1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
11 @c man begin COPYRIGHT
12 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
13 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
15 Permission is granted to copy, distribute and/or modify this document
16 under the terms of the GNU Free Documentation License, Version 1.2 or
17 any later version published by the Free Software Foundation; with the
18 Invariant Sections being ``GNU General Public License'' and ``Funding
19 Free Software'', the Front-Cover texts being (a) (see below), and with
20 the Back-Cover Texts being (b) (see below). A copy of the license is
21 included in the gfdl(7) man page.
23 (a) The FSF's Front-Cover Text is:
27 (b) The FSF's Back-Cover Text is:
29 You have freedom to copy and modify this GNU Manual, like GNU
30 software. Copies published by the Free Software Foundation raise
31 funds for GNU development.
33 @c Set file name and title for the man page.
35 @settitle GNU project C and C++ compiler
37 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
38 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
39 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
40 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
41 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
42 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
43 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
45 Only the most useful options are listed here; see below for the
46 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
49 gpl(7), gfdl(7), fsf-funding(7),
50 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
51 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
52 @file{ld}, @file{binutils} and @file{gdb}.
55 For instructions on reporting bugs, see
56 @w{@uref{http://gcc.gnu.org/bugs.html}}.
59 See the Info entry for @command{gcc}, or
60 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
61 for contributors to GCC@.
66 @chapter GCC Command Options
67 @cindex GCC command options
68 @cindex command options
69 @cindex options, GCC command
71 @c man begin DESCRIPTION
72 When you invoke GCC, it normally does preprocessing, compilation,
73 assembly and linking. The ``overall options'' allow you to stop this
74 process at an intermediate stage. For example, the @option{-c} option
75 says not to run the linker. Then the output consists of object files
76 output by the assembler.
78 Other options are passed on to one stage of processing. Some options
79 control the preprocessor and others the compiler itself. Yet other
80 options control the assembler and linker; most of these are not
81 documented here, since you rarely need to use any of them.
83 @cindex C compilation options
84 Most of the command line options that you can use with GCC are useful
85 for C programs; when an option is only useful with another language
86 (usually C++), the explanation says so explicitly. If the description
87 for a particular option does not mention a source language, you can use
88 that option with all supported languages.
90 @cindex C++ compilation options
91 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
92 options for compiling C++ programs.
94 @cindex grouping options
95 @cindex options, grouping
96 The @command{gcc} program accepts options and file names as operands. Many
97 options have multi-letter names; therefore multiple single-letter options
98 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
101 @cindex order of options
102 @cindex options, order
103 You can mix options and other arguments. For the most part, the order
104 you use doesn't matter. Order does matter when you use several options
105 of the same kind; for example, if you specify @option{-L} more than once,
106 the directories are searched in the order specified.
108 Many options have long names starting with @samp{-f} or with
109 @samp{-W}---for example,
110 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
111 these have both positive and negative forms; the negative form of
112 @option{-ffoo} would be @option{-fno-foo}. This manual documents
113 only one of these two forms, whichever one is not the default.
117 @xref{Option Index}, for an index to GCC's options.
120 * Option Summary:: Brief list of all options, without explanations.
121 * Overall Options:: Controlling the kind of output:
122 an executable, object files, assembler files,
123 or preprocessed source.
124 * Invoking G++:: Compiling C++ programs.
125 * C Dialect Options:: Controlling the variant of C language compiled.
126 * C++ Dialect Options:: Variations on C++.
127 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
129 * Language Independent Options:: Controlling how diagnostics should be
131 * Warning Options:: How picky should the compiler be?
132 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
133 * Optimize Options:: How much optimization?
134 * Preprocessor Options:: Controlling header files and macro definitions.
135 Also, getting dependency information for Make.
136 * Assembler Options:: Passing options to the assembler.
137 * Link Options:: Specifying libraries and so on.
138 * Directory Options:: Where to find header files and libraries.
139 Where to find the compiler executable files.
140 * Spec Files:: How to pass switches to sub-processes.
141 * Target Options:: Running a cross-compiler, or an old version of GCC.
142 * Submodel Options:: Specifying minor hardware or convention variations,
143 such as 68010 vs 68020.
144 * Code Gen Options:: Specifying conventions for function calls, data layout
146 * Environment Variables:: Env vars that affect GCC.
147 * Precompiled Headers:: Compiling a header once, and using it many times.
148 * Running Protoize:: Automatically adding or removing function prototypes.
154 @section Option Summary
156 Here is a summary of all the options, grouped by type. Explanations are
157 in the following sections.
160 @item Overall Options
161 @xref{Overall Options,,Options Controlling the Kind of Output}.
162 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
163 -x @var{language} -v -### --help --target-help --version @@@var{file}}
165 @item C Language Options
166 @xref{C Dialect Options,,Options Controlling C Dialect}.
167 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
168 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
169 -fhosted -ffreestanding -fms-extensions @gol
170 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
171 -fallow-single-precision -fcond-mismatch @gol
172 -fsigned-bitfields -fsigned-char @gol
173 -funsigned-bitfields -funsigned-char}
175 @item C++ Language Options
176 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
177 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
178 -fconserve-space -ffriend-injection @gol
179 -fno-elide-constructors @gol
180 -fno-enforce-eh-specs @gol
181 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
182 -fno-implicit-templates @gol
183 -fno-implicit-inline-templates @gol
184 -fno-implement-inlines -fms-extensions @gol
185 -fno-nonansi-builtins -fno-operator-names @gol
186 -fno-optional-diags -fpermissive @gol
187 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
188 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
189 -fno-default-inline -fvisibility-inlines-hidden @gol
190 -Wabi -Wctor-dtor-privacy @gol
191 -Wnon-virtual-dtor -Wreorder @gol
192 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
193 -Wno-non-template-friend -Wold-style-cast @gol
194 -Woverloaded-virtual -Wno-pmf-conversions @gol
197 @item Objective-C and Objective-C++ Language Options
198 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
199 Objective-C and Objective-C++ Dialects}.
201 -fconstant-string-class=@var{class-name} @gol
202 -fgnu-runtime -fnext-runtime @gol
203 -fno-nil-receivers @gol
204 -fobjc-call-cxx-cdtors @gol
205 -fobjc-direct-dispatch @gol
206 -fobjc-exceptions @gol
208 -freplace-objc-classes @gol
211 -Wassign-intercept @gol
212 -Wno-protocol -Wselector @gol
213 -Wstrict-selector-match @gol
214 -Wundeclared-selector}
216 @item Language Independent Options
217 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
218 @gccoptlist{-fmessage-length=@var{n} @gol
219 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}} @gol
220 -fdiagnostics-show-options
222 @item Warning Options
223 @xref{Warning Options,,Options to Request or Suppress Warnings}.
224 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
225 -w -Wextra -Wall -Waggregate-return -Walways-true -Wno-attributes @gol
226 -Wc++-compat -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
227 -Wconversion -Wno-deprecated-declarations @gol
228 -Wdisabled-optimization -Wno-div-by-zero -Wno-endif-labels @gol
229 -Werror -Werror-* -Werror-implicit-function-declaration @gol
230 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
231 -Wno-format-extra-args -Wformat-nonliteral @gol
232 -Wformat-security -Wformat-y2k @gol
233 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
234 -Wimport -Wno-import -Winit-self -Winline @gol
235 -Wno-int-to-pointer-cast @gol
236 -Wno-invalid-offsetof -Winvalid-pch @gol
237 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
238 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
239 -Wmissing-format-attribute -Wmissing-include-dirs @gol
240 -Wmissing-noreturn @gol
241 -Wno-multichar -Wnonnull -Woverlength-strings -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 -Wpointer-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 -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 -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 @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 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
337 -fstack-protector -fstack-protector-all @gol
338 -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 -fipa-pta -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{GNU/Linux Options}
507 @gccoptlist{-muclibc}
509 @emph{H8/300 Options}
510 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
513 @gccoptlist{-march=@var{architecture-type} @gol
514 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
515 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
516 -mfixed-range=@var{register-range} @gol
517 -mjump-in-delay -mlinker-opt -mlong-calls @gol
518 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
519 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
520 -mno-jump-in-delay -mno-long-load-store @gol
521 -mno-portable-runtime -mno-soft-float @gol
522 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
523 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
524 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
525 -munix=@var{unix-std} -nolibdld -static -threads}
527 @emph{i386 and x86-64 Options}
528 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
529 -mfpmath=@var{unit} @gol
530 -masm=@var{dialect} -mno-fancy-math-387 @gol
531 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
532 -mno-wide-multiply -mrtd -malign-double @gol
533 -mpreferred-stack-boundary=@var{num} @gol
534 -mmmx -msse -msse2 -msse3 -m3dnow -msselibm @gol
535 -mthreads -mno-align-stringops -minline-all-stringops @gol
536 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
537 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
538 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
539 -mcmodel=@var{code-model} @gol
540 -m32 -m64 -mlarge-data-threshold=@var{num}}
543 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
544 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
545 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
546 -minline-float-divide-max-throughput @gol
547 -minline-int-divide-min-latency @gol
548 -minline-int-divide-max-throughput @gol
549 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
550 -mno-dwarf2-asm -mearly-stop-bits @gol
551 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
552 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
553 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
554 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
555 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
556 -mno-sched-prefer-non-data-spec-insns @gol
557 -mno-sched-prefer-non-control-spec-insns @gol
558 -mno-sched-count-spec-in-critical-path}
560 @emph{M32R/D Options}
561 @gccoptlist{-m32r2 -m32rx -m32r @gol
563 -malign-loops -mno-align-loops @gol
564 -missue-rate=@var{number} @gol
565 -mbranch-cost=@var{number} @gol
566 -mmodel=@var{code-size-model-type} @gol
567 -msdata=@var{sdata-type} @gol
568 -mno-flush-func -mflush-func=@var{name} @gol
569 -mno-flush-trap -mflush-trap=@var{number} @gol
573 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
575 @emph{M680x0 Options}
576 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
577 -m68060 -mcpu32 -m5200 -mcfv4e -m68881 -mbitfield @gol
578 -mc68000 -mc68020 @gol
579 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
580 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
581 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
583 @emph{M68hc1x Options}
584 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
585 -mauto-incdec -minmax -mlong-calls -mshort @gol
586 -msoft-reg-count=@var{count}}
589 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
590 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
591 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
592 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
593 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
596 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
597 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
598 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
599 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
600 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
601 -mdsp -mpaired-single -mips3d @gol
602 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
603 -G@var{num} -membedded-data -mno-embedded-data @gol
604 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
605 -msplit-addresses -mno-split-addresses @gol
606 -mexplicit-relocs -mno-explicit-relocs @gol
607 -mcheck-zero-division -mno-check-zero-division @gol
608 -mdivide-traps -mdivide-breaks @gol
609 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
610 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
611 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
612 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
613 -mfix-sb1 -mno-fix-sb1 @gol
614 -mflush-func=@var{func} -mno-flush-func @gol
615 -mbranch-likely -mno-branch-likely @gol
616 -mfp-exceptions -mno-fp-exceptions @gol
617 -mvr4130-align -mno-vr4130-align}
620 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
621 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
622 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
623 -mno-base-addresses -msingle-exit -mno-single-exit}
625 @emph{MN10300 Options}
626 @gccoptlist{-mmult-bug -mno-mult-bug @gol
627 -mam33 -mno-am33 @gol
628 -mam33-2 -mno-am33-2 @gol
629 -mreturn-pointer-on-d0 @gol
633 @gccoptlist{-mno-crt0 -mbacc -msim @gol
634 -march=@var{cpu-type} }
636 @emph{PDP-11 Options}
637 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
638 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
639 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
640 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
641 -mbranch-expensive -mbranch-cheap @gol
642 -msplit -mno-split -munix-asm -mdec-asm}
644 @emph{PowerPC Options}
645 See RS/6000 and PowerPC Options.
647 @emph{RS/6000 and PowerPC Options}
648 @gccoptlist{-mcpu=@var{cpu-type} @gol
649 -mtune=@var{cpu-type} @gol
650 -mpower -mno-power -mpower2 -mno-power2 @gol
651 -mpowerpc -mpowerpc64 -mno-powerpc @gol
652 -maltivec -mno-altivec @gol
653 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
654 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
655 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
656 -mnew-mnemonics -mold-mnemonics @gol
657 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
658 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
659 -malign-power -malign-natural @gol
660 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
661 -mstring -mno-string -mupdate -mno-update @gol
662 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
663 -mstrict-align -mno-strict-align -mrelocatable @gol
664 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
665 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
666 -mdynamic-no-pic -maltivec -mswdiv @gol
667 -mprioritize-restricted-insns=@var{priority} @gol
668 -msched-costly-dep=@var{dependence_type} @gol
669 -minsert-sched-nops=@var{scheme} @gol
670 -mcall-sysv -mcall-netbsd @gol
671 -maix-struct-return -msvr4-struct-return @gol
672 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
673 -misel -mno-isel @gol
674 -misel=yes -misel=no @gol
676 -mspe=yes -mspe=no @gol
677 -mvrsave -mno-vrsave @gol
678 -mmulhw -mno-mulhw @gol
679 -mdlmzb -mno-dlmzb @gol
680 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
681 -mprototype -mno-prototype @gol
682 -msim -mmvme -mads -myellowknife -memb -msdata @gol
683 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
685 @emph{S/390 and zSeries Options}
686 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
687 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
688 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
689 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
690 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
691 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
692 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
695 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
696 -m4-nofpu -m4-single-only -m4-single -m4 @gol
697 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
698 -m5-64media -m5-64media-nofpu @gol
699 -m5-32media -m5-32media-nofpu @gol
700 -m5-compact -m5-compact-nofpu @gol
701 -mb -ml -mdalign -mrelax @gol
702 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
703 -mieee -misize -mpadstruct -mspace @gol
704 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
705 -mdivsi3_libfunc=@var{name} @gol
706 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
710 @gccoptlist{-mcpu=@var{cpu-type} @gol
711 -mtune=@var{cpu-type} @gol
712 -mcmodel=@var{code-model} @gol
713 -m32 -m64 -mapp-regs -mno-app-regs @gol
714 -mfaster-structs -mno-faster-structs @gol
715 -mfpu -mno-fpu -mhard-float -msoft-float @gol
716 -mhard-quad-float -msoft-quad-float @gol
717 -mimpure-text -mno-impure-text -mlittle-endian @gol
718 -mstack-bias -mno-stack-bias @gol
719 -munaligned-doubles -mno-unaligned-doubles @gol
720 -mv8plus -mno-v8plus -mvis -mno-vis
721 -threads -pthreads -pthread}
723 @emph{System V Options}
724 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
726 @emph{TMS320C3x/C4x Options}
727 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
728 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
729 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
730 -mparallel-insns -mparallel-mpy -mpreserve-float}
733 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
734 -mprolog-function -mno-prolog-function -mspace @gol
735 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
736 -mapp-regs -mno-app-regs @gol
737 -mdisable-callt -mno-disable-callt @gol
743 @gccoptlist{-mg -mgnu -munix}
745 @emph{x86-64 Options}
746 See i386 and x86-64 Options.
748 @emph{Xstormy16 Options}
751 @emph{Xtensa Options}
752 @gccoptlist{-mconst16 -mno-const16 @gol
753 -mfused-madd -mno-fused-madd @gol
754 -mtext-section-literals -mno-text-section-literals @gol
755 -mtarget-align -mno-target-align @gol
756 -mlongcalls -mno-longcalls}
758 @emph{zSeries Options}
759 See S/390 and zSeries Options.
761 @item Code Generation Options
762 @xref{Code Gen Options,,Options for Code Generation Conventions}.
763 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
764 -ffixed-@var{reg} -fexceptions @gol
765 -fnon-call-exceptions -funwind-tables @gol
766 -fasynchronous-unwind-tables @gol
767 -finhibit-size-directive -finstrument-functions @gol
768 -fno-common -fno-ident @gol
769 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
770 -fno-jump-tables @gol
771 -freg-struct-return -fshort-enums @gol
772 -fshort-double -fshort-wchar @gol
773 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
774 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
775 -fargument-alias -fargument-noalias @gol
776 -fargument-noalias-global -fleading-underscore @gol
777 -ftls-model=@var{model} @gol
778 -ftrapv -fwrapv -fbounds-check @gol
779 -fvisibility -fopenmp}
783 * Overall Options:: Controlling the kind of output:
784 an executable, object files, assembler files,
785 or preprocessed source.
786 * C Dialect Options:: Controlling the variant of C language compiled.
787 * C++ Dialect Options:: Variations on C++.
788 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
790 * Language Independent Options:: Controlling how diagnostics should be
792 * Warning Options:: How picky should the compiler be?
793 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
794 * Optimize Options:: How much optimization?
795 * Preprocessor Options:: Controlling header files and macro definitions.
796 Also, getting dependency information for Make.
797 * Assembler Options:: Passing options to the assembler.
798 * Link Options:: Specifying libraries and so on.
799 * Directory Options:: Where to find header files and libraries.
800 Where to find the compiler executable files.
801 * Spec Files:: How to pass switches to sub-processes.
802 * Target Options:: Running a cross-compiler, or an old version of GCC.
805 @node Overall Options
806 @section Options Controlling the Kind of Output
808 Compilation can involve up to four stages: preprocessing, compilation
809 proper, assembly and linking, always in that order. GCC is capable of
810 preprocessing and compiling several files either into several
811 assembler input files, or into one assembler input file; then each
812 assembler input file produces an object file, and linking combines all
813 the object files (those newly compiled, and those specified as input)
814 into an executable file.
816 @cindex file name suffix
817 For any given input file, the file name suffix determines what kind of
822 C source code which must be preprocessed.
825 C source code which should not be preprocessed.
828 C++ source code which should not be preprocessed.
831 Objective-C source code. Note that you must link with the @file{libobjc}
832 library to make an Objective-C program work.
835 Objective-C source code which should not be preprocessed.
839 Objective-C++ source code. Note that you must link with the @file{libobjc}
840 library to make an Objective-C++ program work. Note that @samp{.M} refers
841 to a literal capital M@.
844 Objective-C++ source code which should not be preprocessed.
847 C, C++, Objective-C or Objective-C++ header file to be turned into a
852 @itemx @var{file}.cxx
853 @itemx @var{file}.cpp
854 @itemx @var{file}.CPP
855 @itemx @var{file}.c++
857 C++ source code which must be preprocessed. Note that in @samp{.cxx},
858 the last two letters must both be literally @samp{x}. Likewise,
859 @samp{.C} refers to a literal capital C@.
863 Objective-C++ source code which must be preprocessed.
866 Objective-C++ source code which should not be preprocessed.
870 C++ header file to be turned into a precompiled header.
873 @itemx @var{file}.for
874 @itemx @var{file}.FOR
875 Fixed form Fortran source code which should not be preprocessed.
878 @itemx @var{file}.fpp
879 @itemx @var{file}.FPP
880 Fixed form Fortran source code which must be preprocessed (with the traditional
884 @itemx @var{file}.f95
885 Free form Fortran source code which should not be preprocessed.
888 @itemx @var{file}.F95
889 Free form Fortran source code which must be preprocessed (with the
890 traditional preprocessor).
892 @c FIXME: Descriptions of Java file types.
899 Ada source code file which contains a library unit declaration (a
900 declaration of a package, subprogram, or generic, or a generic
901 instantiation), or a library unit renaming declaration (a package,
902 generic, or subprogram renaming declaration). Such files are also
905 @itemx @var{file}.adb
906 Ada source code file containing a library unit body (a subprogram or
907 package body). Such files are also called @dfn{bodies}.
909 @c GCC also knows about some suffixes for languages not yet included:
920 Assembler code which must be preprocessed.
923 An object file to be fed straight into linking.
924 Any file name with no recognized suffix is treated this way.
928 You can specify the input language explicitly with the @option{-x} option:
931 @item -x @var{language}
932 Specify explicitly the @var{language} for the following input files
933 (rather than letting the compiler choose a default based on the file
934 name suffix). This option applies to all following input files until
935 the next @option{-x} option. Possible values for @var{language} are:
937 c c-header c-cpp-output
938 c++ c++-header c++-cpp-output
939 objective-c objective-c-header objective-c-cpp-output
940 objective-c++ objective-c++-header objective-c++-cpp-output
941 assembler assembler-with-cpp
950 Turn off any specification of a language, so that subsequent files are
951 handled according to their file name suffixes (as they are if @option{-x}
952 has not been used at all).
954 @item -pass-exit-codes
955 @opindex pass-exit-codes
956 Normally the @command{gcc} program will exit with the code of 1 if any
957 phase of the compiler returns a non-success return code. If you specify
958 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
959 numerically highest error produced by any phase that returned an error
963 If you only want some of the stages of compilation, you can use
964 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
965 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
966 @command{gcc} is to stop. Note that some combinations (for example,
967 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
972 Compile or assemble the source files, but do not link. The linking
973 stage simply is not done. The ultimate output is in the form of an
974 object file for each source file.
976 By default, the object file name for a source file is made by replacing
977 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
979 Unrecognized input files, not requiring compilation or assembly, are
984 Stop after the stage of compilation proper; do not assemble. The output
985 is in the form of an assembler code file for each non-assembler input
988 By default, the assembler file name for a source file is made by
989 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
991 Input files that don't require compilation are ignored.
995 Stop after the preprocessing stage; do not run the compiler proper. The
996 output is in the form of preprocessed source code, which is sent to the
999 Input files which don't require preprocessing are ignored.
1001 @cindex output file option
1004 Place output in file @var{file}. This applies regardless to whatever
1005 sort of output is being produced, whether it be an executable file,
1006 an object file, an assembler file or preprocessed C code.
1008 If @option{-o} is not specified, the default is to put an executable
1009 file in @file{a.out}, the object file for
1010 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1011 assembler file in @file{@var{source}.s}, a precompiled header file in
1012 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1017 Print (on standard error output) the commands executed to run the stages
1018 of compilation. Also print the version number of the compiler driver
1019 program and of the preprocessor and the compiler proper.
1023 Like @option{-v} except the commands are not executed and all command
1024 arguments are quoted. This is useful for shell scripts to capture the
1025 driver-generated command lines.
1029 Use pipes rather than temporary files for communication between the
1030 various stages of compilation. This fails to work on some systems where
1031 the assembler is unable to read from a pipe; but the GNU assembler has
1036 If you are compiling multiple source files, this option tells the driver
1037 to pass all the source files to the compiler at once (for those
1038 languages for which the compiler can handle this). This will allow
1039 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1040 language for which this is supported is C@. If you pass source files for
1041 multiple languages to the driver, using this option, the driver will invoke
1042 the compiler(s) that support IMA once each, passing each compiler all the
1043 source files appropriate for it. For those languages that do not support
1044 IMA this option will be ignored, and the compiler will be invoked once for
1045 each source file in that language. If you use this option in conjunction
1046 with @option{-save-temps}, the compiler will generate multiple
1048 (one for each source file), but only one (combined) @file{.o} or
1053 Print (on the standard output) a description of the command line options
1054 understood by @command{gcc}. If the @option{-v} option is also specified
1055 then @option{--help} will also be passed on to the various processes
1056 invoked by @command{gcc}, so that they can display the command line options
1057 they accept. If the @option{-Wextra} option is also specified then command
1058 line options which have no documentation associated with them will also
1062 @opindex target-help
1063 Print (on the standard output) a description of target specific command
1064 line options for each tool.
1068 Display the version number and copyrights of the invoked GCC@.
1070 @include @value{srcdir}/../libiberty/at-file.texi
1074 @section Compiling C++ Programs
1076 @cindex suffixes for C++ source
1077 @cindex C++ source file suffixes
1078 C++ source files conventionally use one of the suffixes @samp{.C},
1079 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1080 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1081 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1082 files with these names and compiles them as C++ programs even if you
1083 call the compiler the same way as for compiling C programs (usually
1084 with the name @command{gcc}).
1088 However, C++ programs often require class libraries as well as a
1089 compiler that understands the C++ language---and under some
1090 circumstances, you might want to compile programs or header files from
1091 standard input, or otherwise without a suffix that flags them as C++
1092 programs. You might also like to precompile a C header file with a
1093 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1094 program that calls GCC with the default language set to C++, and
1095 automatically specifies linking against the C++ library. On many
1096 systems, @command{g++} is also installed with the name @command{c++}.
1098 @cindex invoking @command{g++}
1099 When you compile C++ programs, you may specify many of the same
1100 command-line options that you use for compiling programs in any
1101 language; or command-line options meaningful for C and related
1102 languages; or options that are meaningful only for C++ programs.
1103 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1104 explanations of options for languages related to C@.
1105 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1106 explanations of options that are meaningful only for C++ programs.
1108 @node C Dialect Options
1109 @section Options Controlling C Dialect
1110 @cindex dialect options
1111 @cindex language dialect options
1112 @cindex options, dialect
1114 The following options control the dialect of C (or languages derived
1115 from C, such as C++, Objective-C and Objective-C++) that the compiler
1119 @cindex ANSI support
1123 In C mode, support all ISO C90 programs. In C++ mode,
1124 remove GNU extensions that conflict with ISO C++.
1126 This turns off certain features of GCC that are incompatible with ISO
1127 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1128 such as the @code{asm} and @code{typeof} keywords, and
1129 predefined macros such as @code{unix} and @code{vax} that identify the
1130 type of system you are using. It also enables the undesirable and
1131 rarely used ISO trigraph feature. For the C compiler,
1132 it disables recognition of C++ style @samp{//} comments as well as
1133 the @code{inline} keyword.
1135 The alternate keywords @code{__asm__}, @code{__extension__},
1136 @code{__inline__} and @code{__typeof__} continue to work despite
1137 @option{-ansi}. You would not want to use them in an ISO C program, of
1138 course, but it is useful to put them in header files that might be included
1139 in compilations done with @option{-ansi}. Alternate predefined macros
1140 such as @code{__unix__} and @code{__vax__} are also available, with or
1141 without @option{-ansi}.
1143 The @option{-ansi} option does not cause non-ISO programs to be
1144 rejected gratuitously. For that, @option{-pedantic} is required in
1145 addition to @option{-ansi}. @xref{Warning Options}.
1147 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1148 option is used. Some header files may notice this macro and refrain
1149 from declaring certain functions or defining certain macros that the
1150 ISO standard doesn't call for; this is to avoid interfering with any
1151 programs that might use these names for other things.
1153 Functions which would normally be built in but do not have semantics
1154 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1155 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1156 built-in functions provided by GCC}, for details of the functions
1161 Determine the language standard. This option is currently only
1162 supported when compiling C or C++. A value for this option must be
1163 provided; possible values are
1168 ISO C90 (same as @option{-ansi}).
1170 @item iso9899:199409
1171 ISO C90 as modified in amendment 1.
1177 ISO C99. Note that this standard is not yet fully supported; see
1178 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1179 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1182 Default, ISO C90 plus GNU extensions (including some C99 features).
1186 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1187 this will become the default. The name @samp{gnu9x} is deprecated.
1190 The 1998 ISO C++ standard plus amendments.
1193 The same as @option{-std=c++98} plus GNU extensions. This is the
1194 default for C++ code.
1197 Even when this option is not specified, you can still use some of the
1198 features of newer standards in so far as they do not conflict with
1199 previous C standards. For example, you may use @code{__restrict__} even
1200 when @option{-std=c99} is not specified.
1202 The @option{-std} options specifying some version of ISO C have the same
1203 effects as @option{-ansi}, except that features that were not in ISO C90
1204 but are in the specified version (for example, @samp{//} comments and
1205 the @code{inline} keyword in ISO C99) are not disabled.
1207 @xref{Standards,,Language Standards Supported by GCC}, for details of
1208 these standard versions.
1210 @item -aux-info @var{filename}
1212 Output to the given filename prototyped declarations for all functions
1213 declared and/or defined in a translation unit, including those in header
1214 files. This option is silently ignored in any language other than C@.
1216 Besides declarations, the file indicates, in comments, the origin of
1217 each declaration (source file and line), whether the declaration was
1218 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1219 @samp{O} for old, respectively, in the first character after the line
1220 number and the colon), and whether it came from a declaration or a
1221 definition (@samp{C} or @samp{F}, respectively, in the following
1222 character). In the case of function definitions, a K&R-style list of
1223 arguments followed by their declarations is also provided, inside
1224 comments, after the declaration.
1228 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1229 keyword, so that code can use these words as identifiers. You can use
1230 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1231 instead. @option{-ansi} implies @option{-fno-asm}.
1233 In C++, this switch only affects the @code{typeof} keyword, since
1234 @code{asm} and @code{inline} are standard keywords. You may want to
1235 use the @option{-fno-gnu-keywords} flag instead, which has the same
1236 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1237 switch only affects the @code{asm} and @code{typeof} keywords, since
1238 @code{inline} is a standard keyword in ISO C99.
1241 @itemx -fno-builtin-@var{function}
1242 @opindex fno-builtin
1243 @cindex built-in functions
1244 Don't recognize built-in functions that do not begin with
1245 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1246 functions provided by GCC}, for details of the functions affected,
1247 including those which are not built-in functions when @option{-ansi} or
1248 @option{-std} options for strict ISO C conformance are used because they
1249 do not have an ISO standard meaning.
1251 GCC normally generates special code to handle certain built-in functions
1252 more efficiently; for instance, calls to @code{alloca} may become single
1253 instructions that adjust the stack directly, and calls to @code{memcpy}
1254 may become inline copy loops. The resulting code is often both smaller
1255 and faster, but since the function calls no longer appear as such, you
1256 cannot set a breakpoint on those calls, nor can you change the behavior
1257 of the functions by linking with a different library. In addition,
1258 when a function is recognized as a built-in function, GCC may use
1259 information about that function to warn about problems with calls to
1260 that function, or to generate more efficient code, even if the
1261 resulting code still contains calls to that function. For example,
1262 warnings are given with @option{-Wformat} for bad calls to
1263 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1264 known not to modify global memory.
1266 With the @option{-fno-builtin-@var{function}} option
1267 only the built-in function @var{function} is
1268 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1269 function is named this is not built-in in this version of GCC, this
1270 option is ignored. There is no corresponding
1271 @option{-fbuiltin-@var{function}} option; if you wish to enable
1272 built-in functions selectively when using @option{-fno-builtin} or
1273 @option{-ffreestanding}, you may define macros such as:
1276 #define abs(n) __builtin_abs ((n))
1277 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1282 @cindex hosted environment
1284 Assert that compilation takes place in a hosted environment. This implies
1285 @option{-fbuiltin}. A hosted environment is one in which the
1286 entire standard library is available, and in which @code{main} has a return
1287 type of @code{int}. Examples are nearly everything except a kernel.
1288 This is equivalent to @option{-fno-freestanding}.
1290 @item -ffreestanding
1291 @opindex ffreestanding
1292 @cindex hosted environment
1294 Assert that compilation takes place in a freestanding environment. This
1295 implies @option{-fno-builtin}. A freestanding environment
1296 is one in which the standard library may not exist, and program startup may
1297 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1298 This is equivalent to @option{-fno-hosted}.
1300 @xref{Standards,,Language Standards Supported by GCC}, for details of
1301 freestanding and hosted environments.
1303 @item -fms-extensions
1304 @opindex fms-extensions
1305 Accept some non-standard constructs used in Microsoft header files.
1307 Some cases of unnamed fields in structures and unions are only
1308 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1309 fields within structs/unions}, for details.
1313 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1314 options for strict ISO C conformance) implies @option{-trigraphs}.
1316 @item -no-integrated-cpp
1317 @opindex no-integrated-cpp
1318 Performs a compilation in two passes: preprocessing and compiling. This
1319 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1320 @option{-B} option. The user supplied compilation step can then add in
1321 an additional preprocessing step after normal preprocessing but before
1322 compiling. The default is to use the integrated cpp (internal cpp)
1324 The semantics of this option will change if "cc1", "cc1plus", and
1325 "cc1obj" are merged.
1327 @cindex traditional C language
1328 @cindex C language, traditional
1330 @itemx -traditional-cpp
1331 @opindex traditional-cpp
1332 @opindex traditional
1333 Formerly, these options caused GCC to attempt to emulate a pre-standard
1334 C compiler. They are now only supported with the @option{-E} switch.
1335 The preprocessor continues to support a pre-standard mode. See the GNU
1336 CPP manual for details.
1338 @item -fcond-mismatch
1339 @opindex fcond-mismatch
1340 Allow conditional expressions with mismatched types in the second and
1341 third arguments. The value of such an expression is void. This option
1342 is not supported for C++.
1344 @item -funsigned-char
1345 @opindex funsigned-char
1346 Let the type @code{char} be unsigned, like @code{unsigned char}.
1348 Each kind of machine has a default for what @code{char} should
1349 be. It is either like @code{unsigned char} by default or like
1350 @code{signed char} by default.
1352 Ideally, a portable program should always use @code{signed char} or
1353 @code{unsigned char} when it depends on the signedness of an object.
1354 But many programs have been written to use plain @code{char} and
1355 expect it to be signed, or expect it to be unsigned, depending on the
1356 machines they were written for. This option, and its inverse, let you
1357 make such a program work with the opposite default.
1359 The type @code{char} is always a distinct type from each of
1360 @code{signed char} or @code{unsigned char}, even though its behavior
1361 is always just like one of those two.
1364 @opindex fsigned-char
1365 Let the type @code{char} be signed, like @code{signed char}.
1367 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1368 the negative form of @option{-funsigned-char}. Likewise, the option
1369 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1371 @item -fsigned-bitfields
1372 @itemx -funsigned-bitfields
1373 @itemx -fno-signed-bitfields
1374 @itemx -fno-unsigned-bitfields
1375 @opindex fsigned-bitfields
1376 @opindex funsigned-bitfields
1377 @opindex fno-signed-bitfields
1378 @opindex fno-unsigned-bitfields
1379 These options control whether a bit-field is signed or unsigned, when the
1380 declaration does not use either @code{signed} or @code{unsigned}. By
1381 default, such a bit-field is signed, because this is consistent: the
1382 basic integer types such as @code{int} are signed types.
1385 @node C++ Dialect Options
1386 @section Options Controlling C++ Dialect
1388 @cindex compiler options, C++
1389 @cindex C++ options, command line
1390 @cindex options, C++
1391 This section describes the command-line options that are only meaningful
1392 for C++ programs; but you can also use most of the GNU compiler options
1393 regardless of what language your program is in. For example, you
1394 might compile a file @code{firstClass.C} like this:
1397 g++ -g -frepo -O -c firstClass.C
1401 In this example, only @option{-frepo} is an option meant
1402 only for C++ programs; you can use the other options with any
1403 language supported by GCC@.
1405 Here is a list of options that are @emph{only} for compiling C++ programs:
1409 @item -fabi-version=@var{n}
1410 @opindex fabi-version
1411 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1412 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1413 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1414 the version that conforms most closely to the C++ ABI specification.
1415 Therefore, the ABI obtained using version 0 will change as ABI bugs
1418 The default is version 2.
1420 @item -fno-access-control
1421 @opindex fno-access-control
1422 Turn off all access checking. This switch is mainly useful for working
1423 around bugs in the access control code.
1427 Check that the pointer returned by @code{operator new} is non-null
1428 before attempting to modify the storage allocated. This check is
1429 normally unnecessary because the C++ standard specifies that
1430 @code{operator new} will only return @code{0} if it is declared
1431 @samp{throw()}, in which case the compiler will always check the
1432 return value even without this option. In all other cases, when
1433 @code{operator new} has a non-empty exception specification, memory
1434 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1435 @samp{new (nothrow)}.
1437 @item -fconserve-space
1438 @opindex fconserve-space
1439 Put uninitialized or runtime-initialized global variables into the
1440 common segment, as C does. This saves space in the executable at the
1441 cost of not diagnosing duplicate definitions. If you compile with this
1442 flag and your program mysteriously crashes after @code{main()} has
1443 completed, you may have an object that is being destroyed twice because
1444 two definitions were merged.
1446 This option is no longer useful on most targets, now that support has
1447 been added for putting variables into BSS without making them common.
1449 @item -ffriend-injection
1450 @opindex ffriend-injection
1451 Inject friend functions into the enclosing namespace, so that they are
1452 visible outside the scope of the class in which they are declared.
1453 Friend functions were documented to work this way in the old Annotated
1454 C++ Reference Manual, and versions of G++ before 4.1 always worked
1455 that way. However, in ISO C++ a friend function which is not declared
1456 in an enclosing scope can only be found using argument dependent
1457 lookup. This option causes friends to be injected as they were in
1460 This option is for compatibility, and may be removed in a future
1463 @item -fno-elide-constructors
1464 @opindex fno-elide-constructors
1465 The C++ standard allows an implementation to omit creating a temporary
1466 which is only used to initialize another object of the same type.
1467 Specifying this option disables that optimization, and forces G++ to
1468 call the copy constructor in all cases.
1470 @item -fno-enforce-eh-specs
1471 @opindex fno-enforce-eh-specs
1472 Don't generate code to check for violation of exception specifications
1473 at runtime. This option violates the C++ standard, but may be useful
1474 for reducing code size in production builds, much like defining
1475 @samp{NDEBUG}. This does not give user code permission to throw
1476 exceptions in violation of the exception specifications; the compiler
1477 will still optimize based on the specifications, so throwing an
1478 unexpected exception will result in undefined behavior.
1481 @itemx -fno-for-scope
1483 @opindex fno-for-scope
1484 If @option{-ffor-scope} is specified, the scope of variables declared in
1485 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1486 as specified by the C++ standard.
1487 If @option{-fno-for-scope} is specified, the scope of variables declared in
1488 a @i{for-init-statement} extends to the end of the enclosing scope,
1489 as was the case in old versions of G++, and other (traditional)
1490 implementations of C++.
1492 The default if neither flag is given to follow the standard,
1493 but to allow and give a warning for old-style code that would
1494 otherwise be invalid, or have different behavior.
1496 @item -fno-gnu-keywords
1497 @opindex fno-gnu-keywords
1498 Do not recognize @code{typeof} as a keyword, so that code can use this
1499 word as an identifier. You can use the keyword @code{__typeof__} instead.
1500 @option{-ansi} implies @option{-fno-gnu-keywords}.
1502 @item -fno-implicit-templates
1503 @opindex fno-implicit-templates
1504 Never emit code for non-inline templates which are instantiated
1505 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1506 @xref{Template Instantiation}, for more information.
1508 @item -fno-implicit-inline-templates
1509 @opindex fno-implicit-inline-templates
1510 Don't emit code for implicit instantiations of inline templates, either.
1511 The default is to handle inlines differently so that compiles with and
1512 without optimization will need the same set of explicit instantiations.
1514 @item -fno-implement-inlines
1515 @opindex fno-implement-inlines
1516 To save space, do not emit out-of-line copies of inline functions
1517 controlled by @samp{#pragma implementation}. This will cause linker
1518 errors if these functions are not inlined everywhere they are called.
1520 @item -fms-extensions
1521 @opindex fms-extensions
1522 Disable pedantic warnings about constructs used in MFC, such as implicit
1523 int and getting a pointer to member function via non-standard syntax.
1525 @item -fno-nonansi-builtins
1526 @opindex fno-nonansi-builtins
1527 Disable built-in declarations of functions that are not mandated by
1528 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1529 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1531 @item -fno-operator-names
1532 @opindex fno-operator-names
1533 Do not treat the operator name keywords @code{and}, @code{bitand},
1534 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1535 synonyms as keywords.
1537 @item -fno-optional-diags
1538 @opindex fno-optional-diags
1539 Disable diagnostics that the standard says a compiler does not need to
1540 issue. Currently, the only such diagnostic issued by G++ is the one for
1541 a name having multiple meanings within a class.
1544 @opindex fpermissive
1545 Downgrade some diagnostics about nonconformant code from errors to
1546 warnings. Thus, using @option{-fpermissive} will allow some
1547 nonconforming code to compile.
1551 Enable automatic template instantiation at link time. This option also
1552 implies @option{-fno-implicit-templates}. @xref{Template
1553 Instantiation}, for more information.
1557 Disable generation of information about every class with virtual
1558 functions for use by the C++ runtime type identification features
1559 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1560 of the language, you can save some space by using this flag. Note that
1561 exception handling uses the same information, but it will generate it as
1566 Emit statistics about front-end processing at the end of the compilation.
1567 This information is generally only useful to the G++ development team.
1569 @item -ftemplate-depth-@var{n}
1570 @opindex ftemplate-depth
1571 Set the maximum instantiation depth for template classes to @var{n}.
1572 A limit on the template instantiation depth is needed to detect
1573 endless recursions during template class instantiation. ANSI/ISO C++
1574 conforming programs must not rely on a maximum depth greater than 17.
1576 @item -fno-threadsafe-statics
1577 @opindex fno-threadsafe-statics
1578 Do not emit the extra code to use the routines specified in the C++
1579 ABI for thread-safe initialization of local statics. You can use this
1580 option to reduce code size slightly in code that doesn't need to be
1583 @item -fuse-cxa-atexit
1584 @opindex fuse-cxa-atexit
1585 Register destructors for objects with static storage duration with the
1586 @code{__cxa_atexit} function rather than the @code{atexit} function.
1587 This option is required for fully standards-compliant handling of static
1588 destructors, but will only work if your C library supports
1589 @code{__cxa_atexit}.
1591 @item -fno-use-cxa-get-exception-ptr
1592 @opindex fno-use-cxa-get-exception-ptr
1593 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1594 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1595 if the runtime routine is not available.
1597 @item -fvisibility-inlines-hidden
1598 @opindex fvisibility-inlines-hidden
1599 Causes all inlined methods to be marked with
1600 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1601 appear in the export table of a DSO and do not require a PLT indirection
1602 when used within the DSO@. Enabling this option can have a dramatic effect
1603 on load and link times of a DSO as it massively reduces the size of the
1604 dynamic export table when the library makes heavy use of templates. While
1605 it can cause bloating through duplication of code within each DSO where
1606 it is used, often the wastage is less than the considerable space occupied
1607 by a long symbol name in the export table which is typical when using
1608 templates and namespaces. For even more savings, combine with the
1609 @option{-fvisibility=hidden} switch.
1613 Do not use weak symbol support, even if it is provided by the linker.
1614 By default, G++ will use weak symbols if they are available. This
1615 option exists only for testing, and should not be used by end-users;
1616 it will result in inferior code and has no benefits. This option may
1617 be removed in a future release of G++.
1621 Do not search for header files in the standard directories specific to
1622 C++, but do still search the other standard directories. (This option
1623 is used when building the C++ library.)
1626 In addition, these optimization, warning, and code generation options
1627 have meanings only for C++ programs:
1630 @item -fno-default-inline
1631 @opindex fno-default-inline
1632 Do not assume @samp{inline} for functions defined inside a class scope.
1633 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1634 functions will have linkage like inline functions; they just won't be
1637 @item -Wabi @r{(C++ only)}
1639 Warn when G++ generates code that is probably not compatible with the
1640 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1641 all such cases, there are probably some cases that are not warned about,
1642 even though G++ is generating incompatible code. There may also be
1643 cases where warnings are emitted even though the code that is generated
1646 You should rewrite your code to avoid these warnings if you are
1647 concerned about the fact that code generated by G++ may not be binary
1648 compatible with code generated by other compilers.
1650 The known incompatibilities at this point include:
1655 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1656 pack data into the same byte as a base class. For example:
1659 struct A @{ virtual void f(); int f1 : 1; @};
1660 struct B : public A @{ int f2 : 1; @};
1664 In this case, G++ will place @code{B::f2} into the same byte
1665 as@code{A::f1}; other compilers will not. You can avoid this problem
1666 by explicitly padding @code{A} so that its size is a multiple of the
1667 byte size on your platform; that will cause G++ and other compilers to
1668 layout @code{B} identically.
1671 Incorrect handling of tail-padding for virtual bases. G++ does not use
1672 tail padding when laying out virtual bases. For example:
1675 struct A @{ virtual void f(); char c1; @};
1676 struct B @{ B(); char c2; @};
1677 struct C : public A, public virtual B @{@};
1681 In this case, G++ will not place @code{B} into the tail-padding for
1682 @code{A}; other compilers will. You can avoid this problem by
1683 explicitly padding @code{A} so that its size is a multiple of its
1684 alignment (ignoring virtual base classes); that will cause G++ and other
1685 compilers to layout @code{C} identically.
1688 Incorrect handling of bit-fields with declared widths greater than that
1689 of their underlying types, when the bit-fields appear in a union. For
1693 union U @{ int i : 4096; @};
1697 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1698 union too small by the number of bits in an @code{int}.
1701 Empty classes can be placed at incorrect offsets. For example:
1711 struct C : public B, public A @{@};
1715 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1716 it should be placed at offset zero. G++ mistakenly believes that the
1717 @code{A} data member of @code{B} is already at offset zero.
1720 Names of template functions whose types involve @code{typename} or
1721 template template parameters can be mangled incorrectly.
1724 template <typename Q>
1725 void f(typename Q::X) @{@}
1727 template <template <typename> class Q>
1728 void f(typename Q<int>::X) @{@}
1732 Instantiations of these templates may be mangled incorrectly.
1736 @item -Wctor-dtor-privacy @r{(C++ only)}
1737 @opindex Wctor-dtor-privacy
1738 Warn when a class seems unusable because all the constructors or
1739 destructors in that class are private, and it has neither friends nor
1740 public static member functions.
1742 @item -Wnon-virtual-dtor @r{(C++ only)}
1743 @opindex Wnon-virtual-dtor
1744 Warn when a class appears to be polymorphic, thereby requiring a virtual
1745 destructor, yet it declares a non-virtual one. This warning is also
1746 enabled if -Weffc++ is specified.
1748 @item -Wreorder @r{(C++ only)}
1750 @cindex reordering, warning
1751 @cindex warning for reordering of member initializers
1752 Warn when the order of member initializers given in the code does not
1753 match the order in which they must be executed. For instance:
1759 A(): j (0), i (1) @{ @}
1763 The compiler will rearrange the member initializers for @samp{i}
1764 and @samp{j} to match the declaration order of the members, emitting
1765 a warning to that effect. This warning is enabled by @option{-Wall}.
1768 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1771 @item -Weffc++ @r{(C++ only)}
1773 Warn about violations of the following style guidelines from Scott Meyers'
1774 @cite{Effective C++} book:
1778 Item 11: Define a copy constructor and an assignment operator for classes
1779 with dynamically allocated memory.
1782 Item 12: Prefer initialization to assignment in constructors.
1785 Item 14: Make destructors virtual in base classes.
1788 Item 15: Have @code{operator=} return a reference to @code{*this}.
1791 Item 23: Don't try to return a reference when you must return an object.
1795 Also warn about violations of the following style guidelines from
1796 Scott Meyers' @cite{More Effective C++} book:
1800 Item 6: Distinguish between prefix and postfix forms of increment and
1801 decrement operators.
1804 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1808 When selecting this option, be aware that the standard library
1809 headers do not obey all of these guidelines; use @samp{grep -v}
1810 to filter out those warnings.
1812 @item -Wno-deprecated @r{(C++ only)}
1813 @opindex Wno-deprecated
1814 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1816 @item -Wstrict-null-sentinel @r{(C++ only)}
1817 @opindex Wstrict-null-sentinel
1818 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1819 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1820 to @code{__null}. Although it is a null pointer constant not a null pointer,
1821 it is guaranteed to of the same size as a pointer. But this use is
1822 not portable across different compilers.
1824 @item -Wno-non-template-friend @r{(C++ only)}
1825 @opindex Wno-non-template-friend
1826 Disable warnings when non-templatized friend functions are declared
1827 within a template. Since the advent of explicit template specification
1828 support in G++, if the name of the friend is an unqualified-id (i.e.,
1829 @samp{friend foo(int)}), the C++ language specification demands that the
1830 friend declare or define an ordinary, nontemplate function. (Section
1831 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1832 could be interpreted as a particular specialization of a templatized
1833 function. Because this non-conforming behavior is no longer the default
1834 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1835 check existing code for potential trouble spots and is on by default.
1836 This new compiler behavior can be turned off with
1837 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1838 but disables the helpful warning.
1840 @item -Wold-style-cast @r{(C++ only)}
1841 @opindex Wold-style-cast
1842 Warn if an old-style (C-style) cast to a non-void type is used within
1843 a C++ program. The new-style casts (@samp{dynamic_cast},
1844 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1845 less vulnerable to unintended effects and much easier to search for.
1847 @item -Woverloaded-virtual @r{(C++ only)}
1848 @opindex Woverloaded-virtual
1849 @cindex overloaded virtual fn, warning
1850 @cindex warning for overloaded virtual fn
1851 Warn when a function declaration hides virtual functions from a
1852 base class. For example, in:
1859 struct B: public A @{
1864 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1872 will fail to compile.
1874 @item -Wno-pmf-conversions @r{(C++ only)}
1875 @opindex Wno-pmf-conversions
1876 Disable the diagnostic for converting a bound pointer to member function
1879 @item -Wsign-promo @r{(C++ only)}
1880 @opindex Wsign-promo
1881 Warn when overload resolution chooses a promotion from unsigned or
1882 enumerated type to a signed type, over a conversion to an unsigned type of
1883 the same size. Previous versions of G++ would try to preserve
1884 unsignedness, but the standard mandates the current behavior.
1889 A& operator = (int);
1899 In this example, G++ will synthesize a default @samp{A& operator =
1900 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1903 @node Objective-C and Objective-C++ Dialect Options
1904 @section Options Controlling Objective-C and Objective-C++ Dialects
1906 @cindex compiler options, Objective-C and Objective-C++
1907 @cindex Objective-C and Objective-C++ options, command line
1908 @cindex options, Objective-C and Objective-C++
1909 (NOTE: This manual does not describe the Objective-C and Objective-C++
1910 languages themselves. See @xref{Standards,,Language Standards
1911 Supported by GCC}, for references.)
1913 This section describes the command-line options that are only meaningful
1914 for Objective-C and Objective-C++ programs, but you can also use most of
1915 the language-independent GNU compiler options.
1916 For example, you might compile a file @code{some_class.m} like this:
1919 gcc -g -fgnu-runtime -O -c some_class.m
1923 In this example, @option{-fgnu-runtime} is an option meant only for
1924 Objective-C and Objective-C++ programs; you can use the other options with
1925 any language supported by GCC@.
1927 Note that since Objective-C is an extension of the C language, Objective-C
1928 compilations may also use options specific to the C front-end (e.g.,
1929 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1930 C++-specific options (e.g., @option{-Wabi}).
