes to use
13232 when searching for various kinds of files. Some are used to specify other
13233 aspects of the compilation environment.
13235 Note that you can also specify places to search using options such as
13236 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13237 take precedence over places specified using environment variables, which
13238 in turn take precedence over those specified by the configuration of GCC@.
13239 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13240 GNU Compiler Collection (GCC) Internals}.
13245 @c @itemx LC_COLLATE
13247 @c @itemx LC_MONETARY
13248 @c @itemx LC_NUMERIC
13253 @c @findex LC_COLLATE
13254 @findex LC_MESSAGES
13255 @c @findex LC_MONETARY
13256 @c @findex LC_NUMERIC
13260 These environment variables control the way that GCC uses
13261 localization information that allow GCC to work with different
13262 national conventions. GCC inspects the locale categories
13263 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13264 so. These locale categories can be set to any value supported by your
13265 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13266 Kingdom encoded in UTF-8.
13268 The @env{LC_CTYPE} environment variable specifies character
13269 classification. GCC uses it to determine the character boundaries in
13270 a string; this is needed for some multibyte encodings that contain quote
13271 and escape characters that would otherwise be interpreted as a string
13274 The @env{LC_MESSAGES} environment variable specifies the language to
13275 use in diagnostic messages.
13277 If the @env{LC_ALL} environment variable is set, it overrides the value
13278 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13279 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13280 environment variable. If none of these variables are set, GCC
13281 defaults to traditional C English behavior.
13285 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13286 files. GCC uses temporary files to hold the output of one stage of
13287 compilation which is to be used as input to the next stage: for example,
13288 the output of the preprocessor, which is the input to the compiler
13291 @item GCC_EXEC_PREFIX
13292 @findex GCC_EXEC_PREFIX
13293 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13294 names of the subprograms executed by the compiler. No slash is added
13295 when this prefix is combined with the name of a subprogram, but you can
13296 specify a prefix that ends with a slash if you wish.
13298 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13299 an appropriate prefix to use based on the pathname it was invoked with.
13301 If GCC cannot find the subprogram using the specified prefix, it
13302 tries looking in the usual places for the subprogram.
13304 The default value of @env{GCC_EXEC_PREFIX} is
13305 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13306 of @code{prefix} when you ran the @file{configure} script.
13308 Other prefixes specified with @option{-B} take precedence over this prefix.
13310 This prefix is also used for finding files such as @file{crt0.o} that are
13313 In addition, the prefix is used in an unusual way in finding the
13314 directories to search for header files. For each of the standard
13315 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13316 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13317 replacing that beginning with the specified prefix to produce an
13318 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13319 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13320 These alternate directories are searched first; the standard directories
13323 @item COMPILER_PATH
13324 @findex COMPILER_PATH
13325 The value of @env{COMPILER_PATH} is a colon-separated list of
13326 directories, much like @env{PATH}. GCC tries the directories thus
13327 specified when searching for subprograms, if it can't find the
13328 subprograms using @env{GCC_EXEC_PREFIX}.
13331 @findex LIBRARY_PATH
13332 The value of @env{LIBRARY_PATH} is a colon-separated list of
13333 directories, much like @env{PATH}. When configured as a native compiler,
13334 GCC tries the directories thus specified when searching for special
13335 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13336 using GCC also uses these directories when searching for ordinary
13337 libraries for the @option{-l} option (but directories specified with
13338 @option{-L} come first).
13342 @cindex locale definition
13343 This variable is used to pass locale information to the compiler. One way in
13344 which this information is used is to determine the character set to be used
13345 when character literals, string literals and comments are parsed in C and C++.
13346 When the compiler is configured to allow multibyte characters,
13347 the following values for @env{LANG} are recognized:
13351 Recognize JIS characters.
13353 Recognize SJIS characters.
13355 Recognize EUCJP characters.
13358 If @env{LANG} is not defined, or if it has some other value, then the
13359 compiler will use mblen and mbtowc as defined by the default locale to
13360 recognize and translate multibyte characters.
13364 Some additional environments variables affect the behavior of the
13367 @include cppenv.texi
13371 @node Precompiled Headers
13372 @section Using Precompiled Headers
13373 @cindex precompiled headers
13374 @cindex speed of compilation
13376 Often large projects have many header files that are included in every
13377 source file. The time the compiler takes to process these header files
13378 over and over again can account for nearly all of the time required to
13379 build the project. To make builds faster, GCC allows users to
13380 `precompile' a header file; then, if builds can use the precompiled
13381 header file they will be much faster.
13383 To create a precompiled header file, simply compile it as you would any
13384 other file, if necessary using the @option{-x} option to make the driver
13385 treat it as a C or C++ header file. You will probably want to use a
13386 tool like @command{make} to keep the precompiled header up-to-date when
13387 the headers it contains change.