1932 Here is a list of options that are @emph{only} for compiling Objective-C
1933 and Objective-C++ programs:
1936 @item -fconstant-string-class=@var{class-name}
1937 @opindex fconstant-string-class
1938 Use @var{class-name} as the name of the class to instantiate for each
1939 literal string specified with the syntax @code{@@"@dots{}"}. The default
1940 class name is @code{NXConstantString} if the GNU runtime is being used, and
1941 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1942 @option{-fconstant-cfstrings} option, if also present, will override the
1943 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1944 to be laid out as constant CoreFoundation strings.
1947 @opindex fgnu-runtime
1948 Generate object code compatible with the standard GNU Objective-C
1949 runtime. This is the default for most types of systems.
1951 @item -fnext-runtime
1952 @opindex fnext-runtime
1953 Generate output compatible with the NeXT runtime. This is the default
1954 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1955 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1958 @item -fno-nil-receivers
1959 @opindex fno-nil-receivers
1960 Assume that all Objective-C message dispatches (e.g.,
1961 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1962 is not @code{nil}. This allows for more efficient entry points in the runtime
1963 to be used. Currently, this option is only available in conjunction with
1964 the NeXT runtime on Mac OS X 10.3 and later.
1966 @item -fobjc-call-cxx-cdtors
1967 @opindex fobjc-call-cxx-cdtors
1968 For each Objective-C class, check if any of its instance variables is a
1969 C++ object with a non-trivial default constructor. If so, synthesize a
1970 special @code{- (id) .cxx_construct} instance method that will run
1971 non-trivial default constructors on any such instance variables, in order,
1972 and then return @code{self}. Similarly, check if any instance variable
1973 is a C++ object with a non-trivial destructor, and if so, synthesize a
1974 special @code{- (void) .cxx_destruct} method that will run
1975 all such default destructors, in reverse order.
1977 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
1978 thusly generated will only operate on instance variables declared in the
1979 current Objective-C class, and not those inherited from superclasses. It
1980 is the responsibility of the Objective-C runtime to invoke all such methods
1981 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
1982 will be invoked by the runtime immediately after a new object
1983 instance is allocated; the @code{- (void) .cxx_destruct} methods will
1984 be invoked immediately before the runtime deallocates an object instance.
1986 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
1987 support for invoking the @code{- (id) .cxx_construct} and
1988 @code{- (void) .cxx_destruct} methods.
1990 @item -fobjc-direct-dispatch
1991 @opindex fobjc-direct-dispatch
1992 Allow fast jumps to the message dispatcher. On Darwin this is
1993 accomplished via the comm page.
1995 @item -fobjc-exceptions
1996 @opindex fobjc-exceptions
1997 Enable syntactic support for structured exception handling in Objective-C,
1998 similar to what is offered by C++ and Java. This option is
1999 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2008 @@catch (AnObjCClass *exc) @{
2015 @@catch (AnotherClass *exc) @{
2018 @@catch (id allOthers) @{
2028 The @code{@@throw} statement may appear anywhere in an Objective-C or
2029 Objective-C++ program; when used inside of a @code{@@catch} block, the
2030 @code{@@throw} may appear without an argument (as shown above), in which case
2031 the object caught by the @code{@@catch} will be rethrown.
2033 Note that only (pointers to) Objective-C objects may be thrown and
2034 caught using this scheme. When an object is thrown, it will be caught
2035 by the nearest @code{@@catch} clause capable of handling objects of that type,
2036 analogously to how @code{catch} blocks work in C++ and Java. A
2037 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2038 any and all Objective-C exceptions not caught by previous @code{@@catch}
2041 The @code{@@finally} clause, if present, will be executed upon exit from the
2042 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2043 regardless of whether any exceptions are thrown, caught or rethrown
2044 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2045 of the @code{finally} clause in Java.
2047 There are several caveats to using the new exception mechanism:
2051 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2052 idioms provided by the @code{NSException} class, the new
2053 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2054 systems, due to additional functionality needed in the (NeXT) Objective-C
2058 As mentioned above, the new exceptions do not support handling
2059 types other than Objective-C objects. Furthermore, when used from
2060 Objective-C++, the Objective-C exception model does not interoperate with C++
2061 exceptions at this time. This means you cannot @code{@@throw} an exception
2062 from Objective-C and @code{catch} it in C++, or vice versa
2063 (i.e., @code{throw @dots{} @@catch}).
2066 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2067 blocks for thread-safe execution:
2070 @@synchronized (ObjCClass *guard) @{
2075 Upon entering the @code{@@synchronized} block, a thread of execution shall
2076 first check whether a lock has been placed on the corresponding @code{guard}
2077 object by another thread. If it has, the current thread shall wait until
2078 the other thread relinquishes its lock. Once @code{guard} becomes available,
2079 the current thread will place its own lock on it, execute the code contained in
2080 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2081 making @code{guard} available to other threads).
2083 Unlike Java, Objective-C does not allow for entire methods to be marked
2084 @code{@@synchronized}. Note that throwing exceptions out of
2085 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2086 to be unlocked properly.
2090 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2092 @item -freplace-objc-classes
2093 @opindex freplace-objc-classes
2094 Emit a special marker instructing @command{ld(1)} not to statically link in
2095 the resulting object file, and allow @command{dyld(1)} to load it in at
2096 run time instead. This is used in conjunction with the Fix-and-Continue
2097 debugging mode, where the object file in question may be recompiled and
2098 dynamically reloaded in the course of program execution, without the need
2099 to restart the program itself. Currently, Fix-and-Continue functionality
2100 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2105 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2106 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2107 compile time) with static class references that get initialized at load time,
2108 which improves run-time performance. Specifying the @option{-fzero-link} flag
2109 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2110 to be retained. This is useful in Zero-Link debugging mode, since it allows
2111 for individual class implementations to be modified during program execution.
2115 Dump interface declarations for all classes seen in the source file to a
2116 file named @file{@var{sourcename}.decl}.
2118 @item -Wassign-intercept
2119 @opindex Wassign-intercept
2120 Warn whenever an Objective-C assignment is being intercepted by the
2124 @opindex Wno-protocol
2125 If a class is declared to implement a protocol, a warning is issued for
2126 every method in the protocol that is not implemented by the class. The
2127 default behavior is to issue a warning for every method not explicitly
2128 implemented in the class, even if a method implementation is inherited
2129 from the superclass. If you use the @option{-Wno-protocol} option, then
2130 methods inherited from the superclass are considered to be implemented,
2131 and no warning is issued for them.
2135 Warn if multiple methods of different types for the same selector are
2136 found during compilation. The check is performed on the list of methods
2137 in the final stage of compilation. Additionally, a check is performed
2138 for each selector appearing in a @code{@@selector(@dots{})}
2139 expression, and a corresponding method for that selector has been found
2140 during compilation. Because these checks scan the method table only at
2141 the end of compilation, these warnings are not produced if the final
2142 stage of compilation is not reached, for example because an error is
2143 found during compilation, or because the @option{-fsyntax-only} option is
2146 @item -Wstrict-selector-match
2147 @opindex Wstrict-selector-match
2148 Warn if multiple methods with differing argument and/or return types are
2149 found for a given selector when attempting to send a message using this
2150 selector to a receiver of type @code{id} or @code{Class}. When this flag
2151 is off (which is the default behavior), the compiler will omit such warnings
2152 if any differences found are confined to types which share the same size
2155 @item -Wundeclared-selector
2156 @opindex Wundeclared-selector
2157 Warn if a @code{@@selector(@dots{})} expression referring to an
2158 undeclared selector is found. A selector is considered undeclared if no
2159 method with that name has been declared before the
2160 @code{@@selector(@dots{})} expression, either explicitly in an
2161 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2162 an @code{@@implementation} section. This option always performs its
2163 checks as soon as a @code{@@selector(@dots{})} expression is found,
2164 while @option{-Wselector} only performs its checks in the final stage of
2165 compilation. This also enforces the coding style convention
2166 that methods and selectors must be declared before being used.
2168 @item -print-objc-runtime-info
2169 @opindex print-objc-runtime-info
2170 Generate C header describing the largest structure that is passed by
2175 @node Language Independent Options
2176 @section Options to Control Diagnostic Messages Formatting
2177 @cindex options to control diagnostics formatting
2178 @cindex diagnostic messages
2179 @cindex message formatting
2181 Traditionally, diagnostic messages have been formatted irrespective of
2182 the output device's aspect (e.g.@: its width, @dots{}). The options described
2183 below can be used to control the diagnostic messages formatting
2184 algorithm, e.g.@: how many characters per line, how often source location
2185 information should be reported. Right now, only the C++ front end can
2186 honor these options. However it is expected, in the near future, that
2187 the remaining front ends would be able to digest them correctly.
2190 @item -fmessage-length=@var{n}
2191 @opindex fmessage-length
2192 Try to format error messages so that they fit on lines of about @var{n}
2193 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2194 the front ends supported by GCC@. If @var{n} is zero, then no
2195 line-wrapping will be done; each error message will appear on a single
2198 @opindex fdiagnostics-show-location
2199 @item -fdiagnostics-show-location=once
2200 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2201 reporter to emit @emph{once} source location information; that is, in
2202 case the message is too long to fit on a single physical line and has to
2203 be wrapped, the source location won't be emitted (as prefix) again,
2204 over and over, in subsequent continuation lines. This is the default
2207 @item -fdiagnostics-show-location=every-line
2208 Only meaningful in line-wrapping mode. Instructs the diagnostic
2209 messages reporter to emit the same source location information (as
2210 prefix) for physical lines that result from the process of breaking
2211 a message which is too long to fit on a single line.
2213 @item -fdiagnostics-show-options
2214 @opindex fdiagnostics-show-options
2215 This option instructs the diagnostic machinery to add text to each
2216 diagnostic emitted, which indicates which command line option directly
2217 controls that diagnostic, when such an option is known to the
2218 diagnostic machinery.
2222 @node Warning Options
2223 @section Options to Request or Suppress Warnings
2224 @cindex options to control warnings
2225 @cindex warning messages
2226 @cindex messages, warning
2227 @cindex suppressing warnings
2229 Warnings are diagnostic messages that report constructions which
2230 are not inherently erroneous but which are risky or suggest there
2231 may have been an error.
2233 You can request many specific warnings with options beginning @samp{-W},
2234 for example @option{-Wimplicit} to request warnings on implicit
2235 declarations. Each of these specific warning options also has a
2236 negative form beginning @samp{-Wno-} to turn off warnings;
2237 for example, @option{-Wno-implicit}. This manual lists only one of the
2238 two forms, whichever is not the default.
2240 The following options control the amount and kinds of warnings produced
2241 by GCC; for further, language-specific options also refer to
2242 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2246 @cindex syntax checking
2248 @opindex fsyntax-only
2249 Check the code for syntax errors, but don't do anything beyond that.
2253 Issue all the warnings demanded by strict ISO C and ISO C++;
2254 reject all programs that use forbidden extensions, and some other
2255 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2256 version of the ISO C standard specified by any @option{-std} option used.
2258 Valid ISO C and ISO C++ programs should compile properly with or without
2259 this option (though a rare few will require @option{-ansi} or a
2260 @option{-std} option specifying the required version of ISO C)@. However,
2261 without this option, certain GNU extensions and traditional C and C++
2262 features are supported as well. With this option, they are rejected.
2264 @option{-pedantic} does not cause warning messages for use of the
2265 alternate keywords whose names begin and end with @samp{__}. Pedantic
2266 warnings are also disabled in the expression that follows
2267 @code{__extension__}. However, only system header files should use
2268 these escape routes; application programs should avoid them.
2269 @xref{Alternate Keywords}.
2271 Some users try to use @option{-pedantic} to check programs for strict ISO
2272 C conformance. They soon find that it does not do quite what they want:
2273 it finds some non-ISO practices, but not all---only those for which
2274 ISO C @emph{requires} a diagnostic, and some others for which
2275 diagnostics have been added.
2277 A feature to report any failure to conform to ISO C might be useful in
2278 some instances, but would require considerable additional work and would
2279 be quite different from @option{-pedantic}. We don't have plans to
2280 support such a feature in the near future.
2282 Where the standard specified with @option{-std} represents a GNU
2283 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2284 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2285 extended dialect is based. Warnings from @option{-pedantic} are given
2286 where they are required by the base standard. (It would not make sense
2287 for such warnings to be given only for features not in the specified GNU
2288 C dialect, since by definition the GNU dialects of C include all
2289 features the compiler supports with the given option, and there would be
2290 nothing to warn about.)
2292 @item -pedantic-errors
2293 @opindex pedantic-errors
2294 Like @option{-pedantic}, except that errors are produced rather than
2299 Inhibit all warning messages.
2303 Inhibit warning messages about the use of @samp{#import}.
2305 @item -Wchar-subscripts
2306 @opindex Wchar-subscripts
2307 Warn if an array subscript has type @code{char}. This is a common cause
2308 of error, as programmers often forget that this type is signed on some
2310 This warning is enabled by @option{-Wall}.
2314 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2315 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2316 This warning is enabled by @option{-Wall}.
2318 @item -Wfatal-errors
2319 @opindex Wfatal-errors
2320 This option causes the compiler to abort compilation on the first error
2321 occurred rather than trying to keep going and printing further error
2326 @opindex ffreestanding
2327 @opindex fno-builtin
2328 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2329 the arguments supplied have types appropriate to the format string
2330 specified, and that the conversions specified in the format string make
2331 sense. This includes standard functions, and others specified by format
2332 attributes (@pxref{Function Attributes}), in the @code{printf},
2333 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2334 not in the C standard) families (or other target-specific families).
2335 Which functions are checked without format attributes having been
2336 specified depends on the standard version selected, and such checks of
2337 functions without the attribute specified are disabled by
2338 @option{-ffreestanding} or @option{-fno-builtin}.
2340 The formats are checked against the format features supported by GNU
2341 libc version 2.2. These include all ISO C90 and C99 features, as well
2342 as features from the Single Unix Specification and some BSD and GNU
2343 extensions. Other library implementations may not support all these
2344 features; GCC does not support warning about features that go beyond a
2345 particular library's limitations. However, if @option{-pedantic} is used
2346 with @option{-Wformat}, warnings will be given about format features not
2347 in the selected standard version (but not for @code{strfmon} formats,
2348 since those are not in any version of the C standard). @xref{C Dialect
2349 Options,,Options Controlling C Dialect}.
2351 Since @option{-Wformat} also checks for null format arguments for
2352 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2354 @option{-Wformat} is included in @option{-Wall}. For more control over some
2355 aspects of format checking, the options @option{-Wformat-y2k},
2356 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2357 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2358 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2361 @opindex Wformat-y2k
2362 If @option{-Wformat} is specified, also warn about @code{strftime}
2363 formats which may yield only a two-digit year.
2365 @item -Wno-format-extra-args
2366 @opindex Wno-format-extra-args
2367 If @option{-Wformat} is specified, do not warn about excess arguments to a
2368 @code{printf} or @code{scanf} format function. The C standard specifies
2369 that such arguments are ignored.
2371 Where the unused arguments lie between used arguments that are
2372 specified with @samp{$} operand number specifications, normally
2373 warnings are still given, since the implementation could not know what
2374 type to pass to @code{va_arg} to skip the unused arguments. However,
2375 in the case of @code{scanf} formats, this option will suppress the
2376 warning if the unused arguments are all pointers, since the Single
2377 Unix Specification says that such unused arguments are allowed.
2379 @item -Wno-format-zero-length
2380 @opindex Wno-format-zero-length
2381 If @option{-Wformat} is specified, do not warn about zero-length formats.
2382 The C standard specifies that zero-length formats are allowed.
2384 @item -Wformat-nonliteral
2385 @opindex Wformat-nonliteral
2386 If @option{-Wformat} is specified, also warn if the format string is not a
2387 string literal and so cannot be checked, unless the format function
2388 takes its format arguments as a @code{va_list}.
2390 @item -Wformat-security
2391 @opindex Wformat-security
2392 If @option{-Wformat} is specified, also warn about uses of format
2393 functions that represent possible security problems. At present, this
2394 warns about calls to @code{printf} and @code{scanf} functions where the
2395 format string is not a string literal and there are no format arguments,
2396 as in @code{printf (foo);}. This may be a security hole if the format
2397 string came from untrusted input and contains @samp{%n}. (This is
2398 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2399 in future warnings may be added to @option{-Wformat-security} that are not
2400 included in @option{-Wformat-nonliteral}.)
2404 Enable @option{-Wformat} plus format checks not included in
2405 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2406 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2410 Warn about passing a null pointer for arguments marked as
2411 requiring a non-null value by the @code{nonnull} function attribute.
2413 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2414 can be disabled with the @option{-Wno-nonnull} option.
2416 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2418 Warn about uninitialized variables which are initialized with themselves.
2419 Note this option can only be used with the @option{-Wuninitialized} option,
2420 which in turn only works with @option{-O1} and above.
2422 For example, GCC will warn about @code{i} being uninitialized in the
2423 following snippet only when @option{-Winit-self} has been specified:
2434 @item -Wimplicit-int
2435 @opindex Wimplicit-int
2436 Warn when a declaration does not specify a type.
2437 This warning is enabled by @option{-Wall}.
2439 @item -Wimplicit-function-declaration
2440 @itemx -Werror-implicit-function-declaration
2441 @opindex Wimplicit-function-declaration
2442 @opindex Werror-implicit-function-declaration
2443 Give a warning (or error) whenever a function is used before being
2444 declared. The form @option{-Wno-error-implicit-function-declaration}
2446 This warning is enabled by @option{-Wall} (as a warning, not an error).
2450 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2451 This warning is enabled by @option{-Wall}.
2455 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2456 function with external linkage, returning int, taking either zero
2457 arguments, two, or three arguments of appropriate types.
2458 This warning is enabled by @option{-Wall}.
2460 @item -Wmissing-braces
2461 @opindex Wmissing-braces
2462 Warn if an aggregate or union initializer is not fully bracketed. In
2463 the following example, the initializer for @samp{a} is not fully
2464 bracketed, but that for @samp{b} is fully bracketed.
2467 int a[2][2] = @{ 0, 1, 2, 3 @};
2468 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2471 This warning is enabled by @option{-Wall}.
2473 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2474 @opindex Wmissing-include-dirs
2475 Warn if a user-supplied include directory does not exist.
2478 @opindex Wparentheses
2479 Warn if parentheses are omitted in certain contexts, such
2480 as when there is an assignment in a context where a truth value
2481 is expected, or when operators are nested whose precedence people
2482 often get confused about. Only the warning for an assignment used as
2483 a truth value is supported when compiling C++; the other warnings are
2484 only supported when compiling C@.
2486 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2487 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2488 interpretation from that of ordinary mathematical notation.
2490 Also warn about constructions where there may be confusion to which
2491 @code{if} statement an @code{else} branch belongs. Here is an example of
2506 In C, every @code{else} branch belongs to the innermost possible @code{if}
2507 statement, which in this example is @code{if (b)}. This is often not
2508 what the programmer expected, as illustrated in the above example by
2509 indentation the programmer chose. When there is the potential for this
2510 confusion, GCC will issue a warning when this flag is specified.
2511 To eliminate the warning, add explicit braces around the innermost
2512 @code{if} statement so there is no way the @code{else} could belong to
2513 the enclosing @code{if}. The resulting code would look like this:
2529 This warning is enabled by @option{-Wall}.
2531 @item -Wsequence-point
2532 @opindex Wsequence-point
2533 Warn about code that may have undefined semantics because of violations
2534 of sequence point rules in the C and C++ standards.
2536 The C and C++ standards defines the order in which expressions in a C/C++
2537 program are evaluated in terms of @dfn{sequence points}, which represent
2538 a partial ordering between the execution of parts of the program: those
2539 executed before the sequence point, and those executed after it. These
2540 occur after the evaluation of a full expression (one which is not part
2541 of a larger expression), after the evaluation of the first operand of a
2542 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2543 function is called (but after the evaluation of its arguments and the
2544 expression denoting the called function), and in certain other places.
2545 Other than as expressed by the sequence point rules, the order of
2546 evaluation of subexpressions of an expression is not specified. All
2547 these rules describe only a partial order rather than a total order,
2548 since, for example, if two functions are called within one expression
2549 with no sequence point between them, the order in which the functions
2550 are called is not specified. However, the standards committee have
2551 ruled that function calls do not overlap.
2553 It is not specified when between sequence points modifications to the
2554 values of objects take effect. Programs whose behavior depends on this
2555 have undefined behavior; the C and C++ standards specify that ``Between
2556 the previous and next sequence point an object shall have its stored
2557 value modified at most once by the evaluation of an expression.
2558 Furthermore, the prior value shall be read only to determine the value
2559 to be stored.''. If a program breaks these rules, the results on any
2560 particular implementation are entirely unpredictable.
2562 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2563 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2564 diagnosed by this option, and it may give an occasional false positive
2565 result, but in general it has been found fairly effective at detecting
2566 this sort of problem in programs.
2568 The standard is worded confusingly, therefore there is some debate
2569 over the precise meaning of the sequence point rules in subtle cases.
2570 Links to discussions of the problem, including proposed formal
2571 definitions, may be found on the GCC readings page, at
2572 @w{@uref{http://gcc.gnu.org/readings.html}}.
2574 This warning is enabled by @option{-Wall} for C and C++.
2577 @opindex Wreturn-type
2578 Warn whenever a function is defined with a return-type that defaults to
2579 @code{int}. Also warn about any @code{return} statement with no
2580 return-value in a function whose return-type is not @code{void}.
2582 For C, also warn if the return type of a function has a type qualifier
2583 such as @code{const}. Such a type qualifier has no effect, since the
2584 value returned by a function is not an lvalue. ISO C prohibits
2585 qualified @code{void} return types on function definitions, so such
2586 return types always receive a warning even without this option.
2588 For C++, a function without return type always produces a diagnostic
2589 message, even when @option{-Wno-return-type} is specified. The only
2590 exceptions are @samp{main} and functions defined in system headers.
2592 This warning is enabled by @option{-Wall}.
2596 Warn whenever a @code{switch} statement has an index of enumerated type
2597 and lacks a @code{case} for one or more of the named codes of that
2598 enumeration. (The presence of a @code{default} label prevents this
2599 warning.) @code{case} labels outside the enumeration range also
2600 provoke warnings when this option is used.
2601 This warning is enabled by @option{-Wall}.
2603 @item -Wswitch-default
2604 @opindex Wswitch-switch
2605 Warn whenever a @code{switch} statement does not have a @code{default}
2609 @opindex Wswitch-enum
2610 Warn whenever a @code{switch} statement has an index of enumerated type
2611 and lacks a @code{case} for one or more of the named codes of that
2612 enumeration. @code{case} labels outside the enumeration range also
2613 provoke warnings when this option is used.
2617 Warn if any trigraphs are encountered that might change the meaning of
2618 the program (trigraphs within comments are not warned about).
2619 This warning is enabled by @option{-Wall}.
2621 @item -Wunused-function
2622 @opindex Wunused-function
2623 Warn whenever a static function is declared but not defined or a
2624 non-inline static function is unused.
2625 This warning is enabled by @option{-Wall}.
2627 @item -Wunused-label
2628 @opindex Wunused-label
2629 Warn whenever a label is declared but not used.
2630 This warning is enabled by @option{-Wall}.
2632 To suppress this warning use the @samp{unused} attribute
2633 (@pxref{Variable Attributes}).
2635 @item -Wunused-parameter
2636 @opindex Wunused-parameter
2637 Warn whenever a function parameter is unused aside from its declaration.
2639 To suppress this warning use the @samp{unused} attribute
2640 (@pxref{Variable Attributes}).
2642 @item -Wunused-variable
2643 @opindex Wunused-variable
2644 Warn whenever a local variable or non-constant static variable is unused
2645 aside from its declaration
2646 This warning is enabled by @option{-Wall}.
2648 To suppress this warning use the @samp{unused} attribute
2649 (@pxref{Variable Attributes}).
2651 @item -Wunused-value
2652 @opindex Wunused-value
2653 Warn whenever a statement computes a result that is explicitly not used.
2654 This warning is enabled by @option{-Wall}.
2656 To suppress this warning cast the expression to @samp{void}.
2660 All the above @option{-Wunused} options combined.
2662 In order to get a warning about an unused function parameter, you must
2663 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2664 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2666 @item -Wuninitialized
2667 @opindex Wuninitialized
2668 Warn if an automatic variable is used without first being initialized or
2669 if a variable may be clobbered by a @code{setjmp} call.
2671 These warnings are possible only in optimizing compilation,
2672 because they require data flow information that is computed only
2673 when optimizing. If you don't specify @option{-O}, you simply won't
2676 If you want to warn about code which uses the uninitialized value of the
2677 variable in its own initializer, use the @option{-Winit-self} option.
2679 These warnings occur for individual uninitialized or clobbered
2680 elements of structure, union or array variables as well as for
2681 variables which are uninitialized or clobbered as a whole. They do
2682 not occur for variables or elements declared @code{volatile}. Because
2683 these warnings depend on optimization, the exact variables or elements
2684 for which there are warnings will depend on the precise optimization
2685 options and version of GCC used.
2687 Note that there may be no warning about a variable that is used only
2688 to compute a value that itself is never used, because such
2689 computations may be deleted by data flow analysis before the warnings
2692 These warnings are made optional because GCC is not smart
2693 enough to see all the reasons why the code might be correct
2694 despite appearing to have an error. Here is one example of how
2715 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2716 always initialized, but GCC doesn't know this. Here is
2717 another common case:
2722 if (change_y) save_y = y, y = new_y;
2724 if (change_y) y = save_y;
2729 This has no bug because @code{save_y} is used only if it is set.
2731 @cindex @code{longjmp} warnings
2732 This option also warns when a non-volatile automatic variable might be
2733 changed by a call to @code{longjmp}. These warnings as well are possible
2734 only in optimizing compilation.
2736 The compiler sees only the calls to @code{setjmp}. It cannot know
2737 where @code{longjmp} will be called; in fact, a signal handler could
2738 call it at any point in the code. As a result, you may get a warning
2739 even when there is in fact no problem because @code{longjmp} cannot
2740 in fact be called at the place which would cause a problem.
2742 Some spurious warnings can be avoided if you declare all the functions
2743 you use that never return as @code{noreturn}. @xref{Function
2746 This warning is enabled by @option{-Wall}.
2748 @item -Wunknown-pragmas
2749 @opindex Wunknown-pragmas
2750 @cindex warning for unknown pragmas
2751 @cindex unknown pragmas, warning
2752 @cindex pragmas, warning of unknown
2753 Warn when a #pragma directive is encountered which is not understood by
2754 GCC@. If this command line option is used, warnings will even be issued
2755 for unknown pragmas in system header files. This is not the case if
2756 the warnings were only enabled by the @option{-Wall} command line option.
2759 @opindex Wno-pragmas
2761 Do not warn about misuses of pragmas, such as incorrect parameters,
2762 invalid syntax, or conflicts between pragmas. See also
2763 @samp{-Wunknown-pragmas}.
2765 @item -Wstrict-aliasing
2766 @opindex Wstrict-aliasing
2767 This option is only active when @option{-fstrict-aliasing} is active.
2768 It warns about code which might break the strict aliasing rules that the
2769 compiler is using for optimization. The warning does not catch all
2770 cases, but does attempt to catch the more common pitfalls. It is
2771 included in @option{-Wall}.
2773 @item -Wstrict-aliasing=2
2774 @opindex Wstrict-aliasing=2
2775 This option is only active when @option{-fstrict-aliasing} is active.
2776 It warns about code which might break the strict aliasing rules that the
2777 compiler is using for optimization. This warning catches more cases than
2778 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2779 cases that are safe.
2783 All of the above @samp{-W} options combined. This enables all the
2784 warnings about constructions that some users consider questionable, and
2785 that are easy to avoid (or modify to prevent the warning), even in
2786 conjunction with macros. This also enables some language-specific
2787 warnings described in @ref{C++ Dialect Options} and
2788 @ref{Objective-C and Objective-C++ Dialect Options}.
2791 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2792 Some of them warn about constructions that users generally do not
2793 consider questionable, but which occasionally you might wish to check
2794 for; others warn about constructions that are necessary or hard to avoid
2795 in some cases, and there is no simple way to modify the code to suppress
2802 (This option used to be called @option{-W}. The older name is still
2803 supported, but the newer name is more descriptive.) Print extra warning
2804 messages for these events:
2808 A function can return either with or without a value. (Falling
2809 off the end of the function body is considered returning without
2810 a value.) For example, this function would evoke such a
2824 An expression-statement or the left-hand side of a comma expression
2825 contains no side effects.
2826 To suppress the warning, cast the unused expression to void.
2827 For example, an expression such as @samp{x[i,j]} will cause a warning,
2828 but @samp{x[(void)i,j]} will not.
2831 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2834 Storage-class specifiers like @code{static} are not the first things in
2835 a declaration. According to the C Standard, this usage is obsolescent.
2838 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2842 A comparison between signed and unsigned values could produce an
2843 incorrect result when the signed value is converted to unsigned.
2844 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2847 An aggregate has an initializer which does not initialize all members.
2848 This warning can be independently controlled by
2849 @option{-Wmissing-field-initializers}.
2852 A function parameter is declared without a type specifier in K&R-style
2860 An empty body occurs in an @samp{if} or @samp{else} statement.
2863 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2864 @samp{>}, or @samp{>=}.
2867 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2870 Any of several floating-point events that often indicate errors, such as
2871 overflow, underflow, loss of precision, etc.
2873 @item @r{(C++ only)}
2874 An enumerator and a non-enumerator both appear in a conditional expression.
2876 @item @r{(C++ only)}
2877 A non-static reference or non-static @samp{const} member appears in a
2878 class without constructors.
2880 @item @r{(C++ only)}
2881 Ambiguous virtual bases.
2883 @item @r{(C++ only)}
2884 Subscripting an array which has been declared @samp{register}.
2886 @item @r{(C++ only)}
2887 Taking the address of a variable which has been declared @samp{register}.
2889 @item @r{(C++ only)}
2890 A base class is not initialized in a derived class' copy constructor.
2893 @item -Wno-div-by-zero
2894 @opindex Wno-div-by-zero
2895 @opindex Wdiv-by-zero
2896 Do not warn about compile-time integer division by zero. Floating point
2897 division by zero is not warned about, as it can be a legitimate way of
2898 obtaining infinities and NaNs.
2900 @item -Wsystem-headers
2901 @opindex Wsystem-headers
2902 @cindex warnings from system headers
2903 @cindex system headers, warnings from
2904 Print warning messages for constructs found in system header files.
2905 Warnings from system headers are normally suppressed, on the assumption
2906 that they usually do not indicate real problems and would only make the
2907 compiler output harder to read. Using this command line option tells
2908 GCC to emit warnings from system headers as if they occurred in user
2909 code. However, note that using @option{-Wall} in conjunction with this
2910 option will @emph{not} warn about unknown pragmas in system
2911 headers---for that, @option{-Wunknown-pragmas} must also be used.
2914 @opindex Wfloat-equal
2915 Warn if floating point values are used in equality comparisons.
2917 The idea behind this is that sometimes it is convenient (for the
2918 programmer) to consider floating-point values as approximations to
2919 infinitely precise real numbers. If you are doing this, then you need
2920 to compute (by analyzing the code, or in some other way) the maximum or
2921 likely maximum error that the computation introduces, and allow for it
2922 when performing comparisons (and when producing output, but that's a
2923 different problem). In particular, instead of testing for equality, you
2924 would check to see whether the two values have ranges that overlap; and
2925 this is done with the relational operators, so equality comparisons are
2928 @item -Wtraditional @r{(C only)}
2929 @opindex Wtraditional
2930 Warn about certain constructs that behave differently in traditional and
2931 ISO C@. Also warn about ISO C constructs that have no traditional C
2932 equivalent, and/or problematic constructs which should be avoided.
2936 Macro parameters that appear within string literals in the macro body.
2937 In traditional C macro replacement takes place within string literals,
2938 but does not in ISO C@.
2941 In traditional C, some preprocessor directives did not exist.
2942 Traditional preprocessors would only consider a line to be a directive
2943 if the @samp{#} appeared in column 1 on the line. Therefore
2944 @option{-Wtraditional} warns about directives that traditional C
2945 understands but would ignore because the @samp{#} does not appear as the
2946 first character on the line. It also suggests you hide directives like
2947 @samp{#pragma} not understood by traditional C by indenting them. Some
2948 traditional implementations would not recognize @samp{#elif}, so it
2949 suggests avoiding it altogether.
2952 A function-like macro that appears without arguments.
2955 The unary plus operator.
2958 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2959 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2960 constants.) Note, these suffixes appear in macros defined in the system
2961 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2962 Use of these macros in user code might normally lead to spurious
2963 warnings, however GCC's integrated preprocessor has enough context to
2964 avoid warning in these cases.
2967 A function declared external in one block and then used after the end of
2971 A @code{switch} statement has an operand of type @code{long}.
2974 A non-@code{static} function declaration follows a @code{static} one.
2975 This construct is not accepted by some traditional C compilers.
2978 The ISO type of an integer constant has a different width or
2979 signedness from its traditional type. This warning is only issued if
2980 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2981 typically represent bit patterns, are not warned about.
2984 Usage of ISO string concatenation is detected.
2987 Initialization of automatic aggregates.
2990 Identifier conflicts with labels. Traditional C lacks a separate
2991 namespace for labels.
2994 Initialization of unions. If the initializer is zero, the warning is
2995 omitted. This is done under the assumption that the zero initializer in
2996 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2997 initializer warnings and relies on default initialization to zero in the
3001 Conversions by prototypes between fixed/floating point values and vice
3002 versa. The absence of these prototypes when compiling with traditional
3003 C would cause serious problems. This is a subset of the possible
3004 conversion warnings, for the full set use @option{-Wconversion}.
3007 Use of ISO C style function definitions. This warning intentionally is
3008 @emph{not} issued for prototype declarations or variadic functions
3009 because these ISO C features will appear in your code when using
3010 libiberty's traditional C compatibility macros, @code{PARAMS} and
3011 @code{VPARAMS}. This warning is also bypassed for nested functions
3012 because that feature is already a GCC extension and thus not relevant to
3013 traditional C compatibility.
3016 @item -Wdeclaration-after-statement @r{(C only)}
3017 @opindex Wdeclaration-after-statement
3018 Warn when a declaration is found after a statement in a block. This
3019 construct, known from C++, was introduced with ISO C99 and is by default
3020 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3021 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3025 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3027 @item -Wno-endif-labels
3028 @opindex Wno-endif-labels
3029 @opindex Wendif-labels
3030 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3034 Warn whenever a local variable shadows another local variable, parameter or
3035 global variable or whenever a built-in function is shadowed.
3037 @item -Wlarger-than-@var{len}
3038 @opindex Wlarger-than
3039 Warn whenever an object of larger than @var{len} bytes is defined.
3041 @item -Wunsafe-loop-optimizations
3042 @opindex Wunsafe-loop-optimizations
3043 Warn if the loop cannot be optimized because the compiler could not
3044 assume anything on the bounds of the loop indices. With
3045 @option{-funsafe-loop-optimizations} warn if the compiler made
3048 @item -Wpointer-arith
3049 @opindex Wpointer-arith
3050 Warn about anything that depends on the ``size of'' a function type or
3051 of @code{void}. GNU C assigns these types a size of 1, for
3052 convenience in calculations with @code{void *} pointers and pointers
3055 @item -Wbad-function-cast @r{(C only)}
3056 @opindex Wbad-function-cast
3057 Warn whenever a function call is cast to a non-matching type.
3058 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3061 Warn about ISO C constructs that are outside of the common subset of
3062 ISO C and ISO C++, e.g.@: request for implicit conversion from
3063 @code{void *} to a pointer to non-@code{void} type.
3067 Warn whenever a pointer is cast so as to remove a type qualifier from
3068 the target type. For example, warn if a @code{const char *} is cast
3069 to an ordinary @code{char *}.
3072 @opindex Wcast-align
3073 Warn whenever a pointer is cast such that the required alignment of the
3074 target is increased. For example, warn if a @code{char *} is cast to
3075 an @code{int *} on machines where integers can only be accessed at
3076 two- or four-byte boundaries.
3078 @item -Wwrite-strings
3079 @opindex Wwrite-strings
3080 When compiling C, give string constants the type @code{const
3081 char[@var{length}]} so that
3082 copying the address of one into a non-@code{const} @code{char *}
3083 pointer will get a warning; when compiling C++, warn about the
3084 deprecated conversion from string literals to @code{char *}. This
3085 warning, by default, is enabled for C++ programs.
3086 These warnings will help you find at
3087 compile time code that can try to write into a string constant, but
3088 only if you have been very careful about using @code{const} in
3089 declarations and prototypes. Otherwise, it will just be a nuisance;
3090 this is why we did not make @option{-Wall} request these warnings.
3093 @opindex Wconversion
3094 Warn if a prototype causes a type conversion that is different from what
3095 would happen to the same argument in the absence of a prototype. This
3096 includes conversions of fixed point to floating and vice versa, and
3097 conversions changing the width or signedness of a fixed point argument
3098 except when the same as the default promotion.
3100 Also, warn if a negative integer constant expression is implicitly
3101 converted to an unsigned type. For example, warn about the assignment
3102 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3103 casts like @code{(unsigned) -1}.
3105 @item -Wsign-compare
3106 @opindex Wsign-compare
3107 @cindex warning for comparison of signed and unsigned values
3108 @cindex comparison of signed and unsigned values, warning
3109 @cindex signed and unsigned values, comparison warning
3110 Warn when a comparison between signed and unsigned values could produce
3111 an incorrect result when the signed value is converted to unsigned.
3112 This warning is also enabled by @option{-Wextra}; to get the other warnings
3113 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3115 @item -Waggregate-return
3116 @opindex Waggregate-return
3117 Warn if any functions that return structures or unions are defined or
3118 called. (In languages where you can return an array, this also elicits
3122 @opindex Walways-true
3123 Warn about comparisons which are always true such as testing if
3124 unsigned values are greater than or equal to zero. This warning is
3125 enabled by @option{-Wall}.
3127 @item -Wno-attributes
3128 @opindex Wno-attributes
3129 @opindex Wattributes
3130 Do not warn if an unexpected @code{__attribute__} is used, such as
3131 unrecognized attributes, function attributes applied to variables,
3132 etc. This will not stop errors for incorrect use of supported
3135 @item -Wstrict-prototypes @r{(C only)}
3136 @opindex Wstrict-prototypes
3137 Warn if a function is declared or defined without specifying the
3138 argument types. (An old-style function definition is permitted without
3139 a warning if preceded by a declaration which specifies the argument
3142 @item -Wold-style-definition @r{(C only)}
3143 @opindex Wold-style-definition
3144 Warn if an old-style function definition is used. A warning is given
3145 even if there is a previous prototype.
3147 @item -Wmissing-prototypes @r{(C only)}
3148 @opindex Wmissing-prototypes
3149 Warn if a global function is defined without a previous prototype
3150 declaration. This warning is issued even if the definition itself
3151 provides a prototype. The aim is to detect global functions that fail
3152 to be declared in header files.
3154 @item -Wmissing-declarations @r{(C only)}
3155 @opindex Wmissing-declarations
3156 Warn if a global function is defined without a previous declaration.
3157 Do so even if the definition itself provides a prototype.
3158 Use this option to detect global functions that are not declared in
3161 @item -Wmissing-field-initializers
3162 @opindex Wmissing-field-initializers
3165 Warn if a structure's initializer has some fields missing. For
3166 example, the following code would cause such a warning, because
3167 @code{x.h} is implicitly zero:
3170 struct s @{ int f, g, h; @};
3171 struct s x = @{ 3, 4 @};
3174 This option does not warn about designated initializers, so the following
3175 modification would not trigger a warning:
3178 struct s @{ int f, g, h; @};
3179 struct s x = @{ .f = 3, .g = 4 @};
3182 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3183 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3185 @item -Wmissing-noreturn
3186 @opindex Wmissing-noreturn
3187 Warn about functions which might be candidates for attribute @code{noreturn}.
3188 Note these are only possible candidates, not absolute ones. Care should
3189 be taken to manually verify functions actually do not ever return before
3190 adding the @code{noreturn} attribute, otherwise subtle code generation
3191 bugs could be introduced. You will not get a warning for @code{main} in
3192 hosted C environments.
3194 @item -Wmissing-format-attribute
3195 @opindex Wmissing-format-attribute
3197 Warn about function pointers which might be candidates for @code{format}
3198 attributes. Note these are only possible candidates, not absolute ones.
3199 GCC will guess that function pointers with @code{format} attributes that
3200 are used in assignment, initialization, parameter passing or return
3201 statements should have a corresponding @code{format} attribute in the
3202 resulting type. I.e.@: the left-hand side of the assignment or
3203 initialization, the type of the parameter variable, or the return type
3204 of the containing function respectively should also have a @code{format}
3205 attribute to avoid the warning.
3207 GCC will also warn about function definitions which might be
3208 candidates for @code{format} attributes. Again, these are only
3209 possible candidates. GCC will guess that @code{format} attributes
3210 might be appropriate for any function that calls a function like
3211 @code{vprintf} or @code{vscanf}, but this might not always be the
3212 case, and some functions for which @code{format} attributes are
3213 appropriate may not be detected.
3215 @item -Wno-multichar
3216 @opindex Wno-multichar
3218 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3219 Usually they indicate a typo in the user's code, as they have
3220 implementation-defined values, and should not be used in portable code.
3222 @item -Wnormalized=<none|id|nfc|nfkc>
3223 @opindex Wnormalized
3226 @cindex character set, input normalization
3227 In ISO C and ISO C++, two identifiers are different if they are
3228 different sequences of characters. However, sometimes when characters
3229 outside the basic ASCII character set are used, you can have two
3230 different character sequences that look the same. To avoid confusion,
3231 the ISO 10646 standard sets out some @dfn{normalization rules} which
3232 when applied ensure that two sequences that look the same are turned into
3233 the same sequence. GCC can warn you if you are using identifiers which
3234 have not been normalized; this option controls that warning.
3236 There are four levels of warning that GCC supports. The default is
3237 @option{-Wnormalized=nfc}, which warns about any identifier which is
3238 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3239 recommended form for most uses.
3241 Unfortunately, there are some characters which ISO C and ISO C++ allow
3242 in identifiers that when turned into NFC aren't allowable as
3243 identifiers. That is, there's no way to use these symbols in portable
3244 ISO C or C++ and have all your identifiers in NFC.
3245 @option{-Wnormalized=id} suppresses the warning for these characters.
3246 It is hoped that future versions of the standards involved will correct
3247 this, which is why this option is not the default.
3249 You can switch the warning off for all characters by writing
3250 @option{-Wnormalized=none}. You would only want to do this if you
3251 were using some other normalization scheme (like ``D''), because
3252 otherwise you can easily create bugs that are literally impossible to see.
3254 Some characters in ISO 10646 have distinct meanings but look identical
3255 in some fonts or display methodologies, especially once formatting has
3256 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3257 LETTER N'', will display just like a regular @code{n} which has been
3258 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3259 normalisation scheme to convert all these into a standard form as
3260 well, and GCC will warn if your code is not in NFKC if you use
3261 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3262 about every identifier that contains the letter O because it might be
3263 confused with the digit 0, and so is not the default, but may be
3264 useful as a local coding convention if the programming environment is
3265 unable to be fixed to display these characters distinctly.
3267 @item -Wno-deprecated-declarations
3268 @opindex Wno-deprecated-declarations
3269 Do not warn about uses of functions, variables, and types marked as
3270 deprecated by using the @code{deprecated} attribute.
3271 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3272 @pxref{Type Attributes}.)
3276 Warn if a structure is given the packed attribute, but the packed
3277 attribute has no effect on the layout or size of the structure.
3278 Such structures may be mis-aligned for little benefit. For
3279 instance, in this code, the variable @code{f.x} in @code{struct bar}
3280 will be misaligned even though @code{struct bar} does not itself
3281 have the packed attribute:
3288 @} __attribute__((packed));
3298 Warn if padding is included in a structure, either to align an element
3299 of the structure or to align the whole structure. Sometimes when this
3300 happens it is possible to rearrange the fields of the structure to
3301 reduce the padding and so make the structure smaller.
3303 @item -Wredundant-decls
3304 @opindex Wredundant-decls
3305 Warn if anything is declared more than once in the same scope, even in
3306 cases where multiple declaration is valid and changes nothing.
3308 @item -Wnested-externs @r{(C only)}
3309 @opindex Wnested-externs
3310 Warn if an @code{extern} declaration is encountered within a function.
3312 @item -Wunreachable-code
3313 @opindex Wunreachable-code
3314 Warn if the compiler detects that code will never be executed.
3316 This option is intended to warn when the compiler detects that at
3317 least a whole line of source code will never be executed, because
3318 some condition is never satisfied or because it is after a
3319 procedure that never returns.
3321 It is possible for this option to produce a warning even though there
3322 are circumstances under which part of the affected line can be executed,
3323 so care should be taken when removing apparently-unreachable code.
3325 For instance, when a function is inlined, a warning may mean that the
3326 line is unreachable in only one inlined copy of the function.
3328 This option is not made part of @option{-Wall} because in a debugging
3329 version of a program there is often substantial code which checks
3330 correct functioning of the program and is, hopefully, unreachable
3331 because the program does work. Another common use of unreachable
3332 code is to provide behavior which is selectable at compile-time.
3336 Warn if a function can not be inlined and it was declared as inline.
3337 Even with this option, the compiler will not warn about failures to
3338 inline functions declared in system headers.
3340 The compiler uses a variety of heuristics to determine whether or not
3341 to inline a function. For example, the compiler takes into account
3342 the size of the function being inlined and the amount of inlining
3343 that has already been done in the current function. Therefore,
3344 seemingly insignificant changes in the source program can cause the
3345 warnings produced by @option{-Winline} to appear or disappear.
3347 @item -Wno-invalid-offsetof @r{(C++ only)}
3348 @opindex Wno-invalid-offsetof
3349 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3350 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3351 to a non-POD type is undefined. In existing C++ implementations,
3352 however, @samp{offsetof} typically gives meaningful results even when
3353 applied to certain kinds of non-POD types. (Such as a simple
3354 @samp{struct} that fails to be a POD type only by virtue of having a
3355 constructor.) This flag is for users who are aware that they are
3356 writing nonportable code and who have deliberately chosen to ignore the
3359 The restrictions on @samp{offsetof} may be relaxed in a future version
3360 of the C++ standard.
3362 @item -Wno-int-to-pointer-cast @r{(C only)}
3363 @opindex Wno-int-to-pointer-cast
3364 Suppress warnings from casts to pointer type of an integer of a
3367 @item -Wno-pointer-to-int-cast @r{(C only)}
3368 @opindex Wno-pointer-to-int-cast
3369 Suppress warnings from casts from a pointer to an integer type of a
3373 @opindex Winvalid-pch
3374 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3375 the search path but can't be used.
3379 @opindex Wno-long-long
3380 Warn if @samp{long long} type is used. This is default. To inhibit
3381 the warning messages, use @option{-Wno-long-long}. Flags
3382 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3383 only when @option{-pedantic} flag is used.
3385 @item -Wvariadic-macros
3386 @opindex Wvariadic-macros
3387 @opindex Wno-variadic-macros
3388 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3389 alternate syntax when in pedantic ISO C99 mode. This is default.
3390 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3392 @item -Wvolatile-register-var
3393 @opindex Wvolatile-register-var
3394 @opindex Wno-volatile-register-var
3395 Warn if a register variable is declared volatile. The volatile
3396 modifier does not inhibit all optimizations that may eliminate reads
3397 and/or writes to register variables.
3399 @item -Wdisabled-optimization
3400 @opindex Wdisabled-optimization
3401 Warn if a requested optimization pass is disabled. This warning does
3402 not generally indicate that there is anything wrong with your code; it
3403 merely indicates that GCC's optimizers were unable to handle the code
3404 effectively. Often, the problem is that your code is too big or too
3405 complex; GCC will refuse to optimize programs when the optimization
3406 itself is likely to take inordinate amounts of time.
3408 @item -Wpointer-sign
3409 @opindex Wpointer-sign
3410 @opindex Wno-pointer-sign
3411 Warn for pointer argument passing or assignment with different signedness.
3412 This option is only supported for C and Objective-C@. It is implied by
3413 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3414 @option{-Wno-pointer-sign}.
3418 Make all warnings into errors.
3422 Make the specified warning into an errors. The specifier for a
3423 warning is appended, for example @option{-Werror=switch} turns the
3424 warnings controlled by @option{-Wswitch} into errors. This switch
3425 takes a negative form, to be used to negate @option{-Werror} for
3426 specific warnings, for example @option{-Wno-error=switch} makes
3427 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3428 is in effect. You can use the @option{-fdiagnostics-show-option}
3429 option to have each controllable warning amended with the option which
3430 controls it, to determine what to use with this option.
3432 Note that specifying @option{-Werror=}@var{foo} automatically implies
3433 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3436 @item -Wstack-protector
3437 @opindex Wstack-protector
3438 This option is only active when @option{-fstack-protector} is active. It
3439 warns about functions that will not be protected against stack smashing.
3441 @item -Wstring-literal-comparison
3442 @opindex Wstring-literal-comparison
3443 Warn about suspicious comparisons to string literal constants. In C,
3444 direct comparisons against the memory address of a string literal, such
3445 as @code{if (x == "abc")}, typically indicate a programmer error, and
3446 even when intentional, result in unspecified behavior and are not portable.
3447 Usually these warnings alert that the programmer intended to use
3448 @code{strcmp}. This warning is enabled by @option{-Wall}.
3450 @item -Woverlength-strings
3451 @opindex Woverlength-strings
3452 Warn about string constants which are longer than the ``minimum
3453 maximum'' length specified in the C standard. Modern compilers
3454 generally allow string constants which are much longer than the
3455 standard's minimum limit, but very portable programs should avoid
3456 using longer strings.
3458 The limit applies @emph{after} string constant concatenation, and does
3459 not count the trailing NUL@. In C89, the limit was 509 characters; in
3460 C99, it was raised to 4095. C++98 does not specify a normative
3461 minimum maximum, so we do not diagnose overlength strings in C++@.
3463 This option is implied by @option{-pedantic}, and can be disabled with
3464 @option{-Wno-overlength-strings}.
3467 @node Debugging Options
3468 @section Options for Debugging Your Program or GCC
3469 @cindex options, debugging
3470 @cindex debugging information options
3472 GCC has various special options that are used for debugging
3473 either your program or GCC:
3478 Produce debugging information in the operating system's native format
3479 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3482 On most systems that use stabs format, @option{-g} enables use of extra
3483 debugging information that only GDB can use; this extra information
3484 makes debugging work better in GDB but will probably make other debuggers
3486 refuse to read the program. If you want to control for certain whether
3487 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3488 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3490 GCC allows you to use @option{-g} with
3491 @option{-O}. The shortcuts taken by optimized code may occasionally
3492 produce surprising results: some variables you declared may not exist
3493 at all; flow of control may briefly move where you did not expect it;
3494 some statements may not be executed because they compute constant
3495 results or their values were already at hand; some statements may
3496 execute in different places because they were moved out of loops.
3498 Nevertheless it proves possible to debug optimized output. This makes
3499 it reasonable to use the optimizer for programs that might have bugs.
3501 The following options are useful when GCC is generated with the
3502 capability for more than one debugging format.
3506 Produce debugging information for use by GDB@. This means to use the
3507 most expressive format available (DWARF 2, stabs, or the native format
3508 if neither of those are supported), including GDB extensions if at all
3513 Produce debugging information in stabs format (if that is supported),
3514 without GDB extensions. This is the format used by DBX on most BSD
3515 systems. On MIPS, Alpha and System V Release 4 systems this option
3516 produces stabs debugging output which is not understood by DBX or SDB@.
3517 On System V Release 4 systems this option requires the GNU assembler.
3519 @item -feliminate-unused-debug-symbols
3520 @opindex feliminate-unused-debug-symbols
3521 Produce debugging information in stabs format (if that is supported),
3522 for only symbols that are actually used.
3526 Produce debugging information in stabs format (if that is supported),
3527 using GNU extensions understood only by the GNU debugger (GDB)@. The
3528 use of these extensions is likely to make other debuggers crash or
3529 refuse to read the program.
3533 Produce debugging information in COFF format (if that is supported).
3534 This is the format used by SDB on most System V systems prior to
3539 Produce debugging information in XCOFF format (if that is supported).
3540 This is the format used by the DBX debugger on IBM RS/6000 systems.
3544 Produce debugging information in XCOFF format (if that is supported),
3545 using GNU extensions understood only by the GNU debugger (GDB)@. The
3546 use of these extensions is likely to make other debuggers crash or
3547 refuse to read the program, and may cause assemblers other than the GNU
3548 assembler (GAS) to fail with an error.
3552 Produce debugging information in DWARF version 2 format (if that is
3553 supported). This is the format used by DBX on IRIX 6. With this
3554 option, GCC uses features of DWARF version 3 when they are useful;
3555 version 3 is upward compatible with version 2, but may still cause
3556 problems for older debuggers.
3560 Produce debugging information in VMS debug format (if that is
3561 supported). This is the format used by DEBUG on VMS systems.
3564 @itemx -ggdb@var{level}
3565 @itemx -gstabs@var{level}
3566 @itemx -gcoff@var{level}
3567 @itemx -gxcoff@var{level}
3568 @itemx -gvms@var{level}
3569 Request debugging information and also use @var{level} to specify how
3570 much information. The default level is 2.
3572 Level 1 produces minimal information, enough for making backtraces in
3573 parts of the program that you don't plan to debug. This includes
3574 descriptions of functions and external variables, but no information
3575 about local variables and no line numbers.
3577 Level 3 includes extra information, such as all the macro definitions
3578 present in the program. Some debuggers support macro expansion when
3579 you use @option{-g3}.
3581 @option{-gdwarf-2} does not accept a concatenated debug level, because
3582 GCC used to support an option @option{-gdwarf} that meant to generate
3583 debug information in version 1 of the DWARF format (which is very
3584 different from version 2), and it would have been too confusing. That
3585 debug format is long obsolete, but the option cannot be changed now.
3586 Instead use an additional @option{-g@var{level}} option to change the
3587 debug level for DWARF2.
3589 @item -feliminate-dwarf2-dups
3590 @opindex feliminate-dwarf2-dups
3591 Compress DWARF2 debugging information by eliminating duplicated
3592 information about each symbol. This option only makes sense when
3593 generating DWARF2 debugging information with @option{-gdwarf-2}.
3595 @cindex @command{prof}
3598 Generate extra code to write profile information suitable for the
3599 analysis program @command{prof}. You must use this option when compiling
3600 the source files you want data about, and you must also use it when
3603 @cindex @command{gprof}
3606 Generate extra code to write profile information suitable for the
3607 analysis program @command{gprof}. You must use this option when compiling
3608 the source files you want data about, and you must also use it when
3613 Makes the compiler print out each function name as it is compiled, and
3614 print some statistics about each pass when it finishes.
3617 @opindex ftime-report
3618 Makes the compiler print some statistics about the time consumed by each
3619 pass when it finishes.
3622 @opindex fmem-report
3623 Makes the compiler print some statistics about permanent memory
3624 allocation when it finishes.
3626 @item -fprofile-arcs
3627 @opindex fprofile-arcs
3628 Add code so that program flow @dfn{arcs} are instrumented. During
3629 execution the program records how many times each branch and call is
3630 executed and how many times it is taken or returns. When the compiled
3631 program exits it saves this data to a file called
3632 @file{@var{auxname}.gcda} for each source file. The data may be used for
3633 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3634 test coverage analysis (@option{-ftest-coverage}). Each object file's
3635 @var{auxname} is generated from the name of the output file, if
3636 explicitly specified and it is not the final executable, otherwise it is
3637 the basename of the source file. In both cases any suffix is removed
3638 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3639 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3640 @xref{Cross-profiling}.
3642 @cindex @command{gcov}
3646 This option is used to compile and link code instrumented for coverage
3647 analysis. The option is a synonym for @option{-fprofile-arcs}
3648 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3649 linking). See the documentation for those options for more details.
3654 Compile the source files with @option{-fprofile-arcs} plus optimization
3655 and code generation options. For test coverage analysis, use the
3656 additional @option{-ftest-coverage} option. You do not need to profile
3657 every source file in a program.
3660 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3661 (the latter implies the former).
3664 Run the program on a representative workload to generate the arc profile
3665 information. This may be repeated any number of times. You can run
3666 concurrent instances of your program, and provided that the file system
3667 supports locking, the data files will be correctly updated. Also
3668 @code{fork} calls are detected and correctly handled (double counting
3672 For profile-directed optimizations, compile the source files again with
3673 the same optimization and code generation options plus
3674 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3675 Control Optimization}).
3678 For test coverage analysis, use @command{gcov} to produce human readable
3679 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3680 @command{gcov} documentation for further information.
3684 With @option{-fprofile-arcs}, for each function of your program GCC
3685 creates a program flow graph, then finds a spanning tree for the graph.
3686 Only arcs that are not on the spanning tree have to be instrumented: the
3687 compiler adds code to count the number of times that these arcs are
3688 executed. When an arc is the only exit or only entrance to a block, the
3689 instrumentation code can be added to the block; otherwise, a new basic
3690 block must be created to hold the instrumentation code.
3693 @item -ftest-coverage
3694 @opindex ftest-coverage
3695 Produce a notes file that the @command{gcov} code-coverage utility
3696 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3697 show program coverage. Each source file's note file is called
3698 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3699 above for a description of @var{auxname} and instructions on how to
3700 generate test coverage data. Coverage data will match the source files
3701 more closely, if you do not optimize.
3703 @item -d@var{letters}
3704 @item -fdump-rtl-@var{pass}
3706 Says to make debugging dumps during compilation at times specified by
3707 @var{letters}. This is used for debugging the RTL-based passes of the
3708 compiler. The file names for most of the dumps are made by appending a
3709 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3710 from the name of the output file, if explicitly specified and it is not
3711 an executable, otherwise it is the basename of the source file.
3713 Most debug dumps can be enabled either passing a letter to the @option{-d}
3714 option, or with a long @option{-fdump-rtl} switch; here are the possible
3715 letters for use in @var{letters} and @var{pass}, and their meanings:
3720 Annotate the assembler output with miscellaneous debugging information.
3723 @itemx -fdump-rtl-bbro
3725 @opindex fdump-rtl-bbro
3726 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3729 @itemx -fdump-rtl-combine
3731 @opindex fdump-rtl-combine
3732 Dump after instruction combination, to the file @file{@var{file}.129r.combine}.
3735 @itemx -fdump-rtl-ce1
3736 @itemx -fdump-rtl-ce2
3738 @opindex fdump-rtl-ce1
3739 @opindex fdump-rtl-ce2
3740 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3741 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3742 and @option{-fdump-rtl-ce2} enable dumping after the second if
3743 conversion, to the file @file{@var{file}.130r.ce2}.
3746 @itemx -fdump-rtl-btl
3747 @itemx -fdump-rtl-dbr
3749 @opindex fdump-rtl-btl
3750 @opindex fdump-rtl-dbr
3751 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3752 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3753 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3754 scheduling, to @file{@var{file}.36.dbr}.
3758 Dump all macro definitions, at the end of preprocessing, in addition to
3762 @itemx -fdump-rtl-ce3
3764 @opindex fdump-rtl-ce3
3765 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3768 @itemx -fdump-rtl-cfg
3769 @itemx -fdump-rtl-life
3771 @opindex fdump-rtl-cfg
3772 @opindex fdump-rtl-life
3773 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3774 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3775 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3776 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3779 @itemx -fdump-rtl-greg
3781 @opindex fdump-rtl-greg
3782 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3785 @itemx -fdump-rtl-gcse
3786 @itemx -fdump-rtl-bypass
3788 @opindex fdump-rtl-gcse
3789 @opindex fdump-rtl-bypass
3790 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3791 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3792 enable dumping after jump bypassing and control flow optimizations, to
3793 @file{@var{file}.115r.bypass}.
3796 @itemx -fdump-rtl-eh
3798 @opindex fdump-rtl-eh
3799 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3802 @itemx -fdump-rtl-sibling
3804 @opindex fdump-rtl-sibling
3805 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3808 @itemx -fdump-rtl-jump
3810 @opindex fdump-rtl-jump
3811 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3814 @itemx -fdump-rtl-stack
3816 @opindex fdump-rtl-stack
3817 Dump after conversion from registers to stack, to @file{@var{file}.152r.stack}.
3820 @itemx -fdump-rtl-lreg
3822 @opindex fdump-rtl-lreg
3823 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3826 @itemx -fdump-rtl-loop2
3828 @opindex fdump-rtl-loop2
3829 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3830 loop optimization pass, to @file{@var{file}.119r.loop2},
3831 @file{@var{file}.120r.loop2_init},
3832 @file{@var{file}.121r.loop2_invariant}, and
3833 @file{@var{file}.125r.loop2_done}.
3836 @itemx -fdump-rtl-sms
3838 @opindex fdump-rtl-sms
3839 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3842 @itemx -fdump-rtl-mach
3844 @opindex fdump-rtl-mach
3845 Dump after performing the machine dependent reorganization pass, to
3846 @file{@var{file}.155r.mach}.
3849 @itemx -fdump-rtl-rnreg
3851 @opindex fdump-rtl-rnreg
3852 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
3855 @itemx -fdump-rtl-regmove
3857 @opindex fdump-rtl-regmove
3858 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
3861 @itemx -fdump-rtl-postreload
3863 @opindex fdump-rtl-postreload
3864 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3867 @itemx -fdump-rtl-expand
3869 @opindex fdump-rtl-expand
3870 Dump after RTL generation, to @file{@var{file}.104r.expand}.
3873 @itemx -fdump-rtl-sched2
3875 @opindex fdump-rtl-sched2
3876 Dump after the second scheduling pass, to @file{@var{file}.150r.sched2}.
3879 @itemx -fdump-rtl-cse
3881 @opindex fdump-rtl-cse
3882 Dump after CSE (including the jump optimization that sometimes follows
3883 CSE), to @file{@var{file}.113r.cse}.
3886 @itemx -fdump-rtl-sched
3888 @opindex fdump-rtl-sched
3889 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3892 @itemx -fdump-rtl-cse2
3894 @opindex fdump-rtl-cse2
3895 Dump after the second CSE pass (including the jump optimization that
3896 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
3899 @itemx -fdump-rtl-tracer
3901 @opindex fdump-rtl-tracer
3902 Dump after running tracer, to @file{@var{file}.118r.tracer}.
3905 @itemx -fdump-rtl-vpt
3906 @itemx -fdump-rtl-vartrack
3908 @opindex fdump-rtl-vpt
3909 @opindex fdump-rtl-vartrack
3910 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3911 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3912 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3913 to @file{@var{file}.154r.vartrack}.
3916 @itemx -fdump-rtl-flow2
3918 @opindex fdump-rtl-flow2
3919 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
3922 @itemx -fdump-rtl-peephole2
3924 @opindex fdump-rtl-peephole2
3925 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
3928 @itemx -fdump-rtl-web
3930 @opindex fdump-rtl-web
3931 Dump after live range splitting, to @file{@var{file}.126r.web}.
3934 @itemx -fdump-rtl-all
3936 @opindex fdump-rtl-all
3937 Produce all the dumps listed above.
3941 Produce a core dump whenever an error occurs.
3945 Print statistics on memory usage, at the end of the run, to
3950 Annotate the assembler output with a comment indicating which
3951 pattern and alternative was used. The length of each instruction is
3956 Dump the RTL in the assembler output as a comment before each instruction.
3957 Also turns on @option{-dp} annotation.
3961 For each of the other indicated dump files (either with @option{-d} or
3962 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3963 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3967 Just generate RTL for a function instead of compiling it. Usually used
3968 with @samp{r} (@option{-fdump-rtl-expand}).
3972 Dump debugging information during parsing, to standard error.
3975 @item -fdump-unnumbered
3976 @opindex fdump-unnumbered
3977 When doing debugging dumps (see @option{-d} option above), suppress instruction
3978 numbers and line number note output. This makes it more feasible to
3979 use diff on debugging dumps for compiler invocations with different
3980 options, in particular with and without @option{-g}.
3982 @item -fdump-translation-unit @r{(C++ only)}
3983 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
3984 @opindex fdump-translation-unit
3985 Dump a representation of the tree structure for the entire translation
3986 unit to a file. The file name is made by appending @file{.tu} to the
3987 source file name. If the @samp{-@var{options}} form is used, @var{options}
3988 controls the details of the dump as described for the
3989 @option{-fdump-tree} options.
3991 @item -fdump-class-hierarchy @r{(C++ only)}
3992 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3993 @opindex fdump-class-hierarchy
3994 Dump a representation of each class's hierarchy and virtual function
3995 table layout to a file. The file name is made by appending @file{.class}
3996 to the source file name. If the @samp{-@var{options}} form is used,
3997 @var{options} controls the details of the dump as described for the
3998 @option{-fdump-tree} options.
4000 @item -fdump-ipa-@var{switch}
4002 Control the dumping at various stages of inter-procedural analysis
4003 language tree to a file. The file name is generated by appending a switch
4004 specific suffix to the source file name. The following dumps are possible:
4008 Enables all inter-procedural analysis dumps; currently the only produced
4009 dump is the @samp{cgraph} dump.
4012 Dumps information about call-graph optimization, unused function removal,
4013 and inlining decisions.
4016 @item -fdump-tree-@var{switch}
4017 @itemx -fdump-tree-@var{switch}-@var{options}
4019 Control the dumping at various stages of processing the intermediate
4020 language tree to a file. The file name is generated by appending a switch
4021 specific suffix to the source file name. If the @samp{-@var{options}}
4022 form is used, @var{options} is a list of @samp{-} separated options that
4023 control the details of the dump. Not all options are applicable to all
4024 dumps, those which are not meaningful will be ignored. The following
4025 options are available
4029 Print the address of each node. Usually this is not meaningful as it
4030 changes according to the environment and source file. Its primary use
4031 is for tying up a dump file with a debug environment.
4033 Inhibit dumping of members of a scope or body of a function merely
4034 because that scope has been reached. Only dump such items when they
4035 are directly reachable by some other path. When dumping pretty-printed
4036 trees, this option inhibits dumping the bodies of control structures.
4038 Print a raw representation of the tree. By default, trees are
4039 pretty-printed into a C-like representation.
4041 Enable more detailed dumps (not honored by every dump option).
4043 Enable dumping various statistics about the pass (not honored by every dump
4046 Enable showing basic block boundaries (disabled in raw dumps).
4048 Enable showing virtual operands for every statement.
4050 Enable showing line numbers for statements.
4052 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4054 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4057 The following tree dumps are possible:
4061 Dump before any tree based optimization, to @file{@var{file}.original}.
4064 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4067 Dump after function inlining, to @file{@var{file}.inlined}.
4070 @opindex fdump-tree-gimple
4071 Dump each function before and after the gimplification pass to a file. The
4072 file name is made by appending @file{.gimple} to the source file name.
4075 @opindex fdump-tree-cfg
4076 Dump the control flow graph of each function to a file. The file name is
4077 made by appending @file{.cfg} to the source file name.
4080 @opindex fdump-tree-vcg
4081 Dump the control flow graph of each function to a file in VCG format. The
4082 file name is made by appending @file{.vcg} to the source file name. Note
4083 that if the file contains more than one function, the generated file cannot
4084 be used directly by VCG@. You will need to cut and paste each function's
4085 graph into its own separate file first.
4088 @opindex fdump-tree-ch
4089 Dump each function after copying loop headers. The file name is made by
4090 appending @file{.ch} to the source file name.
4093 @opindex fdump-tree-ssa
4094 Dump SSA related information to a file. The file name is made by appending
4095 @file{.ssa} to the source file name.
4098 @opindex fdump-tree-salias
4099 Dump structure aliasing variable information to a file. This file name
4100 is made by appending @file{.salias} to the source file name.
4103 @opindex fdump-tree-alias
4104 Dump aliasing information for each function. The file name is made by
4105 appending @file{.alias} to the source file name.
4108 @opindex fdump-tree-ccp
4109 Dump each function after CCP@. The file name is made by appending
4110 @file{.ccp} to the source file name.
4113 @opindex fdump-tree-storeccp
4114 Dump each function after STORE-CCP. The file name is made by appending
4115 @file{.storeccp} to the source file name.
4118 @opindex fdump-tree-pre
4119 Dump trees after partial redundancy elimination. The file name is made
4120 by appending @file{.pre} to the source file name.
4123 @opindex fdump-tree-fre
4124 Dump trees after full redundancy elimination. The file name is made
4125 by appending @file{.fre} to the source file name.
4128 @opindex fdump-tree-copyprop
4129 Dump trees after copy propagation. The file name is made
4130 by appending @file{.copyprop} to the source file name.
4132 @item store_copyprop
4133 @opindex fdump-tree-store_copyprop
4134 Dump trees after store copy-propagation. The file name is made
4135 by appending @file{.store_copyprop} to the source file name.
4138 @opindex fdump-tree-dce
4139 Dump each function after dead code elimination. The file name is made by
4140 appending @file{.dce} to the source file name.
4143 @opindex fdump-tree-mudflap
4144 Dump each function after adding mudflap instrumentation. The file name is
4145 made by appending @file{.mudflap} to the source file name.
4148 @opindex fdump-tree-sra
4149 Dump each function after performing scalar replacement of aggregates. The
4150 file name is made by appending @file{.sra} to the source file name.
4153 @opindex fdump-tree-sink
4154 Dump each function after performing code sinking. The file name is made
4155 by appending @file{.sink} to the source file name.
4158 @opindex fdump-tree-dom
4159 Dump each function after applying dominator tree optimizations. The file
4160 name is made by appending @file{.dom} to the source file name.
4163 @opindex fdump-tree-dse
4164 Dump each function after applying dead store elimination. The file
4165 name is made by appending @file{.dse} to the source file name.
4168 @opindex fdump-tree-phiopt
4169 Dump each function after optimizing PHI nodes into straightline code. The file
4170 name is made by appending @file{.phiopt} to the source file name.
4173 @opindex fdump-tree-forwprop
4174 Dump each function after forward propagating single use variables. The file
4175 name is made by appending @file{.forwprop} to the source file name.
4178 @opindex fdump-tree-copyrename
4179 Dump each function after applying the copy rename optimization. The file
4180 name is made by appending @file{.copyrename} to the source file name.
4183 @opindex fdump-tree-nrv
4184 Dump each function after applying the named return value optimization on
4185 generic trees. The file name is made by appending @file{.nrv} to the source
4189 @opindex fdump-tree-vect
4190 Dump each function after applying vectorization of loops. The file name is
4191 made by appending @file{.vect} to the source file name.
4194 @opindex fdump-tree-vrp
4195 Dump each function after Value Range Propagation (VRP). The file name
4196 is made by appending @file{.vrp} to the source file name.
4199 @opindex fdump-tree-all
4200 Enable all the available tree dumps with the flags provided in this option.
4203 @item -ftree-vectorizer-verbose=@var{n}
4204 @opindex ftree-vectorizer-verbose
4205 This option controls the amount of debugging output the vectorizer prints.
4206 This information is written to standard error, unless
4207 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4208 in which case it is output to the usual dump listing file, @file{.vect}.
4209 For @var{n}=0 no diagnostic information is reported.
4210 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4211 and the total number of loops that got vectorized.
4212 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4213 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4214 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4215 level that @option{-fdump-tree-vect-stats} uses.
4216 Higher verbosity levels mean either more information dumped for each
4217 reported loop, or same amount of information reported for more loops:
4218 If @var{n}=3, alignment related information is added to the reports.
4219 If @var{n}=4, data-references related information (e.g. memory dependences,
4220 memory access-patterns) is added to the reports.
4221 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4222 that did not pass the first analysis phase (i.e. may not be countable, or
4223 may have complicated control-flow).
4224 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4225 For @var{n}=7, all the information the vectorizer generates during its
4226 analysis and transformation is reported. This is the same verbosity level
4227 that @option{-fdump-tree-vect-details} uses.
4229 @item -frandom-seed=@var{string}
4230 @opindex frandom-string
4231 This option provides a seed that GCC uses when it would otherwise use
4232 random numbers. It is used to generate certain symbol names
4233 that have to be different in every compiled file. It is also used to
4234 place unique stamps in coverage data files and the object files that
4235 produce them. You can use the @option{-frandom-seed} option to produce
4236 reproducibly identical object files.
4238 The @var{string} should be different for every file you compile.
4240 @item -fsched-verbose=@var{n}
4241 @opindex fsched-verbose
4242 On targets that use instruction scheduling, this option controls the
4243 amount of debugging output the scheduler prints. This information is
4244 written to standard error, unless @option{-dS} or @option{-dR} is
4245 specified, in which case it is output to the usual dump
4246 listing file, @file{.sched} or @file{.sched2} respectively. However
4247 for @var{n} greater than nine, the output is always printed to standard
4250 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4251 same information as @option{-dRS}. For @var{n} greater than one, it
4252 also output basic block probabilities, detailed ready list information
4253 and unit/insn info. For @var{n} greater than two, it includes RTL
4254 at abort point, control-flow and regions info. And for @var{n} over
4255 four, @option{-fsched-verbose} also includes dependence info.
4259 Store the usual ``temporary'' intermediate files permanently; place them
4260 in the current directory and name them based on the source file. Thus,
4261 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4262 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4263 preprocessed @file{foo.i} output file even though the compiler now
4264 normally uses an integrated preprocessor.
4266 When used in combination with the @option{-x} command line option,
4267 @option{-save-temps} is sensible enough to avoid over writing an
4268 input source file with the same extension as an intermediate file.
4269 The corresponding intermediate file may be obtained by renaming the
4270 source file before using @option{-save-temps}.
4274 Report the CPU time taken by each subprocess in the compilation
4275 sequence. For C source files, this is the compiler proper and assembler
4276 (plus the linker if linking is done). The output looks like this:
4283 The first number on each line is the ``user time'', that is time spent
4284 executing the program itself. The second number is ``system time'',
4285 time spent executing operating system routines on behalf of the program.
4286 Both numbers are in seconds.
4288 @item -fvar-tracking
4289 @opindex fvar-tracking
4290 Run variable tracking pass. It computes where variables are stored at each
4291 position in code. Better debugging information is then generated
4292 (if the debugging information format supports this information).
4294 It is enabled by default when compiling with optimization (@option{-Os},
4295 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4296 the debug info format supports it.
4298 @item -print-file-name=@var{library}
4299 @opindex print-file-name
4300 Print the full absolute name of the library file @var{library} that
4301 would be used when linking---and don't do anything else. With this
4302 option, GCC does not compile or link anything; it just prints the
4305 @item -print-multi-directory
4306 @opindex print-multi-directory
4307 Print the directory name corresponding to the multilib selected by any
4308 other switches present in the command line. This directory is supposed
4309 to exist in @env{GCC_EXEC_PREFIX}.
4311 @item -print-multi-lib
4312 @opindex print-multi-lib
4313 Print the mapping from multilib directory names to compiler switches
4314 that enable them. The directory name is separated from the switches by
4315 @samp{;}, and each switch starts with an @samp{@@} instead of the
4316 @samp{-}, without spaces between multiple switches. This is supposed to
4317 ease shell-processing.
4319 @item -print-prog-name=@var{program}
4320 @opindex print-prog-name
4321 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4323 @item -print-libgcc-file-name
4324 @opindex print-libgcc-file-name
4325 Same as @option{-print-file-name=libgcc.a}.
4327 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4328 but you do want to link with @file{libgcc.a}. You can do
4331 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4334 @item -print-search-dirs
4335 @opindex print-search-dirs
4336 Print the name of the configured installation directory and a list of
4337 program and library directories @command{gcc} will search---and don't do anything else.
4339 This is useful when @command{gcc} prints the error message
4340 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4341 To resolve this you either need to put @file{cpp0} and the other compiler
4342 components where @command{gcc} expects to find them, or you can set the environment
4343 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4344 Don't forget the trailing @samp{/}.
4345 @xref{Environment Variables}.
4348 @opindex dumpmachine
4349 Print the compiler's target machine (for example,
4350 @samp{i686-pc-linux-gnu})---and don't do anything else.
4353 @opindex dumpversion
4354 Print the compiler version (for example, @samp{3.0})---and don't do
4359 Print the compiler's built-in specs---and don't do anything else. (This
4360 is used when GCC itself is being built.) @xref{Spec Files}.
4362 @item -feliminate-unused-debug-types
4363 @opindex feliminate-unused-debug-types
4364 Normally, when producing DWARF2 output, GCC will emit debugging
4365 information for all types declared in a compilation
4366 unit, regardless of whether or not they are actually used
4367 in that compilation unit. Sometimes this is useful, such as
4368 if, in the debugger, you want to cast a value to a type that is
4369 not actually used in your program (but is declared). More often,
4370 however, this results in a significant amount of wasted space.
4371 With this option, GCC will avoid producing debug symbol output
4372 for types that are nowhere used in the source file being compiled.
4375 @node Optimize Options
4376 @section Options That Control Optimization
4377 @cindex optimize options
4378 @cindex options, optimization
4380 These options control various sorts of optimizations.
4382 Without any optimization option, the compiler's goal is to reduce the
4383 cost of compilation and to make debugging produce the expected
4384 results. Statements are independent: if you stop the program with a
4385 breakpoint between statements, you can then assign a new value to any
4386 variable or change the program counter to any other statement in the
4387 function and get exactly the results you would expect from the source
4390 Turning on optimization flags makes the compiler attempt to improve
4391 the performance and/or code size at the expense of compilation time
4392 and possibly the ability to debug the program.
4394 The compiler performs optimization based on the knowledge it has of
4395 the program. Optimization levels @option{-O2} and above, in
4396 particular, enable @emph{unit-at-a-time} mode, which allows the
4397 compiler to consider information gained from later functions in
4398 the file when compiling a function. Compiling multiple files at
4399 once to a single output file in @emph{unit-at-a-time} mode allows
4400 the compiler to use information gained from all of the files when
4401 compiling each of them.
4403 Not all optimizations are controlled directly by a flag. Only
4404 optimizations that have a flag are listed.
4411 Optimize. Optimizing compilation takes somewhat more time, and a lot
4412 more memory for a large function.
4414 With @option{-O}, the compiler tries to reduce code size and execution
4415 time, without performing any optimizations that take a great deal of
4418 @option{-O} turns on the following optimization flags:
4419 @gccoptlist{-fdefer-pop @gol
4420 -fdelayed-branch @gol
4421 -fguess-branch-probability @gol
4422 -fcprop-registers @gol
4423 -fif-conversion @gol
4424 -fif-conversion2 @gol
4427 -ftree-dominator-opts @gol
4432 -ftree-copyrename @gol
4437 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4438 where doing so does not interfere with debugging.
4442 Optimize even more. GCC performs nearly all supported optimizations
4443 that do not involve a space-speed tradeoff. The compiler does not
4444 perform loop unrolling or function inlining when you specify @option{-O2}.
4445 As compared to @option{-O}, this option increases both compilation time
4446 and the performance of the generated code.
4448 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4449 also turns on the following optimization flags:
4450 @gccoptlist{-fthread-jumps @gol
4452 -foptimize-sibling-calls @gol
4453 -fcse-follow-jumps -fcse-skip-blocks @gol
4454 -fgcse -fgcse-lm @gol
4455 -fexpensive-optimizations @gol
4456 -frerun-cse-after-loop @gol
4459 -fschedule-insns -fschedule-insns2 @gol
4460 -fsched-interblock -fsched-spec @gol
4462 -fstrict-aliasing @gol
4463 -fdelete-null-pointer-checks @gol
4464 -freorder-blocks -freorder-functions @gol
4465 -funit-at-a-time @gol
4466 -falign-functions -falign-jumps @gol
4467 -falign-loops -falign-labels @gol
4471 Please note the warning under @option{-fgcse} about
4472 invoking @option{-O2} on programs that use computed gotos.
4476 Optimize yet more. @option{-O3} turns on all optimizations specified by
4477 @option{-O2} and also turns on the @option{-finline-functions},
4478 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4482 Do not optimize. This is the default.
4486 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4487 do not typically increase code size. It also performs further
4488 optimizations designed to reduce code size.
4490 @option{-Os} disables the following optimization flags:
4491 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4492 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4493 -fprefetch-loop-arrays -ftree-vect-loop-version}
4495 If you use multiple @option{-O} options, with or without level numbers,
4496 the last such option is the one that is effective.
4499 Options of the form @option{-f@var{flag}} specify machine-independent
4500 flags. Most flags have both positive and negative forms; the negative
4501 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4502 below, only one of the forms is listed---the one you typically will
4503 use. You can figure out the other form by either removing @samp{no-}
4506 The following options control specific optimizations. They are either
4507 activated by @option{-O} options or are related to ones that are. You
4508 can use the following flags in the rare cases when ``fine-tuning'' of
4509 optimizations to be performed is desired.
4512 @item -fno-default-inline
4513 @opindex fno-default-inline
4514 Do not make member functions inline by default merely because they are
4515 defined inside the class scope (C++ only). Otherwise, when you specify
4516 @w{@option{-O}}, member functions defined inside class scope are compiled
4517 inline by default; i.e., you don't need to add @samp{inline} in front of
4518 the member function name.
4520 @item -fno-defer-pop
4521 @opindex fno-defer-pop
4522 Always pop the arguments to each function call as soon as that function
4523 returns. For machines which must pop arguments after a function call,
4524 the compiler normally lets arguments accumulate on the stack for several
4525 function calls and pops them all at once.
4527 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4531 Force memory operands to be copied into registers before doing
4532 arithmetic on them. This produces better code by making all memory
4533 references potential common subexpressions. When they are not common
4534 subexpressions, instruction combination should eliminate the separate
4535 register-load. This option is now a nop and will be removed in 4.2.
4538 @opindex fforce-addr
4539 Force memory address constants to be copied into registers before
4540 doing arithmetic on them.
4542 @item -fomit-frame-pointer
4543 @opindex fomit-frame-pointer
4544 Don't keep the frame pointer in a register for functions that
4545 don't need one. This avoids the instructions to save, set up and
4546 restore frame pointers; it also makes an extra register available
4547 in many functions. @strong{It also makes debugging impossible on
4550 On some machines, such as the VAX, this flag has no effect, because
4551 the standard calling sequence automatically handles the frame pointer
4552 and nothing is saved by pretending it doesn't exist. The
4553 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4554 whether a target machine supports this flag. @xref{Registers,,Register
4555 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4557 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4559 @item -foptimize-sibling-calls
4560 @opindex foptimize-sibling-calls
4561 Optimize sibling and tail recursive calls.
4563 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4567 Don't pay attention to the @code{inline} keyword. Normally this option
4568 is used to keep the compiler from expanding any functions inline.
4569 Note that if you are not optimizing, no functions can be expanded inline.
4571 @item -finline-functions
4572 @opindex finline-functions
4573 Integrate all simple functions into their callers. The compiler
4574 heuristically decides which functions are simple enough to be worth
4575 integrating in this way.
4577 If all calls to a given function are integrated, and the function is
4578 declared @code{static}, then the function is normally not output as
4579 assembler code in its own right.
4581 Enabled at level @option{-O3}.
4583 @item -finline-functions-called-once
4584 @opindex finline-functions-called-once
4585 Consider all @code{static} functions called once for inlining into their
4586 caller even if they are not marked @code{inline}. If a call to a given
4587 function is integrated, then the function is not output as assembler code
4590 Enabled if @option{-funit-at-a-time} is enabled.
4592 @item -fearly-inlining
4593 @opindex fearly-inlining
4594 Inline functions marked by @code{always_inline} and functions whose body seems
4595 smaller than the function call overhead early before doing
4596 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4597 makes profiling significantly cheaper and usually inlining faster on programs
4598 having large chains of nested wrapper functions.
4602 @item -finline-limit=@var{n}
4603 @opindex finline-limit
4604 By default, GCC limits the size of functions that can be inlined. This flag
4605 allows the control of this limit for functions that are explicitly marked as
4606 inline (i.e., marked with the inline keyword or defined within the class
4607 definition in c++). @var{n} is the size of functions that can be inlined in
4608 number of pseudo instructions (not counting parameter handling). The default
4609 value of @var{n} is 600.
4610 Increasing this value can result in more inlined code at
4611 the cost of compilation time and memory consumption. Decreasing usually makes
4612 the compilation faster and less code will be inlined (which presumably
4613 means slower programs). This option is particularly useful for programs that
4614 use inlining heavily such as those based on recursive templates with C++.
4616 Inlining is actually controlled by a number of parameters, which may be
4617 specified individually by using @option{--param @var{name}=@var{value}}.
4618 The @option{-finline-limit=@var{n}} option sets some of these parameters
4622 @item max-inline-insns-single
4623 is set to @var{n}/2.
4624 @item max-inline-insns-auto
4625 is set to @var{n}/2.
4626 @item min-inline-insns
4627 is set to 130 or @var{n}/4, whichever is smaller.
4628 @item max-inline-insns-rtl
4632 See below for a documentation of the individual
4633 parameters controlling inlining.
4635 @emph{Note:} pseudo instruction represents, in this particular context, an
4636 abstract measurement of function's size. In no way does it represent a count
4637 of assembly instructions and as such its exact meaning might change from one
4638 release to an another.
4640 @item -fkeep-inline-functions
4641 @opindex fkeep-inline-functions
4642 In C, emit @code{static} functions that are declared @code{inline}
4643 into the object file, even if the function has been inlined into all
4644 of its callers. This switch does not affect functions using the
4645 @code{extern inline} extension in GNU C@. In C++, emit any and all
4646 inline functions into the object file.
4648 @item -fkeep-static-consts
4649 @opindex fkeep-static-consts
4650 Emit variables declared @code{static const} when optimization isn't turned
4651 on, even if the variables aren't referenced.
4653 GCC enables this option by default. If you want to force the compiler to
4654 check if the variable was referenced, regardless of whether or not
4655 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4657 @item -fmerge-constants
4658 Attempt to merge identical constants (string constants and floating point
4659 constants) across compilation units.
4661 This option is the default for optimized compilation if the assembler and
4662 linker support it. Use @option{-fno-merge-constants} to inhibit this
4665 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4667 @item -fmerge-all-constants
4668 Attempt to merge identical constants and identical variables.
4670 This option implies @option{-fmerge-constants}. In addition to
4671 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4672 arrays or initialized constant variables with integral or floating point
4673 types. Languages like C or C++ require each non-automatic variable to
4674 have distinct location, so using this option will result in non-conforming
4677 @item -fmodulo-sched
4678 @opindex fmodulo-sched
4679 Perform swing modulo scheduling immediately before the first scheduling
4680 pass. This pass looks at innermost loops and reorders their
4681 instructions by overlapping different iterations.
4683 @item -fno-branch-count-reg
4684 @opindex fno-branch-count-reg
4685 Do not use ``decrement and branch'' instructions on a count register,
4686 but instead generate a sequence of instructions that decrement a
4687 register, compare it against zero, then branch based upon the result.
4688 This option is only meaningful on architectures that support such
4689 instructions, which include x86, PowerPC, IA-64 and S/390.
4691 The default is @option{-fbranch-count-reg}.
4693 @item -fno-function-cse
4694 @opindex fno-function-cse
4695 Do not put function addresses in registers; make each instruction that
4696 calls a constant function contain the function's address explicitly.
4698 This option results in less efficient code, but some strange hacks
4699 that alter the assembler output may be confused by the optimizations
4700 performed when this option is not used.
4702 The default is @option{-ffunction-cse}
4704 @item -fno-zero-initialized-in-bss
4705 @opindex fno-zero-initialized-in-bss
4706 If the target supports a BSS section, GCC by default puts variables that
4707 are initialized to zero into BSS@. This can save space in the resulting
4710 This option turns off this behavior because some programs explicitly
4711 rely on variables going to the data section. E.g., so that the
4712 resulting executable can find the beginning of that section and/or make
4713 assumptions based on that.
4715 The default is @option{-fzero-initialized-in-bss}.
4717 @item -fbounds-check
4718 @opindex fbounds-check
4719 For front-ends that support it, generate additional code to check that
4720 indices used to access arrays are within the declared range. This is
4721 currently only supported by the Java and Fortran front-ends, where
4722 this option defaults to true and false respectively.
4724 @item -fmudflap -fmudflapth -fmudflapir
4728 @cindex bounds checking
4730 For front-ends that support it (C and C++), instrument all risky
4731 pointer/array dereferencing operations, some standard library
4732 string/heap functions, and some other associated constructs with
4733 range/validity tests. Modules so instrumented should be immune to
4734 buffer overflows, invalid heap use, and some other classes of C/C++
4735 programming errors. The instrumentation relies on a separate runtime
4736 library (@file{libmudflap}), which will be linked into a program if
4737 @option{-fmudflap} is given at link time. Run-time behavior of the
4738 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4739 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4742 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4743 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4744 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4745 instrumentation should ignore pointer reads. This produces less
4746 instrumentation (and therefore faster execution) and still provides
4747 some protection against outright memory corrupting writes, but allows
4748 erroneously read data to propagate within a program.
4750 @item -fthread-jumps
4751 @opindex fthread-jumps
4752 Perform optimizations where we check to see if a jump branches to a
4753 location where another comparison subsumed by the first is found. If
4754 so, the first branch is redirected to either the destination of the
4755 second branch or a point immediately following it, depending on whether
4756 the condition is known to be true or false.
4758 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4760 @item -fcse-follow-jumps
4761 @opindex fcse-follow-jumps
4762 In common subexpression elimination, scan through jump instructions
4763 when the target of the jump is not reached by any other path. For
4764 example, when CSE encounters an @code{if} statement with an
4765 @code{else} clause, CSE will follow the jump when the condition
4768 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4770 @item -fcse-skip-blocks
4771 @opindex fcse-skip-blocks
4772 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4773 follow jumps which conditionally skip over blocks. When CSE
4774 encounters a simple @code{if} statement with no else clause,
4775 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4776 body of the @code{if}.
4778 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4780 @item -frerun-cse-after-loop
4781 @opindex frerun-cse-after-loop
4782 Re-run common subexpression elimination after loop optimizations has been
4785 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4789 Perform a global common subexpression elimination pass.
4790 This pass also performs global constant and copy propagation.
4792 @emph{Note:} When compiling a program using computed gotos, a GCC
4793 extension, you may get better runtime performance if you disable
4794 the global common subexpression elimination pass by adding
4795 @option{-fno-gcse} to the command line.
4797 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4801 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4802 attempt to move loads which are only killed by stores into themselves. This
4803 allows a loop containing a load/store sequence to be changed to a load outside
4804 the loop, and a copy/store within the loop.
4806 Enabled by default when gcse is enabled.
4810 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4811 global common subexpression elimination. This pass will attempt to move
4812 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4813 loops containing a load/store sequence can be changed to a load before
4814 the loop and a store after the loop.
4816 Not enabled at any optimization level.
4820 When @option{-fgcse-las} is enabled, the global common subexpression
4821 elimination pass eliminates redundant loads that come after stores to the
4822 same memory location (both partial and full redundancies).
4824 Not enabled at any optimization level.
4826 @item -fgcse-after-reload
4827 @opindex fgcse-after-reload
4828 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4829 pass is performed after reload. The purpose of this pass is to cleanup
4832 @item -funsafe-loop-optimizations
4833 @opindex funsafe-loop-optimizations
4834 If given, the loop optimizer will assume that loop indices do not
4835 overflow, and that the loops with nontrivial exit condition are not
4836 infinite. This enables a wider range of loop optimizations even if
4837 the loop optimizer itself cannot prove that these assumptions are valid.
4838 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4839 if it finds this kind of loop.
4841 @item -fcrossjumping
4842 @opindex crossjumping
4843 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4844 resulting code may or may not perform better than without cross-jumping.
4846 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4848 @item -fif-conversion
4849 @opindex if-conversion
4850 Attempt to transform conditional jumps into branch-less equivalents. This
4851 include use of conditional moves, min, max, set flags and abs instructions, and
4852 some tricks doable by standard arithmetics. The use of conditional execution
4853 on chips where it is available is controlled by @code{if-conversion2}.
4855 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4857 @item -fif-conversion2
4858 @opindex if-conversion2
4859 Use conditional execution (where available) to transform conditional jumps into
4860 branch-less equivalents.
4862 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4864 @item -fdelete-null-pointer-checks
4865 @opindex fdelete-null-pointer-checks
4866 Use global dataflow analysis to identify and eliminate useless checks
4867 for null pointers. The compiler assumes that dereferencing a null
4868 pointer would have halted the program. If a pointer is checked after
4869 it has already been dereferenced, it cannot be null.
4871 In some environments, this assumption is not true, and programs can
4872 safely dereference null pointers. Use
4873 @option{-fno-delete-null-pointer-checks} to disable this optimization
4874 for programs which depend on that behavior.
4876 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4878 @item -fexpensive-optimizations
4879 @opindex fexpensive-optimizations
4880 Perform a number of minor optimizations that are relatively expensive.
4882 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4884 @item -foptimize-register-move
4886 @opindex foptimize-register-move
4888 Attempt to reassign register numbers in move instructions and as
4889 operands of other simple instructions in order to maximize the amount of
4890 register tying. This is especially helpful on machines with two-operand
4893 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4896 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4898 @item -fdelayed-branch
4899 @opindex fdelayed-branch
4900 If supported for the target machine, attempt to reorder instructions
4901 to exploit instruction slots available after delayed branch
4904 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4906 @item -fschedule-insns
4907 @opindex fschedule-insns
4908 If supported for the target machine, attempt to reorder instructions to
4909 eliminate execution stalls due to required data being unavailable. This
4910 helps machines that have slow floating point or memory load instructions
4911 by allowing other instructions to be issued until the result of the load
4912 or floating point instruction is required.
4914 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4916 @item -fschedule-insns2
4917 @opindex fschedule-insns2
4918 Similar to @option{-fschedule-insns}, but requests an additional pass of
4919 instruction scheduling after register allocation has been done. This is
4920 especially useful on machines with a relatively small number of
4921 registers and where memory load instructions take more than one cycle.
4923 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4925 @item -fno-sched-interblock
4926 @opindex fno-sched-interblock
4927 Don't schedule instructions across basic blocks. This is normally
4928 enabled by default when scheduling before register allocation, i.e.@:
4929 with @option{-fschedule-insns} or at @option{-O2} or higher.
4931 @item -fno-sched-spec
4932 @opindex fno-sched-spec
4933 Don't allow speculative motion of non-load instructions. 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 -fsched-spec-load
4938 @opindex fsched-spec-load
4939 Allow speculative motion of some load instructions. This only makes
4940 sense when scheduling before register allocation, i.e.@: with
4941 @option{-fschedule-insns} or at @option{-O2} or higher.
4943 @item -fsched-spec-load-dangerous
4944 @opindex fsched-spec-load-dangerous
4945 Allow speculative motion of more 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-stalled-insns=@var{n}
4950 @opindex fsched-stalled-insns
4951 Define how many insns (if any) can be moved prematurely from the queue
4952 of stalled insns into the ready list, during the second scheduling pass.
4954 @item -fsched-stalled-insns-dep=@var{n}
4955 @opindex fsched-stalled-insns-dep
4956 Define how many insn groups (cycles) will be examined for a dependency
4957 on a stalled insn that is candidate for premature removal from the queue
4958 of stalled insns. Has an effect only during the second scheduling pass,
4959 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4961 @item -fsched2-use-superblocks
4962 @opindex fsched2-use-superblocks
4963 When scheduling after register allocation, do use superblock scheduling
4964 algorithm. Superblock scheduling allows motion across basic block boundaries
4965 resulting on faster schedules. This option is experimental, as not all machine
4966 descriptions used by GCC model the CPU closely enough to avoid unreliable
4967 results from the algorithm.
4969 This only makes sense when scheduling after register allocation, i.e.@: with
4970 @option{-fschedule-insns2} or at @option{-O2} or higher.
4972 @item -fsched2-use-traces
4973 @opindex fsched2-use-traces
4974 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4975 allocation and additionally perform code duplication in order to increase the
4976 size of superblocks using tracer pass. See @option{-ftracer} for details on
4979 This mode should produce faster but significantly longer programs. Also
4980 without @option{-fbranch-probabilities} the traces constructed may not
4981 match the reality and hurt the performance. This only makes
4982 sense when scheduling after register allocation, i.e.@: with
4983 @option{-fschedule-insns2} or at @option{-O2} or higher.
4985 @item -freschedule-modulo-scheduled-loops
4986 @opindex fscheduling-in-modulo-scheduled-loops
4987 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4988 we may want to prevent the later scheduling passes from changing its schedule, we use this
4989 option to control that.
4991 @item -fcaller-saves
4992 @opindex fcaller-saves
4993 Enable values to be allocated in registers that will be clobbered by
4994 function calls, by emitting extra instructions to save and restore the
4995 registers around such calls. Such allocation is done only when it
4996 seems to result in better code than would otherwise be produced.
4998 This option is always enabled by default on certain machines, usually
4999 those which have no call-preserved registers to use instead.
5001 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5004 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5005 enabled by default at @option{-O2} and @option{-O3}.
5008 Perform Full Redundancy Elimination (FRE) on trees. The difference
5009 between FRE and PRE is that FRE only considers expressions
5010 that are computed on all paths leading to the redundant computation.
5011 This analysis faster than PRE, though it exposes fewer redundancies.
5012 This flag is enabled by default at @option{-O} and higher.
5014 @item -ftree-copy-prop
5015 Perform copy propagation on trees. This pass eliminates unnecessary
5016 copy operations. This flag is enabled by default at @option{-O} and
5019 @item -ftree-store-copy-prop
5020 Perform copy propagation of memory loads and stores. This pass
5021 eliminates unnecessary copy operations in memory references
5022 (structures, global variables, arrays, etc). This flag is enabled by
5023 default at @option{-O2} and higher.
5026 Perform structural alias analysis on trees. This flag
5027 is enabled by default at @option{-O} and higher.
5030 Perform interprocedural pointer analysis.
5033 Perform forward store motion on trees. This flag is
5034 enabled by default at @option{-O} and higher.
5037 Perform sparse conditional constant propagation (CCP) on trees. This
5038 pass only operates on local scalar variables and is enabled by default
5039 at @option{-O} and higher.
5041 @item -ftree-store-ccp
5042 Perform sparse conditional constant propagation (CCP) on trees. This
5043 pass operates on both local scalar variables and memory stores and
5044 loads (global variables, structures, arrays, etc). This flag is
5045 enabled by default at @option{-O2} and higher.
5048 Perform dead code elimination (DCE) on trees. This flag is enabled by
5049 default at @option{-O} and higher.
5051 @item -ftree-dominator-opts
5052 Perform a variety of simple scalar cleanups (constant/copy
5053 propagation, redundancy elimination, range propagation and expression
5054 simplification) based on a dominator tree traversal. This also
5055 performs jump threading (to reduce jumps to jumps). This flag is
5056 enabled by default at @option{-O} and higher.
5059 Perform loop header copying on trees. This is beneficial since it increases
5060 effectiveness of code motion optimizations. It also saves one jump. This flag
5061 is enabled by default at @option{-O} and higher. It is not enabled
5062 for @option{-Os}, since it usually increases code size.
5064 @item -ftree-loop-optimize
5065 Perform loop optimizations on trees. This flag is enabled by default
5066 at @option{-O} and higher.
5068 @item -ftree-loop-linear
5069 Perform linear loop transformations on tree. This flag can improve cache
5070 performance and allow further loop optimizations to take place.
5072 @item -ftree-loop-im
5073 Perform loop invariant motion on trees. This pass moves only invariants that
5074 would be hard to handle at RTL level (function calls, operations that expand to
5075 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5076 operands of conditions that are invariant out of the loop, so that we can use
5077 just trivial invariantness analysis in loop unswitching. The pass also includes
5080 @item -ftree-loop-ivcanon
5081 Create a canonical counter for number of iterations in the loop for that
5082 determining number of iterations requires complicated analysis. Later
5083 optimizations then may determine the number easily. Useful especially
5084 in connection with unrolling.
5087 Perform induction variable optimizations (strength reduction, induction
5088 variable merging and induction variable elimination) on trees.
5091 Perform scalar replacement of aggregates. This pass replaces structure
5092 references with scalars to prevent committing structures to memory too
5093 early. This flag is enabled by default at @option{-O} and higher.
5095 @item -ftree-copyrename
5096 Perform copy renaming on trees. This pass attempts to rename compiler
5097 temporaries to other variables at copy locations, usually resulting in
5098 variable names which more closely resemble the original variables. This flag
5099 is enabled by default at @option{-O} and higher.