13389 A precompiled header file will be searched for when @code{#include} is
13390 seen in the compilation. As it searches for the included file
13391 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13392 compiler looks for a precompiled header in each directory just before it
13393 looks for the include file in that directory. The name searched for is
13394 the name specified in the @code{#include} with @samp{.gch} appended. If
13395 the precompiled header file can't be used, it is ignored.
13397 For instance, if you have @code{#include "all.h"}, and you have
13398 @file{all.h.gch} in the same directory as @file{all.h}, then the
13399 precompiled header file will be used if possible, and the original
13400 header will be used otherwise.
13402 Alternatively, you might decide to put the precompiled header file in a
13403 directory and use @option{-I} to ensure that directory is searched
13404 before (or instead of) the directory containing the original header.
13405 Then, if you want to check that the precompiled header file is always
13406 used, you can put a file of the same name as the original header in this
13407 directory containing an @code{#error} command.
13409 This also works with @option{-include}. So yet another way to use
13410 precompiled headers, good for projects not designed with precompiled
13411 header files in mind, is to simply take most of the header files used by
13412 a project, include them from another header file, precompile that header
13413 file, and @option{-include} the precompiled header. If the header files
13414 have guards against multiple inclusion, they will be skipped because
13415 they've already been included (in the precompiled header).
13417 If you need to precompile the same header file for different
13418 languages, targets, or compiler options, you can instead make a
13419 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13420 header in the directory, perhaps using @option{-o}. It doesn't matter
13421 what you call the files in the directory, every precompiled header in
13422 the directory will be considered. The first precompiled header
13423 encountered in the directory that is valid for this compilation will
13424 be used; they're searched in no particular order.
13426 There are many other possibilities, limited only by your imagination,
13427 good sense, and the constraints of your build system.
13429 A precompiled header file can be used only when these conditions apply:
13433 Only one precompiled header can be used in a particular compilation.
13436 A precompiled header can't be used once the first C token is seen. You
13437 can have preprocessor directives before a precompiled header; you can
13438 even include a precompiled header from inside another header, so long as
13439 there are no C tokens before the @code{#include}.
13442 The precompiled header file must be produced for the same language as
13443 the current compilation. You can't use a C precompiled header for a C++
13447 The precompiled header file must have been produced by the same compiler
13448 binary as the current compilation is using.
13451 Any macros defined before the precompiled header is included must
13452 either be defined in the same way as when the precompiled header was
13453 generated, or must not affect the precompiled header, which usually
13454 means that they don't appear in the precompiled header at all.
13456 The @option{-D} option is one way to define a macro before a
13457 precompiled header is included; using a @code{#define} can also do it.
13458 There are also some options that define macros implicitly, like
13459 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13462 @item If debugging information is output when using the precompiled
13463 header, using @option{-g} or similar, the same kind of debugging information
13464 must have been output when building the precompiled header. However,
13465 a precompiled header built using @option{-g} can be used in a compilation
13466 when no debugging information is being output.
13468 @item The same @option{-m} options must generally be used when building
13469 and using the precompiled header. @xref{Submodel Options},
13470 for any cases where this rule is relaxed.
13472 @item Each of the following options must be the same when building and using
13473 the precompiled header:
13475 @gccoptlist{-fexceptions -funit-at-a-time}
13478 Some other command-line options starting with @option{-f},
13479 @option{-p}, or @option{-O} must be defined in the same way as when
13480 the precompiled header was generated. At present, it's not clear
13481 which options are safe to change and which are not; the safest choice
13482 is to use exactly the same options when generating and using the
13483 precompiled header. The following are known to be safe:
13485 @gccoptlist{-fmessage-length= -fpreprocessed
13486 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13487 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13492 For all of these except the last, the compiler will automatically
13493 ignore the precompiled header if the conditions aren't met. If you
13494 find an option combination that doesn't work and doesn't cause the
13495 precompiled header to be ignored, please consider filing a bug report,
13498 If you do use differing options when generating and using the
13499 precompiled header, the actual behavior will be a mixture of the
13500 behavior for the options. For instance, if you use @option{-g} to
13501 generate the precompiled header but not when using it, you may or may
13502 not get debugging information for routines in the precompiled header.
13504 @node Running Protoize
13505 @section Running Protoize
13507 The program @code{protoize} is an optional part of GCC@. You can use
13508 it to add prototypes to a program, thus converting the program to ISO
13509 C in one respect. The companion program @code{unprotoize} does the
13510 reverse: it removes argument types from any prototypes that are found.
13512 When you run these programs, you must specify a set of source files as
13513 command line arguments. The conversion programs start out by compiling
13514 these files to see what functions they define. The information gathered
13515 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13517 After scanning comes actual conversion. The specified files are all
13518 eligible to be converted; any files they include (whether sources or
13519 just headers) are eligible as well.