5102 Perform temporary expression replacement during the SSA->normal phase. Single
5103 use/single def temporaries are replaced at their use location with their
5104 defining expression. This results in non-GIMPLE code, but gives the expanders
5105 much more complex trees to work on resulting in better RTL generation. This is
5106 enabled by default at @option{-O} and higher.
5109 Perform live range splitting during the SSA->normal phase. Distinct live
5110 ranges of a variable are split into unique variables, allowing for better
5111 optimization later. This is enabled by default at @option{-O} and higher.
5113 @item -ftree-vectorize
5114 Perform loop vectorization on trees.
5116 @item -ftree-vect-loop-version
5117 @opindex ftree-vect-loop-version
5118 Perform loop versioning when doing loop vectorization on trees. When a loop
5119 appears to be vectorizable except that data alignment or data dependence cannot
5120 be determined at compile time then vectorized and non-vectorized versions of
5121 the loop are generated along with runtime checks for alignment or dependence
5122 to control which version is executed. This option is enabled by default
5123 except at level @option{-Os} where it is disabled.
5126 Perform Value Range Propagation on trees. This is similar to the
5127 constant propagation pass, but instead of values, ranges of values are
5128 propagated. This allows the optimizers to remove unnecessary range
5129 checks like array bound checks and null pointer checks. This is
5130 enabled by default at @option{-O2} and higher. Null pointer check
5131 elimination is only done if @option{-fdelete-null-pointer-checks} is
5136 Perform tail duplication to enlarge superblock size. This transformation
5137 simplifies the control flow of the function allowing other optimizations to do
5140 @item -funroll-loops
5141 @opindex funroll-loops
5142 Unroll loops whose number of iterations can be determined at compile
5143 time or upon entry to the loop. @option{-funroll-loops} implies
5144 @option{-frerun-cse-after-loop}. This option makes code larger,
5145 and may or may not make it run faster.
5147 @item -funroll-all-loops
5148 @opindex funroll-all-loops
5149 Unroll all loops, even if their number of iterations is uncertain when
5150 the loop is entered. This usually makes programs run more slowly.
5151 @option{-funroll-all-loops} implies the same options as
5152 @option{-funroll-loops},
5154 @item -fsplit-ivs-in-unroller
5155 @opindex -fsplit-ivs-in-unroller
5156 Enables expressing of values of induction variables in later iterations
5157 of the unrolled loop using the value in the first iteration. This breaks
5158 long dependency chains, thus improving efficiency of the scheduling passes.
5160 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5161 same effect. However in cases the loop body is more complicated than
5162 a single basic block, this is not reliable. It also does not work at all
5163 on some of the architectures due to restrictions in the CSE pass.
5165 This optimization is enabled by default.
5167 @item -fvariable-expansion-in-unroller
5168 @opindex -fvariable-expansion-in-unroller
5169 With this option, the compiler will create multiple copies of some
5170 local variables when unrolling a loop which can result in superior code.
5172 @item -fprefetch-loop-arrays
5173 @opindex fprefetch-loop-arrays
5174 If supported by the target machine, generate instructions to prefetch
5175 memory to improve the performance of loops that access large arrays.
5177 This option may generate better or worse code; results are highly
5178 dependent on the structure of loops within the source code.
5180 Disabled at level @option{-Os}.
5183 @itemx -fno-peephole2
5184 @opindex fno-peephole
5185 @opindex fno-peephole2
5186 Disable any machine-specific peephole optimizations. The difference
5187 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5188 are implemented in the compiler; some targets use one, some use the
5189 other, a few use both.
5191 @option{-fpeephole} is enabled by default.
5192 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5194 @item -fno-guess-branch-probability
5195 @opindex fno-guess-branch-probability
5196 Do not guess branch probabilities using heuristics.
5198 GCC will use heuristics to guess branch probabilities if they are
5199 not provided by profiling feedback (@option{-fprofile-arcs}). These
5200 heuristics are based on the control flow graph. If some branch probabilities
5201 are specified by @samp{__builtin_expect}, then the heuristics will be
5202 used to guess branch probabilities for the rest of the control flow graph,
5203 taking the @samp{__builtin_expect} info into account. The interactions
5204 between the heuristics and @samp{__builtin_expect} can be complex, and in
5205 some cases, it may be useful to disable the heuristics so that the effects
5206 of @samp{__builtin_expect} are easier to understand.
5208 The default is @option{-fguess-branch-probability} at levels
5209 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5211 @item -freorder-blocks
5212 @opindex freorder-blocks
5213 Reorder basic blocks in the compiled function in order to reduce number of
5214 taken branches and improve code locality.
5216 Enabled at levels @option{-O2}, @option{-O3}.
5218 @item -freorder-blocks-and-partition
5219 @opindex freorder-blocks-and-partition
5220 In addition to reordering basic blocks in the compiled function, in order
5221 to reduce number of taken branches, partitions hot and cold basic blocks
5222 into separate sections of the assembly and .o files, to improve
5223 paging and cache locality performance.
5225 This optimization is automatically turned off in the presence of
5226 exception handling, for linkonce sections, for functions with a user-defined
5227 section attribute and on any architecture that does not support named
5230 @item -freorder-functions
5231 @opindex freorder-functions
5232 Reorder functions in the object file in order to
5233 improve code locality. This is implemented by using special
5234 subsections @code{.text.hot} for most frequently executed functions and
5235 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5236 the linker so object file format must support named sections and linker must
5237 place them in a reasonable way.
5239 Also profile feedback must be available in to make this option effective. See
5240 @option{-fprofile-arcs} for details.
5242 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5244 @item -fstrict-aliasing
5245 @opindex fstrict-aliasing
5246 Allows the compiler to assume the strictest aliasing rules applicable to
5247 the language being compiled. For C (and C++), this activates
5248 optimizations based on the type of expressions. In particular, an
5249 object of one type is assumed never to reside at the same address as an
5250 object of a different type, unless the types are almost the same. For
5251 example, an @code{unsigned int} can alias an @code{int}, but not a
5252 @code{void*} or a @code{double}. A character type may alias any other
5255 Pay special attention to code like this:
5268 The practice of reading from a different union member than the one most
5269 recently written to (called ``type-punning'') is common. Even with
5270 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5271 is accessed through the union type. So, the code above will work as
5272 expected. However, this code might not:
5283 Every language that wishes to perform language-specific alias analysis
5284 should define a function that computes, given an @code{tree}
5285 node, an alias set for the node. Nodes in different alias sets are not
5286 allowed to alias. For an example, see the C front-end function
5287 @code{c_get_alias_set}.
5289 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5291 @item -falign-functions
5292 @itemx -falign-functions=@var{n}
5293 @opindex falign-functions
5294 Align the start of functions to the next power-of-two greater than
5295 @var{n}, skipping up to @var{n} bytes. For instance,
5296 @option{-falign-functions=32} aligns functions to the next 32-byte
5297 boundary, but @option{-falign-functions=24} would align to the next
5298 32-byte boundary only if this can be done by skipping 23 bytes or less.
5300 @option{-fno-align-functions} and @option{-falign-functions=1} are
5301 equivalent and mean that functions will not be aligned.
5303 Some assemblers only support this flag when @var{n} is a power of two;
5304 in that case, it is rounded up.
5306 If @var{n} is not specified or is zero, use a machine-dependent default.
5308 Enabled at levels @option{-O2}, @option{-O3}.
5310 @item -falign-labels
5311 @itemx -falign-labels=@var{n}
5312 @opindex falign-labels
5313 Align all branch targets to a power-of-two boundary, skipping up to
5314 @var{n} bytes like @option{-falign-functions}. This option can easily
5315 make code slower, because it must insert dummy operations for when the
5316 branch target is reached in the usual flow of the code.
5318 @option{-fno-align-labels} and @option{-falign-labels=1} are
5319 equivalent and mean that labels will not be aligned.
5321 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5322 are greater than this value, then their values are used instead.
5324 If @var{n} is not specified or is zero, use a machine-dependent default
5325 which is very likely to be @samp{1}, meaning no alignment.
5327 Enabled at levels @option{-O2}, @option{-O3}.
5330 @itemx -falign-loops=@var{n}
5331 @opindex falign-loops
5332 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5333 like @option{-falign-functions}. The hope is that the loop will be
5334 executed many times, which will make up for any execution of the dummy
5337 @option{-fno-align-loops} and @option{-falign-loops=1} are
5338 equivalent and mean that loops will not be aligned.
5340 If @var{n} is not specified or is zero, use a machine-dependent default.
5342 Enabled at levels @option{-O2}, @option{-O3}.
5345 @itemx -falign-jumps=@var{n}
5346 @opindex falign-jumps
5347 Align branch targets to a power-of-two boundary, for branch targets
5348 where the targets can only be reached by jumping, skipping up to @var{n}
5349 bytes like @option{-falign-functions}. In this case, no dummy operations
5352 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5353 equivalent and mean that loops will not be aligned.
5355 If @var{n} is not specified or is zero, use a machine-dependent default.
5357 Enabled at levels @option{-O2}, @option{-O3}.
5359 @item -funit-at-a-time
5360 @opindex funit-at-a-time
5361 Parse the whole compilation unit before starting to produce code.
5362 This allows some extra optimizations to take place but consumes
5363 more memory (in general). There are some compatibility issues
5364 with @emph{unit-at-a-time} mode:
5367 enabling @emph{unit-at-a-time} mode may change the order
5368 in which functions, variables, and top-level @code{asm} statements
5369 are emitted, and will likely break code relying on some particular
5370 ordering. The majority of such top-level @code{asm} statements,
5371 though, can be replaced by @code{section} attributes. The
5372 @option{fno-toplevel-reorder} option may be used to keep the ordering
5373 used in the input file, at the cost of some optimizations.
5376 @emph{unit-at-a-time} mode removes unreferenced static variables
5377 and functions. This may result in undefined references
5378 when an @code{asm} statement refers directly to variables or functions
5379 that are otherwise unused. In that case either the variable/function
5380 shall be listed as an operand of the @code{asm} statement operand or,
5381 in the case of top-level @code{asm} statements the attribute @code{used}
5382 shall be used on the declaration.
5385 Static functions now can use non-standard passing conventions that
5386 may break @code{asm} statements calling functions directly. Again,
5387 attribute @code{used} will prevent this behavior.
5390 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5391 but this scheme may not be supported by future releases of GCC@.
5393 Enabled at levels @option{-O2}, @option{-O3}.
5395 @item -fno-toplevel-reorder
5396 Do not reorder top-level functions, variables, and @code{asm}
5397 statements. Output them in the same order that they appear in the
5398 input file. When this option is used, unreferenced static variables
5399 will not be removed. This option is intended to support existing code
5400 which relies on a particular ordering. For new code, it is better to
5405 Constructs webs as commonly used for register allocation purposes and assign
5406 each web individual pseudo register. This allows the register allocation pass
5407 to operate on pseudos directly, but also strengthens several other optimization
5408 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5409 however, make debugging impossible, since variables will no longer stay in a
5412 Enabled by default with @option{-funroll-loops}.
5414 @item -fwhole-program
5415 @opindex fwhole-program
5416 Assume that the current compilation unit represents whole program being
5417 compiled. All public functions and variables with the exception of @code{main}
5418 and those merged by attribute @code{externally_visible} become static functions
5419 and in a affect gets more aggressively optimized by interprocedural optimizers.
5420 While this option is equivalent to proper use of @code{static} keyword for
5421 programs consisting of single file, in combination with option
5422 @option{--combine} this flag can be used to compile most of smaller scale C
5423 programs since the functions and variables become local for the whole combined
5424 compilation unit, not for the single source file itself.
5427 @item -fno-cprop-registers
5428 @opindex fno-cprop-registers
5429 After register allocation and post-register allocation instruction splitting,
5430 we perform a copy-propagation pass to try to reduce scheduling dependencies
5431 and occasionally eliminate the copy.
5433 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5435 @item -fprofile-generate
5436 @opindex fprofile-generate
5438 Enable options usually used for instrumenting application to produce
5439 profile useful for later recompilation with profile feedback based
5440 optimization. You must use @option{-fprofile-generate} both when
5441 compiling and when linking your program.
5443 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5446 @opindex fprofile-use
5447 Enable profile feedback directed optimizations, and optimizations
5448 generally profitable only with profile feedback available.
5450 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5451 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5455 The following options control compiler behavior regarding floating
5456 point arithmetic. These options trade off between speed and
5457 correctness. All must be specifically enabled.
5461 @opindex ffloat-store
5462 Do not store floating point variables in registers, and inhibit other
5463 options that might change whether a floating point value is taken from a
5466 @cindex floating point precision
5467 This option prevents undesirable excess precision on machines such as
5468 the 68000 where the floating registers (of the 68881) keep more
5469 precision than a @code{double} is supposed to have. Similarly for the
5470 x86 architecture. For most programs, the excess precision does only
5471 good, but a few programs rely on the precise definition of IEEE floating
5472 point. Use @option{-ffloat-store} for such programs, after modifying
5473 them to store all pertinent intermediate computations into variables.
5477 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5478 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5479 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5480 and @option{fcx-limited-range}.
5482 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5484 This option should never be turned on by any @option{-O} option since
5485 it can result in incorrect output for programs which depend on
5486 an exact implementation of IEEE or ISO rules/specifications for
5489 @item -fno-math-errno
5490 @opindex fno-math-errno
5491 Do not set ERRNO after calling math functions that are executed
5492 with a single instruction, e.g., sqrt. A program that relies on
5493 IEEE exceptions for math error handling may want to use this flag
5494 for speed while maintaining IEEE arithmetic compatibility.
5496 This option should never be turned on by any @option{-O} option since
5497 it can result in incorrect output for programs which depend on
5498 an exact implementation of IEEE or ISO rules/specifications for
5501 The default is @option{-fmath-errno}.
5503 On Darwin systems, the math library never sets @code{errno}. There is therefore
5504 no reason for the compiler to consider the possibility that it might,
5505 and @option{-fno-math-errno} is the default.
5507 @item -funsafe-math-optimizations
5508 @opindex funsafe-math-optimizations
5509 Allow optimizations for floating-point arithmetic that (a) assume
5510 that arguments and results are valid and (b) may violate IEEE or
5511 ANSI standards. When used at link-time, it may include libraries
5512 or startup files that change the default FPU control word or other
5513 similar optimizations.
5515 This option should never be turned on by any @option{-O} option since
5516 it can result in incorrect output for programs which depend on
5517 an exact implementation of IEEE or ISO rules/specifications for
5520 The default is @option{-fno-unsafe-math-optimizations}.
5522 @item -ffinite-math-only
5523 @opindex ffinite-math-only
5524 Allow optimizations for floating-point arithmetic that assume
5525 that arguments and results are not NaNs or +-Infs.
5527 This option should never be turned on by any @option{-O} option since
5528 it can result in incorrect output for programs which depend on
5529 an exact implementation of IEEE or ISO rules/specifications.
5531 The default is @option{-fno-finite-math-only}.
5533 @item -fno-trapping-math
5534 @opindex fno-trapping-math
5535 Compile code assuming that floating-point operations cannot generate
5536 user-visible traps. These traps include division by zero, overflow,
5537 underflow, inexact result and invalid operation. This option implies
5538 @option{-fno-signaling-nans}. Setting this option may allow faster
5539 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5541 This option should never be turned on by any @option{-O} option since
5542 it can result in incorrect output for programs which depend on
5543 an exact implementation of IEEE or ISO rules/specifications for
5546 The default is @option{-ftrapping-math}.
5548 @item -frounding-math
5549 @opindex frounding-math
5550 Disable transformations and optimizations that assume default floating
5551 point rounding behavior. This is round-to-zero for all floating point
5552 to integer conversions, and round-to-nearest for all other arithmetic
5553 truncations. This option should be specified for programs that change
5554 the FP rounding mode dynamically, or that may be executed with a
5555 non-default rounding mode. This option disables constant folding of
5556 floating point expressions at compile-time (which may be affected by
5557 rounding mode) and arithmetic transformations that are unsafe in the
5558 presence of sign-dependent rounding modes.
5560 The default is @option{-fno-rounding-math}.
5562 This option is experimental and does not currently guarantee to
5563 disable all GCC optimizations that are affected by rounding mode.
5564 Future versions of GCC may provide finer control of this setting
5565 using C99's @code{FENV_ACCESS} pragma. This command line option
5566 will be used to specify the default state for @code{FENV_ACCESS}.
5568 @item -frtl-abstract-sequences
5569 @opindex frtl-abstract-sequences
5570 It is a size optimization method. This option is to find identical
5571 sequences of code, which can be turned into pseudo-procedures and
5572 then replace all occurrences with calls to the newly created
5573 subroutine. It is kind of an opposite of @option{-finline-functions}.
5574 This optimization runs at RTL level.
5576 @item -fsignaling-nans
5577 @opindex fsignaling-nans
5578 Compile code assuming that IEEE signaling NaNs may generate user-visible
5579 traps during floating-point operations. Setting this option disables
5580 optimizations that may change the number of exceptions visible with
5581 signaling NaNs. This option implies @option{-ftrapping-math}.
5583 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5586 The default is @option{-fno-signaling-nans}.
5588 This option is experimental and does not currently guarantee to
5589 disable all GCC optimizations that affect signaling NaN behavior.
5591 @item -fsingle-precision-constant
5592 @opindex fsingle-precision-constant
5593 Treat floating point constant as single precision constant instead of
5594 implicitly converting it to double precision constant.
5596 @item -fcx-limited-range
5597 @itemx -fno-cx-limited-range
5598 @opindex fcx-limited-range
5599 @opindex fno-cx-limited-range
5600 When enabled, this option states that a range reduction step is not
5601 needed when performing complex division. The default is
5602 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5604 This option controls the default setting of the ISO C99
5605 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5610 The following options control optimizations that may improve
5611 performance, but are not enabled by any @option{-O} options. This
5612 section includes experimental options that may produce broken code.
5615 @item -fbranch-probabilities
5616 @opindex fbranch-probabilities
5617 After running a program compiled with @option{-fprofile-arcs}
5618 (@pxref{Debugging Options,, Options for Debugging Your Program or
5619 @command{gcc}}), you can compile it a second time using
5620 @option{-fbranch-probabilities}, to improve optimizations based on
5621 the number of times each branch was taken. When the program
5622 compiled with @option{-fprofile-arcs} exits it saves arc execution
5623 counts to a file called @file{@var{sourcename}.gcda} for each source
5624 file The information in this data file is very dependent on the
5625 structure of the generated code, so you must use the same source code
5626 and the same optimization options for both compilations.
5628 With @option{-fbranch-probabilities}, GCC puts a
5629 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5630 These can be used to improve optimization. Currently, they are only
5631 used in one place: in @file{reorg.c}, instead of guessing which path a
5632 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5633 exactly determine which path is taken more often.
5635 @item -fprofile-values
5636 @opindex fprofile-values
5637 If combined with @option{-fprofile-arcs}, it adds code so that some
5638 data about values of expressions in the program is gathered.
5640 With @option{-fbranch-probabilities}, it reads back the data gathered
5641 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5642 notes to instructions for their later usage in optimizations.
5644 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5648 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5649 a code to gather information about values of expressions.
5651 With @option{-fbranch-probabilities}, it reads back the data gathered
5652 and actually performs the optimizations based on them.
5653 Currently the optimizations include specialization of division operation
5654 using the knowledge about the value of the denominator.
5656 @item -frename-registers
5657 @opindex frename-registers
5658 Attempt to avoid false dependencies in scheduled code by making use
5659 of registers left over after register allocation. This optimization
5660 will most benefit processors with lots of registers. Depending on the
5661 debug information format adopted by the target, however, it can
5662 make debugging impossible, since variables will no longer stay in
5663 a ``home register''.
5665 Enabled by default with @option{-funroll-loops}.
5669 Perform tail duplication to enlarge superblock size. This transformation
5670 simplifies the control flow of the function allowing other optimizations to do
5673 Enabled with @option{-fprofile-use}.
5675 @item -funroll-loops
5676 @opindex funroll-loops
5677 Unroll loops whose number of iterations can be determined at compile time or
5678 upon entry to the loop. @option{-funroll-loops} implies
5679 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5680 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5681 small constant number of iterations). This option makes code larger, and may
5682 or may not make it run faster.
5684 Enabled with @option{-fprofile-use}.
5686 @item -funroll-all-loops
5687 @opindex funroll-all-loops
5688 Unroll all loops, even if their number of iterations is uncertain when
5689 the loop is entered. This usually makes programs run more slowly.
5690 @option{-funroll-all-loops} implies the same options as
5691 @option{-funroll-loops}.
5694 @opindex fpeel-loops
5695 Peels the loops for that there is enough information that they do not
5696 roll much (from profile feedback). It also turns on complete loop peeling
5697 (i.e.@: complete removal of loops with small constant number of iterations).
5699 Enabled with @option{-fprofile-use}.
5701 @item -fmove-loop-invariants
5702 @opindex fmove-loop-invariants
5703 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5704 at level @option{-O1}
5706 @item -funswitch-loops
5707 @opindex funswitch-loops
5708 Move branches with loop invariant conditions out of the loop, with duplicates
5709 of the loop on both branches (modified according to result of the condition).
5711 @item -ffunction-sections
5712 @itemx -fdata-sections
5713 @opindex ffunction-sections
5714 @opindex fdata-sections
5715 Place each function or data item into its own section in the output
5716 file if the target supports arbitrary sections. The name of the
5717 function or the name of the data item determines the section's name
5720 Use these options on systems where the linker can perform optimizations
5721 to improve locality of reference in the instruction space. Most systems
5722 using the ELF object format and SPARC processors running Solaris 2 have
5723 linkers with such optimizations. AIX may have these optimizations in
5726 Only use these options when there are significant benefits from doing
5727 so. When you specify these options, the assembler and linker will
5728 create larger object and executable files and will also be slower.
5729 You will not be able to use @code{gprof} on all systems if you
5730 specify this option and you may have problems with debugging if
5731 you specify both this option and @option{-g}.
5733 @item -fbranch-target-load-optimize
5734 @opindex fbranch-target-load-optimize
5735 Perform branch target register load optimization before prologue / epilogue
5737 The use of target registers can typically be exposed only during reload,
5738 thus hoisting loads out of loops and doing inter-block scheduling needs
5739 a separate optimization pass.
5741 @item -fbranch-target-load-optimize2
5742 @opindex fbranch-target-load-optimize2
5743 Perform branch target register load optimization after prologue / epilogue
5746 @item -fbtr-bb-exclusive
5747 @opindex fbtr-bb-exclusive
5748 When performing branch target register load optimization, don't reuse
5749 branch target registers in within any basic block.
5751 @item -fstack-protector
5752 Emit extra code to check for buffer overflows, such as stack smashing
5753 attacks. This is done by adding a guard variable to functions with
5754 vulnerable objects. This includes functions that call alloca, and
5755 functions with buffers larger than 8 bytes. The guards are initialized
5756 when a function is entered and then checked when the function exits.
5757 If a guard check fails, an error message is printed and the program exits.
5759 @item -fstack-protector-all
5760 Like @option{-fstack-protector} except that all functions are protected.
5762 @item -fsection-anchors
5763 @opindex fsection-anchors
5764 Try to reduce the number of symbolic address calculations by using
5765 shared ``anchor'' symbols to address nearby objects. This transformation
5766 can help to reduce the number of GOT entries and GOT accesses on some
5769 For example, the implementation of the following function @code{foo}:
5773 int foo (void) @{ return a + b + c; @}
5776 would usually calculate the addresses of all three variables, but if you
5777 compile it with @option{-fsection-anchors}, it will access the variables
5778 from a common anchor point instead. The effect is similar to the
5779 following pseudocode (which isn't valid C):
5784 register int *xr = &x;
5785 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5789 Not all targets support this option.
5791 @item --param @var{name}=@var{value}
5793 In some places, GCC uses various constants to control the amount of
5794 optimization that is done. For example, GCC will not inline functions
5795 that contain more that a certain number of instructions. You can
5796 control some of these constants on the command-line using the
5797 @option{--param} option.
5799 The names of specific parameters, and the meaning of the values, are
5800 tied to the internals of the compiler, and are subject to change
5801 without notice in future releases.
5803 In each case, the @var{value} is an integer. The allowable choices for
5804 @var{name} are given in the following table:
5807 @item salias-max-implicit-fields
5808 The maximum number of fields in a variable without direct
5809 structure accesses for which structure aliasing will consider trying
5810 to track each field. The default is 5
5812 @item salias-max-array-elements
5813 The maximum number of elements an array can have and its elements
5814 still be tracked individually by structure aliasing. The default is 4
5816 @item sra-max-structure-size
5817 The maximum structure size, in bytes, at which the scalar replacement
5818 of aggregates (SRA) optimization will perform block copies. The
5819 default value, 0, implies that GCC will select the most appropriate
5822 @item sra-field-structure-ratio
5823 The threshold ratio (as a percentage) between instantiated fields and
5824 the complete structure size. We say that if the ratio of the number
5825 of bytes in instantiated fields to the number of bytes in the complete
5826 structure exceeds this parameter, then block copies are not used. The
5829 @item max-crossjump-edges
5830 The maximum number of incoming edges to consider for crossjumping.
5831 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5832 the number of edges incoming to each block. Increasing values mean
5833 more aggressive optimization, making the compile time increase with
5834 probably small improvement in executable size.
5836 @item min-crossjump-insns
5837 The minimum number of instructions which must be matched at the end
5838 of two blocks before crossjumping will be performed on them. This
5839 value is ignored in the case where all instructions in the block being
5840 crossjumped from are matched. The default value is 5.
5842 @item max-grow-copy-bb-insns
5843 The maximum code size expansion factor when copying basic blocks
5844 instead of jumping. The expansion is relative to a jump instruction.
5845 The default value is 8.
5847 @item max-goto-duplication-insns
5848 The maximum number of instructions to duplicate to a block that jumps
5849 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5850 passes, GCC factors computed gotos early in the compilation process,
5851 and unfactors them as late as possible. Only computed jumps at the
5852 end of a basic blocks with no more than max-goto-duplication-insns are
5853 unfactored. The default value is 8.
5855 @item max-delay-slot-insn-search
5856 The maximum number of instructions to consider when looking for an
5857 instruction to fill a delay slot. If more than this arbitrary number of
5858 instructions is searched, the time savings from filling the delay slot
5859 will be minimal so stop searching. Increasing values mean more
5860 aggressive optimization, making the compile time increase with probably
5861 small improvement in executable run time.
5863 @item max-delay-slot-live-search
5864 When trying to fill delay slots, the maximum number of instructions to
5865 consider when searching for a block with valid live register
5866 information. Increasing this arbitrarily chosen value means more
5867 aggressive optimization, increasing the compile time. This parameter
5868 should be removed when the delay slot code is rewritten to maintain the
5871 @item max-gcse-memory
5872 The approximate maximum amount of memory that will be allocated in
5873 order to perform the global common subexpression elimination
5874 optimization. If more memory than specified is required, the
5875 optimization will not be done.
5877 @item max-gcse-passes
5878 The maximum number of passes of GCSE to run. The default is 1.
5880 @item max-pending-list-length
5881 The maximum number of pending dependencies scheduling will allow
5882 before flushing the current state and starting over. Large functions
5883 with few branches or calls can create excessively large lists which
5884 needlessly consume memory and resources.
5886 @item max-inline-insns-single
5887 Several parameters control the tree inliner used in gcc.
5888 This number sets the maximum number of instructions (counted in GCC's
5889 internal representation) in a single function that the tree inliner
5890 will consider for inlining. This only affects functions declared
5891 inline and methods implemented in a class declaration (C++).
5892 The default value is 450.
5894 @item max-inline-insns-auto
5895 When you use @option{-finline-functions} (included in @option{-O3}),
5896 a lot of functions that would otherwise not be considered for inlining
5897 by the compiler will be investigated. To those functions, a different
5898 (more restrictive) limit compared to functions declared inline can
5900 The default value is 90.
5902 @item large-function-insns
5903 The limit specifying really large functions. For functions larger than this
5904 limit after inlining inlining is constrained by
5905 @option{--param large-function-growth}. This parameter is useful primarily
5906 to avoid extreme compilation time caused by non-linear algorithms used by the
5908 This parameter is ignored when @option{-funit-at-a-time} is not used.
5909 The default value is 2700.
5911 @item large-function-growth
5912 Specifies maximal growth of large function caused by inlining in percents.
5913 This parameter is ignored when @option{-funit-at-a-time} is not used.
5914 The default value is 100 which limits large function growth to 2.0 times
5917 @item large-unit-insns
5918 The limit specifying large translation unit. Growth caused by inlining of
5919 units larger than this limit is limited by @option{--param inline-unit-growth}.
5920 For small units this might be too tight (consider unit consisting of function A
5921 that is inline and B that just calls A three time. If B is small relative to
5922 A, the growth of unit is 300\% and yet such inlining is very sane. For very
5923 large units consisting of small inlininable functions however the overall unit
5924 growth limit is needed to avoid exponential explosion of code size. Thus for
5925 smaller units, the size is increased to @option{--param large-unit-insns}
5926 before applying @option{--param inline-unit-growth}. The default is 10000
5928 @item inline-unit-growth
5929 Specifies maximal overall growth of the compilation unit caused by inlining.
5930 This parameter is ignored when @option{-funit-at-a-time} is not used.
5931 The default value is 50 which limits unit growth to 1.5 times the original
5934 @item max-inline-insns-recursive
5935 @itemx max-inline-insns-recursive-auto
5936 Specifies maximum number of instructions out-of-line copy of self recursive inline
5937 function can grow into by performing recursive inlining.
5939 For functions declared inline @option{--param max-inline-insns-recursive} is
5940 taken into acount. For function not declared inline, recursive inlining
5941 happens only when @option{-finline-functions} (included in @option{-O3}) is
5942 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5943 default value is 450.
5945 @item max-inline-recursive-depth
5946 @itemx max-inline-recursive-depth-auto
5947 Specifies maximum recursion depth used by the recursive inlining.
5949 For functions declared inline @option{--param max-inline-recursive-depth} is
5950 taken into acount. For function not declared inline, recursive inlining
5951 happens only when @option{-finline-functions} (included in @option{-O3}) is
5952 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5953 default value is 450.
5955 @item min-inline-recursive-probability
5956 Recursive inlining is profitable only for function having deep recursion
5957 in average and can hurt for function having little recursion depth by
5958 increasing the prologue size or complexity of function body to other
5961 When profile feedback is available (see @option{-fprofile-generate}) the actual
5962 recursion depth can be guessed from probability that function will recurse via
5963 given call expression. This parameter limits inlining only to call expression
5964 whose probability exceeds given threshold (in percents). The default value is
5967 @item inline-call-cost
5968 Specify cost of call instruction relative to simple arithmetics operations
5969 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
5970 functions and at the same time increases size of leaf function that is believed to
5971 reduce function size by being inlined. In effect it increases amount of
5972 inlining for code having large abstraction penalty (many functions that just
5973 pass the arguments to other functions) and decrease inlining for code with low
5974 abstraction penalty. The default value is 16.
5976 @item max-unrolled-insns
5977 The maximum number of instructions that a loop should have if that loop
5978 is unrolled, and if the loop is unrolled, it determines how many times
5979 the loop code is unrolled.
5981 @item max-average-unrolled-insns
5982 The maximum number of instructions biased by probabilities of their execution
5983 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5984 it determines how many times the loop code is unrolled.
5986 @item max-unroll-times
5987 The maximum number of unrollings of a single loop.
5989 @item max-peeled-insns
5990 The maximum number of instructions that a loop should have if that loop
5991 is peeled, and if the loop is peeled, it determines how many times
5992 the loop code is peeled.
5994 @item max-peel-times
5995 The maximum number of peelings of a single loop.
5997 @item max-completely-peeled-insns
5998 The maximum number of insns of a completely peeled loop.
6000 @item max-completely-peel-times
6001 The maximum number of iterations of a loop to be suitable for complete peeling.
6003 @item max-unswitch-insns
6004 The maximum number of insns of an unswitched loop.
6006 @item max-unswitch-level
6007 The maximum number of branches unswitched in a single loop.
6010 The minimum cost of an expensive expression in the loop invariant motion.
6012 @item iv-consider-all-candidates-bound
6013 Bound on number of candidates for induction variables below that
6014 all candidates are considered for each use in induction variable
6015 optimizations. Only the most relevant candidates are considered
6016 if there are more candidates, to avoid quadratic time complexity.
6018 @item iv-max-considered-uses
6019 The induction variable optimizations give up on loops that contain more
6020 induction variable uses.
6022 @item iv-always-prune-cand-set-bound
6023 If number of candidates in the set is smaller than this value,
6024 we always try to remove unnecessary ivs from the set during its
6025 optimization when a new iv is added to the set.
6027 @item scev-max-expr-size
6028 Bound on size of expressions used in the scalar evolutions analyzer.
6029 Large expressions slow the analyzer.
6031 @item vect-max-version-checks
6032 The maximum number of runtime checks that can be performed when doing
6033 loop versioning in the vectorizer. See option ftree-vect-loop-version
6034 for more information.
6036 @item max-iterations-to-track
6038 The maximum number of iterations of a loop the brute force algorithm
6039 for analysis of # of iterations of the loop tries to evaluate.
6041 @item hot-bb-count-fraction
6042 Select fraction of the maximal count of repetitions of basic block in program
6043 given basic block needs to have to be considered hot.
6045 @item hot-bb-frequency-fraction
6046 Select fraction of the maximal frequency of executions of basic block in
6047 function given basic block needs to have to be considered hot
6049 @item max-predicted-iterations
6050 The maximum number of loop iterations we predict statically. This is useful
6051 in cases where function contain single loop with known bound and other loop
6052 with unknown. We predict the known number of iterations correctly, while
6053 the unknown number of iterations average to roughly 10. This means that the
6054 loop without bounds would appear artificially cold relative to the other one.
6056 @item tracer-dynamic-coverage
6057 @itemx tracer-dynamic-coverage-feedback
6059 This value is used to limit superblock formation once the given percentage of
6060 executed instructions is covered. This limits unnecessary code size
6063 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6064 feedback is available. The real profiles (as opposed to statically estimated
6065 ones) are much less balanced allowing the threshold to be larger value.
6067 @item tracer-max-code-growth
6068 Stop tail duplication once code growth has reached given percentage. This is
6069 rather hokey argument, as most of the duplicates will be eliminated later in
6070 cross jumping, so it may be set to much higher values than is the desired code
6073 @item tracer-min-branch-ratio
6075 Stop reverse growth when the reverse probability of best edge is less than this
6076 threshold (in percent).
6078 @item tracer-min-branch-ratio
6079 @itemx tracer-min-branch-ratio-feedback
6081 Stop forward growth if the best edge do have probability lower than this
6084 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6085 compilation for profile feedback and one for compilation without. The value
6086 for compilation with profile feedback needs to be more conservative (higher) in
6087 order to make tracer effective.
6089 @item max-cse-path-length
6091 Maximum number of basic blocks on path that cse considers. The default is 10.
6094 The maximum instructions CSE process before flushing. The default is 1000.
6096 @item global-var-threshold
6098 Counts the number of function calls (@var{n}) and the number of
6099 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
6100 single artificial variable will be created to represent all the
6101 call-clobbered variables at function call sites. This artificial
6102 variable will then be made to alias every call-clobbered variable.
6103 (done as @code{int * size_t} on the host machine; beware overflow).
6105 @item max-aliased-vops
6107 Maximum number of virtual operands allowed to represent aliases
6108 before triggering the alias grouping heuristic. Alias grouping
6109 reduces compile times and memory consumption needed for aliasing at
6110 the expense of precision loss in alias information.
6112 @item ggc-min-expand
6114 GCC uses a garbage collector to manage its own memory allocation. This
6115 parameter specifies the minimum percentage by which the garbage
6116 collector's heap should be allowed to expand between collections.
6117 Tuning this may improve compilation speed; it has no effect on code
6120 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6121 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6122 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6123 GCC is not able to calculate RAM on a particular platform, the lower
6124 bound of 30% is used. Setting this parameter and
6125 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6126 every opportunity. This is extremely slow, but can be useful for
6129 @item ggc-min-heapsize
6131 Minimum size of the garbage collector's heap before it begins bothering
6132 to collect garbage. The first collection occurs after the heap expands
6133 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6134 tuning this may improve compilation speed, and has no effect on code
6137 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6138 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6139 with a lower bound of 4096 (four megabytes) and an upper bound of
6140 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6141 particular platform, the lower bound is used. Setting this parameter
6142 very large effectively disables garbage collection. Setting this
6143 parameter and @option{ggc-min-expand} to zero causes a full collection
6144 to occur at every opportunity.
6146 @item max-reload-search-insns
6147 The maximum number of instruction reload should look backward for equivalent
6148 register. Increasing values mean more aggressive optimization, making the
6149 compile time increase with probably slightly better performance. The default
6152 @item max-cselib-memory-location
6153 The maximum number of memory locations cselib should take into acount.
6154 Increasing values mean more aggressive optimization, making the compile time
6155 increase with probably slightly better performance. The default value is 500.
6157 @item max-flow-memory-location
6158 Similar as @option{max-cselib-memory-location} but for dataflow liveness.
6159 The default value is 100.
6161 @item reorder-blocks-duplicate
6162 @itemx reorder-blocks-duplicate-feedback
6164 Used by basic block reordering pass to decide whether to use unconditional
6165 branch or duplicate the code on its destination. Code is duplicated when its
6166 estimated size is smaller than this value multiplied by the estimated size of
6167 unconditional jump in the hot spots of the program.
6169 The @option{reorder-block-duplicate-feedback} is used only when profile
6170 feedback is available and may be set to higher values than
6171 @option{reorder-block-duplicate} since information about the hot spots is more
6174 @item max-sched-region-blocks
6175 The maximum number of blocks in a region to be considered for
6176 interblock scheduling. The default value is 10.
6178 @item max-sched-region-insns
6179 The maximum number of insns in a region to be considered for
6180 interblock scheduling. The default value is 100.
6183 The minimum probability (in percents) of reaching a source block
6184 for interblock speculative scheduling. The default value is 40.
6186 @item max-sched-extend-regions-iters
6187 The maximum number of iterations through CFG to extend regions.
6188 0 - disable region extension,
6189 N - do at most N iterations.
6190 The default value is 2.
6192 @item max-sched-insn-conflict-delay
6193 The maximum conflict delay for an insn to be considered for speculative motion.
6194 The default value is 3.
6196 @item sched-spec-prob-cutoff
6197 The minimal probability of speculation success (in percents), so that
6198 speculative insn will be scheduled.
6199 The default value is 40.
6201 @item max-last-value-rtl
6203 The maximum size measured as number of RTLs that can be recorded in an expression
6204 in combiner for a pseudo register as last known value of that register. The default
6207 @item integer-share-limit
6208 Small integer constants can use a shared data structure, reducing the
6209 compiler's memory usage and increasing its speed. This sets the maximum
6210 value of a shared integer constant's. The default value is 256.
6212 @item min-virtual-mappings
6213 Specifies the minimum number of virtual mappings in the incremental
6214 SSA updater that should be registered to trigger the virtual mappings
6215 heuristic defined by virtual-mappings-ratio. The default value is
6218 @item virtual-mappings-ratio
6219 If the number of virtual mappings is virtual-mappings-ratio bigger
6220 than the number of virtual symbols to be updated, then the incremental
6221 SSA updater switches to a full update for those symbols. The default
6224 @item ssp-buffer-size
6225 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6226 protection when @option{-fstack-protection} is used.
6228 @item max-jump-thread-duplication-stmts
6229 Maximum number of statements allowed in a block that needs to be
6230 duplicated when threading jumps.
6232 @item max-fields-for-field-sensitive
6233 Maximum number of fields in a structure we will treat in
6234 a field sensitive manner during pointer analysis.
6239 @node Preprocessor Options
6240 @section Options Controlling the Preprocessor
6241 @cindex preprocessor options
6242 @cindex options, preprocessor
6244 These options control the C preprocessor, which is run on each C source
6245 file before actual compilation.
6247 If you use the @option{-E} option, nothing is done except preprocessing.
6248 Some of these options make sense only together with @option{-E} because
6249 they cause the preprocessor output to be unsuitable for actual
6254 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6255 and pass @var{option} directly through to the preprocessor. If
6256 @var{option} contains commas, it is split into multiple options at the
6257 commas. However, many options are modified, translated or interpreted
6258 by the compiler driver before being passed to the preprocessor, and
6259 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6260 interface is undocumented and subject to change, so whenever possible
6261 you should avoid using @option{-Wp} and let the driver handle the
6264 @item -Xpreprocessor @var{option}
6265 @opindex preprocessor
6266 Pass @var{option} as an option to the preprocessor. You can use this to
6267 supply system-specific preprocessor options which GCC does not know how to
6270 If you want to pass an option that takes an argument, you must use
6271 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6274 @include cppopts.texi
6276 @node Assembler Options
6277 @section Passing Options to the Assembler
6279 @c prevent bad page break with this line
6280 You can pass options to the assembler.
6283 @item -Wa,@var{option}
6285 Pass @var{option} as an option to the assembler. If @var{option}
6286 contains commas, it is split into multiple options at the commas.
6288 @item -Xassembler @var{option}
6290 Pass @var{option} as an option to the assembler. You can use this to
6291 supply system-specific assembler options which GCC does not know how to
6294 If you want to pass an option that takes an argument, you must use
6295 @option{-Xassembler} twice, once for the option and once for the argument.
6300 @section Options for Linking
6301 @cindex link options
6302 @cindex options, linking
6304 These options come into play when the compiler links object files into
6305 an executable output file. They are meaningless if the compiler is
6306 not doing a link step.
6310 @item @var{object-file-name}
6311 A file name that does not end in a special recognized suffix is
6312 considered to name an object file or library. (Object files are
6313 distinguished from libraries by the linker according to the file
6314 contents.) If linking is done, these object files are used as input
6323 If any of these options is used, then the linker is not run, and
6324 object file names should not be used as arguments. @xref{Overall
6328 @item -l@var{library}
6329 @itemx -l @var{library}
6331 Search the library named @var{library} when linking. (The second
6332 alternative with the library as a separate argument is only for
6333 POSIX compliance and is not recommended.)
6335 It makes a difference where in the command you write this option; the
6336 linker searches and processes libraries and object files in the order they
6337 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6338 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6339 to functions in @samp{z}, those functions may not be loaded.
6341 The linker searches a standard list of directories for the library,
6342 which is actually a file named @file{lib@var{library}.a}. The linker
6343 then uses this file as if it had been specified precisely by name.
6345 The directories searched include several standard system directories
6346 plus any that you specify with @option{-L}.
6348 Normally the files found this way are library files---archive files
6349 whose members are object files. The linker handles an archive file by
6350 scanning through it for members which define symbols that have so far
6351 been referenced but not defined. But if the file that is found is an
6352 ordinary object file, it is linked in the usual fashion. The only
6353 difference between using an @option{-l} option and specifying a file name
6354 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6355 and searches several directories.
6359 You need this special case of the @option{-l} option in order to
6360 link an Objective-C or Objective-C++ program.
6363 @opindex nostartfiles
6364 Do not use the standard system startup files when linking.
6365 The standard system libraries are used normally, unless @option{-nostdlib}
6366 or @option{-nodefaultlibs} is used.
6368 @item -nodefaultlibs
6369 @opindex nodefaultlibs
6370 Do not use the standard system libraries when linking.
6371 Only the libraries you specify will be passed to the linker.
6372 The standard startup files are used normally, unless @option{-nostartfiles}
6373 is used. The compiler may generate calls to @code{memcmp},
6374 @code{memset}, @code{memcpy} and @code{memmove}.
6375 These entries are usually resolved by entries in
6376 libc. These entry points should be supplied through some other
6377 mechanism when this option is specified.
6381 Do not use the standard system startup files or libraries when linking.
6382 No startup files and only the libraries you specify will be passed to
6383 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6384 @code{memcpy} and @code{memmove}.
6385 These entries are usually resolved by entries in
6386 libc. These entry points should be supplied through some other
6387 mechanism when this option is specified.
6389 @cindex @option{-lgcc}, use with @option{-nostdlib}
6390 @cindex @option{-nostdlib} and unresolved references
6391 @cindex unresolved references and @option{-nostdlib}
6392 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6393 @cindex @option{-nodefaultlibs} and unresolved references
6394 @cindex unresolved references and @option{-nodefaultlibs}
6395 One of the standard libraries bypassed by @option{-nostdlib} and
6396 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6397 that GCC uses to overcome shortcomings of particular machines, or special
6398 needs for some languages.
6399 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6400 Collection (GCC) Internals},
6401 for more discussion of @file{libgcc.a}.)
6402 In most cases, you need @file{libgcc.a} even when you want to avoid
6403 other standard libraries. In other words, when you specify @option{-nostdlib}
6404 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6405 This ensures that you have no unresolved references to internal GCC
6406 library subroutines. (For example, @samp{__main}, used to ensure C++
6407 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6408 GNU Compiler Collection (GCC) Internals}.)
6412 Produce a position independent executable on targets which support it.
6413 For predictable results, you must also specify the same set of options
6414 that were used to generate code (@option{-fpie}, @option{-fPIE},
6415 or model suboptions) when you specify this option.
6419 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6420 that support it. This instructs the linker to add all symbols, not
6421 only used ones, to the dynamic symbol table. This option is needed
6422 for some uses of @code{dlopen} or to allow obtaining backtraces
6423 from within a program.
6427 Remove all symbol table and relocation information from the executable.
6431 On systems that support dynamic linking, this prevents linking with the shared
6432 libraries. On other systems, this option has no effect.
6436 Produce a shared object which can then be linked with other objects to
6437 form an executable. Not all systems support this option. For predictable
6438 results, you must also specify the same set of options that were used to
6439 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6440 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6441 needs to build supplementary stub code for constructors to work. On
6442 multi-libbed systems, @samp{gcc -shared} must select the correct support
6443 libraries to link against. Failing to supply the correct flags may lead
6444 to subtle defects. Supplying them in cases where they are not necessary
6447 @item -shared-libgcc
6448 @itemx -static-libgcc
6449 @opindex shared-libgcc
6450 @opindex static-libgcc
6451 On systems that provide @file{libgcc} as a shared library, these options
6452 force the use of either the shared or static version respectively.
6453 If no shared version of @file{libgcc} was built when the compiler was
6454 configured, these options have no effect.
6456 There are several situations in which an application should use the
6457 shared @file{libgcc} instead of the static version. The most common
6458 of these is when the application wishes to throw and catch exceptions
6459 across different shared libraries. In that case, each of the libraries
6460 as well as the application itself should use the shared @file{libgcc}.
6462 Therefore, the G++ and GCJ drivers automatically add
6463 @option{-shared-libgcc} whenever you build a shared library or a main
6464 executable, because C++ and Java programs typically use exceptions, so
6465 this is the right thing to do.
6467 If, instead, you use the GCC driver to create shared libraries, you may
6468 find that they will not always be linked with the shared @file{libgcc}.
6469 If GCC finds, at its configuration time, that you have a non-GNU linker
6470 or a GNU linker that does not support option @option{--eh-frame-hdr},
6471 it will link the shared version of @file{libgcc} into shared libraries
6472 by default. Otherwise, it will take advantage of the linker and optimize
6473 away the linking with the shared version of @file{libgcc}, linking with
6474 the static version of libgcc by default. This allows exceptions to
6475 propagate through such shared libraries, without incurring relocation
6476 costs at library load time.
6478 However, if a library or main executable is supposed to throw or catch
6479 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6480 for the languages used in the program, or using the option
6481 @option{-shared-libgcc}, such that it is linked with the shared
6486 Bind references to global symbols when building a shared object. Warn
6487 about any unresolved references (unless overridden by the link editor
6488 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6491 @item -Xlinker @var{option}
6493 Pass @var{option} as an option to the linker. You can use this to
6494 supply system-specific linker options which GCC does not know how to
6497 If you want to pass an option that takes an argument, you must use
6498 @option{-Xlinker} twice, once for the option and once for the argument.
6499 For example, to pass @option{-assert definitions}, you must write
6500 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6501 @option{-Xlinker "-assert definitions"}, because this passes the entire
6502 string as a single argument, which is not what the linker expects.
6504 @item -Wl,@var{option}
6506 Pass @var{option} as an option to the linker. If @var{option} contains
6507 commas, it is split into multiple options at the commas.
6509 @item -u @var{symbol}
6511 Pretend the symbol @var{symbol} is undefined, to force linking of
6512 library modules to define it. You can use @option{-u} multiple times with
6513 different symbols to force loading of additional library modules.
6516 @node Directory Options
6517 @section Options for Directory Search
6518 @cindex directory options
6519 @cindex options, directory search
6522 These options specify directories to search for header files, for
6523 libraries and for parts of the compiler:
6528 Add the directory @var{dir} to the head of the list of directories to be
6529 searched for header files. This can be used to override a system header
6530 file, substituting your own version, since these directories are
6531 searched before the system header file directories. However, you should
6532 not use this option to add directories that contain vendor-supplied
6533 system header files (use @option{-isystem} for that). If you use more than
6534 one @option{-I} option, the directories are scanned in left-to-right
6535 order; the standard system directories come after.
6537 If a standard system include directory, or a directory specified with
6538 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6539 option will be ignored. The directory will still be searched but as a
6540 system directory at its normal position in the system include chain.
6541 This is to ensure that GCC's procedure to fix buggy system headers and
6542 the ordering for the include_next directive are not inadvertently changed.
6543 If you really need to change the search order for system directories,
6544 use the @option{-nostdinc} and/or @option{-isystem} options.
6546 @item -iquote@var{dir}
6548 Add the directory @var{dir} to the head of the list of directories to
6549 be searched for header files only for the case of @samp{#include
6550 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6551 otherwise just like @option{-I}.
6555 Add directory @var{dir} to the list of directories to be searched
6558 @item -B@var{prefix}
6560 This option specifies where to find the executables, libraries,
6561 include files, and data files of the compiler itself.
6563 The compiler driver program runs one or more of the subprograms
6564 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6565 @var{prefix} as a prefix for each program it tries to run, both with and
6566 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6568 For each subprogram to be run, the compiler driver first tries the
6569 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6570 was not specified, the driver tries two standard prefixes, which are
6571 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6572 those results in a file name that is found, the unmodified program
6573 name is searched for using the directories specified in your
6574 @env{PATH} environment variable.
6576 The compiler will check to see if the path provided by the @option{-B}
6577 refers to a directory, and if necessary it will add a directory
6578 separator character at the end of the path.
6580 @option{-B} prefixes that effectively specify directory names also apply
6581 to libraries in the linker, because the compiler translates these
6582 options into @option{-L} options for the linker. They also apply to
6583 includes files in the preprocessor, because the compiler translates these
6584 options into @option{-isystem} options for the preprocessor. In this case,
6585 the compiler appends @samp{include} to the prefix.
6587 The run-time support file @file{libgcc.a} can also be searched for using
6588 the @option{-B} prefix, if needed. If it is not found there, the two
6589 standard prefixes above are tried, and that is all. The file is left
6590 out of the link if it is not found by those means.
6592 Another way to specify a prefix much like the @option{-B} prefix is to use
6593 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6596 As a special kludge, if the path provided by @option{-B} is
6597 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6598 9, then it will be replaced by @file{[dir/]include}. This is to help
6599 with boot-strapping the compiler.
6601 @item -specs=@var{file}
6603 Process @var{file} after the compiler reads in the standard @file{specs}
6604 file, in order to override the defaults that the @file{gcc} driver
6605 program uses when determining what switches to pass to @file{cc1},
6606 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6607 @option{-specs=@var{file}} can be specified on the command line, and they
6608 are processed in order, from left to right.
6610 @item --sysroot=@var{dir}
6612 Use @var{dir} as the logical root directory for headers and libraries.
6613 For example, if the compiler would normally search for headers in
6614 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6615 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6617 If you use both this option and the @option{-isysroot} option, then
6618 the @option{--sysroot} option will apply to libraries, but the
6619 @option{-isysroot} option will apply to header files.
6621 The GNU linker (beginning with version 2.16) has the necessary support
6622 for this option. If your linker does not support this option, the
6623 header file aspect of @option{--sysroot} will still work, but the
6624 library aspect will not.
6628 This option has been deprecated. Please use @option{-iquote} instead for
6629 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6630 Any directories you specify with @option{-I} options before the @option{-I-}
6631 option are searched only for the case of @samp{#include "@var{file}"};
6632 they are not searched for @samp{#include <@var{file}>}.
6634 If additional directories are specified with @option{-I} options after
6635 the @option{-I-}, these directories are searched for all @samp{#include}
6636 directives. (Ordinarily @emph{all} @option{-I} directories are used
6639 In addition, the @option{-I-} option inhibits the use of the current
6640 directory (where the current input file came from) as the first search
6641 directory for @samp{#include "@var{file}"}. There is no way to
6642 override this effect of @option{-I-}. With @option{-I.} you can specify
6643 searching the directory which was current when the compiler was
6644 invoked. That is not exactly the same as what the preprocessor does
6645 by default, but it is often satisfactory.
6647 @option{-I-} does not inhibit the use of the standard system directories
6648 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6655 @section Specifying subprocesses and the switches to pass to them
6658 @command{gcc} is a driver program. It performs its job by invoking a
6659 sequence of other programs to do the work of compiling, assembling and
6660 linking. GCC interprets its command-line parameters and uses these to
6661 deduce which programs it should invoke, and which command-line options
6662 it ought to place on their command lines. This behavior is controlled
6663 by @dfn{spec strings}. In most cases there is one spec string for each
6664 program that GCC can invoke, but a few programs have multiple spec
6665 strings to control their behavior. The spec strings built into GCC can
6666 be overridden by using the @option{-specs=} command-line switch to specify
6669 @dfn{Spec files} are plaintext files that are used to construct spec
6670 strings. They consist of a sequence of directives separated by blank
6671 lines. The type of directive is determined by the first non-whitespace
6672 character on the line and it can be one of the following:
6675 @item %@var{command}
6676 Issues a @var{command} to the spec file processor. The commands that can
6680 @item %include <@var{file}>
6682 Search for @var{file} and insert its text at the current point in the
6685 @item %include_noerr <@var{file}>
6686 @cindex %include_noerr
6687 Just like @samp{%include}, but do not generate an error message if the include
6688 file cannot be found.
6690 @item %rename @var{old_name} @var{new_name}
6692 Rename the spec string @var{old_name} to @var{new_name}.
6696 @item *[@var{spec_name}]:
6697 This tells the compiler to create, override or delete the named spec
6698 string. All lines after this directive up to the next directive or
6699 blank line are considered to be the text for the spec string. If this
6700 results in an empty string then the spec will be deleted. (Or, if the
6701 spec did not exist, then nothing will happened.) Otherwise, if the spec
6702 does not currently exist a new spec will be created. If the spec does
6703 exist then its contents will be overridden by the text of this
6704 directive, unless the first character of that text is the @samp{+}
6705 character, in which case the text will be appended to the spec.
6707 @item [@var{suffix}]:
6708 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6709 and up to the next directive or blank line are considered to make up the
6710 spec string for the indicated suffix. When the compiler encounters an
6711 input file with the named suffix, it will processes the spec string in
6712 order to work out how to compile that file. For example:
6719 This says that any input file whose name ends in @samp{.ZZ} should be
6720 passed to the program @samp{z-compile}, which should be invoked with the
6721 command-line switch @option{-input} and with the result of performing the
6722 @samp{%i} substitution. (See below.)
6724 As an alternative to providing a spec string, the text that follows a
6725 suffix directive can be one of the following:
6728 @item @@@var{language}
6729 This says that the suffix is an alias for a known @var{language}. This is
6730 similar to using the @option{-x} command-line switch to GCC to specify a
6731 language explicitly. For example:
6738 Says that .ZZ files are, in fact, C++ source files.
6741 This causes an error messages saying:
6744 @var{name} compiler not installed on this system.
6748 GCC already has an extensive list of suffixes built into it.
6749 This directive will add an entry to the end of the list of suffixes, but
6750 since the list is searched from the end backwards, it is effectively
6751 possible to override earlier entries using this technique.
6755 GCC has the following spec strings built into it. Spec files can
6756 override these strings or create their own. Note that individual
6757 targets can also add their own spec strings to this list.
6760 asm Options to pass to the assembler
6761 asm_final Options to pass to the assembler post-processor
6762 cpp Options to pass to the C preprocessor
6763 cc1 Options to pass to the C compiler
6764 cc1plus Options to pass to the C++ compiler
6765 endfile Object files to include at the end of the link
6766 link Options to pass to the linker
6767 lib Libraries to include on the command line to the linker
6768 libgcc Decides which GCC support library to pass to the linker
6769 linker Sets the name of the linker
6770 predefines Defines to be passed to the C preprocessor
6771 signed_char Defines to pass to CPP to say whether @code{char} is signed
6773 startfile Object files to include at the start of the link
6776 Here is a small example of a spec file:
6782 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6785 This example renames the spec called @samp{lib} to @samp{old_lib} and
6786 then overrides the previous definition of @samp{lib} with a new one.
6787 The new definition adds in some extra command-line options before
6788 including the text of the old definition.
6790 @dfn{Spec strings} are a list of command-line options to be passed to their
6791 corresponding program. In addition, the spec strings can contain
6792 @samp{%}-prefixed sequences to substitute variable text or to
6793 conditionally insert text into the command line. Using these constructs
6794 it is possible to generate quite complex command lines.
6796 Here is a table of all defined @samp{%}-sequences for spec
6797 strings. Note that spaces are not generated automatically around the
6798 results of expanding these sequences. Therefore you can concatenate them
6799 together or combine them with constant text in a single argument.
6803 Substitute one @samp{%} into the program name or argument.
6806 Substitute the name of the input file being processed.
6809 Substitute the basename of the input file being processed.
6810 This is the substring up to (and not including) the last period
6811 and not including the directory.
6814 This is the same as @samp{%b}, but include the file suffix (text after
6818 Marks the argument containing or following the @samp{%d} as a
6819 temporary file name, so that that file will be deleted if GCC exits
6820 successfully. Unlike @samp{%g}, this contributes no text to the
6823 @item %g@var{suffix}
6824 Substitute a file name that has suffix @var{suffix} and is chosen
6825 once per compilation, and mark the argument in the same way as
6826 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6827 name is now chosen in a way that is hard to predict even when previously
6828 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6829 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6830 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6831 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6832 was simply substituted with a file name chosen once per compilation,
6833 without regard to any appended suffix (which was therefore treated
6834 just like ordinary text), making such attacks more likely to succeed.
6836 @item %u@var{suffix}
6837 Like @samp{%g}, but generates a new temporary file name even if
6838 @samp{%u@var{suffix}} was already seen.
6840 @item %U@var{suffix}
6841 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6842 new one if there is no such last file name. In the absence of any
6843 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6844 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6845 would involve the generation of two distinct file names, one
6846 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6847 simply substituted with a file name chosen for the previous @samp{%u},
6848 without regard to any appended suffix.
6850 @item %j@var{suffix}
6851 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6852 writable, and if save-temps is off; otherwise, substitute the name
6853 of a temporary file, just like @samp{%u}. This temporary file is not
6854 meant for communication between processes, but rather as a junk
6857 @item %|@var{suffix}
6858 @itemx %m@var{suffix}
6859 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6860 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6861 all. These are the two most common ways to instruct a program that it
6862 should read from standard input or write to standard output. If you
6863 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6864 construct: see for example @file{f/lang-specs.h}.
6866 @item %.@var{SUFFIX}
6867 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6868 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6869 terminated by the next space or %.
6872 Marks the argument containing or following the @samp{%w} as the
6873 designated output file of this compilation. This puts the argument
6874 into the sequence of arguments that @samp{%o} will substitute later.
6877 Substitutes the names of all the output files, with spaces
6878 automatically placed around them. You should write spaces
6879 around the @samp{%o} as well or the results are undefined.
6880 @samp{%o} is for use in the specs for running the linker.
6881 Input files whose names have no recognized suffix are not compiled
6882 at all, but they are included among the output files, so they will
6886 Substitutes the suffix for object files. Note that this is
6887 handled specially when it immediately follows @samp{%g, %u, or %U},
6888 because of the need for those to form complete file names. The
6889 handling is such that @samp{%O} is treated exactly as if it had already
6890 been substituted, except that @samp{%g, %u, and %U} do not currently
6891 support additional @var{suffix} characters following @samp{%O} as they would
6892 following, for example, @samp{.o}.
6895 Substitutes the standard macro predefinitions for the
6896 current target machine. Use this when running @code{cpp}.
6899 Like @samp{%p}, but puts @samp{__} before and after the name of each
6900 predefined macro, except for macros that start with @samp{__} or with
6901 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6905 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6906 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
6907 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6908 and @option{-imultilib} as necessary.
6911 Current argument is the name of a library or startup file of some sort.
6912 Search for that file in a standard list of directories and substitute
6913 the full name found.
6916 Print @var{str} as an error message. @var{str} is terminated by a newline.
6917 Use this when inconsistent options are detected.
6920 Substitute the contents of spec string @var{name} at this point.
6923 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6925 @item %x@{@var{option}@}
6926 Accumulate an option for @samp{%X}.
6929 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6933 Output the accumulated assembler options specified by @option{-Wa}.
6936 Output the accumulated preprocessor options specified by @option{-Wp}.
6939 Process the @code{asm} spec. This is used to compute the
6940 switches to be passed to the assembler.
6943 Process the @code{asm_final} spec. This is a spec string for
6944 passing switches to an assembler post-processor, if such a program is
6948 Process the @code{link} spec. This is the spec for computing the
6949 command line passed to the linker. Typically it will make use of the
6950 @samp{%L %G %S %D and %E} sequences.
6953 Dump out a @option{-L} option for each directory that GCC believes might
6954 contain startup files. If the target supports multilibs then the
6955 current multilib directory will be prepended to each of these paths.
6958 Process the @code{lib} spec. This is a spec string for deciding which
6959 libraries should be included on the command line to the linker.
6962 Process the @code{libgcc} spec. This is a spec string for deciding
6963 which GCC support library should be included on the command line to the linker.
6966 Process the @code{startfile} spec. This is a spec for deciding which
6967 object files should be the first ones passed to the linker. Typically
6968 this might be a file named @file{crt0.o}.
6971 Process the @code{endfile} spec. This is a spec string that specifies
6972 the last object files that will be passed to the linker.
6975 Process the @code{cpp} spec. This is used to construct the arguments
6976 to be passed to the C preprocessor.
6979 Process the @code{cc1} spec. This is used to construct the options to be
6980 passed to the actual C compiler (@samp{cc1}).
6983 Process the @code{cc1plus} spec. This is used to construct the options to be
6984 passed to the actual C++ compiler (@samp{cc1plus}).
6987 Substitute the variable part of a matched option. See below.
6988 Note that each comma in the substituted string is replaced by
6992 Remove all occurrences of @code{-S} from the command line. Note---this
6993 command is position dependent. @samp{%} commands in the spec string
6994 before this one will see @code{-S}, @samp{%} commands in the spec string
6995 after this one will not.
6997 @item %:@var{function}(@var{args})
6998 Call the named function @var{function}, passing it @var{args}.
6999 @var{args} is first processed as a nested spec string, then split
7000 into an argument vector in the usual fashion. The function returns
7001 a string which is processed as if it had appeared literally as part
7002 of the current spec.
7004 The following built-in spec functions are provided:
7007 @item @code{if-exists}
7008 The @code{if-exists} spec function takes one argument, an absolute
7009 pathname to a file. If the file exists, @code{if-exists} returns the
7010 pathname. Here is a small example of its usage:
7014 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7017 @item @code{if-exists-else}
7018 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7019 spec function, except that it takes two arguments. The first argument is
7020 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7021 returns the pathname. If it does not exist, it returns the second argument.
7022 This way, @code{if-exists-else} can be used to select one file or another,
7023 based on the existence of the first. Here is a small example of its usage:
7027 crt0%O%s %:if-exists(crti%O%s) \
7028 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7031 @item @code{replace-outfile}
7032 The @code{replace-outfile} spec function takes two arguments. It looks for the
7033 first argument in the outfiles array and replaces it with the second argument. Here
7034 is a small example of its usage:
7037 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7043 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7044 If that switch was not specified, this substitutes nothing. Note that
7045 the leading dash is omitted when specifying this option, and it is
7046 automatically inserted if the substitution is performed. Thus the spec
7047 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7048 and would output the command line option @option{-foo}.
7050 @item %W@{@code{S}@}
7051 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7054 @item %@{@code{S}*@}
7055 Substitutes all the switches specified to GCC whose names start
7056 with @code{-S}, but which also take an argument. This is used for
7057 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7058 GCC considers @option{-o foo} as being
7059 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7060 text, including the space. Thus two arguments would be generated.
7062 @item %@{@code{S}*&@code{T}*@}
7063 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7064 (the order of @code{S} and @code{T} in the spec is not significant).
7065 There can be any number of ampersand-separated variables; for each the
7066 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7068 @item %@{@code{S}:@code{X}@}
7069 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7071 @item %@{!@code{S}:@code{X}@}
7072 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7074 @item %@{@code{S}*:@code{X}@}
7075 Substitutes @code{X} if one or more switches whose names start with
7076 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7077 once, no matter how many such switches appeared. However, if @code{%*}
7078 appears somewhere in @code{X}, then @code{X} will be substituted once
7079 for each matching switch, with the @code{%*} replaced by the part of
7080 that switch that matched the @code{*}.
7082 @item %@{.@code{S}:@code{X}@}
7083 Substitutes @code{X}, if processing a file with suffix @code{S}.
7085 @item %@{!.@code{S}:@code{X}@}
7086 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7088 @item %@{@code{S}|@code{P}:@code{X}@}
7089 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7090 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7091 although they have a stronger binding than the @samp{|}. If @code{%*}
7092 appears in @code{X}, all of the alternatives must be starred, and only
7093 the first matching alternative is substituted.
7095 For example, a spec string like this:
7098 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7101 will output the following command-line options from the following input
7102 command-line options:
7107 -d fred.c -foo -baz -boggle
7108 -d jim.d -bar -baz -boggle
7111 @item %@{S:X; T:Y; :D@}
7113 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7114 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7115 be as many clauses as you need. This may be combined with @code{.},
7116 @code{!}, @code{|}, and @code{*} as needed.
7121 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7122 construct may contain other nested @samp{%} constructs or spaces, or
7123 even newlines. They are processed as usual, as described above.
7124 Trailing white space in @code{X} is ignored. White space may also
7125 appear anywhere on the left side of the colon in these constructs,
7126 except between @code{.} or @code{*} and the corresponding word.
7128 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7129 handled specifically in these constructs. If another value of
7130 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7131 @option{-W} switch is found later in the command line, the earlier
7132 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7133 just one letter, which passes all matching options.
7135 The character @samp{|} at the beginning of the predicate text is used to
7136 indicate that a command should be piped to the following command, but
7137 only if @option{-pipe} is specified.
7139 It is built into GCC which switches take arguments and which do not.
7140 (You might think it would be useful to generalize this to allow each
7141 compiler's spec to say which switches take arguments. But this cannot
7142 be done in a consistent fashion. GCC cannot even decide which input
7143 files have been specified without knowing which switches take arguments,
7144 and it must know which input files to compile in order to tell which
7147 GCC also knows implicitly that arguments starting in @option{-l} are to be
7148 treated as compiler output files, and passed to the linker in their
7149 proper position among the other output files.
7151 @c man begin OPTIONS
7153 @node Target Options
7154 @section Specifying Target Machine and Compiler Version
7155 @cindex target options
7156 @cindex cross compiling
7157 @cindex specifying machine version
7158 @cindex specifying compiler version and target machine
7159 @cindex compiler version, specifying
7160 @cindex target machine, specifying
7162 The usual way to run GCC is to run the executable called @file{gcc}, or
7163 @file{<machine>-gcc} when cross-compiling, or
7164 @file{<machine>-gcc-<version>} to run a version other than the one that
7165 was installed last. Sometimes this is inconvenient, so GCC provides
7166 options that will switch to another cross-compiler or version.
7169 @item -b @var{machine}
7171 The argument @var{machine} specifies the target machine for compilation.
7173 The value to use for @var{machine} is the same as was specified as the
7174 machine type when configuring GCC as a cross-compiler. For
7175 example, if a cross-compiler was configured with @samp{configure
7176 arm-elf}, meaning to compile for an arm processor with elf binaries,
7177 then you would specify @option{-b arm-elf} to run that cross compiler.
7178 Because there are other options beginning with @option{-b}, the
7179 configuration must contain a hyphen.
7181 @item -V @var{version}
7183 The argument @var{version} specifies which version of GCC to run.
7184 This is useful when multiple versions are installed. For example,
7185 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7188 The @option{-V} and @option{-b} options work by running the
7189 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7190 use them if you can just run that directly.
7192 @node Submodel Options
7193 @section Hardware Models and Configurations
7194 @cindex submodel options
7195 @cindex specifying hardware config
7196 @cindex hardware models and configurations, specifying
7197 @cindex machine dependent options
7199 Earlier we discussed the standard option @option{-b} which chooses among
7200 different installed compilers for completely different target
7201 machines, such as VAX vs.@: 68000 vs.@: 80386.
7203 In addition, each of these target machine types can have its own
7204 special options, starting with @samp{-m}, to choose among various
7205 hardware models or configurations---for example, 68010 vs 68020,
7206 floating coprocessor or none. A single installed version of the
7207 compiler can compile for any model or configuration, according to the
7210 Some configurations of the compiler also support additional special
7211 options, usually for compatibility with other compilers on the same
7214 @c This list is ordered alphanumerically by subsection name.
7215 @c It should be the same order and spelling as these options are listed
7216 @c in Machine Dependent Options
7222 * Blackfin Options::
7226 * DEC Alpha Options::
7227 * DEC Alpha/VMS Options::
7229 * GNU/Linux Options::
7232 * i386 and x86-64 Options::
7245 * RS/6000 and PowerPC Options::
7246 * S/390 and zSeries Options::
7249 * System V Options::
7250 * TMS320C3x/C4x Options::
7254 * Xstormy16 Options::
7260 @subsection ARC Options
7263 These options are defined for ARC implementations:
7268 Compile code for little endian mode. This is the default.
7272 Compile code for big endian mode.
7275 @opindex mmangle-cpu
7276 Prepend the name of the cpu to all public symbol names.
7277 In multiple-processor systems, there are many ARC variants with different
7278 instruction and register set characteristics. This flag prevents code
7279 compiled for one cpu to be linked with code compiled for another.
7280 No facility exists for handling variants that are ``almost identical''.
7281 This is an all or nothing option.
7283 @item -mcpu=@var{cpu}
7285 Compile code for ARC variant @var{cpu}.
7286 Which variants are supported depend on the configuration.
7287 All variants support @option{-mcpu=base}, this is the default.
7289 @item -mtext=@var{text-section}
7290 @itemx -mdata=@var{data-section}
7291 @itemx -mrodata=@var{readonly-data-section}
7295 Put functions, data, and readonly data in @var{text-section},
7296 @var{data-section}, and @var{readonly-data-section} respectively
7297 by default. This can be overridden with the @code{section} attribute.
7298 @xref{Variable Attributes}.
7303 @subsection ARM Options
7306 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7310 @item -mabi=@var{name}
7312 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7313 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7316 @opindex mapcs-frame
7317 Generate a stack frame that is compliant with the ARM Procedure Call
7318 Standard for all functions, even if this is not strictly necessary for
7319 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7320 with this option will cause the stack frames not to be generated for
7321 leaf functions. The default is @option{-mno-apcs-frame}.
7325 This is a synonym for @option{-mapcs-frame}.
7328 @c not currently implemented
7329 @item -mapcs-stack-check
7330 @opindex mapcs-stack-check
7331 Generate code to check the amount of stack space available upon entry to
7332 every function (that actually uses some stack space). If there is
7333 insufficient space available then either the function
7334 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7335 called, depending upon the amount of stack space required. The run time
7336 system is required to provide these functions. The default is
7337 @option{-mno-apcs-stack-check}, since this produces smaller code.
7339 @c not currently implemented
7341 @opindex mapcs-float
7342 Pass floating point arguments using the float point registers. This is
7343 one of the variants of the APCS@. This option is recommended if the
7344 target hardware has a floating point unit or if a lot of floating point
7345 arithmetic is going to be performed by the code. The default is
7346 @option{-mno-apcs-float}, since integer only code is slightly increased in
7347 size if @option{-mapcs-float} is used.
7349 @c not currently implemented
7350 @item -mapcs-reentrant
7351 @opindex mapcs-reentrant
7352 Generate reentrant, position independent code. The default is
7353 @option{-mno-apcs-reentrant}.
7356 @item -mthumb-interwork
7357 @opindex mthumb-interwork
7358 Generate code which supports calling between the ARM and Thumb
7359 instruction sets. Without this option the two instruction sets cannot
7360 be reliably used inside one program. The default is
7361 @option{-mno-thumb-interwork}, since slightly larger code is generated
7362 when @option{-mthumb-interwork} is specified.
7364 @item -mno-sched-prolog
7365 @opindex mno-sched-prolog
7366 Prevent the reordering of instructions in the function prolog, or the
7367 merging of those instruction with the instructions in the function's
7368 body. This means that all functions will start with a recognizable set
7369 of instructions (or in fact one of a choice from a small set of
7370 different function prologues), and this information can be used to
7371 locate the start if functions inside an executable piece of code. The
7372 default is @option{-msched-prolog}.
7375 @opindex mhard-float
7376 Generate output containing floating point instructions. This is the
7380 @opindex msoft-float
7381 Generate output containing library calls for floating point.
7382 @strong{Warning:} the requisite libraries are not available for all ARM
7383 targets. Normally the facilities of the machine's usual C compiler are
7384 used, but this cannot be done directly in cross-compilation. You must make
7385 your own arrangements to provide suitable library functions for
7388 @option{-msoft-float} changes the calling convention in the output file;
7389 therefore, it is only useful if you compile @emph{all} of a program with
7390 this option. In particular, you need to compile @file{libgcc.a}, the
7391 library that comes with GCC, with @option{-msoft-float} in order for
7394 @item -mfloat-abi=@var{name}
7396 Specifies which ABI to use for floating point values. Permissible values
7397 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7399 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7400 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7401 of floating point instructions, but still uses the soft-float calling
7404 @item -mlittle-endian
7405 @opindex mlittle-endian
7406 Generate code for a processor running in little-endian mode. This is
7407 the default for all standard configurations.
7410 @opindex mbig-endian
7411 Generate code for a processor running in big-endian mode; the default is
7412 to compile code for a little-endian processor.
7414 @item -mwords-little-endian
7415 @opindex mwords-little-endian
7416 This option only applies when generating code for big-endian processors.
7417 Generate code for a little-endian word order but a big-endian byte
7418 order. That is, a byte order of the form @samp{32107654}. Note: this
7419 option should only be used if you require compatibility with code for
7420 big-endian ARM processors generated by versions of the compiler prior to
7423 @item -mcpu=@var{name}
7425 This specifies the name of the target ARM processor. GCC uses this name
7426 to determine what kind of instructions it can emit when generating
7427 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7428 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7429 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7430 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7431 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7432 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7433 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7434 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7435 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7436 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7437 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7438 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7439 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7440 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7443 @itemx -mtune=@var{name}
7445 This option is very similar to the @option{-mcpu=} option, except that
7446 instead of specifying the actual target processor type, and hence
7447 restricting which instructions can be used, it specifies that GCC should
7448 tune the performance of the code as if the target were of the type
7449 specified in this option, but still choosing the instructions that it
7450 will generate based on the cpu specified by a @option{-mcpu=} option.
7451 For some ARM implementations better performance can be obtained by using
7454 @item -march=@var{name}
7456 This specifies the name of the target ARM architecture. GCC uses this
7457 name to determine what kind of instructions it can emit when generating
7458 assembly code. This option can be used in conjunction with or instead
7459 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7460 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7461 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7462 @samp{iwmmxt}, @samp{ep9312}.
7464 @item -mfpu=@var{name}
7465 @itemx -mfpe=@var{number}
7466 @itemx -mfp=@var{number}
7470 This specifies what floating point hardware (or hardware emulation) is
7471 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7472 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7473 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7474 with older versions of GCC@.
7476 If @option{-msoft-float} is specified this specifies the format of
7477 floating point values.
7479 @item -mstructure-size-boundary=@var{n}
7480 @opindex mstructure-size-boundary
7481 The size of all structures and unions will be rounded up to a multiple
7482 of the number of bits set by this option. Permissible values are 8, 32
7483 and 64. The default value varies for different toolchains. For the COFF
7484 targeted toolchain the default value is 8. A value of 64 is only allowed
7485 if the underlying ABI supports it.
7487 Specifying the larger number can produce faster, more efficient code, but
7488 can also increase the size of the program. Different values are potentially
7489 incompatible. Code compiled with one value cannot necessarily expect to
7490 work with code or libraries compiled with another value, if they exchange
7491 information using structures or unions.
7493 @item -mabort-on-noreturn
7494 @opindex mabort-on-noreturn
7495 Generate a call to the function @code{abort} at the end of a
7496 @code{noreturn} function. It will be executed if the function tries to
7500 @itemx -mno-long-calls
7501 @opindex mlong-calls
7502 @opindex mno-long-calls
7503 Tells the compiler to perform function calls by first loading the
7504 address of the function into a register and then performing a subroutine
7505 call on this register. This switch is needed if the target function
7506 will lie outside of the 64 megabyte addressing range of the offset based
7507 version of subroutine call instruction.
7509 Even if this switch is enabled, not all function calls will be turned
7510 into long calls. The heuristic is that static functions, functions
7511 which have the @samp{short-call} attribute, functions that are inside
7512 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7513 definitions have already been compiled within the current compilation
7514 unit, will not be turned into long calls. The exception to this rule is
7515 that weak function definitions, functions with the @samp{long-call}
7516 attribute or the @samp{section} attribute, and functions that are within
7517 the scope of a @samp{#pragma long_calls} directive, will always be
7518 turned into long calls.
7520 This feature is not enabled by default. Specifying
7521 @option{-mno-long-calls} will restore the default behavior, as will
7522 placing the function calls within the scope of a @samp{#pragma
7523 long_calls_off} directive. Note these switches have no effect on how
7524 the compiler generates code to handle function calls via function
7527 @item -mnop-fun-dllimport
7528 @opindex mnop-fun-dllimport
7529 Disable support for the @code{dllimport} attribute.
7531 @item -msingle-pic-base
7532 @opindex msingle-pic-base
7533 Treat the register used for PIC addressing as read-only, rather than
7534 loading it in the prologue for each function. The run-time system is
7535 responsible for initializing this register with an appropriate value
7536 before execution begins.
7538 @item -mpic-register=@var{reg}
7539 @opindex mpic-register
7540 Specify the register to be used for PIC addressing. The default is R10
7541 unless stack-checking is enabled, when R9 is used.
7543 @item -mcirrus-fix-invalid-insns
7544 @opindex mcirrus-fix-invalid-insns
7545 @opindex mno-cirrus-fix-invalid-insns
7546 Insert NOPs into the instruction stream to in order to work around
7547 problems with invalid Maverick instruction combinations. This option
7548 is only valid if the @option{-mcpu=ep9312} option has been used to
7549 enable generation of instructions for the Cirrus Maverick floating
7550 point co-processor. This option is not enabled by default, since the
7551 problem is only present in older Maverick implementations. The default
7552 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7555 @item -mpoke-function-name
7556 @opindex mpoke-function-name
7557 Write the name of each function into the text section, directly
7558 preceding the function prologue. The generated code is similar to this:
7562 .ascii "arm_poke_function_name", 0
7565 .word 0xff000000 + (t1 - t0)
7566 arm_poke_function_name
7568 stmfd sp!, @{fp, ip, lr, pc@}
7572 When performing a stack backtrace, code can inspect the value of
7573 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7574 location @code{pc - 12} and the top 8 bits are set, then we know that
7575 there is a function name embedded immediately preceding this location
7576 and has length @code{((pc[-3]) & 0xff000000)}.
7580 Generate code for the 16-bit Thumb instruction set. The default is to
7581 use the 32-bit ARM instruction set.
7584 @opindex mtpcs-frame
7585 Generate a stack frame that is compliant with the Thumb Procedure Call
7586 Standard for all non-leaf functions. (A leaf function is one that does
7587 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7589 @item -mtpcs-leaf-frame
7590 @opindex mtpcs-leaf-frame
7591 Generate a stack frame that is compliant with the Thumb Procedure Call
7592 Standard for all leaf functions. (A leaf function is one that does
7593 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7595 @item -mcallee-super-interworking
7596 @opindex mcallee-super-interworking
7597 Gives all externally visible functions in the file being compiled an ARM
7598 instruction set header which switches to Thumb mode before executing the
7599 rest of the function. This allows these functions to be called from
7600 non-interworking code.
7602 @item -mcaller-super-interworking
7603 @opindex mcaller-super-interworking
7604 Allows calls via function pointers (including virtual functions) to
7605 execute correctly regardless of whether the target code has been
7606 compiled for interworking or not. There is a small overhead in the cost
7607 of executing a function pointer if this option is enabled.
7609 @item -mtp=@var{name}
7611 Specify the access model for the thread local storage pointer. The valid
7612 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7613 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7614 (supported in the arm6k architecture), and @option{auto}, which uses the
7615 best available method for the selected processor. The default setting is
7621 @subsection AVR Options
7624 These options are defined for AVR implementations:
7627 @item -mmcu=@var{mcu}
7629 Specify ATMEL AVR instruction set or MCU type.
7631 Instruction set avr1 is for the minimal AVR core, not supported by the C
7632 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7633 attiny11, attiny12, attiny15, attiny28).
7635 Instruction set avr2 (default) is for the classic AVR core with up to
7636 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7637 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7638 at90c8534, at90s8535).
7640 Instruction set avr3 is for the classic AVR core with up to 128K program
7641 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7643 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7644 memory space (MCU types: atmega8, atmega83, atmega85).
7646 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7647 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7648 atmega64, atmega128, at43usb355, at94k).
7652 Output instruction sizes to the asm file.
7654 @item -minit-stack=@var{N}
7655 @opindex minit-stack
7656 Specify the initial stack address, which may be a symbol or numeric value,
7657 @samp{__stack} is the default.
7659 @item -mno-interrupts
7660 @opindex mno-interrupts
7661 Generated code is not compatible with hardware interrupts.
7662 Code size will be smaller.
7664 @item -mcall-prologues
7665 @opindex mcall-prologues
7666 Functions prologues/epilogues expanded as call to appropriate
7667 subroutines. Code size will be smaller.
7669 @item -mno-tablejump
7670 @opindex mno-tablejump
7671 Do not generate tablejump insns which sometimes increase code size.
7674 @opindex mtiny-stack
7675 Change only the low 8 bits of the stack pointer.
7679 Assume int to be 8 bit integer. This affects the sizes of all types: A
7680 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7681 and long long will be 4 bytes. Please note that this option does not
7682 comply to the C standards, but it will provide you with smaller code
7686 @node Blackfin Options
7687 @subsection Blackfin Options
7688 @cindex Blackfin Options
7691 @item -momit-leaf-frame-pointer
7692 @opindex momit-leaf-frame-pointer
7693 Don't keep the frame pointer in a register for leaf functions. This
7694 avoids the instructions to save, set up and restore frame pointers and
7695 makes an extra register available in leaf functions. The option
7696 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7697 which might make debugging harder.
7699 @item -mspecld-anomaly
7700 @opindex mspecld-anomaly
7701 When enabled, the compiler will ensure that the generated code does not
7702 contain speculative loads after jump instructions. This option is enabled
7705 @item -mno-specld-anomaly
7706 @opindex mno-specld-anomaly
7707 Don't generate extra code to prevent speculative loads from occurring.
7709 @item -mcsync-anomaly
7710 @opindex mcsync-anomaly
7711 When enabled, the compiler will ensure that the generated code does not
7712 contain CSYNC or SSYNC instructions too soon after conditional branches.
7713 This option is enabled by default.
7715 @item -mno-csync-anomaly
7716 @opindex mno-csync-anomaly
7717 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7718 occurring too soon after a conditional branch.
7722 When enabled, the compiler is free to take advantage of the knowledge that
7723 the entire program fits into the low 64k of memory.
7726 @opindex mno-low-64k
7727 Assume that the program is arbitrarily large. This is the default.
7729 @item -mid-shared-library
7730 @opindex mid-shared-library
7731 Generate code that supports shared libraries via the library ID method.
7732 This allows for execute in place and shared libraries in an environment
7733 without virtual memory management. This option implies @option{-fPIC}.
7735 @item -mno-id-shared-library
7736 @opindex mno-id-shared-library
7737 Generate code that doesn't assume ID based shared libraries are being used.
7738 This is the default.
7740 @item -mshared-library-id=n
7741 @opindex mshared-library-id
7742 Specified the identification number of the ID based shared library being
7743 compiled. Specifying a value of 0 will generate more compact code, specifying
7744 other values will force the allocation of that number to the current
7745 library but is no more space or time efficient than omitting this option.
7748 @itemx -mno-long-calls
7749 @opindex mlong-calls
7750 @opindex mno-long-calls
7751 Tells the compiler to perform function calls by first loading the
7752 address of the function into a register and then performing a subroutine
7753 call on this register. This switch is needed if the target function
7754 will lie outside of the 24 bit addressing range of the offset based
7755 version of subroutine call instruction.
7757 This feature is not enabled by default. Specifying
7758 @option{-mno-long-calls} will restore the default behavior. Note these
7759 switches have no effect on how the compiler generates code to handle
7760 function calls via function pointers.
7764 @subsection CRIS Options
7765 @cindex CRIS Options
7767 These options are defined specifically for the CRIS ports.
7770 @item -march=@var{architecture-type}
7771 @itemx -mcpu=@var{architecture-type}
7774 Generate code for the specified architecture. The choices for
7775 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7776 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7777 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7780 @item -mtune=@var{architecture-type}
7782 Tune to @var{architecture-type} everything applicable about the generated
7783 code, except for the ABI and the set of available instructions. The
7784 choices for @var{architecture-type} are the same as for
7785 @option{-march=@var{architecture-type}}.
7787 @item -mmax-stack-frame=@var{n}
7788 @opindex mmax-stack-frame
7789 Warn when the stack frame of a function exceeds @var{n} bytes.
7791 @item -melinux-stacksize=@var{n}
7792 @opindex melinux-stacksize
7793 Only available with the @samp{cris-axis-aout} target. Arranges for
7794 indications in the program to the kernel loader that the stack of the
7795 program should be set to @var{n} bytes.
7801 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7802 @option{-march=v3} and @option{-march=v8} respectively.
7804 @item -mmul-bug-workaround
7805 @itemx -mno-mul-bug-workaround
7806 @opindex mmul-bug-workaround
7807 @opindex mno-mul-bug-workaround
7808 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7809 models where it applies. This option is active by default.
7813 Enable CRIS-specific verbose debug-related information in the assembly
7814 code. This option also has the effect to turn off the @samp{#NO_APP}
7815 formatted-code indicator to the assembler at the beginning of the
7820 Do not use condition-code results from previous instruction; always emit
7821 compare and test instructions before use of condition codes.
7823 @item -mno-side-effects
7824 @opindex mno-side-effects
7825 Do not emit instructions with side-effects in addressing modes other than
7829 @itemx -mno-stack-align
7831 @itemx -mno-data-align
7832 @itemx -mconst-align
7833 @itemx -mno-const-align
7834 @opindex mstack-align
7835 @opindex mno-stack-align
7836 @opindex mdata-align
7837 @opindex mno-data-align
7838 @opindex mconst-align
7839 @opindex mno-const-align
7840 These options (no-options) arranges (eliminate arrangements) for the
7841 stack-frame, individual data and constants to be aligned for the maximum
7842 single data access size for the chosen CPU model. The default is to
7843 arrange for 32-bit alignment. ABI details such as structure layout are
7844 not affected by these options.
7852 Similar to the stack- data- and const-align options above, these options
7853 arrange for stack-frame, writable data and constants to all be 32-bit,
7854 16-bit or 8-bit aligned. The default is 32-bit alignment.
7856 @item -mno-prologue-epilogue
7857 @itemx -mprologue-epilogue
7858 @opindex mno-prologue-epilogue
7859 @opindex mprologue-epilogue
7860 With @option{-mno-prologue-epilogue}, the normal function prologue and
7861 epilogue that sets up the stack-frame are omitted and no return
7862 instructions or return sequences are generated in the code. Use this
7863 option only together with visual inspection of the compiled code: no
7864 warnings or errors are generated when call-saved registers must be saved,
7865 or storage for local variable needs to be allocated.
7871 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7872 instruction sequences that load addresses for functions from the PLT part
7873 of the GOT rather than (traditional on other architectures) calls to the
7874 PLT@. The default is @option{-mgotplt}.
7878 Legacy no-op option only recognized with the cris-axis-aout target.
7882 Legacy no-op option only recognized with the cris-axis-elf and
7883 cris-axis-linux-gnu targets.
7887 Only recognized with the cris-axis-aout target, where it selects a
7888 GNU/linux-like multilib, include files and instruction set for
7893 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7897 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7898 to link with input-output functions from a simulator library. Code,
7899 initialized data and zero-initialized data are allocated consecutively.
7903 Like @option{-sim}, but pass linker options to locate initialized data at
7904 0x40000000 and zero-initialized data at 0x80000000.
7908 @subsection CRX Options
7911 These options are defined specifically for the CRX ports.
7917 Enable the use of multiply-accumulate instructions. Disabled by default.
7921 Push instructions will be used to pass outgoing arguments when functions
7922 are called. Enabled by default.
7925 @node Darwin Options
7926 @subsection Darwin Options
7927 @cindex Darwin options
7929 These options are defined for all architectures running the Darwin operating
7932 FSF GCC on Darwin does not create ``fat'' object files; it will create
7933 an object file for the single architecture that it was built to
7934 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7935 @option{-arch} options are used; it does so by running the compiler or
7936 linker multiple times and joining the results together with
7939 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7940 @samp{i686}) is determined by the flags that specify the ISA
7941 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7942 @option{-force_cpusubtype_ALL} option can be used to override this.
7944 The Darwin tools vary in their behavior when presented with an ISA
7945 mismatch. The assembler, @file{as}, will only permit instructions to
7946 be used that are valid for the subtype of the file it is generating,
7947 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7948 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7949 and print an error if asked to create a shared library with a less
7950 restrictive subtype than its input files (for instance, trying to put
7951 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7952 for executables, @file{ld}, will quietly give the executable the most
7953 restrictive subtype of any of its input files.
7958 Add the framework directory @var{dir} to the head of the list of
7959 directories to be searched for header files. These directories are
7960 interleaved with those specified by @option{-I} options and are
7961 scanned in a left-to-right order.
7963 A framework directory is a directory with frameworks in it. A
7964 framework is a directory with a @samp{"Headers"} and/or
7965 @samp{"PrivateHeaders"} directory contained directly in it that ends
7966 in @samp{".framework"}. The name of a framework is the name of this
7967 directory excluding the @samp{".framework"}. Headers associated with
7968 the framework are found in one of those two directories, with
7969 @samp{"Headers"} being searched first. A subframework is a framework
7970 directory that is in a framework's @samp{"Frameworks"} directory.
7971 Includes of subframework headers can only appear in a header of a
7972 framework that contains the subframework, or in a sibling subframework
7973 header. Two subframeworks are siblings if they occur in the same
7974 framework. A subframework should not have the same name as a
7975 framework, a warning will be issued if this is violated. Currently a
7976 subframework cannot have subframeworks, in the future, the mechanism
7977 may be extended to support this. The standard frameworks can be found
7978 in @samp{"/System/Library/Frameworks"} and
7979 @samp{"/Library/Frameworks"}. An example include looks like
7980 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7981 the name of the framework and header.h is found in the
7982 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7986 Emit debugging information for symbols that are used. For STABS
7987 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7988 This is by default ON@.
7992 Emit debugging information for all symbols and types.
7994 @item -mmacosx-version-min=@var{version}
7995 The earliest version of MacOS X that this executable will run on
7996 is @var{version}. Typical values of @var{version} include @code{10.1},
7997 @code{10.2}, and @code{10.3.9}.
7999 The default for this option is to make choices that seem to be most
8002 @item -mone-byte-bool
8003 @opindex -mone-byte-bool
8004 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8005 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8006 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8007 option has no effect on x86.
8009 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8010 to generate code that is not binary compatible with code generated
8011 without that switch. Using this switch may require recompiling all
8012 other modules in a program, including system libraries. Use this
8013 switch to conform to a non-default data model.
8015 @item -mfix-and-continue
8016 @itemx -ffix-and-continue
8017 @itemx -findirect-data
8018 @opindex mfix-and-continue
8019 @opindex ffix-and-continue
8020 @opindex findirect-data
8021 Generate code suitable for fast turn around development. Needed to
8022 enable gdb to dynamically load @code{.o} files into already running
8023 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8024 are provided for backwards compatibility.
8028 Loads all members of static archive libraries.
8029 See man ld(1) for more information.
8031 @item -arch_errors_fatal
8032 @opindex arch_errors_fatal
8033 Cause the errors having to do with files that have the wrong architecture
8037 @opindex bind_at_load
8038 Causes the output file to be marked such that the dynamic linker will
8039 bind all undefined references when the file is loaded or launched.
8043 Produce a Mach-o bundle format file.
8044 See man ld(1) for more information.
8046 @item -bundle_loader @var{executable}
8047 @opindex bundle_loader
8048 This option specifies the @var{executable} that will be loading the build
8049 output file being linked. See man ld(1) for more information.
8052 @opindex -dynamiclib
8053 When passed this option, GCC will produce a dynamic library instead of
8054 an executable when linking, using the Darwin @file{libtool} command.
8056 @item -force_cpusubtype_ALL
8057 @opindex -force_cpusubtype_ALL
8058 This causes GCC's output file to have the @var{ALL} subtype, instead of
8059 one controlled by the @option{-mcpu} or @option{-march} option.
8061 @item -allowable_client @var{client_name}
8063 @itemx -compatibility_version
8064 @itemx -current_version
8066 @itemx -dependency-file
8068 @itemx -dylinker_install_name
8070 @itemx -exported_symbols_list
8072 @itemx -flat_namespace
8073 @itemx -force_flat_namespace
8074 @itemx -headerpad_max_install_names
8077 @itemx -install_name
8078 @itemx -keep_private_externs
8079 @itemx -multi_module
8080 @itemx -multiply_defined
8081 @itemx -multiply_defined_unused
8083 @itemx -no_dead_strip_inits_and_terms
8084 @itemx -nofixprebinding
8087 @itemx -noseglinkedit
8088 @itemx -pagezero_size
8090 @itemx -prebind_all_twolevel_modules
8091 @itemx -private_bundle
8092 @itemx -read_only_relocs
8094 @itemx -sectobjectsymbols
8098 @itemx -sectobjectsymbols
8101 @itemx -segs_read_only_addr
8102 @itemx -segs_read_write_addr
8103 @itemx -seg_addr_table
8104 @itemx -seg_addr_table_filename
8107 @itemx -segs_read_only_addr
8108 @itemx -segs_read_write_addr
8109 @itemx -single_module
8112 @itemx -sub_umbrella
8113 @itemx -twolevel_namespace
8116 @itemx -unexported_symbols_list
8117 @itemx -weak_reference_mismatches
8120 @opindex allowable_client
8121 @opindex client_name
8122 @opindex compatibility_version
8123 @opindex current_version
8125 @opindex dependency-file
8127 @opindex dylinker_install_name
8129 @opindex exported_symbols_list
8131 @opindex flat_namespace
8132 @opindex force_flat_namespace
8133 @opindex headerpad_max_install_names
8136 @opindex install_name
8137 @opindex keep_private_externs
8138 @opindex multi_module
8139 @opindex multiply_defined
8140 @opindex multiply_defined_unused
8142 @opindex no_dead_strip_inits_and_terms
8143 @opindex nofixprebinding
8144 @opindex nomultidefs
8146 @opindex noseglinkedit
8147 @opindex pagezero_size
8149 @opindex prebind_all_twolevel_modules
8150 @opindex private_bundle
8151 @opindex read_only_relocs
8153 @opindex sectobjectsymbols
8157 @opindex sectobjectsymbols
8160 @opindex segs_read_only_addr
8161 @opindex segs_read_write_addr
8162 @opindex seg_addr_table
8163 @opindex seg_addr_table_filename
8164 @opindex seglinkedit
8166 @opindex segs_read_only_addr
8167 @opindex segs_read_write_addr
8168 @opindex single_module
8170 @opindex sub_library
8171 @opindex sub_umbrella
8172 @opindex twolevel_namespace
8175 @opindex unexported_symbols_list
8176 @opindex weak_reference_mismatches
8177 @opindex whatsloaded
8179 These options are passed to the Darwin linker. The Darwin linker man page
8180 describes them in detail.
8183 @node DEC Alpha Options
8184 @subsection DEC Alpha Options
8186 These @samp{-m} options are defined for the DEC Alpha implementations:
8189 @item -mno-soft-float
8191 @opindex mno-soft-float
8192 @opindex msoft-float
8193 Use (do not use) the hardware floating-point instructions for
8194 floating-point operations. When @option{-msoft-float} is specified,
8195 functions in @file{libgcc.a} will be used to perform floating-point
8196 operations. Unless they are replaced by routines that emulate the
8197 floating-point operations, or compiled in such a way as to call such
8198 emulations routines, these routines will issue floating-point
8199 operations. If you are compiling for an Alpha without floating-point
8200 operations, you must ensure that the library is built so as not to call
8203 Note that Alpha implementations without floating-point operations are
8204 required to have floating-point registers.
8209 @opindex mno-fp-regs
8210 Generate code that uses (does not use) the floating-point register set.
8211 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8212 register set is not used, floating point operands are passed in integer
8213 registers as if they were integers and floating-point results are passed
8214 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8215 so any function with a floating-point argument or return value called by code
8216 compiled with @option{-mno-fp-regs} must also be compiled with that
8219 A typical use of this option is building a kernel that does not use,
8220 and hence need not save and restore, any floating-point registers.
8224 The Alpha architecture implements floating-point hardware optimized for
8225 maximum performance. It is mostly compliant with the IEEE floating
8226 point standard. However, for full compliance, software assistance is
8227 required. This option generates code fully IEEE compliant code
8228 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8229 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8230 defined during compilation. The resulting code is less efficient but is
8231 able to correctly support denormalized numbers and exceptional IEEE
8232 values such as not-a-number and plus/minus infinity. Other Alpha
8233 compilers call this option @option{-ieee_with_no_inexact}.
8235 @item -mieee-with-inexact
8236 @opindex mieee-with-inexact
8237 This is like @option{-mieee} except the generated code also maintains
8238 the IEEE @var{inexact-flag}. Turning on this option causes the
8239 generated code to implement fully-compliant IEEE math. In addition to
8240 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8241 macro. On some Alpha implementations the resulting code may execute
8242 significantly slower than the code generated by default. Since there is
8243 very little code that depends on the @var{inexact-flag}, you should
8244 normally not specify this option. Other Alpha compilers call this
8245 option @option{-ieee_with_inexact}.
8247 @item -mfp-trap-mode=@var{trap-mode}
8248 @opindex mfp-trap-mode
8249 This option controls what floating-point related traps are enabled.
8250 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8251 The trap mode can be set to one of four values:
8255 This is the default (normal) setting. The only traps that are enabled
8256 are the ones that cannot be disabled in software (e.g., division by zero
8260 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8264 Like @samp{su}, but the instructions are marked to be safe for software
8265 completion (see Alpha architecture manual for details).
8268 Like @samp{su}, but inexact traps are enabled as well.
8271 @item -mfp-rounding-mode=@var{rounding-mode}
8272 @opindex mfp-rounding-mode
8273 Selects the IEEE rounding mode. Other Alpha compilers call this option
8274 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8279 Normal IEEE rounding mode. Floating point numbers are rounded towards
8280 the nearest machine number or towards the even machine number in case
8284 Round towards minus infinity.
8287 Chopped rounding mode. Floating point numbers are rounded towards zero.
8290 Dynamic rounding mode. A field in the floating point control register
8291 (@var{fpcr}, see Alpha architecture reference manual) controls the
8292 rounding mode in effect. The C library initializes this register for
8293 rounding towards plus infinity. Thus, unless your program modifies the
8294 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8297 @item -mtrap-precision=@var{trap-precision}
8298 @opindex mtrap-precision
8299 In the Alpha architecture, floating point traps are imprecise. This
8300 means without software assistance it is impossible to recover from a
8301 floating trap and program execution normally needs to be terminated.