13521 But not all the eligible files are converted. By default,
13522 @code{protoize} and @code{unprotoize} convert only source and header
13523 files in the current directory. You can specify additional directories
13524 whose files should be converted with the @option{-d @var{directory}}
13525 option. You can also specify particular files to exclude with the
13526 @option{-x @var{file}} option. A file is converted if it is eligible, its
13527 directory name matches one of the specified directory names, and its
13528 name within the directory has not been excluded.
13530 Basic conversion with @code{protoize} consists of rewriting most
13531 function definitions and function declarations to specify the types of
13532 the arguments. The only ones not rewritten are those for varargs
13535 @code{protoize} optionally inserts prototype declarations at the
13536 beginning of the source file, to make them available for any calls that
13537 precede the function's definition. Or it can insert prototype
13538 declarations with block scope in the blocks where undeclared functions
13541 Basic conversion with @code{unprotoize} consists of rewriting most
13542 function declarations to remove any argument types, and rewriting
13543 function definitions to the old-style pre-ISO form.
13545 Both conversion programs print a warning for any function declaration or
13546 definition that they can't convert. You can suppress these warnings
13549 The output from @code{protoize} or @code{unprotoize} replaces the
13550 original source file. The original file is renamed to a name ending
13551 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13552 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13553 for DOS) file already exists, then the source file is simply discarded.
13555 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13556 scan the program and collect information about the functions it uses.
13557 So neither of these programs will work until GCC is installed.
13559 Here is a table of the options you can use with @code{protoize} and
13560 @code{unprotoize}. Each option works with both programs unless
13564 @item -B @var{directory}
13565 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13566 usual directory (normally @file{/usr/local/lib}). This file contains
13567 prototype information about standard system functions. This option
13568 applies only to @code{protoize}.
13570 @item -c @var{compilation-options}
13571 Use @var{compilation-options} as the options when running @command{gcc} to
13572 produce the @samp{.X} files. The special option @option{-aux-info} is
13573 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13575 Note that the compilation options must be given as a single argument to
13576 @code{protoize} or @code{unprotoize}. If you want to specify several
13577 @command{gcc} options, you must quote the entire set of compilation options
13578 to make them a single word in the shell.
13580 There are certain @command{gcc} arguments that you cannot use, because they
13581 would produce the wrong kind of output. These include @option{-g},
13582 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13583 the @var{compilation-options}, they are ignored.
13586 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13587 systems) instead of @samp{.c}. This is convenient if you are converting
13588 a C program to C++. This option applies only to @code{protoize}.
13591 Add explicit global declarations. This means inserting explicit
13592 declarations at the beginning of each source file for each function
13593 that is called in the file and was not declared. These declarations
13594 precede the first function definition that contains a call to an
13595 undeclared function. This option applies only to @code{protoize}.
13597 @item -i @var{string}
13598 Indent old-style parameter declarations with the string @var{string}.
13599 This option applies only to @code{protoize}.
13601 @code{unprotoize} converts prototyped function definitions to old-style
13602 function definitions, where the arguments are declared between the
13603 argument list and the initial @samp{@{}. By default, @code{unprotoize}
13604 uses five spaces as the indentation. If you want to indent with just
13605 one space instead, use @option{-i " "}.
13608 Keep the @samp{.X} files. Normally, they are deleted after conversion
13612 Add explicit local declarations. @code{protoize} with @option{-l} inserts
13613 a prototype declaration for each function in each block which calls the
13614 function without any declaration. This option applies only to
13618 Make no real changes. This mode just prints information about the conversions
13619 that would have been done without @option{-n}.
13622 Make no @samp{.save} files. The original files are simply deleted.
13623 Use this option with caution.
13625 @item -p @var{program}
13626 Use the program @var{program} as the compiler. Normally, the name
13627 @file{gcc} is used.
13630 Work quietly. Most warnings are suppressed.
13633 Print the version number, just like @option{-v} for @command{gcc}.
13636 If you need special compiler options to compile one of your program's
13637 source files, then you should generate that file's @samp{.X} file
13638 specially, by running @command{gcc} on that source file with the
13639 appropriate options and the option @option{-aux-info}. Then run
13640 @code{protoize} on the entire set of files. @code{protoize} will use
13641 the existing @samp{.X} file because it is newer than the source file.
13645 gcc -Dfoo=bar file1.c -aux-info file1.X
13650 You need to include the special files along with the rest in the
13651 @code{protoize} command, even though their @samp{.X} files already
13652 exist, because otherwise they won't get converted.
13654 @xref{Protoize Caveats}, for more information on how to use
13655 @code{protoize} successfully.