8302 GCC can generate code that can assist operating system trap handlers
8303 in determining the exact location that caused a floating point trap.
8304 Depending on the requirements of an application, different levels of
8305 precisions can be selected:
8309 Program precision. This option is the default and means a trap handler
8310 can only identify which program caused a floating point exception.
8313 Function precision. The trap handler can determine the function that
8314 caused a floating point exception.
8317 Instruction precision. The trap handler can determine the exact
8318 instruction that caused a floating point exception.
8321 Other Alpha compilers provide the equivalent options called
8322 @option{-scope_safe} and @option{-resumption_safe}.
8324 @item -mieee-conformant
8325 @opindex mieee-conformant
8326 This option marks the generated code as IEEE conformant. You must not
8327 use this option unless you also specify @option{-mtrap-precision=i} and either
8328 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8329 is to emit the line @samp{.eflag 48} in the function prologue of the
8330 generated assembly file. Under DEC Unix, this has the effect that
8331 IEEE-conformant math library routines will be linked in.
8333 @item -mbuild-constants
8334 @opindex mbuild-constants
8335 Normally GCC examines a 32- or 64-bit integer constant to
8336 see if it can construct it from smaller constants in two or three
8337 instructions. If it cannot, it will output the constant as a literal and
8338 generate code to load it from the data segment at runtime.
8340 Use this option to require GCC to construct @emph{all} integer constants
8341 using code, even if it takes more instructions (the maximum is six).
8343 You would typically use this option to build a shared library dynamic
8344 loader. Itself a shared library, it must relocate itself in memory
8345 before it can find the variables and constants in its own data segment.
8351 Select whether to generate code to be assembled by the vendor-supplied
8352 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8370 Indicate whether GCC should generate code to use the optional BWX,
8371 CIX, FIX and MAX instruction sets. The default is to use the instruction
8372 sets supported by the CPU type specified via @option{-mcpu=} option or that
8373 of the CPU on which GCC was built if none was specified.
8378 @opindex mfloat-ieee
8379 Generate code that uses (does not use) VAX F and G floating point
8380 arithmetic instead of IEEE single and double precision.
8382 @item -mexplicit-relocs
8383 @itemx -mno-explicit-relocs
8384 @opindex mexplicit-relocs
8385 @opindex mno-explicit-relocs
8386 Older Alpha assemblers provided no way to generate symbol relocations
8387 except via assembler macros. Use of these macros does not allow
8388 optimal instruction scheduling. GNU binutils as of version 2.12
8389 supports a new syntax that allows the compiler to explicitly mark
8390 which relocations should apply to which instructions. This option
8391 is mostly useful for debugging, as GCC detects the capabilities of
8392 the assembler when it is built and sets the default accordingly.
8396 @opindex msmall-data
8397 @opindex mlarge-data
8398 When @option{-mexplicit-relocs} is in effect, static data is
8399 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8400 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8401 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8402 16-bit relocations off of the @code{$gp} register. This limits the
8403 size of the small data area to 64KB, but allows the variables to be
8404 directly accessed via a single instruction.
8406 The default is @option{-mlarge-data}. With this option the data area
8407 is limited to just below 2GB@. Programs that require more than 2GB of
8408 data must use @code{malloc} or @code{mmap} to allocate the data in the
8409 heap instead of in the program's data segment.
8411 When generating code for shared libraries, @option{-fpic} implies
8412 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8416 @opindex msmall-text
8417 @opindex mlarge-text
8418 When @option{-msmall-text} is used, the compiler assumes that the
8419 code of the entire program (or shared library) fits in 4MB, and is
8420 thus reachable with a branch instruction. When @option{-msmall-data}
8421 is used, the compiler can assume that all local symbols share the
8422 same @code{$gp} value, and thus reduce the number of instructions
8423 required for a function call from 4 to 1.
8425 The default is @option{-mlarge-text}.
8427 @item -mcpu=@var{cpu_type}
8429 Set the instruction set and instruction scheduling parameters for
8430 machine type @var{cpu_type}. You can specify either the @samp{EV}
8431 style name or the corresponding chip number. GCC supports scheduling
8432 parameters for the EV4, EV5 and EV6 family of processors and will
8433 choose the default values for the instruction set from the processor
8434 you specify. If you do not specify a processor type, GCC will default
8435 to the processor on which the compiler was built.
8437 Supported values for @var{cpu_type} are
8443 Schedules as an EV4 and has no instruction set extensions.
8447 Schedules as an EV5 and has no instruction set extensions.
8451 Schedules as an EV5 and supports the BWX extension.
8456 Schedules as an EV5 and supports the BWX and MAX extensions.
8460 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8464 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8467 @item -mtune=@var{cpu_type}
8469 Set only the instruction scheduling parameters for machine type
8470 @var{cpu_type}. The instruction set is not changed.
8472 @item -mmemory-latency=@var{time}
8473 @opindex mmemory-latency
8474 Sets the latency the scheduler should assume for typical memory
8475 references as seen by the application. This number is highly
8476 dependent on the memory access patterns used by the application
8477 and the size of the external cache on the machine.
8479 Valid options for @var{time} are
8483 A decimal number representing clock cycles.
8489 The compiler contains estimates of the number of clock cycles for
8490 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8491 (also called Dcache, Scache, and Bcache), as well as to main memory.
8492 Note that L3 is only valid for EV5.
8497 @node DEC Alpha/VMS Options
8498 @subsection DEC Alpha/VMS Options
8500 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8503 @item -mvms-return-codes
8504 @opindex mvms-return-codes
8505 Return VMS condition codes from main. The default is to return POSIX
8506 style condition (e.g.@ error) codes.
8510 @subsection FRV Options
8517 Only use the first 32 general purpose registers.
8522 Use all 64 general purpose registers.
8527 Use only the first 32 floating point registers.
8532 Use all 64 floating point registers
8535 @opindex mhard-float
8537 Use hardware instructions for floating point operations.
8540 @opindex msoft-float
8542 Use library routines for floating point operations.
8547 Dynamically allocate condition code registers.
8552 Do not try to dynamically allocate condition code registers, only
8553 use @code{icc0} and @code{fcc0}.
8558 Change ABI to use double word insns.
8563 Do not use double word instructions.
8568 Use floating point double instructions.
8573 Do not use floating point double instructions.
8578 Use media instructions.
8583 Do not use media instructions.
8588 Use multiply and add/subtract instructions.
8593 Do not use multiply and add/subtract instructions.
8598 Select the FDPIC ABI, that uses function descriptors to represent
8599 pointers to functions. Without any PIC/PIE-related options, it
8600 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8601 assumes GOT entries and small data are within a 12-bit range from the
8602 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8603 are computed with 32 bits.
8606 @opindex minline-plt
8608 Enable inlining of PLT entries in function calls to functions that are
8609 not known to bind locally. It has no effect without @option{-mfdpic}.
8610 It's enabled by default if optimizing for speed and compiling for
8611 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8612 optimization option such as @option{-O3} or above is present in the
8618 Assume a large TLS segment when generating thread-local code.
8623 Do not assume a large TLS segment when generating thread-local code.
8628 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8629 that is known to be in read-only sections. It's enabled by default,
8630 except for @option{-fpic} or @option{-fpie}: even though it may help
8631 make the global offset table smaller, it trades 1 instruction for 4.
8632 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8633 one of which may be shared by multiple symbols, and it avoids the need
8634 for a GOT entry for the referenced symbol, so it's more likely to be a
8635 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8637 @item -multilib-library-pic
8638 @opindex multilib-library-pic
8640 Link with the (library, not FD) pic libraries. It's implied by
8641 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8642 @option{-fpic} without @option{-mfdpic}. You should never have to use
8648 Follow the EABI requirement of always creating a frame pointer whenever
8649 a stack frame is allocated. This option is enabled by default and can
8650 be disabled with @option{-mno-linked-fp}.
8653 @opindex mlong-calls
8655 Use indirect addressing to call functions outside the current
8656 compilation unit. This allows the functions to be placed anywhere
8657 within the 32-bit address space.
8659 @item -malign-labels
8660 @opindex malign-labels
8662 Try to align labels to an 8-byte boundary by inserting nops into the
8663 previous packet. This option only has an effect when VLIW packing
8664 is enabled. It doesn't create new packets; it merely adds nops to
8668 @opindex mlibrary-pic
8670 Generate position-independent EABI code.
8675 Use only the first four media accumulator registers.
8680 Use all eight media accumulator registers.
8685 Pack VLIW instructions.
8690 Do not pack VLIW instructions.
8695 Do not mark ABI switches in e_flags.
8700 Enable the use of conditional-move instructions (default).
8702 This switch is mainly for debugging the compiler and will likely be removed
8703 in a future version.
8705 @item -mno-cond-move
8706 @opindex mno-cond-move
8708 Disable the use of conditional-move instructions.
8710 This switch is mainly for debugging the compiler and will likely be removed
8711 in a future version.
8716 Enable the use of conditional set instructions (default).
8718 This switch is mainly for debugging the compiler and will likely be removed
8719 in a future version.
8724 Disable the use of conditional set instructions.
8726 This switch is mainly for debugging the compiler and will likely be removed
8727 in a future version.
8732 Enable the use of conditional execution (default).
8734 This switch is mainly for debugging the compiler and will likely be removed
8735 in a future version.
8737 @item -mno-cond-exec
8738 @opindex mno-cond-exec
8740 Disable the use of conditional execution.
8742 This switch is mainly for debugging the compiler and will likely be removed
8743 in a future version.
8746 @opindex mvliw-branch
8748 Run a pass to pack branches into VLIW instructions (default).
8750 This switch is mainly for debugging the compiler and will likely be removed
8751 in a future version.
8753 @item -mno-vliw-branch
8754 @opindex mno-vliw-branch
8756 Do not run a pass to pack branches into VLIW instructions.
8758 This switch is mainly for debugging the compiler and will likely be removed
8759 in a future version.
8761 @item -mmulti-cond-exec
8762 @opindex mmulti-cond-exec
8764 Enable optimization of @code{&&} and @code{||} in conditional execution
8767 This switch is mainly for debugging the compiler and will likely be removed
8768 in a future version.
8770 @item -mno-multi-cond-exec
8771 @opindex mno-multi-cond-exec
8773 Disable optimization of @code{&&} and @code{||} in conditional execution.
8775 This switch is mainly for debugging the compiler and will likely be removed
8776 in a future version.
8778 @item -mnested-cond-exec
8779 @opindex mnested-cond-exec
8781 Enable nested conditional execution optimizations (default).
8783 This switch is mainly for debugging the compiler and will likely be removed
8784 in a future version.
8786 @item -mno-nested-cond-exec
8787 @opindex mno-nested-cond-exec
8789 Disable nested conditional execution optimizations.
8791 This switch is mainly for debugging the compiler and will likely be removed
8792 in a future version.
8794 @item -moptimize-membar
8795 @opindex moptimize-membar
8797 This switch removes redundant @code{membar} instructions from the
8798 compiler generated code. It is enabled by default.
8800 @item -mno-optimize-membar
8801 @opindex mno-optimize-membar
8803 This switch disables the automatic removal of redundant @code{membar}
8804 instructions from the generated code.
8806 @item -mtomcat-stats
8807 @opindex mtomcat-stats
8809 Cause gas to print out tomcat statistics.
8811 @item -mcpu=@var{cpu}
8814 Select the processor type for which to generate code. Possible values are
8815 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8816 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8820 @node GNU/Linux Options
8821 @subsection GNU/Linux Options
8823 These @samp{-m} options are defined for GNU/Linux targets:
8828 Use the GNU C library instead of uClibc. This is the default except
8829 on @samp{*-*-linux-*uclibc*} targets.
8833 Use uClibc instead of the GNU C library. This is the default on
8834 @samp{*-*-linux-*uclibc*} targets.
8837 @node H8/300 Options
8838 @subsection H8/300 Options
8840 These @samp{-m} options are defined for the H8/300 implementations:
8845 Shorten some address references at link time, when possible; uses the
8846 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8847 ld, Using ld}, for a fuller description.
8851 Generate code for the H8/300H@.
8855 Generate code for the H8S@.
8859 Generate code for the H8S and H8/300H in the normal mode. This switch
8860 must be used either with @option{-mh} or @option{-ms}.
8864 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8868 Make @code{int} data 32 bits by default.
8872 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8873 The default for the H8/300H and H8S is to align longs and floats on 4
8875 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8876 This option has no effect on the H8/300.
8880 @subsection HPPA Options
8881 @cindex HPPA Options
8883 These @samp{-m} options are defined for the HPPA family of computers:
8886 @item -march=@var{architecture-type}
8888 Generate code for the specified architecture. The choices for
8889 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8890 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8891 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8892 architecture option for your machine. Code compiled for lower numbered
8893 architectures will run on higher numbered architectures, but not the
8897 @itemx -mpa-risc-1-1
8898 @itemx -mpa-risc-2-0
8899 @opindex mpa-risc-1-0
8900 @opindex mpa-risc-1-1
8901 @opindex mpa-risc-2-0
8902 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8905 @opindex mbig-switch
8906 Generate code suitable for big switch tables. Use this option only if
8907 the assembler/linker complain about out of range branches within a switch
8910 @item -mjump-in-delay
8911 @opindex mjump-in-delay
8912 Fill delay slots of function calls with unconditional jump instructions
8913 by modifying the return pointer for the function call to be the target
8914 of the conditional jump.
8916 @item -mdisable-fpregs
8917 @opindex mdisable-fpregs
8918 Prevent floating point registers from being used in any manner. This is
8919 necessary for compiling kernels which perform lazy context switching of
8920 floating point registers. If you use this option and attempt to perform
8921 floating point operations, the compiler will abort.
8923 @item -mdisable-indexing
8924 @opindex mdisable-indexing
8925 Prevent the compiler from using indexing address modes. This avoids some
8926 rather obscure problems when compiling MIG generated code under MACH@.
8928 @item -mno-space-regs
8929 @opindex mno-space-regs
8930 Generate code that assumes the target has no space registers. This allows
8931 GCC to generate faster indirect calls and use unscaled index address modes.
8933 Such code is suitable for level 0 PA systems and kernels.
8935 @item -mfast-indirect-calls
8936 @opindex mfast-indirect-calls
8937 Generate code that assumes calls never cross space boundaries. This
8938 allows GCC to emit code which performs faster indirect calls.
8940 This option will not work in the presence of shared libraries or nested
8943 @item -mfixed-range=@var{register-range}
8944 @opindex mfixed-range
8945 Generate code treating the given register range as fixed registers.
8946 A fixed register is one that the register allocator can not use. This is
8947 useful when compiling kernel code. A register range is specified as
8948 two registers separated by a dash. Multiple register ranges can be
8949 specified separated by a comma.
8951 @item -mlong-load-store
8952 @opindex mlong-load-store
8953 Generate 3-instruction load and store sequences as sometimes required by
8954 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8957 @item -mportable-runtime
8958 @opindex mportable-runtime
8959 Use the portable calling conventions proposed by HP for ELF systems.
8963 Enable the use of assembler directives only GAS understands.
8965 @item -mschedule=@var{cpu-type}
8967 Schedule code according to the constraints for the machine type
8968 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8969 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8970 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8971 proper scheduling option for your machine. The default scheduling is
8975 @opindex mlinker-opt
8976 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8977 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8978 linkers in which they give bogus error messages when linking some programs.
8981 @opindex msoft-float
8982 Generate output containing library calls for floating point.
8983 @strong{Warning:} the requisite libraries are not available for all HPPA
8984 targets. Normally the facilities of the machine's usual C compiler are
8985 used, but this cannot be done directly in cross-compilation. You must make
8986 your own arrangements to provide suitable library functions for
8987 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8988 does provide software floating point support.
8990 @option{-msoft-float} changes the calling convention in the output file;
8991 therefore, it is only useful if you compile @emph{all} of a program with
8992 this option. In particular, you need to compile @file{libgcc.a}, the
8993 library that comes with GCC, with @option{-msoft-float} in order for
8998 Generate the predefine, @code{_SIO}, for server IO@. The default is
8999 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9000 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9001 options are available under HP-UX and HI-UX@.
9005 Use GNU ld specific options. This passes @option{-shared} to ld when
9006 building a shared library. It is the default when GCC is configured,
9007 explicitly or implicitly, with the GNU linker. This option does not
9008 have any affect on which ld is called, it only changes what parameters
9009 are passed to that ld. The ld that is called is determined by the
9010 @option{--with-ld} configure option, GCC's program search path, and
9011 finally by the user's @env{PATH}. The linker used by GCC can be printed
9012 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9013 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9017 Use HP ld specific options. This passes @option{-b} to ld when building
9018 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9019 links. It is the default when GCC is configured, explicitly or
9020 implicitly, with the HP linker. This option does not have any affect on
9021 which ld is called, it only changes what parameters are passed to that
9022 ld. The ld that is called is determined by the @option{--with-ld}
9023 configure option, GCC's program search path, and finally by the user's
9024 @env{PATH}. The linker used by GCC can be printed using @samp{which
9025 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9026 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9029 @opindex mno-long-calls
9030 Generate code that uses long call sequences. This ensures that a call
9031 is always able to reach linker generated stubs. The default is to generate
9032 long calls only when the distance from the call site to the beginning
9033 of the function or translation unit, as the case may be, exceeds a
9034 predefined limit set by the branch type being used. The limits for
9035 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9036 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9039 Distances are measured from the beginning of functions when using the
9040 @option{-ffunction-sections} option, or when using the @option{-mgas}
9041 and @option{-mno-portable-runtime} options together under HP-UX with
9044 It is normally not desirable to use this option as it will degrade
9045 performance. However, it may be useful in large applications,
9046 particularly when partial linking is used to build the application.
9048 The types of long calls used depends on the capabilities of the
9049 assembler and linker, and the type of code being generated. The
9050 impact on systems that support long absolute calls, and long pic
9051 symbol-difference or pc-relative calls should be relatively small.
9052 However, an indirect call is used on 32-bit ELF systems in pic code
9053 and it is quite long.
9055 @item -munix=@var{unix-std}
9057 Generate compiler predefines and select a startfile for the specified
9058 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9059 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9060 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9061 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9062 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9065 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9066 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9067 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9068 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9069 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9070 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9072 It is @emph{important} to note that this option changes the interfaces
9073 for various library routines. It also affects the operational behavior
9074 of the C library. Thus, @emph{extreme} care is needed in using this
9077 Library code that is intended to operate with more than one UNIX
9078 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9079 as appropriate. Most GNU software doesn't provide this capability.
9083 Suppress the generation of link options to search libdld.sl when the
9084 @option{-static} option is specified on HP-UX 10 and later.
9088 The HP-UX implementation of setlocale in libc has a dependency on
9089 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9090 when the @option{-static} option is specified, special link options
9091 are needed to resolve this dependency.
9093 On HP-UX 10 and later, the GCC driver adds the necessary options to
9094 link with libdld.sl when the @option{-static} option is specified.
9095 This causes the resulting binary to be dynamic. On the 64-bit port,
9096 the linkers generate dynamic binaries by default in any case. The
9097 @option{-nolibdld} option can be used to prevent the GCC driver from
9098 adding these link options.
9102 Add support for multithreading with the @dfn{dce thread} library
9103 under HP-UX@. This option sets flags for both the preprocessor and
9107 @node i386 and x86-64 Options
9108 @subsection Intel 386 and AMD x86-64 Options
9109 @cindex i386 Options
9110 @cindex x86-64 Options
9111 @cindex Intel 386 Options
9112 @cindex AMD x86-64 Options
9114 These @samp{-m} options are defined for the i386 and x86-64 family of
9118 @item -mtune=@var{cpu-type}
9120 Tune to @var{cpu-type} everything applicable about the generated code, except
9121 for the ABI and the set of available instructions. The choices for
9125 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9126 If you know the CPU on which your code will run, then you should use
9127 the corresponding @option{-mtune} option instead of
9128 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9129 of your application will have, then you should use this option.
9131 As new processors are deployed in the marketplace, the behavior of this
9132 option will change. Therefore, if you upgrade to a newer version of
9133 GCC, the code generated option will change to reflect the processors
9134 that were most common when that version of GCC was released.
9136 There is no @option{-march=generic} option because @option{-march}
9137 indicates the instruction set the compiler can use, and there is no
9138 generic instruction set applicable to all processors. In contrast,
9139 @option{-mtune} indicates the processor (or, in this case, collection of
9140 processors) for which the code is optimized.
9142 Original Intel's i386 CPU@.
9144 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9146 Intel Pentium CPU with no MMX support.
9148 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9150 Intel PentiumPro CPU@.
9152 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9153 instruction set will be used, so the code will run on all i686 familly chips.
9155 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9156 @item pentium3, pentium3m
9157 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9160 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9161 support. Used by Centrino notebooks.
9162 @item pentium4, pentium4m
9163 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9165 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9168 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9169 SSE2 and SSE3 instruction set support.
9171 AMD K6 CPU with MMX instruction set support.
9173 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9174 @item athlon, athlon-tbird
9175 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9177 @item athlon-4, athlon-xp, athlon-mp
9178 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9179 instruction set support.
9180 @item k8, opteron, athlon64, athlon-fx
9181 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9182 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9184 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9187 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9188 instruction set support.
9190 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9191 implemented for this chip.)
9193 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9194 implemented for this chip.)
9197 While picking a specific @var{cpu-type} will schedule things appropriately
9198 for that particular chip, the compiler will not generate any code that
9199 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9202 @item -march=@var{cpu-type}
9204 Generate instructions for the machine type @var{cpu-type}. The choices
9205 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9206 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9208 @item -mcpu=@var{cpu-type}
9210 A deprecated synonym for @option{-mtune}.
9219 @opindex mpentiumpro
9220 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9221 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9222 These synonyms are deprecated.
9224 @item -mfpmath=@var{unit}
9226 Generate floating point arithmetics for selected unit @var{unit}. The choices
9231 Use the standard 387 floating point coprocessor present majority of chips and
9232 emulated otherwise. Code compiled with this option will run almost everywhere.
9233 The temporary results are computed in 80bit precision instead of precision
9234 specified by the type resulting in slightly different results compared to most
9235 of other chips. See @option{-ffloat-store} for more detailed description.
9237 This is the default choice for i386 compiler.
9240 Use scalar floating point instructions present in the SSE instruction set.
9241 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9242 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9243 instruction set supports only single precision arithmetics, thus the double and
9244 extended precision arithmetics is still done using 387. Later version, present
9245 only in Pentium4 and the future AMD x86-64 chips supports double precision
9248 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9249 or @option{-msse2} switches to enable SSE extensions and make this option
9250 effective. For the x86-64 compiler, these extensions are enabled by default.
9252 The resulting code should be considerably faster in the majority of cases and avoid
9253 the numerical instability problems of 387 code, but may break some existing
9254 code that expects temporaries to be 80bit.
9256 This is the default choice for the x86-64 compiler.
9259 Attempt to utilize both instruction sets at once. This effectively double the
9260 amount of available registers and on chips with separate execution units for
9261 387 and SSE the execution resources too. Use this option with care, as it is
9262 still experimental, because the GCC register allocator does not model separate
9263 functional units well resulting in instable performance.
9266 @item -masm=@var{dialect}
9267 @opindex masm=@var{dialect}
9268 Output asm instructions using selected @var{dialect}. Supported
9269 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9270 not support @samp{intel}.
9275 @opindex mno-ieee-fp
9276 Control whether or not the compiler uses IEEE floating point
9277 comparisons. These handle correctly the case where the result of a
9278 comparison is unordered.
9281 @opindex msoft-float
9282 Generate output containing library calls for floating point.
9283 @strong{Warning:} the requisite libraries are not part of GCC@.
9284 Normally the facilities of the machine's usual C compiler are used, but
9285 this can't be done directly in cross-compilation. You must make your
9286 own arrangements to provide suitable library functions for
9289 On machines where a function returns floating point results in the 80387
9290 register stack, some floating point opcodes may be emitted even if
9291 @option{-msoft-float} is used.
9293 @item -mno-fp-ret-in-387
9294 @opindex mno-fp-ret-in-387
9295 Do not use the FPU registers for return values of functions.
9297 The usual calling convention has functions return values of types
9298 @code{float} and @code{double} in an FPU register, even if there
9299 is no FPU@. The idea is that the operating system should emulate
9302 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9303 in ordinary CPU registers instead.
9305 @item -mno-fancy-math-387
9306 @opindex mno-fancy-math-387
9307 Some 387 emulators do not support the @code{sin}, @code{cos} and
9308 @code{sqrt} instructions for the 387. Specify this option to avoid
9309 generating those instructions. This option is the default on FreeBSD,
9310 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9311 indicates that the target cpu will always have an FPU and so the
9312 instruction will not need emulation. As of revision 2.6.1, these
9313 instructions are not generated unless you also use the
9314 @option{-funsafe-math-optimizations} switch.
9316 @item -malign-double
9317 @itemx -mno-align-double
9318 @opindex malign-double
9319 @opindex mno-align-double
9320 Control whether GCC aligns @code{double}, @code{long double}, and
9321 @code{long long} variables on a two word boundary or a one word
9322 boundary. Aligning @code{double} variables on a two word boundary will
9323 produce code that runs somewhat faster on a @samp{Pentium} at the
9324 expense of more memory.
9326 @strong{Warning:} if you use the @option{-malign-double} switch,
9327 structures containing the above types will be aligned differently than
9328 the published application binary interface specifications for the 386
9329 and will not be binary compatible with structures in code compiled
9330 without that switch.
9332 @item -m96bit-long-double
9333 @itemx -m128bit-long-double
9334 @opindex m96bit-long-double
9335 @opindex m128bit-long-double
9336 These switches control the size of @code{long double} type. The i386
9337 application binary interface specifies the size to be 96 bits,
9338 so @option{-m96bit-long-double} is the default in 32 bit mode.
9340 Modern architectures (Pentium and newer) would prefer @code{long double}
9341 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9342 conforming to the ABI, this would not be possible. So specifying a
9343 @option{-m128bit-long-double} will align @code{long double}
9344 to a 16 byte boundary by padding the @code{long double} with an additional
9347 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9348 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9350 Notice that neither of these options enable any extra precision over the x87
9351 standard of 80 bits for a @code{long double}.
9353 @strong{Warning:} if you override the default value for your target ABI, the
9354 structures and arrays containing @code{long double} variables will change
9355 their size as well as function calling convention for function taking
9356 @code{long double} will be modified. Hence they will not be binary
9357 compatible with arrays or structures in code compiled without that switch.
9359 @item -mmlarge-data-threshold=@var{number}
9360 @opindex mlarge-data-threshold=@var{number}
9361 When @option{-mcmodel=medium} is specified, the data greater than
9362 @var{threshold} are placed in large data section. This value must be the
9363 same across all object linked into the binary and defaults to 65535.
9366 @itemx -mno-svr3-shlib
9367 @opindex msvr3-shlib
9368 @opindex mno-svr3-shlib
9369 Control whether GCC places uninitialized local variables into the
9370 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9371 into @code{bss}. These options are meaningful only on System V Release 3.
9375 Use a different function-calling convention, in which functions that
9376 take a fixed number of arguments return with the @code{ret} @var{num}
9377 instruction, which pops their arguments while returning. This saves one
9378 instruction in the caller since there is no need to pop the arguments
9381 You can specify that an individual function is called with this calling
9382 sequence with the function attribute @samp{stdcall}. You can also
9383 override the @option{-mrtd} option by using the function attribute
9384 @samp{cdecl}. @xref{Function Attributes}.
9386 @strong{Warning:} this calling convention is incompatible with the one
9387 normally used on Unix, so you cannot use it if you need to call
9388 libraries compiled with the Unix compiler.
9390 Also, you must provide function prototypes for all functions that
9391 take variable numbers of arguments (including @code{printf});
9392 otherwise incorrect code will be generated for calls to those
9395 In addition, seriously incorrect code will result if you call a
9396 function with too many arguments. (Normally, extra arguments are
9397 harmlessly ignored.)
9399 @item -mregparm=@var{num}
9401 Control how many registers are used to pass integer arguments. By
9402 default, no registers are used to pass arguments, and at most 3
9403 registers can be used. You can control this behavior for a specific
9404 function by using the function attribute @samp{regparm}.
9405 @xref{Function Attributes}.
9407 @strong{Warning:} if you use this switch, and
9408 @var{num} is nonzero, then you must build all modules with the same
9409 value, including any libraries. This includes the system libraries and
9413 @opindex msseregparm
9414 Use SSE register passing conventions for float and double arguments
9415 and return values. You can control this behavior for a specific
9416 function by using the function attribute @samp{sseregparm}.
9417 @xref{Function Attributes}.
9419 @strong{Warning:} if you use this switch then you must build all
9420 modules with the same value, including any libraries. This includes
9421 the system libraries and startup modules.
9423 @item -mpreferred-stack-boundary=@var{num}
9424 @opindex mpreferred-stack-boundary
9425 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9426 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9427 the default is 4 (16 bytes or 128 bits), except when optimizing for code
9428 size (@option{-Os}), in which case the default is the minimum correct
9429 alignment (4 bytes for x86, and 8 bytes for x86-64).
9431 On Pentium and PentiumPro, @code{double} and @code{long double} values
9432 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9433 suffer significant run time performance penalties. On Pentium III, the
9434 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
9435 penalties if it is not 16 byte aligned.
9437 To ensure proper alignment of this values on the stack, the stack boundary
9438 must be as aligned as that required by any value stored on the stack.
9439 Further, every function must be generated such that it keeps the stack
9440 aligned. Thus calling a function compiled with a higher preferred
9441 stack boundary from a function compiled with a lower preferred stack
9442 boundary will most likely misalign the stack. It is recommended that
9443 libraries that use callbacks always use the default setting.
9445 This extra alignment does consume extra stack space, and generally
9446 increases code size. Code that is sensitive to stack space usage, such
9447 as embedded systems and operating system kernels, may want to reduce the
9448 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9466 These switches enable or disable the use of instructions in the MMX,
9467 SSE, SSE2 or 3DNow! extended instruction sets. These extensions are
9468 also available as built-in functions: see @ref{X86 Built-in Functions},
9469 for details of the functions enabled and disabled by these switches.
9471 To have SSE/SSE2 instructions generated automatically from floating-point
9472 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9474 These options will enable GCC to use these extended instructions in
9475 generated code, even without @option{-mfpmath=sse}. Applications which
9476 perform runtime CPU detection must compile separate files for each
9477 supported architecture, using the appropriate flags. In particular,
9478 the file containing the CPU detection code should be compiled without
9483 Use special versions of certain libm routines that come with an SSE
9484 ABI and an SSE implementation. Useful together with @option{-mfpmath=sse}
9485 to avoid moving values between SSE registers and the x87 FP stack.
9488 @itemx -mno-push-args
9490 @opindex mno-push-args
9491 Use PUSH operations to store outgoing parameters. This method is shorter
9492 and usually equally fast as method using SUB/MOV operations and is enabled
9493 by default. In some cases disabling it may improve performance because of
9494 improved scheduling and reduced dependencies.
9496 @item -maccumulate-outgoing-args
9497 @opindex maccumulate-outgoing-args
9498 If enabled, the maximum amount of space required for outgoing arguments will be
9499 computed in the function prologue. This is faster on most modern CPUs
9500 because of reduced dependencies, improved scheduling and reduced stack usage
9501 when preferred stack boundary is not equal to 2. The drawback is a notable
9502 increase in code size. This switch implies @option{-mno-push-args}.
9506 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9507 on thread-safe exception handling must compile and link all code with the
9508 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9509 @option{-D_MT}; when linking, it links in a special thread helper library
9510 @option{-lmingwthrd} which cleans up per thread exception handling data.
9512 @item -mno-align-stringops
9513 @opindex mno-align-stringops
9514 Do not align destination of inlined string operations. This switch reduces
9515 code size and improves performance in case the destination is already aligned,
9516 but GCC doesn't know about it.
9518 @item -minline-all-stringops
9519 @opindex minline-all-stringops
9520 By default GCC inlines string operations only when destination is known to be
9521 aligned at least to 4 byte boundary. This enables more inlining, increase code
9522 size, but may improve performance of code that depends on fast memcpy, strlen
9523 and memset for short lengths.
9525 @item -momit-leaf-frame-pointer
9526 @opindex momit-leaf-frame-pointer
9527 Don't keep the frame pointer in a register for leaf functions. This
9528 avoids the instructions to save, set up and restore frame pointers and
9529 makes an extra register available in leaf functions. The option
9530 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9531 which might make debugging harder.
9533 @item -mtls-direct-seg-refs
9534 @itemx -mno-tls-direct-seg-refs
9535 @opindex mtls-direct-seg-refs
9536 Controls whether TLS variables may be accessed with offsets from the
9537 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9538 or whether the thread base pointer must be added. Whether or not this
9539 is legal depends on the operating system, and whether it maps the
9540 segment to cover the entire TLS area.
9542 For systems that use GNU libc, the default is on.
9545 These @samp{-m} switches are supported in addition to the above
9546 on AMD x86-64 processors in 64-bit environments.
9553 Generate code for a 32-bit or 64-bit environment.
9554 The 32-bit environment sets int, long and pointer to 32 bits and
9555 generates code that runs on any i386 system.
9556 The 64-bit environment sets int to 32 bits and long and pointer
9557 to 64 bits and generates code for AMD's x86-64 architecture.
9560 @opindex no-red-zone
9561 Do not use a so called red zone for x86-64 code. The red zone is mandated
9562 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9563 stack pointer that will not be modified by signal or interrupt handlers
9564 and therefore can be used for temporary data without adjusting the stack
9565 pointer. The flag @option{-mno-red-zone} disables this red zone.
9567 @item -mcmodel=small
9568 @opindex mcmodel=small
9569 Generate code for the small code model: the program and its symbols must
9570 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9571 Programs can be statically or dynamically linked. This is the default
9574 @item -mcmodel=kernel
9575 @opindex mcmodel=kernel
9576 Generate code for the kernel code model. The kernel runs in the
9577 negative 2 GB of the address space.
9578 This model has to be used for Linux kernel code.
9580 @item -mcmodel=medium
9581 @opindex mcmodel=medium
9582 Generate code for the medium model: The program is linked in the lower 2
9583 GB of the address space but symbols can be located anywhere in the
9584 address space. Programs can be statically or dynamically linked, but
9585 building of shared libraries are not supported with the medium model.
9587 @item -mcmodel=large
9588 @opindex mcmodel=large
9589 Generate code for the large model: This model makes no assumptions
9590 about addresses and sizes of sections. Currently GCC does not implement
9595 @subsection IA-64 Options
9596 @cindex IA-64 Options
9598 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9602 @opindex mbig-endian
9603 Generate code for a big endian target. This is the default for HP-UX@.
9605 @item -mlittle-endian
9606 @opindex mlittle-endian
9607 Generate code for a little endian target. This is the default for AIX5
9614 Generate (or don't) code for the GNU assembler. This is the default.
9615 @c Also, this is the default if the configure option @option{--with-gnu-as}
9622 Generate (or don't) code for the GNU linker. This is the default.
9623 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9628 Generate code that does not use a global pointer register. The result
9629 is not position independent code, and violates the IA-64 ABI@.
9631 @item -mvolatile-asm-stop
9632 @itemx -mno-volatile-asm-stop
9633 @opindex mvolatile-asm-stop
9634 @opindex mno-volatile-asm-stop
9635 Generate (or don't) a stop bit immediately before and after volatile asm
9638 @item -mregister-names
9639 @itemx -mno-register-names
9640 @opindex mregister-names
9641 @opindex mno-register-names
9642 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9643 the stacked registers. This may make assembler output more readable.
9649 Disable (or enable) optimizations that use the small data section. This may
9650 be useful for working around optimizer bugs.
9653 @opindex mconstant-gp
9654 Generate code that uses a single constant global pointer value. This is
9655 useful when compiling kernel code.
9659 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9660 This is useful when compiling firmware code.
9662 @item -minline-float-divide-min-latency
9663 @opindex minline-float-divide-min-latency
9664 Generate code for inline divides of floating point values
9665 using the minimum latency algorithm.
9667 @item -minline-float-divide-max-throughput
9668 @opindex minline-float-divide-max-throughput
9669 Generate code for inline divides of floating point values
9670 using the maximum throughput algorithm.
9672 @item -minline-int-divide-min-latency
9673 @opindex minline-int-divide-min-latency
9674 Generate code for inline divides of integer values
9675 using the minimum latency algorithm.
9677 @item -minline-int-divide-max-throughput
9678 @opindex minline-int-divide-max-throughput
9679 Generate code for inline divides of integer values
9680 using the maximum throughput algorithm.
9682 @item -minline-sqrt-min-latency
9683 @opindex minline-sqrt-min-latency
9684 Generate code for inline square roots
9685 using the minimum latency algorithm.
9687 @item -minline-sqrt-max-throughput
9688 @opindex minline-sqrt-max-throughput
9689 Generate code for inline square roots
9690 using the maximum throughput algorithm.
9692 @item -mno-dwarf2-asm
9694 @opindex mno-dwarf2-asm
9695 @opindex mdwarf2-asm
9696 Don't (or do) generate assembler code for the DWARF2 line number debugging
9697 info. This may be useful when not using the GNU assembler.
9699 @item -mearly-stop-bits
9700 @itemx -mno-early-stop-bits
9701 @opindex mearly-stop-bits
9702 @opindex mno-early-stop-bits
9703 Allow stop bits to be placed earlier than immediately preceding the
9704 instruction that triggered the stop bit. This can improve instruction
9705 scheduling, but does not always do so.
9707 @item -mfixed-range=@var{register-range}
9708 @opindex mfixed-range
9709 Generate code treating the given register range as fixed registers.
9710 A fixed register is one that the register allocator can not use. This is
9711 useful when compiling kernel code. A register range is specified as
9712 two registers separated by a dash. Multiple register ranges can be
9713 specified separated by a comma.
9715 @item -mtls-size=@var{tls-size}
9717 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9720 @item -mtune=@var{cpu-type}
9722 Tune the instruction scheduling for a particular CPU, Valid values are
9723 itanium, itanium1, merced, itanium2, and mckinley.
9729 Add support for multithreading using the POSIX threads library. This
9730 option sets flags for both the preprocessor and linker. It does
9731 not affect the thread safety of object code produced by the compiler or
9732 that of libraries supplied with it. These are HP-UX specific flags.
9738 Generate code for a 32-bit or 64-bit environment.
9739 The 32-bit environment sets int, long and pointer to 32 bits.
9740 The 64-bit environment sets int to 32 bits and long and pointer
9741 to 64 bits. These are HP-UX specific flags.
9743 @item -mno-sched-br-data-spec
9744 @itemx -msched-br-data-spec
9745 @opindex -mno-sched-br-data-spec
9746 @opindex -msched-br-data-spec
9747 (Dis/En)able data speculative scheduling before reload.
9748 This will result in generation of the ld.a instructions and
9749 the corresponding check instructions (ld.c / chk.a).
9750 The default is 'disable'.
9752 @item -msched-ar-data-spec
9753 @itemx -mno-sched-ar-data-spec
9754 @opindex -msched-ar-data-spec
9755 @opindex -mno-sched-ar-data-spec
9756 (En/Dis)able data speculative scheduling after reload.
9757 This will result in generation of the ld.a instructions and
9758 the corresponding check instructions (ld.c / chk.a).
9759 The default is 'enable'.
9761 @item -mno-sched-control-spec
9762 @itemx -msched-control-spec
9763 @opindex -mno-sched-control-spec
9764 @opindex -msched-control-spec
9765 (Dis/En)able control speculative scheduling. This feature is
9766 available only during region scheduling (i.e. before reload).
9767 This will result in generation of the ld.s instructions and
9768 the corresponding check instructions chk.s .
9769 The default is 'disable'.
9771 @item -msched-br-in-data-spec
9772 @itemx -mno-sched-br-in-data-spec
9773 @opindex -msched-br-in-data-spec
9774 @opindex -mno-sched-br-in-data-spec
9775 (En/Dis)able speculative scheduling of the instructions that
9776 are dependent on the data speculative loads before reload.
9777 This is effective only with @option{-msched-br-data-spec} enabled.
9778 The default is 'enable'.
9780 @item -msched-ar-in-data-spec
9781 @itemx -mno-sched-ar-in-data-spec
9782 @opindex -msched-ar-in-data-spec
9783 @opindex -mno-sched-ar-in-data-spec
9784 (En/Dis)able speculative scheduling of the instructions that
9785 are dependent on the data speculative loads after reload.
9786 This is effective only with @option{-msched-ar-data-spec} enabled.
9787 The default is 'enable'.
9789 @item -msched-in-control-spec
9790 @itemx -mno-sched-in-control-spec
9791 @opindex -msched-in-control-spec
9792 @opindex -mno-sched-in-control-spec
9793 (En/Dis)able speculative scheduling of the instructions that
9794 are dependent on the control speculative loads.
9795 This is effective only with @option{-msched-control-spec} enabled.
9796 The default is 'enable'.
9799 @itemx -mno-sched-ldc
9800 @opindex -msched-ldc
9801 @opindex -mno-sched-ldc
9802 (En/Dis)able use of simple data speculation checks ld.c .
9803 If disabled, only chk.a instructions will be emitted to check
9804 data speculative loads.
9805 The default is 'enable'.
9807 @item -mno-sched-control-ldc
9808 @itemx -msched-control-ldc
9809 @opindex -mno-sched-control-ldc
9810 @opindex -msched-control-ldc
9811 (Dis/En)able use of ld.c instructions to check control speculative loads.
9812 If enabled, in case of control speculative load with no speculatively
9813 scheduled dependent instructions this load will be emitted as ld.sa and
9814 ld.c will be used to check it.
9815 The default is 'disable'.
9817 @item -mno-sched-spec-verbose
9818 @itemx -msched-spec-verbose
9819 @opindex -mno-sched-spec-verbose
9820 @opindex -msched-spec-verbose
9821 (Dis/En)able printing of the information about speculative motions.
9823 @item -mno-sched-prefer-non-data-spec-insns
9824 @itemx -msched-prefer-non-data-spec-insns
9825 @opindex -mno-sched-prefer-non-data-spec-insns
9826 @opindex -msched-prefer-non-data-spec-insns
9827 If enabled, data speculative instructions will be choosen for schedule
9828 only if there are no other choices at the moment. This will make
9829 the use of the data speculation much more conservative.
9830 The default is 'disable'.
9832 @item -mno-sched-prefer-non-control-spec-insns
9833 @itemx -msched-prefer-non-control-spec-insns
9834 @opindex -mno-sched-prefer-non-control-spec-insns
9835 @opindex -msched-prefer-non-control-spec-insns
9836 If enabled, control speculative instructions will be choosen for schedule
9837 only if there are no other choices at the moment. This will make
9838 the use of the control speculation much more conservative.
9839 The default is 'disable'.
9841 @item -mno-sched-count-spec-in-critical-path
9842 @itemx -msched-count-spec-in-critical-path
9843 @opindex -mno-sched-count-spec-in-critical-path
9844 @opindex -msched-count-spec-in-critical-path
9845 If enabled, speculative depedencies will be considered during
9846 computation of the instructions priorities. This will make the use of the
9847 speculation a bit more conservative.
9848 The default is 'disable'.
9853 @subsection M32C Options
9854 @cindex M32C options
9857 @item -mcpu=@var{name}
9859 Select the CPU for which code is generated. @var{name} may be one of
9860 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9861 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9866 Specifies that the program will be run on the simulator. This causes
9867 an alternate runtime library to be linked in which supports, for
9868 example, file I/O. You must not use this option when generating
9869 programs that will run on real hardware; you must provide your own
9870 runtime library for whatever I/O functions are needed.
9872 @item -memregs=@var{number}
9874 Specifies the number of memory-based pseudo-registers GCC will use
9875 during code generation. These pseudo-registers will be used like real
9876 registers, so there is a tradeoff between GCC's ability to fit the
9877 code into available registers, and the performance penalty of using
9878 memory instead of registers. Note that all modules in a program must
9879 be compiled with the same value for this option. Because of that, you
9880 must not use this option with the default runtime libraries gcc
9885 @node M32R/D Options
9886 @subsection M32R/D Options
9887 @cindex M32R/D options
9889 These @option{-m} options are defined for Renesas M32R/D architectures:
9894 Generate code for the M32R/2@.
9898 Generate code for the M32R/X@.
9902 Generate code for the M32R@. This is the default.
9905 @opindex mmodel=small
9906 Assume all objects live in the lower 16MB of memory (so that their addresses
9907 can be loaded with the @code{ld24} instruction), and assume all subroutines
9908 are reachable with the @code{bl} instruction.
9909 This is the default.
9911 The addressability of a particular object can be set with the
9912 @code{model} attribute.
9914 @item -mmodel=medium
9915 @opindex mmodel=medium
9916 Assume objects may be anywhere in the 32-bit address space (the compiler
9917 will generate @code{seth/add3} instructions to load their addresses), and
9918 assume all subroutines are reachable with the @code{bl} instruction.
9921 @opindex mmodel=large
9922 Assume objects may be anywhere in the 32-bit address space (the compiler
9923 will generate @code{seth/add3} instructions to load their addresses), and
9924 assume subroutines may not be reachable with the @code{bl} instruction
9925 (the compiler will generate the much slower @code{seth/add3/jl}
9926 instruction sequence).
9929 @opindex msdata=none
9930 Disable use of the small data area. Variables will be put into
9931 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
9932 @code{section} attribute has been specified).
9933 This is the default.
9935 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
9936 Objects may be explicitly put in the small data area with the
9937 @code{section} attribute using one of these sections.
9940 @opindex msdata=sdata
9941 Put small global and static data in the small data area, but do not
9942 generate special code to reference them.
9946 Put small global and static data in the small data area, and generate
9947 special instructions to reference them.
9951 @cindex smaller data references
9952 Put global and static objects less than or equal to @var{num} bytes
9953 into the small data or bss sections instead of the normal data or bss
9954 sections. The default value of @var{num} is 8.
9955 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
9956 for this option to have any effect.
9958 All modules should be compiled with the same @option{-G @var{num}} value.
9959 Compiling with different values of @var{num} may or may not work; if it
9960 doesn't the linker will give an error message---incorrect code will not be
9965 Makes the M32R specific code in the compiler display some statistics
9966 that might help in debugging programs.
9969 @opindex malign-loops
9970 Align all loops to a 32-byte boundary.
9972 @item -mno-align-loops
9973 @opindex mno-align-loops
9974 Do not enforce a 32-byte alignment for loops. This is the default.
9976 @item -missue-rate=@var{number}
9977 @opindex missue-rate=@var{number}
9978 Issue @var{number} instructions per cycle. @var{number} can only be 1
9981 @item -mbranch-cost=@var{number}
9982 @opindex mbranch-cost=@var{number}
9983 @var{number} can only be 1 or 2. If it is 1 then branches will be
9984 preferred over conditional code, if it is 2, then the opposite will
9987 @item -mflush-trap=@var{number}
9988 @opindex mflush-trap=@var{number}
9989 Specifies the trap number to use to flush the cache. The default is
9990 12. Valid numbers are between 0 and 15 inclusive.
9992 @item -mno-flush-trap
9993 @opindex mno-flush-trap
9994 Specifies that the cache cannot be flushed by using a trap.
9996 @item -mflush-func=@var{name}
9997 @opindex mflush-func=@var{name}
9998 Specifies the name of the operating system function to call to flush
9999 the cache. The default is @emph{_flush_cache}, but a function call
10000 will only be used if a trap is not available.
10002 @item -mno-flush-func
10003 @opindex mno-flush-func
10004 Indicates that there is no OS function for flushing the cache.
10008 @node M680x0 Options
10009 @subsection M680x0 Options
10010 @cindex M680x0 options
10012 These are the @samp{-m} options defined for the 68000 series. The default
10013 values for these options depends on which style of 68000 was selected when
10014 the compiler was configured; the defaults for the most common choices are
10022 Generate output for a 68000. This is the default
10023 when the compiler is configured for 68000-based systems.
10025 Use this option for microcontrollers with a 68000 or EC000 core,
10026 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10032 Generate output for a 68020. This is the default
10033 when the compiler is configured for 68020-based systems.
10037 Generate output containing 68881 instructions for floating point.
10038 This is the default for most 68020 systems unless @option{--nfp} was
10039 specified when the compiler was configured.
10043 Generate output for a 68030. This is the default when the compiler is
10044 configured for 68030-based systems.
10048 Generate output for a 68040. This is the default when the compiler is
10049 configured for 68040-based systems.
10051 This option inhibits the use of 68881/68882 instructions that have to be
10052 emulated by software on the 68040. Use this option if your 68040 does not
10053 have code to emulate those instructions.
10057 Generate output for a 68060. This is the default when the compiler is
10058 configured for 68060-based systems.
10060 This option inhibits the use of 68020 and 68881/68882 instructions that
10061 have to be emulated by software on the 68060. Use this option if your 68060
10062 does not have code to emulate those instructions.
10066 Generate output for a CPU32. This is the default
10067 when the compiler is configured for CPU32-based systems.
10069 Use this option for microcontrollers with a
10070 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10071 68336, 68340, 68341, 68349 and 68360.
10075 Generate output for a 520X ``coldfire'' family cpu. This is the default
10076 when the compiler is configured for 520X-based systems.
10078 Use this option for microcontroller with a 5200 core, including
10079 the MCF5202, MCF5203, MCF5204 and MCF5202.
10083 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
10084 This includes use of hardware floating point instructions.
10088 Generate output for a 68040, without using any of the new instructions.
10089 This results in code which can run relatively efficiently on either a
10090 68020/68881 or a 68030 or a 68040. The generated code does use the
10091 68881 instructions that are emulated on the 68040.
10095 Generate output for a 68060, without using any of the new instructions.
10096 This results in code which can run relatively efficiently on either a
10097 68020/68881 or a 68030 or a 68040. The generated code does use the
10098 68881 instructions that are emulated on the 68060.
10101 @opindex msoft-float
10102 Generate output containing library calls for floating point.
10103 @strong{Warning:} the requisite libraries are not available for all m68k
10104 targets. Normally the facilities of the machine's usual C compiler are
10105 used, but this can't be done directly in cross-compilation. You must
10106 make your own arrangements to provide suitable library functions for
10107 cross-compilation. The embedded targets @samp{m68k-*-aout} and
10108 @samp{m68k-*-coff} do provide software floating point support.
10112 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10113 Additionally, parameters passed on the stack are also aligned to a
10114 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10117 @opindex mnobitfield
10118 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10119 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10123 Do use the bit-field instructions. The @option{-m68020} option implies
10124 @option{-mbitfield}. This is the default if you use a configuration
10125 designed for a 68020.
10129 Use a different function-calling convention, in which functions
10130 that take a fixed number of arguments return with the @code{rtd}
10131 instruction, which pops their arguments while returning. This
10132 saves one instruction in the caller since there is no need to pop
10133 the arguments there.
10135 This calling convention is incompatible with the one normally
10136 used on Unix, so you cannot use it if you need to call libraries
10137 compiled with the Unix compiler.
10139 Also, you must provide function prototypes for all functions that
10140 take variable numbers of arguments (including @code{printf});
10141 otherwise incorrect code will be generated for calls to those
10144 In addition, seriously incorrect code will result if you call a
10145 function with too many arguments. (Normally, extra arguments are
10146 harmlessly ignored.)
10148 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10149 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10152 @itemx -mno-align-int
10153 @opindex malign-int
10154 @opindex mno-align-int
10155 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10156 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10157 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10158 Aligning variables on 32-bit boundaries produces code that runs somewhat
10159 faster on processors with 32-bit busses at the expense of more memory.
10161 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10162 align structures containing the above types differently than
10163 most published application binary interface specifications for the m68k.
10167 Use the pc-relative addressing mode of the 68000 directly, instead of
10168 using a global offset table. At present, this option implies @option{-fpic},
10169 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10170 not presently supported with @option{-mpcrel}, though this could be supported for
10171 68020 and higher processors.
10173 @item -mno-strict-align
10174 @itemx -mstrict-align
10175 @opindex mno-strict-align
10176 @opindex mstrict-align
10177 Do not (do) assume that unaligned memory references will be handled by
10181 Generate code that allows the data segment to be located in a different
10182 area of memory from the text segment. This allows for execute in place in
10183 an environment without virtual memory management. This option implies
10186 @item -mno-sep-data
10187 Generate code that assumes that the data segment follows the text segment.
10188 This is the default.
10190 @item -mid-shared-library
10191 Generate code that supports shared libraries via the library ID method.
10192 This allows for execute in place and shared libraries in an environment
10193 without virtual memory management. This option implies @option{-fPIC}.
10195 @item -mno-id-shared-library
10196 Generate code that doesn't assume ID based shared libraries are being used.
10197 This is the default.
10199 @item -mshared-library-id=n
10200 Specified the identification number of the ID based shared library being
10201 compiled. Specifying a value of 0 will generate more compact code, specifying
10202 other values will force the allocation of that number to the current
10203 library but is no more space or time efficient than omitting this option.
10207 @node M68hc1x Options
10208 @subsection M68hc1x Options
10209 @cindex M68hc1x options
10211 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10212 microcontrollers. The default values for these options depends on
10213 which style of microcontroller was selected when the compiler was configured;
10214 the defaults for the most common choices are given below.
10221 Generate output for a 68HC11. This is the default
10222 when the compiler is configured for 68HC11-based systems.
10228 Generate output for a 68HC12. This is the default
10229 when the compiler is configured for 68HC12-based systems.
10235 Generate output for a 68HCS12.
10237 @item -mauto-incdec
10238 @opindex mauto-incdec
10239 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10246 Enable the use of 68HC12 min and max instructions.
10249 @itemx -mno-long-calls
10250 @opindex mlong-calls
10251 @opindex mno-long-calls
10252 Treat all calls as being far away (near). If calls are assumed to be
10253 far away, the compiler will use the @code{call} instruction to
10254 call a function and the @code{rtc} instruction for returning.
10258 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10260 @item -msoft-reg-count=@var{count}
10261 @opindex msoft-reg-count
10262 Specify the number of pseudo-soft registers which are used for the
10263 code generation. The maximum number is 32. Using more pseudo-soft
10264 register may or may not result in better code depending on the program.
10265 The default is 4 for 68HC11 and 2 for 68HC12.
10269 @node MCore Options
10270 @subsection MCore Options
10271 @cindex MCore options
10273 These are the @samp{-m} options defined for the Motorola M*Core
10279 @itemx -mno-hardlit
10281 @opindex mno-hardlit
10282 Inline constants into the code stream if it can be done in two
10283 instructions or less.
10289 Use the divide instruction. (Enabled by default).
10291 @item -mrelax-immediate
10292 @itemx -mno-relax-immediate
10293 @opindex mrelax-immediate
10294 @opindex mno-relax-immediate
10295 Allow arbitrary sized immediates in bit operations.
10297 @item -mwide-bitfields
10298 @itemx -mno-wide-bitfields
10299 @opindex mwide-bitfields
10300 @opindex mno-wide-bitfields
10301 Always treat bit-fields as int-sized.
10303 @item -m4byte-functions
10304 @itemx -mno-4byte-functions
10305 @opindex m4byte-functions
10306 @opindex mno-4byte-functions
10307 Force all functions to be aligned to a four byte boundary.
10309 @item -mcallgraph-data
10310 @itemx -mno-callgraph-data
10311 @opindex mcallgraph-data
10312 @opindex mno-callgraph-data
10313 Emit callgraph information.
10316 @itemx -mno-slow-bytes
10317 @opindex mslow-bytes
10318 @opindex mno-slow-bytes
10319 Prefer word access when reading byte quantities.
10321 @item -mlittle-endian
10322 @itemx -mbig-endian
10323 @opindex mlittle-endian
10324 @opindex mbig-endian
10325 Generate code for a little endian target.
10331 Generate code for the 210 processor.
10335 @subsection MIPS Options
10336 @cindex MIPS options
10342 Generate big-endian code.
10346 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10349 @item -march=@var{arch}
10351 Generate code that will run on @var{arch}, which can be the name of a
10352 generic MIPS ISA, or the name of a particular processor.
10354 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10355 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10356 The processor names are:
10357 @samp{4kc}, @samp{4km}, @samp{4kp},
10358 @samp{5kc}, @samp{5kf},
10360 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10363 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10364 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10365 @samp{rm7000}, @samp{rm9000},
10368 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10369 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10370 The special value @samp{from-abi} selects the
10371 most compatible architecture for the selected ABI (that is,
10372 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10374 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10375 (for example, @samp{-march=r2k}). Prefixes are optional, and
10376 @samp{vr} may be written @samp{r}.
10378 GCC defines two macros based on the value of this option. The first
10379 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10380 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10381 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10382 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10383 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10385 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10386 above. In other words, it will have the full prefix and will not
10387 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10388 the macro names the resolved architecture (either @samp{"mips1"} or
10389 @samp{"mips3"}). It names the default architecture when no
10390 @option{-march} option is given.
10392 @item -mtune=@var{arch}
10394 Optimize for @var{arch}. Among other things, this option controls
10395 the way instructions are scheduled, and the perceived cost of arithmetic
10396 operations. The list of @var{arch} values is the same as for
10399 When this option is not used, GCC will optimize for the processor
10400 specified by @option{-march}. By using @option{-march} and
10401 @option{-mtune} together, it is possible to generate code that will
10402 run on a family of processors, but optimize the code for one
10403 particular member of that family.
10405 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10406 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10407 @samp{-march} ones described above.
10411 Equivalent to @samp{-march=mips1}.
10415 Equivalent to @samp{-march=mips2}.
10419 Equivalent to @samp{-march=mips3}.
10423 Equivalent to @samp{-march=mips4}.
10427 Equivalent to @samp{-march=mips32}.
10431 Equivalent to @samp{-march=mips32r2}.
10435 Equivalent to @samp{-march=mips64}.
10440 @opindex mno-mips16
10441 Generate (do not generate) MIPS16 code. If GCC is targetting a
10442 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10454 Generate code for the given ABI@.
10456 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10457 generates 64-bit code when you select a 64-bit architecture, but you
10458 can use @option{-mgp32} to get 32-bit code instead.
10460 For information about the O64 ABI, see
10461 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10464 @itemx -mno-abicalls
10466 @opindex mno-abicalls
10467 Generate (do not generate) SVR4-style position-independent code.
10468 @option{-mabicalls} is the default for SVR4-based systems.
10474 Lift (do not lift) the usual restrictions on the size of the global
10477 GCC normally uses a single instruction to load values from the GOT@.
10478 While this is relatively efficient, it will only work if the GOT
10479 is smaller than about 64k. Anything larger will cause the linker
10480 to report an error such as:
10482 @cindex relocation truncated to fit (MIPS)
10484 relocation truncated to fit: R_MIPS_GOT16 foobar
10487 If this happens, you should recompile your code with @option{-mxgot}.
10488 It should then work with very large GOTs, although it will also be
10489 less efficient, since it will take three instructions to fetch the
10490 value of a global symbol.
10492 Note that some linkers can create multiple GOTs. If you have such a
10493 linker, you should only need to use @option{-mxgot} when a single object
10494 file accesses more than 64k's worth of GOT entries. Very few do.
10496 These options have no effect unless GCC is generating position
10501 Assume that general-purpose registers are 32 bits wide.
10505 Assume that general-purpose registers are 64 bits wide.
10509 Assume that floating-point registers are 32 bits wide.
10513 Assume that floating-point registers are 64 bits wide.
10516 @opindex mhard-float
10517 Use floating-point coprocessor instructions.
10520 @opindex msoft-float
10521 Do not use floating-point coprocessor instructions. Implement
10522 floating-point calculations using library calls instead.
10524 @item -msingle-float
10525 @opindex msingle-float
10526 Assume that the floating-point coprocessor only supports single-precision
10529 @itemx -mdouble-float
10530 @opindex mdouble-float
10531 Assume that the floating-point coprocessor supports double-precision
10532 operations. This is the default.
10538 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10540 @itemx -mpaired-single
10541 @itemx -mno-paired-single
10542 @opindex mpaired-single
10543 @opindex mno-paired-single
10544 Use (do not use) paired-single floating-point instructions.
10545 @xref{MIPS Paired-Single Support}. This option can only be used
10546 when generating 64-bit code and requires hardware floating-point
10547 support to be enabled.
10552 @opindex mno-mips3d
10553 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10554 The option @option{-mips3d} implies @option{-mpaired-single}.
10558 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10559 an explanation of the default and the way that the pointer size is
10564 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10566 The default size of @code{int}s, @code{long}s and pointers depends on
10567 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10568 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10569 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10570 or the same size as integer registers, whichever is smaller.
10576 Assume (do not assume) that all symbols have 32-bit values, regardless
10577 of the selected ABI@. This option is useful in combination with
10578 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10579 to generate shorter and faster references to symbolic addresses.
10583 @cindex smaller data references (MIPS)
10584 @cindex gp-relative references (MIPS)
10585 Put global and static items less than or equal to @var{num} bytes into
10586 the small data or bss section instead of the normal data or bss section.
10587 This allows the data to be accessed using a single instruction.
10589 All modules should be compiled with the same @option{-G @var{num}}
10592 @item -membedded-data
10593 @itemx -mno-embedded-data
10594 @opindex membedded-data
10595 @opindex mno-embedded-data
10596 Allocate variables to the read-only data section first if possible, then
10597 next in the small data section if possible, otherwise in data. This gives
10598 slightly slower code than the default, but reduces the amount of RAM required
10599 when executing, and thus may be preferred for some embedded systems.
10601 @item -muninit-const-in-rodata
10602 @itemx -mno-uninit-const-in-rodata
10603 @opindex muninit-const-in-rodata
10604 @opindex mno-uninit-const-in-rodata
10605 Put uninitialized @code{const} variables in the read-only data section.
10606 This option is only meaningful in conjunction with @option{-membedded-data}.
10608 @item -msplit-addresses
10609 @itemx -mno-split-addresses
10610 @opindex msplit-addresses
10611 @opindex mno-split-addresses
10612 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10613 relocation operators. This option has been superseded by
10614 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10616 @item -mexplicit-relocs
10617 @itemx -mno-explicit-relocs
10618 @opindex mexplicit-relocs
10619 @opindex mno-explicit-relocs
10620 Use (do not use) assembler relocation operators when dealing with symbolic
10621 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10622 is to use assembler macros instead.
10624 @option{-mexplicit-relocs} is the default if GCC was configured
10625 to use an assembler that supports relocation operators.
10627 @item -mcheck-zero-division
10628 @itemx -mno-check-zero-division
10629 @opindex mcheck-zero-division
10630 @opindex mno-check-zero-division
10631 Trap (do not trap) on integer division by zero. The default is
10632 @option{-mcheck-zero-division}.
10634 @item -mdivide-traps
10635 @itemx -mdivide-breaks
10636 @opindex mdivide-traps
10637 @opindex mdivide-breaks
10638 MIPS systems check for division by zero by generating either a
10639 conditional trap or a break instruction. Using traps results in
10640 smaller code, but is only supported on MIPS II and later. Also, some
10641 versions of the Linux kernel have a bug that prevents trap from
10642 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10643 allow conditional traps on architectures that support them and
10644 @option{-mdivide-breaks} to force the use of breaks.
10646 The default is usually @option{-mdivide-traps}, but this can be
10647 overridden at configure time using @option{--with-divide=breaks}.
10648 Divide-by-zero checks can be completely disabled using
10649 @option{-mno-check-zero-division}.
10654 @opindex mno-memcpy
10655 Force (do not force) the use of @code{memcpy()} for non-trivial block
10656 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10657 most constant-sized copies.
10660 @itemx -mno-long-calls
10661 @opindex mlong-calls
10662 @opindex mno-long-calls
10663 Disable (do not disable) use of the @code{jal} instruction. Calling
10664 functions using @code{jal} is more efficient but requires the caller
10665 and callee to be in the same 256 megabyte segment.
10667 This option has no effect on abicalls code. The default is
10668 @option{-mno-long-calls}.
10674 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10675 instructions, as provided by the R4650 ISA@.
10678 @itemx -mno-fused-madd
10679 @opindex mfused-madd
10680 @opindex mno-fused-madd
10681 Enable (disable) use of the floating point multiply-accumulate
10682 instructions, when they are available. The default is
10683 @option{-mfused-madd}.
10685 When multiply-accumulate instructions are used, the intermediate
10686 product is calculated to infinite precision and is not subject to
10687 the FCSR Flush to Zero bit. This may be undesirable in some
10692 Tell the MIPS assembler to not run its preprocessor over user
10693 assembler files (with a @samp{.s} suffix) when assembling them.
10696 @itemx -mno-fix-r4000
10697 @opindex mfix-r4000
10698 @opindex mno-fix-r4000
10699 Work around certain R4000 CPU errata:
10702 A double-word or a variable shift may give an incorrect result if executed
10703 immediately after starting an integer division.
10705 A double-word or a variable shift may give an incorrect result if executed
10706 while an integer multiplication is in progress.
10708 An integer division may give an incorrect result if started in a delay slot
10709 of a taken branch or a jump.
10713 @itemx -mno-fix-r4400
10714 @opindex mfix-r4400
10715 @opindex mno-fix-r4400
10716 Work around certain R4400 CPU errata:
10719 A double-word or a variable shift may give an incorrect result if executed
10720 immediately after starting an integer division.
10724 @itemx -mno-fix-vr4120
10725 @opindex mfix-vr4120
10726 Work around certain VR4120 errata:
10729 @code{dmultu} does not always produce the correct result.
10731 @code{div} and @code{ddiv} do not always produce the correct result if one
10732 of the operands is negative.
10734 The workarounds for the division errata rely on special functions in
10735 @file{libgcc.a}. At present, these functions are only provided by
10736 the @code{mips64vr*-elf} configurations.
10738 Other VR4120 errata require a nop to be inserted between certain pairs of
10739 instructions. These errata are handled by the assembler, not by GCC itself.
10742 @opindex mfix-vr4130
10743 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10744 workarounds are implemented by the assembler rather than by GCC,
10745 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10746 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10747 instructions are available instead.
10750 @itemx -mno-fix-sb1
10752 Work around certain SB-1 CPU core errata.
10753 (This flag currently works around the SB-1 revision 2
10754 ``F1'' and ``F2'' floating point errata.)
10756 @item -mflush-func=@var{func}
10757 @itemx -mno-flush-func
10758 @opindex mflush-func
10759 Specifies the function to call to flush the I and D caches, or to not
10760 call any such function. If called, the function must take the same
10761 arguments as the common @code{_flush_func()}, that is, the address of the
10762 memory range for which the cache is being flushed, the size of the
10763 memory range, and the number 3 (to flush both caches). The default
10764 depends on the target GCC was configured for, but commonly is either
10765 @samp{_flush_func} or @samp{__cpu_flush}.
10767 @item -mbranch-likely
10768 @itemx -mno-branch-likely
10769 @opindex mbranch-likely
10770 @opindex mno-branch-likely
10771 Enable or disable use of Branch Likely instructions, regardless of the
10772 default for the selected architecture. By default, Branch Likely
10773 instructions may be generated if they are supported by the selected
10774 architecture. An exception is for the MIPS32 and MIPS64 architectures
10775 and processors which implement those architectures; for those, Branch
10776 Likely instructions will not be generated by default because the MIPS32
10777 and MIPS64 architectures specifically deprecate their use.
10779 @item -mfp-exceptions
10780 @itemx -mno-fp-exceptions
10781 @opindex mfp-exceptions
10782 Specifies whether FP exceptions are enabled. This affects how we schedule
10783 FP instructions for some processors. The default is that FP exceptions are
10786 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10787 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10790 @item -mvr4130-align
10791 @itemx -mno-vr4130-align
10792 @opindex mvr4130-align
10793 The VR4130 pipeline is two-way superscalar, but can only issue two
10794 instructions together if the first one is 8-byte aligned. When this
10795 option is enabled, GCC will align pairs of instructions that it
10796 thinks should execute in parallel.
10798 This option only has an effect when optimizing for the VR4130.
10799 It normally makes code faster, but at the expense of making it bigger.
10800 It is enabled by default at optimization level @option{-O3}.
10804 @subsection MMIX Options
10805 @cindex MMIX Options
10807 These options are defined for the MMIX:
10811 @itemx -mno-libfuncs
10813 @opindex mno-libfuncs
10814 Specify that intrinsic library functions are being compiled, passing all
10815 values in registers, no matter the size.
10818 @itemx -mno-epsilon
10820 @opindex mno-epsilon
10821 Generate floating-point comparison instructions that compare with respect
10822 to the @code{rE} epsilon register.
10824 @item -mabi=mmixware
10826 @opindex mabi-mmixware
10828 Generate code that passes function parameters and return values that (in
10829 the called function) are seen as registers @code{$0} and up, as opposed to
10830 the GNU ABI which uses global registers @code{$231} and up.
10832 @item -mzero-extend
10833 @itemx -mno-zero-extend
10834 @opindex mzero-extend
10835 @opindex mno-zero-extend
10836 When reading data from memory in sizes shorter than 64 bits, use (do not
10837 use) zero-extending load instructions by default, rather than
10838 sign-extending ones.
10841 @itemx -mno-knuthdiv
10843 @opindex mno-knuthdiv
10844 Make the result of a division yielding a remainder have the same sign as
10845 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10846 remainder follows the sign of the dividend. Both methods are
10847 arithmetically valid, the latter being almost exclusively used.
10849 @item -mtoplevel-symbols
10850 @itemx -mno-toplevel-symbols
10851 @opindex mtoplevel-symbols
10852 @opindex mno-toplevel-symbols
10853 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10854 code can be used with the @code{PREFIX} assembly directive.
10858 Generate an executable in the ELF format, rather than the default
10859 @samp{mmo} format used by the @command{mmix} simulator.
10861 @item -mbranch-predict
10862 @itemx -mno-branch-predict
10863 @opindex mbranch-predict
10864 @opindex mno-branch-predict
10865 Use (do not use) the probable-branch instructions, when static branch
10866 prediction indicates a probable branch.
10868 @item -mbase-addresses
10869 @itemx -mno-base-addresses
10870 @opindex mbase-addresses
10871 @opindex mno-base-addresses
10872 Generate (do not generate) code that uses @emph{base addresses}. Using a
10873 base address automatically generates a request (handled by the assembler
10874 and the linker) for a constant to be set up in a global register. The
10875 register is used for one or more base address requests within the range 0
10876 to 255 from the value held in the register. The generally leads to short
10877 and fast code, but the number of different data items that can be
10878 addressed is limited. This means that a program that uses lots of static
10879 data may require @option{-mno-base-addresses}.
10881 @item -msingle-exit
10882 @itemx -mno-single-exit
10883 @opindex msingle-exit
10884 @opindex mno-single-exit
10885 Force (do not force) generated code to have a single exit point in each
10889 @node MN10300 Options
10890 @subsection MN10300 Options
10891 @cindex MN10300 options
10893 These @option{-m} options are defined for Matsushita MN10300 architectures:
10898 Generate code to avoid bugs in the multiply instructions for the MN10300
10899 processors. This is the default.
10901 @item -mno-mult-bug
10902 @opindex mno-mult-bug
10903 Do not generate code to avoid bugs in the multiply instructions for the
10904 MN10300 processors.
10908 Generate code which uses features specific to the AM33 processor.
10912 Do not generate code which uses features specific to the AM33 processor. This
10915 @item -mreturn-pointer-on-d0
10916 @opindex mreturn-pointer-on-d0
10917 When generating a function which returns a pointer, return the pointer
10918 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
10919 only in a0, and attempts to call such functions without a prototype
10920 would result in errors. Note that this option is on by default; use
10921 @option{-mno-return-pointer-on-d0} to disable it.
10925 Do not link in the C run-time initialization object file.
10929 Indicate to the linker that it should perform a relaxation optimization pass
10930 to shorten branches, calls and absolute memory addresses. This option only
10931 has an effect when used on the command line for the final link step.
10933 This option makes symbolic debugging impossible.
10937 @subsection MT Options
10940 These @option{-m} options are defined for Morpho MT architectures:
10944 @item -march=@var{cpu-type}
10946 Generate code that will run on @var{cpu-type}, which is the name of a system
10947 representing a certain processor type. Possible values for
10948 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
10949 @samp{ms1-16-003} and @samp{ms2}.
10951 When this option is not used, the default is @option{-march=ms1-16-002}.
10955 Use byte loads and stores when generating code.
10959 Do not use byte loads and stores when generating code.
10963 Use simulator runtime
10967 Do not link in the C run-time initialization object file
10968 @file{crti.o}. Other run-time initialization and termination files
10969 such as @file{startup.o} and @file{exit.o} are still included on the
10970 linker command line.
10974 @node PDP-11 Options
10975 @subsection PDP-11 Options
10976 @cindex PDP-11 Options
10978 These options are defined for the PDP-11:
10983 Use hardware FPP floating point. This is the default. (FIS floating
10984 point on the PDP-11/40 is not supported.)
10987 @opindex msoft-float
10988 Do not use hardware floating point.
10992 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10996 Return floating-point results in memory. This is the default.
11000 Generate code for a PDP-11/40.
11004 Generate code for a PDP-11/45. This is the default.
11008 Generate code for a PDP-11/10.
11010 @item -mbcopy-builtin
11011 @opindex bcopy-builtin
11012 Use inline @code{movmemhi} patterns for copying memory. This is the
11017 Do not use inline @code{movmemhi} patterns for copying memory.
11023 Use 16-bit @code{int}. This is the default.
11029 Use 32-bit @code{int}.
11032 @itemx -mno-float32
11034 @opindex mno-float32
11035 Use 64-bit @code{float}. This is the default.
11038 @itemx -mno-float64
11040 @opindex mno-float64
11041 Use 32-bit @code{float}.
11045 Use @code{abshi2} pattern. This is the default.
11049 Do not use @code{abshi2} pattern.
11051 @item -mbranch-expensive
11052 @opindex mbranch-expensive
11053 Pretend that branches are expensive. This is for experimenting with
11054 code generation only.
11056 @item -mbranch-cheap
11057 @opindex mbranch-cheap
11058 Do not pretend that branches are expensive. This is the default.
11062 Generate code for a system with split I&D@.
11066 Generate code for a system without split I&D@. This is the default.
11070 Use Unix assembler syntax. This is the default when configured for
11071 @samp{pdp11-*-bsd}.
11075 Use DEC assembler syntax. This is the default when configured for any
11076 PDP-11 target other than @samp{pdp11-*-bsd}.
11079 @node PowerPC Options
11080 @subsection PowerPC Options
11081 @cindex PowerPC options
11083 These are listed under @xref{RS/6000 and PowerPC Options}.
11085 @node RS/6000 and PowerPC Options
11086 @subsection IBM RS/6000 and PowerPC Options
11087 @cindex RS/6000 and PowerPC Options
11088 @cindex IBM RS/6000 and PowerPC Options
11090 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11097 @itemx -mno-powerpc
11098 @itemx -mpowerpc-gpopt
11099 @itemx -mno-powerpc-gpopt
11100 @itemx -mpowerpc-gfxopt
11101 @itemx -mno-powerpc-gfxopt
11103 @itemx -mno-powerpc64
11107 @itemx -mno-popcntb
11113 @opindex mno-power2
11115 @opindex mno-powerpc
11116 @opindex mpowerpc-gpopt
11117 @opindex mno-powerpc-gpopt
11118 @opindex mpowerpc-gfxopt
11119 @opindex mno-powerpc-gfxopt
11120 @opindex mpowerpc64
11121 @opindex mno-powerpc64
11125 @opindex mno-popcntb
11128 GCC supports two related instruction set architectures for the
11129 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11130 instructions supported by the @samp{rios} chip set used in the original
11131 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11132 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11133 the IBM 4xx, 6xx, and follow-on microprocessors.
11135 Neither architecture is a subset of the other. However there is a
11136 large common subset of instructions supported by both. An MQ
11137 register is included in processors supporting the POWER architecture.
11139 You use these options to specify which instructions are available on the
11140 processor you are using. The default value of these options is
11141 determined when configuring GCC@. Specifying the
11142 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11143 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11144 rather than the options listed above.
11146 The @option{-mpower} option allows GCC to generate instructions that
11147 are found only in the POWER architecture and to use the MQ register.
11148 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11149 to generate instructions that are present in the POWER2 architecture but
11150 not the original POWER architecture.
11152 The @option{-mpowerpc} option allows GCC to generate instructions that
11153 are found only in the 32-bit subset of the PowerPC architecture.
11154 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11155 GCC to use the optional PowerPC architecture instructions in the
11156 General Purpose group, including floating-point square root. Specifying
11157 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11158 use the optional PowerPC architecture instructions in the Graphics
11159 group, including floating-point select.
11161 The @option{-mmfcrf} option allows GCC to generate the move from
11162 condition register field instruction implemented on the POWER4
11163 processor and other processors that support the PowerPC V2.01
11165 The @option{-mpopcntb} option allows GCC to generate the popcount and
11166 double precision FP reciprocal estimate instruction implemented on the
11167 POWER5 processor and other processors that support the PowerPC V2.02
11169 The @option{-mfprnd} option allows GCC to generate the FP round to
11170 integer instructions implemented on the POWER5+ processor and other
11171 processors that support the PowerPC V2.03 architecture.
11173 The @option{-mpowerpc64} option allows GCC to generate the additional
11174 64-bit instructions that are found in the full PowerPC64 architecture
11175 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11176 @option{-mno-powerpc64}.
11178 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11179 will use only the instructions in the common subset of both
11180 architectures plus some special AIX common-mode calls, and will not use
11181 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11182 permits GCC to use any instruction from either architecture and to
11183 allow use of the MQ register; specify this for the Motorola MPC601.
11185 @item -mnew-mnemonics
11186 @itemx -mold-mnemonics
11187 @opindex mnew-mnemonics
11188 @opindex mold-mnemonics
11189 Select which mnemonics to use in the generated assembler code. With
11190 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11191 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11192 assembler mnemonics defined for the POWER architecture. Instructions
11193 defined in only one architecture have only one mnemonic; GCC uses that
11194 mnemonic irrespective of which of these options is specified.
11196 GCC defaults to the mnemonics appropriate for the architecture in
11197 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11198 value of these option. Unless you are building a cross-compiler, you
11199 should normally not specify either @option{-mnew-mnemonics} or
11200 @option{-mold-mnemonics}, but should instead accept the default.
11202 @item -mcpu=@var{cpu_type}
11204 Set architecture type, register usage, choice of mnemonics, and
11205 instruction scheduling parameters for machine type @var{cpu_type}.
11206 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11207 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11208 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11209 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11210 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11211 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11212 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11213 @samp{power4}, @samp{power5}, @samp{power5+},
11214 @samp{common}, @samp{powerpc}, @samp{powerpc64},
11215 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11217 @option{-mcpu=common} selects a completely generic processor. Code
11218 generated under this option will run on any POWER or PowerPC processor.
11219 GCC will use only the instructions in the common subset of both
11220 architectures, and will not use the MQ register. GCC assumes a generic
11221 processor model for scheduling purposes.
11223 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11224 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11225 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11226 types, with an appropriate, generic processor model assumed for
11227 scheduling purposes.
11229 The other options specify a specific processor. Code generated under
11230 those options will run best on that processor, and may not run at all on
11233 The @option{-mcpu} options automatically enable or disable the
11234 following options: @option{-maltivec}, @option{-mfprnd},
11235 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11236 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11237 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11238 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw}, @option{dlmzb}.
11239 The particular options
11240 set for any particular CPU will vary between compiler versions,
11241 depending on what setting seems to produce optimal code for that CPU;
11242 it doesn't necessarily reflect the actual hardware's capabilities. If
11243 you wish to set an individual option to a particular value, you may
11244 specify it after the @option{-mcpu} option, like @samp{-mcpu=970
11247 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11248 not enabled or disabled by the @option{-mcpu} option at present because
11249 AIX does not have full support for these options. You may still
11250 enable or disable them individually if you're sure it'll work in your
11253 @item -mtune=@var{cpu_type}
11255 Set the instruction scheduling parameters for machine type
11256 @var{cpu_type}, but do not set the architecture type, register usage, or
11257 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11258 values for @var{cpu_type} are used for @option{-mtune} as for
11259 @option{-mcpu}. If both are specified, the code generated will use the
11260 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11261 scheduling parameters set by @option{-mtune}.
11267 Generate code to compute division as reciprocal estimate and iterative
11268 refinement, creating opportunities for increased throughput. This
11269 feature requires: optional PowerPC Graphics instruction set for single
11270 precision and FRE instruction for double precision, assuming divides
11271 cannot generate user-visible traps, and the domain values not include
11272 Infinities, denormals or zero denominator.
11275 @itemx -mno-altivec
11277 @opindex mno-altivec
11278 Generate code that uses (does not use) AltiVec instructions, and also
11279 enable the use of built-in functions that allow more direct access to
11280 the AltiVec instruction set. You may also need to set
11281 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11287 @opindex mno-vrsave
11288 Generate VRSAVE instructions when generating AltiVec code.
11291 @opindex msecure-plt
11292 Generate code that allows ld and ld.so to build executables and shared
11293 libraries with non-exec .plt and .got sections. This is a PowerPC
11294 32-bit SYSV ABI option.
11298 Generate code that uses a BSS .plt section that ld.so fills in, and
11299 requires .plt and .got sections that are both writable and executable.
11300 This is a PowerPC 32-bit SYSV ABI option.
11306 This switch enables or disables the generation of ISEL instructions.
11308 @item -misel=@var{yes/no}
11309 This switch has been deprecated. Use @option{-misel} and
11310 @option{-mno-isel} instead.
11316 This switch enables or disables the generation of SPE simd
11319 @item -mspe=@var{yes/no}
11320 This option has been deprecated. Use @option{-mspe} and
11321 @option{-mno-spe} instead.
11323 @item -mfloat-gprs=@var{yes/single/double/no}
11324 @itemx -mfloat-gprs
11325 @opindex mfloat-gprs
11326 This switch enables or disables the generation of floating point
11327 operations on the general purpose registers for architectures that
11330 The argument @var{yes} or @var{single} enables the use of
11331 single-precision floating point operations.
11333 The argument @var{double} enables the use of single and
11334 double-precision floating point operations.
11336 The argument @var{no} disables floating point operations on the
11337 general purpose registers.
11339 This option is currently only available on the MPC854x.
11345 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11346 targets (including GNU/Linux). The 32-bit environment sets int, long
11347 and pointer to 32 bits and generates code that runs on any PowerPC
11348 variant. The 64-bit environment sets int to 32 bits and long and
11349 pointer to 64 bits, and generates code for PowerPC64, as for
11350 @option{-mpowerpc64}.
11353 @itemx -mno-fp-in-toc
11354 @itemx -mno-sum-in-toc
11355 @itemx -mminimal-toc
11357 @opindex mno-fp-in-toc
11358 @opindex mno-sum-in-toc
11359 @opindex mminimal-toc
11360 Modify generation of the TOC (Table Of Contents), which is created for
11361 every executable file. The @option{-mfull-toc} option is selected by
11362 default. In that case, GCC will allocate at least one TOC entry for
11363 each unique non-automatic variable reference in your program. GCC
11364 will also place floating-point constants in the TOC@. However, only
11365 16,384 entries are available in the TOC@.
11367 If you receive a linker error message that saying you have overflowed
11368 the available TOC space, you can reduce the amount of TOC space used
11369 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11370 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11371 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11372 generate code to calculate the sum of an address and a constant at
11373 run-time instead of putting that sum into the TOC@. You may specify one
11374 or both of these options. Each causes GCC to produce very slightly
11375 slower and larger code at the expense of conserving TOC space.
11377 If you still run out of space in the TOC even when you specify both of
11378 these options, specify @option{-mminimal-toc} instead. This option causes
11379 GCC to make only one TOC entry for every file. When you specify this
11380 option, GCC will produce code that is slower and larger but which
11381 uses extremely little TOC space. You may wish to use this option
11382 only on files that contain less frequently executed code.
11388 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11389 @code{long} type, and the infrastructure needed to support them.
11390 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11391 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11392 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11395 @itemx -mno-xl-compat
11396 @opindex mxl-compat
11397 @opindex mno-xl-compat
11398 Produce code that conforms more closely to IBM XL compiler semantics
11399 when using AIX-compatible ABI. Pass floating-point arguments to
11400 prototyped functions beyond the register save area (RSA) on the stack
11401 in addition to argument FPRs. Do not assume that most significant
11402 double in 128-bit long double value is properly rounded when comparing
11403 values and converting to double. Use XL symbol names for long double
11406 The AIX calling convention was extended but not initially documented to
11407 handle an obscure K&R C case of calling a function that takes the
11408 address of its arguments with fewer arguments than declared. IBM XL
11409 compilers access floating point arguments which do not fit in the
11410 RSA from the stack when a subroutine is compiled without
11411 optimization. Because always storing floating-point arguments on the
11412 stack is inefficient and rarely needed, this option is not enabled by
11413 default and only is necessary when calling subroutines compiled by IBM
11414 XL compilers without optimization.
11418 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11419 application written to use message passing with special startup code to
11420 enable the application to run. The system must have PE installed in the
11421 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11422 must be overridden with the @option{-specs=} option to specify the
11423 appropriate directory location. The Parallel Environment does not
11424 support threads, so the @option{-mpe} option and the @option{-pthread}
11425 option are incompatible.
11427 @item -malign-natural
11428 @itemx -malign-power
11429 @opindex malign-natural
11430 @opindex malign-power
11431 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11432 @option{-malign-natural} overrides the ABI-defined alignment of larger
11433 types, such as floating-point doubles, on their natural size-based boundary.
11434 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11435 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11437 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11441 @itemx -mhard-float
11442 @opindex msoft-float
11443 @opindex mhard-float
11444 Generate code that does not use (uses) the floating-point register set.
11445 Software floating point emulation is provided if you use the
11446 @option{-msoft-float} option, and pass the option to GCC when linking.
11449 @itemx -mno-multiple
11451 @opindex mno-multiple
11452 Generate code that uses (does not use) the load multiple word
11453 instructions and the store multiple word instructions. These
11454 instructions are generated by default on POWER systems, and not
11455 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11456 endian PowerPC systems, since those instructions do not work when the
11457 processor is in little endian mode. The exceptions are PPC740 and
11458 PPC750 which permit the instructions usage in little endian mode.
11463 @opindex mno-string
11464 Generate code that uses (does not use) the load string instructions
11465 and the store string word instructions to save multiple registers and
11466 do small block moves. These instructions are generated by default on
11467 POWER systems, and not generated on PowerPC systems. Do not use
11468 @option{-mstring} on little endian PowerPC systems, since those
11469 instructions do not work when the processor is in little endian mode.
11470 The exceptions are PPC740 and PPC750 which permit the instructions
11471 usage in little endian mode.
11476 @opindex mno-update
11477 Generate code that uses (does not use) the load or store instructions
11478 that update the base register to the address of the calculated memory
11479 location. These instructions are generated by default. If you use
11480 @option{-mno-update}, there is a small window between the time that the
11481 stack pointer is updated and the address of the previous frame is
11482 stored, which means code that walks the stack frame across interrupts or
11483 signals may get corrupted data.
11486 @itemx -mno-fused-madd
11487 @opindex mfused-madd
11488 @opindex mno-fused-madd
11489 Generate code that uses (does not use) the floating point multiply and
11490 accumulate instructions. These instructions are generated by default if
11491 hardware floating is used.
11497 Generate code that uses (does not use) the half-word multiply and
11498 multiply-accumulate instructions on the IBM 405 and 440 processors.
11499 These instructions are generated by default when targetting those
11506 Generate code that uses (does not use) the string-search @samp{dlmzb}
11507 instruction on the IBM 405 and 440 processors. This instruction is
11508 generated by default when targetting those processors.
11510 @item -mno-bit-align
11512 @opindex mno-bit-align
11513 @opindex mbit-align
11514 On System V.4 and embedded PowerPC systems do not (do) force structures
11515 and unions that contain bit-fields to be aligned to the base type of the
11518 For example, by default a structure containing nothing but 8
11519 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11520 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11521 the structure would be aligned to a 1 byte boundary and be one byte in
11524 @item -mno-strict-align
11525 @itemx -mstrict-align
11526 @opindex mno-strict-align
11527 @opindex mstrict-align
11528 On System V.4 and embedded PowerPC systems do not (do) assume that
11529 unaligned memory references will be handled by the system.
11531 @item -mrelocatable
11532 @itemx -mno-relocatable
11533 @opindex mrelocatable
11534 @opindex mno-relocatable
11535 On embedded PowerPC systems generate code that allows (does not allow)
11536 the program to be relocated to a different address at runtime. If you
11537 use @option{-mrelocatable} on any module, all objects linked together must
11538 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11540 @item -mrelocatable-lib
11541 @itemx -mno-relocatable-lib
11542 @opindex mrelocatable-lib
11543 @opindex mno-relocatable-lib
11544 On embedded PowerPC systems generate code that allows (does not allow)
11545 the program to be relocated to a different address at runtime. Modules
11546 compiled with @option{-mrelocatable-lib} can be linked with either modules
11547 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11548 with modules compiled with the @option{-mrelocatable} options.
11554 On System V.4 and embedded PowerPC systems do not (do) assume that
11555 register 2 contains a pointer to a global area pointing to the addresses
11556 used in the program.
11559 @itemx -mlittle-endian
11561 @opindex mlittle-endian
11562 On System V.4 and embedded PowerPC systems compile code for the
11563 processor in little endian mode. The @option{-mlittle-endian} option is
11564 the same as @option{-mlittle}.
11567 @itemx -mbig-endian
11569 @opindex mbig-endian
11570 On System V.4 and embedded PowerPC systems compile code for the
11571 processor in big endian mode. The @option{-mbig-endian} option is
11572 the same as @option{-mbig}.
11574 @item -mdynamic-no-pic
11575 @opindex mdynamic-no-pic
11576 On Darwin and Mac OS X systems, compile code so that it is not
11577 relocatable, but that its external references are relocatable. The
11578 resulting code is suitable for applications, but not shared
11581 @item -mprioritize-restricted-insns=@var{priority}
11582 @opindex mprioritize-restricted-insns
11583 This option controls the priority that is assigned to
11584 dispatch-slot restricted instructions during the second scheduling
11585 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11586 @var{no/highest/second-highest} priority to dispatch slot restricted
11589 @item -msched-costly-dep=@var{dependence_type}
11590 @opindex msched-costly-dep
11591 This option controls which dependences are considered costly
11592 by the target during instruction scheduling. The argument
11593 @var{dependence_type} takes one of the following values:
11594 @var{no}: no dependence is costly,
11595 @var{all}: all dependences are costly,
11596 @var{true_store_to_load}: a true dependence from store to load is costly,
11597 @var{store_to_load}: any dependence from store to load is costly,
11598 @var{number}: any dependence which latency >= @var{number} is costly.
11600 @item -minsert-sched-nops=@var{scheme}
11601 @opindex minsert-sched-nops
11602 This option controls which nop insertion scheme will be used during
11603 the second scheduling pass. The argument @var{scheme} takes one of the
11605 @var{no}: Don't insert nops.
11606 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11607 according to the scheduler's grouping.
11608 @var{regroup_exact}: Insert nops to force costly dependent insns into
11609 separate groups. Insert exactly as many nops as needed to force an insn
11610 to a new group, according to the estimated processor grouping.
11611 @var{number}: Insert nops to force costly dependent insns into
11612 separate groups. Insert @var{number} nops to force an insn to a new group.
11615 @opindex mcall-sysv
11616 On System V.4 and embedded PowerPC systems compile code using calling
11617 conventions that adheres to the March 1995 draft of the System V
11618 Application Binary Interface, PowerPC processor supplement. This is the
11619 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11621 @item -mcall-sysv-eabi
11622 @opindex mcall-sysv-eabi
11623 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11625 @item -mcall-sysv-noeabi
11626 @opindex mcall-sysv-noeabi
11627 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11629 @item -mcall-solaris
11630 @opindex mcall-solaris
11631 On System V.4 and embedded PowerPC systems compile code for the Solaris
11635 @opindex mcall-linux
11636 On System V.4 and embedded PowerPC systems compile code for the
11637 Linux-based GNU system.
11641 On System V.4 and embedded PowerPC systems compile code for the
11642 Hurd-based GNU system.
11644 @item -mcall-netbsd
11645 @opindex mcall-netbsd
11646 On System V.4 and embedded PowerPC systems compile code for the
11647 NetBSD operating system.
11649 @item -maix-struct-return
11650 @opindex maix-struct-return
11651 Return all structures in memory (as specified by the AIX ABI)@.
11653 @item -msvr4-struct-return
11654 @opindex msvr4-struct-return
11655 Return structures smaller than 8 bytes in registers (as specified by the
11658 @item -mabi=@var{abi-type}
11660 Extend the current ABI with a particular extension, or remove such extension.
11661 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11662 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
11666 Extend the current ABI with SPE ABI extensions. This does not change
11667 the default ABI, instead it adds the SPE ABI extensions to the current
11671 @opindex mabi=no-spe
11672 Disable Booke SPE ABI extensions for the current ABI@.
11674 @item -mabi=ibmlongdouble
11675 @opindex mabi=ibmlongdouble
11676 Change the current ABI to use IBM extended precision long double.
11677 This is a PowerPC 32-bit SYSV ABI option.
11679 @item -mabi=ieeelongdouble
11680 @opindex mabi=ieeelongdouble
11681 Change the current ABI to use IEEE extended precision long double.
11682 This is a PowerPC 32-bit Linux ABI option.
11685 @itemx -mno-prototype
11686 @opindex mprototype
11687 @opindex mno-prototype
11688 On System V.4 and embedded PowerPC systems assume that all calls to
11689 variable argument functions are properly prototyped. Otherwise, the
11690 compiler must insert an instruction before every non prototyped call to
11691 set or clear bit 6 of the condition code register (@var{CR}) to
11692 indicate whether floating point values were passed in the floating point
11693 registers in case the function takes a variable arguments. With
11694 @option{-mprototype}, only calls to prototyped variable argument functions
11695 will set or clear the bit.
11699 On embedded PowerPC systems, assume that the startup module is called
11700 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11701 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11706 On embedded PowerPC systems, assume that the startup module is called
11707 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11712 On embedded PowerPC systems, assume that the startup module is called
11713 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11716 @item -myellowknife
11717 @opindex myellowknife
11718 On embedded PowerPC systems, assume that the startup module is called
11719 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11724 On System V.4 and embedded PowerPC systems, specify that you are
11725 compiling for a VxWorks system.
11729 Specify that you are compiling for the WindISS simulation environment.
11733 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11734 header to indicate that @samp{eabi} extended relocations are used.
11740 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11741 Embedded Applications Binary Interface (eabi) which is a set of
11742 modifications to the System V.4 specifications. Selecting @option{-meabi}
11743 means that the stack is aligned to an 8 byte boundary, a function
11744 @code{__eabi} is called to from @code{main} to set up the eabi
11745 environment, and the @option{-msdata} option can use both @code{r2} and
11746 @code{r13} to point to two separate small data areas. Selecting
11747 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11748 do not call an initialization function from @code{main}, and the
11749 @option{-msdata} option will only use @code{r13} to point to a single
11750 small data area. The @option{-meabi} option is on by default if you
11751 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11754 @opindex msdata=eabi
11755 On System V.4 and embedded PowerPC systems, put small initialized
11756 @code{const} global and static data in the @samp{.sdata2} section, which
11757 is pointed to by register @code{r2}. Put small initialized
11758 non-@code{const} global and static data in the @samp{.sdata} section,
11759 which is pointed to by register @code{r13}. Put small uninitialized
11760 global and static data in the @samp{.sbss} section, which is adjacent to
11761 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11762 incompatible with the @option{-mrelocatable} option. The
11763 @option{-msdata=eabi} option also sets the @option{-memb} option.
11766 @opindex msdata=sysv
11767 On System V.4 and embedded PowerPC systems, put small global and static
11768 data in the @samp{.sdata} section, which is pointed to by register
11769 @code{r13}. Put small uninitialized global and static data in the
11770 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11771 The @option{-msdata=sysv} option is incompatible with the
11772 @option{-mrelocatable} option.
11774 @item -msdata=default
11776 @opindex msdata=default
11778 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11779 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11780 same as @option{-msdata=sysv}.
11783 @opindex msdata-data
11784 On System V.4 and embedded PowerPC systems, put small global
11785 data in the @samp{.sdata} section. Put small uninitialized global
11786 data in the @samp{.sbss} section. Do not use register @code{r13}
11787 to address small data however. This is the default behavior unless
11788 other @option{-msdata} options are used.
11792 @opindex msdata=none
11794 On embedded PowerPC systems, put all initialized global and static data
11795 in the @samp{.data} section, and all uninitialized data in the
11796 @samp{.bss} section.
11800 @cindex smaller data references (PowerPC)
11801 @cindex .sdata/.sdata2 references (PowerPC)
11802 On embedded PowerPC systems, put global and static items less than or
11803 equal to @var{num} bytes into the small data or bss sections instead of
11804 the normal data or bss section. By default, @var{num} is 8. The
11805 @option{-G @var{num}} switch is also passed to the linker.
11806 All modules should be compiled with the same @option{-G @var{num}} value.
11809 @itemx -mno-regnames
11811 @opindex mno-regnames
11812 On System V.4 and embedded PowerPC systems do (do not) emit register
11813 names in the assembly language output using symbolic forms.
11816 @itemx -mno-longcall
11818 @opindex mno-longcall
11819 Default to making all function calls indirectly, using a register, so
11820 that functions which reside further than 32 megabytes (33,554,432
11821 bytes) from the current location can be called. This setting can be
11822 overridden by the @code{shortcall} function attribute, or by
11823 @code{#pragma longcall(0)}.
11825 Some linkers are capable of detecting out-of-range calls and generating
11826 glue code on the fly. On these systems, long calls are unnecessary and
11827 generate slower code. As of this writing, the AIX linker can do this,
11828 as can the GNU linker for PowerPC/64. It is planned to add this feature
11829 to the GNU linker for 32-bit PowerPC systems as well.
11831 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11832 callee, L42'', plus a ``branch island'' (glue code). The two target
11833 addresses represent the callee and the ``branch island''. The
11834 Darwin/PPC linker will prefer the first address and generate a ``bl
11835 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11836 otherwise, the linker will generate ``bl L42'' to call the ``branch
11837 island''. The ``branch island'' is appended to the body of the
11838 calling function; it computes the full 32-bit address of the callee
11841 On Mach-O (Darwin) systems, this option directs the compiler emit to
11842 the glue for every direct call, and the Darwin linker decides whether
11843 to use or discard it.
11845 In the future, we may cause GCC to ignore all longcall specifications
11846 when the linker is known to generate glue.
11850 Adds support for multithreading with the @dfn{pthreads} library.
11851 This option sets flags for both the preprocessor and linker.
11855 @node S/390 and zSeries Options
11856 @subsection S/390 and zSeries Options
11857 @cindex S/390 and zSeries Options
11859 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
11863 @itemx -msoft-float
11864 @opindex mhard-float
11865 @opindex msoft-float
11866 Use (do not use) the hardware floating-point instructions and registers
11867 for floating-point operations. When @option{-msoft-float} is specified,
11868 functions in @file{libgcc.a} will be used to perform floating-point
11869 operations. When @option{-mhard-float} is specified, the compiler
11870 generates IEEE floating-point instructions. This is the default.
11872 @item -mlong-double-64
11873 @itemx -mlong-double-128
11874 @opindex mlong-double-64
11875 @opindex mlong-double-128
11876 These switches control the size of @code{long double} type. A size
11877 of 64bit makes the @code{long double} type equivalent to the @code{double}
11878 type. This is the default.
11881 @itemx -mno-backchain
11882 @opindex mbackchain
11883 @opindex mno-backchain
11884 Store (do not store) the address of the caller's frame as backchain pointer
11885 into the callee's stack frame.
11886 A backchain may be needed to allow debugging using tools that do not understand
11887 DWARF-2 call frame information.
11888 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
11889 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
11890 the backchain is placed into the topmost word of the 96/160 byte register
11893 In general, code compiled with @option{-mbackchain} is call-compatible with
11894 code compiled with @option{-mmo-backchain}; however, use of the backchain
11895 for debugging purposes usually requires that the whole binary is built with
11896 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
11897 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11898 to build a linux kernel use @option{-msoft-float}.
11900 The default is to not maintain the backchain.
11902 @item -mpacked-stack
11903 @item -mno-packed-stack
11904 @opindex mpacked-stack
11905 @opindex mno-packed-stack
11906 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
11907 specified, the compiler uses the all fields of the 96/160 byte register save
11908 area only for their default purpose; unused fields still take up stack space.
11909 When @option{-mpacked-stack} is specified, register save slots are densely
11910 packed at the top of the register save area; unused space is reused for other
11911 purposes, allowing for more efficient use of the available stack space.
11912 However, when @option{-mbackchain} is also in effect, the topmost word of
11913 the save area is always used to store the backchain, and the return address
11914 register is always saved two words below the backchain.
11916 As long as the stack frame backchain is not used, code generated with
11917 @option{-mpacked-stack} is call-compatible with code generated with
11918 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
11919 S/390 or zSeries generated code that uses the stack frame backchain at run
11920 time, not just for debugging purposes. Such code is not call-compatible
11921 with code compiled with @option{-mpacked-stack}. Also, note that the
11922 combination of @option{-mbackchain},
11923 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11924 to build a linux kernel use @option{-msoft-float}.
11926 The default is to not use the packed stack layout.
11929 @itemx -mno-small-exec
11930 @opindex msmall-exec
11931 @opindex mno-small-exec
11932 Generate (or do not generate) code using the @code{bras} instruction
11933 to do subroutine calls.
11934 This only works reliably if the total executable size does not
11935 exceed 64k. The default is to use the @code{basr} instruction instead,
11936 which does not have this limitation.
11942 When @option{-m31} is specified, generate code compliant to the
11943 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
11944 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
11945 particular to generate 64-bit instructions. For the @samp{s390}
11946 targets, the default is @option{-m31}, while the @samp{s390x}
11947 targets default to @option{-m64}.
11953 When @option{-mzarch} is specified, generate code using the
11954 instructions available on z/Architecture.
11955 When @option{-mesa} is specified, generate code using the
11956 instructions available on ESA/390. Note that @option{-mesa} is
11957 not possible with @option{-m64}.
11958 When generating code compliant to the GNU/Linux for S/390 ABI,
11959 the default is @option{-mesa}. When generating code compliant
11960 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
11966 Generate (or do not generate) code using the @code{mvcle} instruction
11967 to perform block moves. When @option{-mno-mvcle} is specified,
11968 use a @code{mvc} loop instead. This is the default unless optimizing for
11975 Print (or do not print) additional debug information when compiling.
11976 The default is to not print debug information.
11978 @item -march=@var{cpu-type}
11980 Generate code that will run on @var{cpu-type}, which is the name of a system
11981 representing a certain processor type. Possible values for
11982 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
11983 When generating code using the instructions available on z/Architecture,
11984 the default is @option{-march=z900}. Otherwise, the default is
11985 @option{-march=g5}.
11987 @item -mtune=@var{cpu-type}
11989 Tune to @var{cpu-type} everything applicable about the generated code,
11990 except for the ABI and the set of available instructions.
11991 The list of @var{cpu-type} values is the same as for @option{-march}.
11992 The default is the value used for @option{-march}.
11995 @itemx -mno-tpf-trace
11996 @opindex mtpf-trace
11997 @opindex mno-tpf-trace
11998 Generate code that adds (does not add) in TPF OS specific branches to trace
11999 routines in the operating system. This option is off by default, even
12000 when compiling for the TPF OS@.
12003 @itemx -mno-fused-madd
12004 @opindex mfused-madd
12005 @opindex mno-fused-madd
12006 Generate code that uses (does not use) the floating point multiply and
12007 accumulate instructions. These instructions are generated by default if
12008 hardware floating point is used.
12010 @item -mwarn-framesize=@var{framesize}
12011 @opindex mwarn-framesize
12012 Emit a warning if the current function exceeds the given frame size. Because
12013 this is a compile time check it doesn't need to be a real problem when the program
12014 runs. It is intended to identify functions which most probably cause
12015 a stack overflow. It is useful to be used in an environment with limited stack
12016 size e.g.@: the linux kernel.
12018 @item -mwarn-dynamicstack
12019 @opindex mwarn-dynamicstack
12020 Emit a warning if the function calls alloca or uses dynamically
12021 sized arrays. This is generally a bad idea with a limited stack size.
12023 @item -mstack-guard=@var{stack-guard}
12024 @item -mstack-size=@var{stack-size}
12025 @opindex mstack-guard
12026 @opindex mstack-size
12027 These arguments always have to be used in conjunction. If they are present the s390
12028 back end emits additional instructions in the function prologue which trigger a trap
12029 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12030 (remember that the stack on s390 grows downward). These options are intended to
12031 be used to help debugging stack overflow problems. The additionally emitted code
12032 causes only little overhead and hence can also be used in production like systems
12033 without greater performance degradation. The given values have to be exact
12034 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12036 In order to be efficient the extra code makes the assumption that the stack starts
12037 at an address aligned to the value given by @var{stack-size}.
12041 @subsection SH Options
12043 These @samp{-m} options are defined for the SH implementations:
12048 Generate code for the SH1.
12052 Generate code for the SH2.
12055 Generate code for the SH2e.
12059 Generate code for the SH3.
12063 Generate code for the SH3e.
12067 Generate code for the SH4 without a floating-point unit.
12069 @item -m4-single-only
12070 @opindex m4-single-only
12071 Generate code for the SH4 with a floating-point unit that only
12072 supports single-precision arithmetic.
12076 Generate code for the SH4 assuming the floating-point unit is in
12077 single-precision mode by default.
12081 Generate code for the SH4.
12085 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12086 floating-point unit is not used.
12088 @item -m4a-single-only
12089 @opindex m4a-single-only
12090 Generate code for the SH4a, in such a way that no double-precision
12091 floating point operations are used.
12094 @opindex m4a-single
12095 Generate code for the SH4a assuming the floating-point unit is in
12096 single-precision mode by default.
12100 Generate code for the SH4a.
12104 Same as @option{-m4a-nofpu}, except that it implicitly passes
12105 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12106 instructions at the moment.
12110 Compile code for the processor in big endian mode.
12114 Compile code for the processor in little endian mode.
12118 Align doubles at 64-bit boundaries. Note that this changes the calling
12119 conventions, and thus some functions from the standard C library will
12120 not work unless you recompile it first with @option{-mdalign}.
12124 Shorten some address references at link time, when possible; uses the
12125 linker option @option{-relax}.
12129 Use 32-bit offsets in @code{switch} tables. The default is to use
12134 Enable the use of the instruction @code{fmovd}.
12138 Comply with the calling conventions defined by Renesas.
12142 Comply with the calling conventions defined by Renesas.
12146 Comply with the calling conventions defined for GCC before the Renesas
12147 conventions were available. This option is the default for all
12148 targets of the SH toolchain except for @samp{sh-symbianelf}.
12151 @opindex mnomacsave
12152 Mark the @code{MAC} register as call-clobbered, even if
12153 @option{-mhitachi} is given.
12157 Increase IEEE-compliance of floating-point code.
12158 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12159 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12160 comparisons of NANs / infinities incurs extra overhead in every
12161 floating point comparison, therefore the default is set to
12162 @option{-ffinite-math-only}.
12166 Dump instruction size and location in the assembly code.
12169 @opindex mpadstruct
12170 This option is deprecated. It pads structures to multiple of 4 bytes,
12171 which is incompatible with the SH ABI@.
12175 Optimize for space instead of speed. Implied by @option{-Os}.
12178 @opindex mprefergot
12179 When generating position-independent code, emit function calls using
12180 the Global Offset Table instead of the Procedure Linkage Table.
12184 Generate a library function call to invalidate instruction cache
12185 entries, after fixing up a trampoline. This library function call
12186 doesn't assume it can write to the whole memory address space. This
12187 is the default when the target is @code{sh-*-linux*}.
12189 @item -multcost=@var{number}
12190 @opindex multcost=@var{number}
12191 Set the cost to assume for a multiply insn.
12193 @item -mdiv=@var{strategy}
12194 @opindex mdiv=@var{strategy}
12195 Set the division strategy to use for SHmedia code. @var{strategy} must be
12196 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12197 inv:call2, inv:fp .
12198 "fp" performs the operation in floating point. This has a very high latency,
12199 but needs only a few instructions, so it might be a good choice if
12200 your code has enough easily exploitable ILP to allow the compiler to
12201 schedule the floating point instructions together with other instructions.
12202 Division by zero causes a floating point exception.
12203 "inv" uses integer operations to calculate the inverse of the divisor,
12204 and then multiplies the dividend with the inverse. This strategy allows
12205 cse and hoisting of the inverse calculation. Division by zero calculates
12206 an unspecified result, but does not trap.
12207 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12208 have been found, or if the entire operation has been hoisted to the same
12209 place, the last stages of the inverse calculation are intertwined with the
12210 final multiply to reduce the overall latency, at the expense of using a few
12211 more instructions, and thus offering fewer scheduling opportunities with
12213 "call" calls a library function that usually implements the inv:minlat
12215 This gives high code density for m5-*media-nofpu compilations.
12216 "call2" uses a different entry point of the same library function, where it
12217 assumes that a pointer to a lookup table has already been set up, which
12218 exposes the pointer load to cse / code hoisting optimizations.
12219 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12220 code generation, but if the code stays unoptimized, revert to the "call",
12221 "call2", or "fp" strategies, respectively. Note that the
12222 potentially-trapping side effect of division by zero is carried by a
12223 separate instruction, so it is possible that all the integer instructions
12224 are hoisted out, but the marker for the side effect stays where it is.
12225 A recombination to fp operations or a call is not possible in that case.
12226 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12227 that the inverse calculation was nor separated from the multiply, they speed
12228 up division where the dividend fits into 20 bits (plus sign where applicable),
12229 by inserting a test to skip a number of operations in this case; this test
12230 slows down the case of larger dividends. inv20u assumes the case of a such
12231 a small dividend to be unlikely, and inv20l assumes it to be likely.
12233 @item -mdivsi3_libfunc=@var{name}
12234 @opindex mdivsi3_libfunc=@var{name}
12235 Set the name of the library function used for 32 bit signed division to
12236 @var{name}. This only affect the name used in the call and inv:call
12237 division strategies, and the compiler will still expect the same
12238 sets of input/output/clobbered registers as if this option was not present.
12240 @item -madjust-unroll
12241 @opindex madjust-unroll
12242 Throttle unrolling to avoid thrashing target registers.
12243 This option only has an effect if the gcc code base supports the
12244 TARGET_ADJUST_UNROLL_MAX target hook.
12246 @item -mindexed-addressing
12247 @opindex mindexed-addressing
12248 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12249 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12250 semantics for the indexed addressing mode. The architecture allows the
12251 implementation of processors with 64 bit MMU, which the OS could use to
12252 get 32 bit addressing, but since no current hardware implementation supports
12253 this or any other way to make the indexed addressing mode safe to use in
12254 the 32 bit ABI, the default is -mno-indexed-addressing.
12256 @item -mgettrcost=@var{number}
12257 @opindex mgettrcost=@var{number}
12258 Set the cost assumed for the gettr instruction to @var{number}.
12259 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12263 Assume pt* instructions won't trap. This will generally generate better
12264 scheduled code, but is unsafe on current hardware. The current architecture
12265 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12266 This has the unintentional effect of making it unsafe to schedule ptabs /
12267 ptrel before a branch, or hoist it out of a loop. For example,
12268 __do_global_ctors, a part of libgcc that runs constructors at program
12269 startup, calls functions in a list which is delimited by -1. With the
12270 -mpt-fixed option, the ptabs will be done before testing against -1.
12271 That means that all the constructors will be run a bit quicker, but when
12272 the loop comes to the end of the list, the program crashes because ptabs
12273 loads -1 into a target register. Since this option is unsafe for any
12274 hardware implementing the current architecture specification, the default
12275 is -mno-pt-fixed. Unless the user specifies a specific cost with
12276 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12277 this deters register allocation using target registers for storing
12280 @item -minvalid-symbols
12281 @opindex minvalid-symbols
12282 Assume symbols might be invalid. Ordinary function symbols generated by
12283 the compiler will always be valid to load with movi/shori/ptabs or
12284 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12285 to generate symbols that will cause ptabs / ptrel to trap.
12286 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12287 It will then prevent cross-basic-block cse, hoisting and most scheduling
12288 of symbol loads. The default is @option{-mno-invalid-symbols}.
12291 @node SPARC Options
12292 @subsection SPARC Options
12293 @cindex SPARC options
12295 These @samp{-m} options are supported on the SPARC:
12298 @item -mno-app-regs
12300 @opindex mno-app-regs
12302 Specify @option{-mapp-regs} to generate output using the global registers
12303 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12306 To be fully SVR4 ABI compliant at the cost of some performance loss,
12307 specify @option{-mno-app-regs}. You should compile libraries and system
12308 software with this option.
12311 @itemx -mhard-float
12313 @opindex mhard-float
12314 Generate output containing floating point instructions. This is the
12318 @itemx -msoft-float
12320 @opindex msoft-float
12321 Generate output containing library calls for floating point.
12322 @strong{Warning:} the requisite libraries are not available for all SPARC
12323 targets. Normally the facilities of the machine's usual C compiler are
12324 used, but this cannot be done directly in cross-compilation. You must make
12325 your own arrangements to provide suitable library functions for
12326 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12327 @samp{sparclite-*-*} do provide software floating point support.
12329 @option{-msoft-float} changes the calling convention in the output file;
12330 therefore, it is only useful if you compile @emph{all} of a program with
12331 this option. In particular, you need to compile @file{libgcc.a}, the
12332 library that comes with GCC, with @option{-msoft-float} in order for
12335 @item -mhard-quad-float
12336 @opindex mhard-quad-float
12337 Generate output containing quad-word (long double) floating point
12340 @item -msoft-quad-float
12341 @opindex msoft-quad-float
12342 Generate output containing library calls for quad-word (long double)
12343 floating point instructions. The functions called are those specified
12344 in the SPARC ABI@. This is the default.
12346 As of this writing, there are no SPARC implementations that have hardware
12347 support for the quad-word floating point instructions. They all invoke
12348 a trap handler for one of these instructions, and then the trap handler
12349 emulates the effect of the instruction. Because of the trap handler overhead,
12350 this is much slower than calling the ABI library routines. Thus the
12351 @option{-msoft-quad-float} option is the default.
12353 @item -mno-unaligned-doubles
12354 @itemx -munaligned-doubles
12355 @opindex mno-unaligned-doubles
12356 @opindex munaligned-doubles
12357 Assume that doubles have 8 byte alignment. This is the default.
12359 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12360 alignment only if they are contained in another type, or if they have an
12361 absolute address. Otherwise, it assumes they have 4 byte alignment.
12362 Specifying this option avoids some rare compatibility problems with code
12363 generated by other compilers. It is not the default because it results
12364 in a performance loss, especially for floating point code.
12366 @item -mno-faster-structs
12367 @itemx -mfaster-structs
12368 @opindex mno-faster-structs
12369 @opindex mfaster-structs
12370 With @option{-mfaster-structs}, the compiler assumes that structures
12371 should have 8 byte alignment. This enables the use of pairs of
12372 @code{ldd} and @code{std} instructions for copies in structure
12373 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12374 However, the use of this changed alignment directly violates the SPARC
12375 ABI@. Thus, it's intended only for use on targets where the developer
12376 acknowledges that their resulting code will not be directly in line with
12377 the rules of the ABI@.
12379 @item -mimpure-text
12380 @opindex mimpure-text
12381 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12382 the compiler to not pass @option{-z text} to the linker when linking a
12383 shared object. Using this option, you can link position-dependent
12384 code into a shared object.
12386 @option{-mimpure-text} suppresses the ``relocations remain against
12387 allocatable but non-writable sections'' linker error message.
12388 However, the necessary relocations will trigger copy-on-write, and the
12389 shared object is not actually shared across processes. Instead of
12390 using @option{-mimpure-text}, you should compile all source code with
12391 @option{-fpic} or @option{-fPIC}.
12393 This option is only available on SunOS and Solaris.
12395 @item -mcpu=@var{cpu_type}
12397 Set the instruction set, register set, and instruction scheduling parameters
12398 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12399 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12400 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12401 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12402 @samp{ultrasparc3}, and @samp{niagara}.
12404 Default instruction scheduling parameters are used for values that select
12405 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12406 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12408 Here is a list of each supported architecture and their supported
12413 v8: supersparc, hypersparc
12414 sparclite: f930, f934, sparclite86x
12416 v9: ultrasparc, ultrasparc3, niagara
12419 By default (unless configured otherwise), GCC generates code for the V7
12420 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12421 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12422 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12423 SPARCStation 1, 2, IPX etc.
12425 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12426 architecture. The only difference from V7 code is that the compiler emits
12427 the integer multiply and integer divide instructions which exist in SPARC-V8
12428 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12429 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12432 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12433 the SPARC architecture. This adds the integer multiply, integer divide step
12434 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12435 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12436 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12437 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12438 MB86934 chip, which is the more recent SPARClite with FPU@.
12440 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12441 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12442 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12443 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12444 optimizes it for the TEMIC SPARClet chip.
12446 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12447 architecture. This adds 64-bit integer and floating-point move instructions,
12448 3 additional floating-point condition code registers and conditional move
12449 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12450 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12451 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12452 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12453 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12454 Sun UltraSPARC T1 chips.
12456 @item -mtune=@var{cpu_type}
12458 Set the instruction scheduling parameters for machine type
12459 @var{cpu_type}, but do not set the instruction set or register set that the
12460 option @option{-mcpu=@var{cpu_type}} would.
12462 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12463 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12464 that select a particular cpu implementation. Those are @samp{cypress},
12465 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12466 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12467 @samp{ultrasparc3}, and @samp{niagara}.
12472 @opindex mno-v8plus
12473 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12474 difference from the V8 ABI is that the global and out registers are
12475 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12476 mode for all SPARC-V9 processors.
12482 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12483 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12486 These @samp{-m} options are supported in addition to the above
12487 on SPARC-V9 processors in 64-bit environments:
12490 @item -mlittle-endian
12491 @opindex mlittle-endian
12492 Generate code for a processor running in little-endian mode. It is only
12493 available for a few configurations and most notably not on Solaris and Linux.
12499 Generate code for a 32-bit or 64-bit environment.
12500 The 32-bit environment sets int, long and pointer to 32 bits.
12501 The 64-bit environment sets int to 32 bits and long and pointer
12504 @item -mcmodel=medlow
12505 @opindex mcmodel=medlow
12506 Generate code for the Medium/Low code model: 64-bit addresses, programs
12507 must be linked in the low 32 bits of memory. Programs can be statically
12508 or dynamically linked.
12510 @item -mcmodel=medmid
12511 @opindex mcmodel=medmid
12512 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12513 must be linked in the low 44 bits of memory, the text and data segments must
12514 be less than 2GB in size and the data segment must be located within 2GB of
12517 @item -mcmodel=medany
12518 @opindex mcmodel=medany
12519 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12520 may be linked anywhere in memory, the text and data segments must be less
12521 than 2GB in size and the data segment must be located within 2GB of the
12524 @item -mcmodel=embmedany
12525 @opindex mcmodel=embmedany
12526 Generate code for the Medium/Anywhere code model for embedded systems:
12527 64-bit addresses, the text and data segments must be less than 2GB in
12528 size, both starting anywhere in memory (determined at link time). The
12529 global register %g4 points to the base of the data segment. Programs
12530 are statically linked and PIC is not supported.
12533 @itemx -mno-stack-bias
12534 @opindex mstack-bias
12535 @opindex mno-stack-bias
12536 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12537 frame pointer if present, are offset by @minus{}2047 which must be added back
12538 when making stack frame references. This is the default in 64-bit mode.
12539 Otherwise, assume no such offset is present.
12542 These switches are supported in addition to the above on Solaris:
12547 Add support for multithreading using the Solaris threads library. This
12548 option sets flags for both the preprocessor and linker. This option does
12549 not affect the thread safety of object code produced by the compiler or
12550 that of libraries supplied with it.
12554 Add support for multithreading using the POSIX threads library. This
12555 option sets flags for both the preprocessor and linker. This option does
12556 not affect the thread safety of object code produced by the compiler or
12557 that of libraries supplied with it.
12561 This is a synonym for @option{-pthreads}.
12564 @node System V Options
12565 @subsection Options for System V
12567 These additional options are available on System V Release 4 for
12568 compatibility with other compilers on those systems:
12573 Create a shared object.
12574 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12578 Identify the versions of each tool used by the compiler, in a
12579 @code{.ident} assembler directive in the output.
12583 Refrain from adding @code{.ident} directives to the output file (this is
12586 @item -YP,@var{dirs}
12588 Search the directories @var{dirs}, and no others, for libraries
12589 specified with @option{-l}.
12591 @item -Ym,@var{dir}
12593 Look in the directory @var{dir} to find the M4 preprocessor.
12594 The assembler uses this option.
12595 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12596 @c the generic assembler that comes with Solaris takes just -Ym.
12599 @node TMS320C3x/C4x Options
12600 @subsection TMS320C3x/C4x Options
12601 @cindex TMS320C3x/C4x Options
12603 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12607 @item -mcpu=@var{cpu_type}
12609 Set the instruction set, register set, and instruction scheduling
12610 parameters for machine type @var{cpu_type}. Supported values for
12611 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12612 @samp{c44}. The default is @samp{c40} to generate code for the
12617 @itemx -msmall-memory
12619 @opindex mbig-memory
12621 @opindex msmall-memory
12623 Generates code for the big or small memory model. The small memory
12624 model assumed that all data fits into one 64K word page. At run-time
12625 the data page (DP) register must be set to point to the 64K page
12626 containing the .bss and .data program sections. The big memory model is
12627 the default and requires reloading of the DP register for every direct
12634 Allow (disallow) allocation of general integer operands into the block
12635 count register BK@.
12641 Enable (disable) generation of code using decrement and branch,
12642 DBcond(D), instructions. This is enabled by default for the C4x. To be
12643 on the safe side, this is disabled for the C3x, since the maximum
12644 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12645 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12646 that it can utilize the decrement and branch instruction, but will give
12647 up if there is more than one memory reference in the loop. Thus a loop
12648 where the loop counter is decremented can generate slightly more
12649 efficient code, in cases where the RPTB instruction cannot be utilized.
12651 @item -mdp-isr-reload
12653 @opindex mdp-isr-reload
12655 Force the DP register to be saved on entry to an interrupt service
12656 routine (ISR), reloaded to point to the data section, and restored on
12657 exit from the ISR@. This should not be required unless someone has
12658 violated the small memory model by modifying the DP register, say within
12665 For the C3x use the 24-bit MPYI instruction for integer multiplies
12666 instead of a library call to guarantee 32-bit results. Note that if one
12667 of the operands is a constant, then the multiplication will be performed
12668 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12669 then squaring operations are performed inline instead of a library call.
12672 @itemx -mno-fast-fix
12674 @opindex mno-fast-fix
12675 The C3x/C4x FIX instruction to convert a floating point value to an
12676 integer value chooses the nearest integer less than or equal to the
12677 floating point value rather than to the nearest integer. Thus if the
12678 floating point number is negative, the result will be incorrectly
12679 truncated an additional code is necessary to detect and correct this
12680 case. This option can be used to disable generation of the additional
12681 code required to correct the result.
12687 Enable (disable) generation of repeat block sequences using the RPTB
12688 instruction for zero overhead looping. The RPTB construct is only used
12689 for innermost loops that do not call functions or jump across the loop
12690 boundaries. There is no advantage having nested RPTB loops due to the
12691 overhead required to save and restore the RC, RS, and RE registers.
12692 This is enabled by default with @option{-O2}.
12694 @item -mrpts=@var{count}
12698 Enable (disable) the use of the single instruction repeat instruction
12699 RPTS@. If a repeat block contains a single instruction, and the loop
12700 count can be guaranteed to be less than the value @var{count}, GCC will
12701 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12702 then a RPTS will be emitted even if the loop count cannot be determined
12703 at compile time. Note that the repeated instruction following RPTS does
12704 not have to be reloaded from memory each iteration, thus freeing up the
12705 CPU buses for operands. However, since interrupts are blocked by this
12706 instruction, it is disabled by default.
12708 @item -mloop-unsigned
12709 @itemx -mno-loop-unsigned
12710 @opindex mloop-unsigned
12711 @opindex mno-loop-unsigned
12712 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12713 is @math{2^{31} + 1} since these instructions test if the iteration count is
12714 negative to terminate the loop. If the iteration count is unsigned
12715 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12716 exceeded. This switch allows an unsigned iteration count.
12720 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12721 with. This also enforces compatibility with the API employed by the TI
12722 C3x C compiler. For example, long doubles are passed as structures
12723 rather than in floating point registers.
12729 Generate code that uses registers (stack) for passing arguments to functions.
12730 By default, arguments are passed in registers where possible rather
12731 than by pushing arguments on to the stack.
12733 @item -mparallel-insns
12734 @itemx -mno-parallel-insns
12735 @opindex mparallel-insns
12736 @opindex mno-parallel-insns
12737 Allow the generation of parallel instructions. This is enabled by
12738 default with @option{-O2}.
12740 @item -mparallel-mpy
12741 @itemx -mno-parallel-mpy
12742 @opindex mparallel-mpy
12743 @opindex mno-parallel-mpy
12744 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12745 provided @option{-mparallel-insns} is also specified. These instructions have
12746 tight register constraints which can pessimize the code generation
12747 of large functions.
12752 @subsection V850 Options
12753 @cindex V850 Options
12755 These @samp{-m} options are defined for V850 implementations:
12759 @itemx -mno-long-calls
12760 @opindex mlong-calls
12761 @opindex mno-long-calls
12762 Treat all calls as being far away (near). If calls are assumed to be
12763 far away, the compiler will always load the functions address up into a
12764 register, and call indirect through the pointer.
12770 Do not optimize (do optimize) basic blocks that use the same index
12771 pointer 4 or more times to copy pointer into the @code{ep} register, and
12772 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12773 option is on by default if you optimize.
12775 @item -mno-prolog-function
12776 @itemx -mprolog-function
12777 @opindex mno-prolog-function
12778 @opindex mprolog-function
12779 Do not use (do use) external functions to save and restore registers
12780 at the prologue and epilogue of a function. The external functions
12781 are slower, but use less code space if more than one function saves
12782 the same number of registers. The @option{-mprolog-function} option
12783 is on by default if you optimize.
12787 Try to make the code as small as possible. At present, this just turns
12788 on the @option{-mep} and @option{-mprolog-function} options.
12790 @item -mtda=@var{n}
12792 Put static or global variables whose size is @var{n} bytes or less into
12793 the tiny data area that register @code{ep} points to. The tiny data
12794 area can hold up to 256 bytes in total (128 bytes for byte references).
12796 @item -msda=@var{n}
12798 Put static or global variables whose size is @var{n} bytes or less into
12799 the small data area that register @code{gp} points to. The small data
12800 area can hold up to 64 kilobytes.
12802 @item -mzda=@var{n}
12804 Put static or global variables whose size is @var{n} bytes or less into
12805 the first 32 kilobytes of memory.
12809 Specify that the target processor is the V850.
12812 @opindex mbig-switch
12813 Generate code suitable for big switch tables. Use this option only if
12814 the assembler/linker complain about out of range branches within a switch
12819 This option will cause r2 and r5 to be used in the code generated by
12820 the compiler. This setting is the default.
12822 @item -mno-app-regs
12823 @opindex mno-app-regs
12824 This option will cause r2 and r5 to be treated as fixed registers.
12828 Specify that the target processor is the V850E1. The preprocessor
12829 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
12830 this option is used.
12834 Specify that the target processor is the V850E@. The preprocessor
12835 constant @samp{__v850e__} will be defined if this option is used.
12837 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
12838 are defined then a default target processor will be chosen and the
12839 relevant @samp{__v850*__} preprocessor constant will be defined.
12841 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
12842 defined, regardless of which processor variant is the target.
12844 @item -mdisable-callt
12845 @opindex mdisable-callt
12846 This option will suppress generation of the CALLT instruction for the
12847 v850e and v850e1 flavors of the v850 architecture. The default is
12848 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
12853 @subsection VAX Options
12854 @cindex VAX options
12856 These @samp{-m} options are defined for the VAX:
12861 Do not output certain jump instructions (@code{aobleq} and so on)
12862 that the Unix assembler for the VAX cannot handle across long
12867 Do output those jump instructions, on the assumption that you
12868 will assemble with the GNU assembler.
12872 Output code for g-format floating point numbers instead of d-format.
12875 @node x86-64 Options
12876 @subsection x86-64 Options
12877 @cindex x86-64 options
12879 These are listed under @xref{i386 and x86-64 Options}.
12881 @node Xstormy16 Options
12882 @subsection Xstormy16 Options
12883 @cindex Xstormy16 Options
12885 These options are defined for Xstormy16:
12890 Choose startup files and linker script suitable for the simulator.
12893 @node Xtensa Options
12894 @subsection Xtensa Options
12895 @cindex Xtensa Options
12897 These options are supported for Xtensa targets:
12901 @itemx -mno-const16
12903 @opindex mno-const16
12904 Enable or disable use of @code{CONST16} instructions for loading
12905 constant values. The @code{CONST16} instruction is currently not a
12906 standard option from Tensilica. When enabled, @code{CONST16}
12907 instructions are always used in place of the standard @code{L32R}
12908 instructions. The use of @code{CONST16} is enabled by default only if
12909 the @code{L32R} instruction is not available.
12912 @itemx -mno-fused-madd
12913 @opindex mfused-madd
12914 @opindex mno-fused-madd
12915 Enable or disable use of fused multiply/add and multiply/subtract
12916 instructions in the floating-point option. This has no effect if the
12917 floating-point option is not also enabled. Disabling fused multiply/add
12918 and multiply/subtract instructions forces the compiler to use separate
12919 instructions for the multiply and add/subtract operations. This may be
12920 desirable in some cases where strict IEEE 754-compliant results are
12921 required: the fused multiply add/subtract instructions do not round the
12922 intermediate result, thereby producing results with @emph{more} bits of
12923 precision than specified by the IEEE standard. Disabling fused multiply
12924 add/subtract instructions also ensures that the program output is not
12925 sensitive to the compiler's ability to combine multiply and add/subtract
12928 @item -mtext-section-literals
12929 @itemx -mno-text-section-literals
12930 @opindex mtext-section-literals
12931 @opindex mno-text-section-literals
12932 Control the treatment of literal pools. The default is
12933 @option{-mno-text-section-literals}, which places literals in a separate
12934 section in the output file. This allows the literal pool to be placed
12935 in a data RAM/ROM, and it also allows the linker to combine literal
12936 pools from separate object files to remove redundant literals and
12937 improve code size. With @option{-mtext-section-literals}, the literals
12938 are interspersed in the text section in order to keep them as close as
12939 possible to their references. This may be necessary for large assembly
12942 @item -mtarget-align
12943 @itemx -mno-target-align
12944 @opindex mtarget-align
12945 @opindex mno-target-align
12946 When this option is enabled, GCC instructs the assembler to
12947 automatically align instructions to reduce branch penalties at the
12948 expense of some code density. The assembler attempts to widen density
12949 instructions to align branch targets and the instructions following call
12950 instructions. If there are not enough preceding safe density
12951 instructions to align a target, no widening will be performed. The
12952 default is @option{-mtarget-align}. These options do not affect the
12953 treatment of auto-aligned instructions like @code{LOOP}, which the
12954 assembler will always align, either by widening density instructions or
12955 by inserting no-op instructions.
12958 @itemx -mno-longcalls
12959 @opindex mlongcalls
12960 @opindex mno-longcalls
12961 When this option is enabled, GCC instructs the assembler to translate
12962 direct calls to indirect calls unless it can determine that the target
12963 of a direct call is in the range allowed by the call instruction. This
12964 translation typically occurs for calls to functions in other source
12965 files. Specifically, the assembler translates a direct @code{CALL}
12966 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
12967 The default is @option{-mno-longcalls}. This option should be used in
12968 programs where the call target can potentially be out of range. This
12969 option is implemented in the assembler, not the compiler, so the
12970 assembly code generated by GCC will still show direct call
12971 instructions---look at the disassembled object code to see the actual
12972 instructions. Note that the assembler will use an indirect call for
12973 every cross-file call, not just those that really will be out of range.
12976 @node zSeries Options
12977 @subsection zSeries Options
12978 @cindex zSeries options
12980 These are listed under @xref{S/390 and zSeries Options}.
12982 @node Code Gen Options
12983 @section Options for Code Generation Conventions
12984 @cindex code generation conventions
12985 @cindex options, code generation
12986 @cindex run-time options
12988 These machine-independent options control the interface conventions
12989 used in code generation.
12991 Most of them have both positive and negative forms; the negative form
12992 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
12993 one of the forms is listed---the one which is not the default. You
12994 can figure out the other form by either removing @samp{no-} or adding
12998 @item -fbounds-check
12999 @opindex fbounds-check
13000 For front-ends that support it, generate additional code to check that
13001 indices used to access arrays are within the declared range. This is
13002 currently only supported by the Java and Fortran 77 front-ends, where
13003 this option defaults to true and false respectively.
13007 This option generates traps for signed overflow on addition, subtraction,
13008 multiplication operations.
13012 This option instructs the compiler to assume that signed arithmetic
13013 overflow of addition, subtraction and multiplication wraps around
13014 using twos-complement representation. This flag enables some optimizations
13015 and disables others. This option is enabled by default for the Java
13016 front-end, as required by the Java language specification.
13019 @opindex fexceptions
13020 Enable exception handling. Generates extra code needed to propagate
13021 exceptions. For some targets, this implies GCC will generate frame
13022 unwind information for all functions, which can produce significant data
13023 size overhead, although it does not affect execution. If you do not
13024 specify this option, GCC will enable it by default for languages like
13025 C++ which normally require exception handling, and disable it for
13026 languages like C that do not normally require it. However, you may need
13027 to enable this option when compiling C code that needs to interoperate
13028 properly with exception handlers written in C++. You may also wish to
13029 disable this option if you are compiling older C++ programs that don't
13030 use exception handling.
13032 @item -fnon-call-exceptions
13033 @opindex fnon-call-exceptions
13034 Generate code that allows trapping instructions to throw exceptions.
13035 Note that this requires platform-specific runtime support that does
13036 not exist everywhere. Moreover, it only allows @emph{trapping}
13037 instructions to throw exceptions, i.e.@: memory references or floating
13038 point instructions. It does not allow exceptions to be thrown from
13039 arbitrary signal handlers such as @code{SIGALRM}.
13041 @item -funwind-tables
13042 @opindex funwind-tables
13043 Similar to @option{-fexceptions}, except that it will just generate any needed
13044 static data, but will not affect the generated code in any other way.
13045 You will normally not enable this option; instead, a language processor
13046 that needs this handling would enable it on your behalf.
13048 @item -fasynchronous-unwind-tables
13049 @opindex fasynchronous-unwind-tables
13050 Generate unwind table in dwarf2 format, if supported by target machine. The
13051 table is exact at each instruction boundary, so it can be used for stack
13052 unwinding from asynchronous events (such as debugger or garbage collector).
13054 @item -fpcc-struct-return
13055 @opindex fpcc-struct-return
13056 Return ``short'' @code{struct} and @code{union} values in memory like
13057 longer ones, rather than in registers. This convention is less
13058 efficient, but it has the advantage of allowing intercallability between
13059 GCC-compiled files and files compiled with other compilers, particularly
13060 the Portable C Compiler (pcc).
13062 The precise convention for returning structures in memory depends
13063 on the target configuration macros.
13065 Short structures and unions are those whose size and alignment match
13066 that of some integer type.
13068 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13069 switch is not binary compatible with code compiled with the
13070 @option{-freg-struct-return} switch.
13071 Use it to conform to a non-default application binary interface.
13073 @item -freg-struct-return
13074 @opindex freg-struct-return
13075 Return @code{struct} and @code{union} values in registers when possible.
13076 This is more efficient for small structures than
13077 @option{-fpcc-struct-return}.
13079 If you specify neither @option{-fpcc-struct-return} nor
13080 @option{-freg-struct-return}, GCC defaults to whichever convention is
13081 standard for the target. If there is no standard convention, GCC
13082 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13083 the principal compiler. In those cases, we can choose the standard, and
13084 we chose the more efficient register return alternative.
13086 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13087 switch is not binary compatible with code compiled with the
13088 @option{-fpcc-struct-return} switch.
13089 Use it to conform to a non-default application binary interface.
13091 @item -fshort-enums
13092 @opindex fshort-enums
13093 Allocate to an @code{enum} type only as many bytes as it needs for the
13094 declared range of possible values. Specifically, the @code{enum} type
13095 will be equivalent to the smallest integer type which has enough room.
13097 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13098 code that is not binary compatible with code generated without that switch.
13099 Use it to conform to a non-default application binary interface.
13101 @item -fshort-double
13102 @opindex fshort-double
13103 Use the same size for @code{double} as for @code{float}.
13105 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13106 code that is not binary compatible with code generated without that switch.
13107 Use it to conform to a non-default application binary interface.
13109 @item -fshort-wchar
13110 @opindex fshort-wchar
13111 Override the underlying type for @samp{wchar_t} to be @samp{short
13112 unsigned int} instead of the default for the target. This option is
13113 useful for building programs to run under WINE@.
13115 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13116 code that is not binary compatible with code generated without that switch.
13117 Use it to conform to a non-default application binary interface.
13120 @opindex fno-common
13121 In C, allocate even uninitialized global variables in the data section of the
13122 object file, rather than generating them as common blocks. This has the
13123 effect that if the same variable is declared (without @code{extern}) in
13124 two different compilations, you will get an error when you link them.
13125 The only reason this might be useful is if you wish to verify that the
13126 program will work on other systems which always work this way.
13130 Ignore the @samp{#ident} directive.
13132 @item -finhibit-size-directive
13133 @opindex finhibit-size-directive
13134 Don't output a @code{.size} assembler directive, or anything else that
13135 would cause trouble if the function is split in the middle, and the
13136 two halves are placed at locations far apart in memory. This option is
13137 used when compiling @file{crtstuff.c}; you should not need to use it
13140 @item -fverbose-asm
13141 @opindex fverbose-asm
13142 Put extra commentary information in the generated assembly code to
13143 make it more readable. This option is generally only of use to those
13144 who actually need to read the generated assembly code (perhaps while
13145 debugging the compiler itself).
13147 @option{-fno-verbose-asm}, the default, causes the
13148 extra information to be omitted and is useful when comparing two assembler
13153 @cindex global offset table
13155 Generate position-independent code (PIC) suitable for use in a shared
13156 library, if supported for the target machine. Such code accesses all
13157 constant addresses through a global offset table (GOT)@. The dynamic
13158 loader resolves the GOT entries when the program starts (the dynamic
13159 loader is not part of GCC; it is part of the operating system). If
13160 the GOT size for the linked executable exceeds a machine-specific
13161 maximum size, you get an error message from the linker indicating that
13162 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13163 instead. (These maximums are 8k on the SPARC and 32k
13164 on the m68k and RS/6000. The 386 has no such limit.)
13166 Position-independent code requires special support, and therefore works
13167 only on certain machines. For the 386, GCC supports PIC for System V
13168 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13169 position-independent.
13171 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13176 If supported for the target machine, emit position-independent code,
13177 suitable for dynamic linking and avoiding any limit on the size of the
13178 global offset table. This option makes a difference on the m68k,
13179 PowerPC and SPARC@.
13181 Position-independent code requires special support, and therefore works
13182 only on certain machines.
13184 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13191 These options are similar to @option{-fpic} and @option{-fPIC}, but
13192 generated position independent code can be only linked into executables.
13193 Usually these options are used when @option{-pie} GCC option will be
13194 used during linking.
13196 @item -fno-jump-tables
13197 @opindex fno-jump-tables
13198 Do not use jump tables for switch statements even where it would be
13199 more efficient than other code generation strategies. This option is
13200 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13201 building code which forms part of a dynamic linker and cannot
13202 reference the address of a jump table. On some targets, jump tables
13203 do not require a GOT and this option is not needed.
13205 @item -ffixed-@var{reg}
13207 Treat the register named @var{reg} as a fixed register; generated code
13208 should never refer to it (except perhaps as a stack pointer, frame
13209 pointer or in some other fixed role).
13211 @var{reg} must be the name of a register. The register names accepted
13212 are machine-specific and are defined in the @code{REGISTER_NAMES}
13213 macro in the machine description macro file.
13215 This flag does not have a negative form, because it specifies a
13218 @item -fcall-used-@var{reg}
13219 @opindex fcall-used
13220 Treat the register named @var{reg} as an allocable register that is
13221 clobbered by function calls. It may be allocated for temporaries or
13222 variables that do not live across a call. Functions compiled this way
13223 will not save and restore the register @var{reg}.
13225 It is an error to used this flag with the frame pointer or stack pointer.
13226 Use of this flag for other registers that have fixed pervasive roles in
13227 the machine's execution model will produce disastrous results.
13229 This flag does not have a negative form, because it specifies a
13232 @item -fcall-saved-@var{reg}
13233 @opindex fcall-saved
13234 Treat the register named @var{reg} as an allocable register saved by
13235 functions. It may be allocated even for temporaries or variables that
13236 live across a call. Functions compiled this way will save and restore
13237 the register @var{reg} if they use it.
13239 It is an error to used this flag with the frame pointer or stack pointer.
13240 Use of this flag for other registers that have fixed pervasive roles in
13241 the machine's execution model will produce disastrous results.
13243 A different sort of disaster will result from the use of this flag for
13244 a register in which function values may be returned.
13246 This flag does not have a negative form, because it specifies a
13249 @item -fpack-struct[=@var{n}]
13250 @opindex fpack-struct
13251 Without a value specified, pack all structure members together without
13252 holes. When a value is specified (which must be a small power of two), pack
13253 structure members according to this value, representing the maximum
13254 alignment (that is, objects with default alignment requirements larger than
13255 this will be output potentially unaligned at the next fitting location.
13257 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13258 code that is not binary compatible with code generated without that switch.
13259 Additionally, it makes the code suboptimal.
13260 Use it to conform to a non-default application binary interface.
13262 @item -finstrument-functions
13263 @opindex finstrument-functions
13264 Generate instrumentation calls for entry and exit to functions. Just
13265 after function entry and just before function exit, the following
13266 profiling functions will be called with the address of the current
13267 function and its call site. (On some platforms,
13268 @code{__builtin_return_address} does not work beyond the current
13269 function, so the call site information may not be available to the
13270 profiling functions otherwise.)
13273 void __cyg_profile_func_enter (void *this_fn,
13275 void __cyg_profile_func_exit (void *this_fn,
13279 The first argument is the address of the start of the current function,
13280 which may be looked up exactly in the symbol table.
13282 This instrumentation is also done for functions expanded inline in other
13283 functions. The profiling calls will indicate where, conceptually, the
13284 inline function is entered and exited. This means that addressable
13285 versions of such functions must be available. If all your uses of a
13286 function are expanded inline, this may mean an additional expansion of
13287 code size. If you use @samp{extern inline} in your C code, an
13288 addressable version of such functions must be provided. (This is
13289 normally the case anyways, but if you get lucky and the optimizer always
13290 expands the functions inline, you might have gotten away without
13291 providing static copies.)
13293 A function may be given the attribute @code{no_instrument_function}, in
13294 which case this instrumentation will not be done. This can be used, for
13295 example, for the profiling functions listed above, high-priority
13296 interrupt routines, and any functions from which the profiling functions
13297 cannot safely be called (perhaps signal handlers, if the profiling
13298 routines generate output or allocate memory).
13300 @item -fstack-check
13301 @opindex fstack-check
13302 Generate code to verify that you do not go beyond the boundary of the
13303 stack. You should specify this flag if you are running in an
13304 environment with multiple threads, but only rarely need to specify it in
13305 a single-threaded environment since stack overflow is automatically
13306 detected on nearly all systems if there is only one stack.
13308 Note that this switch does not actually cause checking to be done; the
13309 operating system must do that. The switch causes generation of code
13310 to ensure that the operating system sees the stack being extended.
13312 @item -fstack-limit-register=@var{reg}
13313 @itemx -fstack-limit-symbol=@var{sym}
13314 @itemx -fno-stack-limit
13315 @opindex fstack-limit-register
13316 @opindex fstack-limit-symbol
13317 @opindex fno-stack-limit
13318 Generate code to ensure that the stack does not grow beyond a certain value,
13319 either the value of a register or the address of a symbol. If the stack
13320 would grow beyond the value, a signal is raised. For most targets,
13321 the signal is raised before the stack overruns the boundary, so
13322 it is possible to catch the signal without taking special precautions.
13324 For instance, if the stack starts at absolute address @samp{0x80000000}
13325 and grows downwards, you can use the flags
13326 @option{-fstack-limit-symbol=__stack_limit} and
13327 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13328 of 128KB@. Note that this may only work with the GNU linker.
13330 @cindex aliasing of parameters
13331 @cindex parameters, aliased
13332 @item -fargument-alias
13333 @itemx -fargument-noalias
13334 @itemx -fargument-noalias-global
13335 @opindex fargument-alias
13336 @opindex fargument-noalias
13337 @opindex fargument-noalias-global
13338 Specify the possible relationships among parameters and between
13339 parameters and global data.
13341 @option{-fargument-alias} specifies that arguments (parameters) may
13342 alias each other and may alias global storage.@*
13343 @option{-fargument-noalias} specifies that arguments do not alias
13344 each other, but may alias global storage.@*
13345 @option{-fargument-noalias-global} specifies that arguments do not
13346 alias each other and do not alias global storage.
13348 Each language will automatically use whatever option is required by
13349 the language standard. You should not need to use these options yourself.
13351 @item -fleading-underscore
13352 @opindex fleading-underscore
13353 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13354 change the way C symbols are represented in the object file. One use
13355 is to help link with legacy assembly code.
13357 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13358 generate code that is not binary compatible with code generated without that
13359 switch. Use it to conform to a non-default application binary interface.
13360 Not all targets provide complete support for this switch.
13362 @item -ftls-model=@var{model}
13363 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13364 The @var{model} argument should be one of @code{global-dynamic},
13365 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13367 The default without @option{-fpic} is @code{initial-exec}; with
13368 @option{-fpic} the default is @code{global-dynamic}.
13370 @item -fvisibility=@var{default|internal|hidden|protected}
13371 @opindex fvisibility
13372 Set the default ELF image symbol visibility to the specified option---all
13373 symbols will be marked with this unless overridden within the code.
13374 Using this feature can very substantially improve linking and
13375 load times of shared object libraries, produce more optimized
13376 code, provide near-perfect API export and prevent symbol clashes.
13377 It is @strong{strongly} recommended that you use this in any shared objects
13380 Despite the nomenclature, @code{default} always means public ie;
13381 available to be linked against from outside the shared object.
13382 @code{protected} and @code{internal} are pretty useless in real-world
13383 usage so the only other commonly used option will be @code{hidden}.
13384 The default if @option{-fvisibility} isn't specified is
13385 @code{default}, i.e., make every
13386 symbol public---this causes the same behavior as previous versions of
13389 A good explanation of the benefits offered by ensuring ELF
13390 symbols have the correct visibility is given by ``How To Write
13391 Shared Libraries'' by Ulrich Drepper (which can be found at
13392 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13393 solution made possible by this option to marking things hidden when
13394 the default is public is to make the default hidden and mark things
13395 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13396 and @code{__attribute__ ((visibility("default")))} instead of
13397 @code{__declspec(dllexport)} you get almost identical semantics with
13398 identical syntax. This is a great boon to those working with
13399 cross-platform projects.
13401 For those adding visibility support to existing code, you may find
13402 @samp{#pragma GCC visibility} of use. This works by you enclosing
13403 the declarations you wish to set visibility for with (for example)
13404 @samp{#pragma GCC visibility push(hidden)} and
13405 @samp{#pragma GCC visibility pop}.
13406 Bear in mind that symbol visibility should be viewed @strong{as
13407 part of the API interface contract} and thus all new code should
13408 always specify visibility when it is not the default ie; declarations
13409 only for use within the local DSO should @strong{always} be marked explicitly
13410 as hidden as so to avoid PLT indirection overheads---making this
13411 abundantly clear also aids readability and self-documentation of the code.
13412 Note that due to ISO C++ specification requirements, operator new and
13413 operator delete must always be of default visibility.
13415 An overview of these techniques, their benefits and how to use them
13416 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13420 @cindex openmp parallel
13421 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
13422 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
13423 compiler generates parallel code according to the OpenMP Application
13424 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
13430 @node Environment Variables
13431 @section Environment Variables Affecting GCC
13432 @cindex environment variables
13434 @c man begin ENVIRONMENT
13435 This section describes several environment variables that affect how GCC
13436 operates. Some of them work by specifying directories or prefixes to use
13437 when searching for various kinds of files. Some are used to specify other
13438 aspects of the compilation environment.
13440 Note that you can also specify places to search using options such as
13441 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13442 take precedence over places specified using environment variables, which
13443 in turn take precedence over those specified by the configuration of GCC@.
13444 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13445 GNU Compiler Collection (GCC) Internals}.
13450 @c @itemx LC_COLLATE
13452 @c @itemx LC_MONETARY
13453 @c @itemx LC_NUMERIC
13458 @c @findex LC_COLLATE
13459 @findex LC_MESSAGES
13460 @c @findex LC_MONETARY
13461 @c @findex LC_NUMERIC
13465 These environment variables control the way that GCC uses
13466 localization information that allow GCC to work with different
13467 national conventions. GCC inspects the locale categories
13468 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13469 so. These locale categories can be set to any value supported by your
13470 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13471 Kingdom encoded in UTF-8.
13473 The @env{LC_CTYPE} environment variable specifies character
13474 classification. GCC uses it to determine the character boundaries in
13475 a string; this is needed for some multibyte encodings that contain quote
13476 and escape characters that would otherwise be interpreted as a string
13479 The @env{LC_MESSAGES} environment variable specifies the language to
13480 use in diagnostic messages.
13482 If the @env{LC_ALL} environment variable is set, it overrides the value
13483 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13484 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13485 environment variable. If none of these variables are set, GCC
13486 defaults to traditional C English behavior.
13490 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13491 files. GCC uses temporary files to hold the output of one stage of
13492 compilation which is to be used as input to the next stage: for example,
13493 the output of the preprocessor, which is the input to the compiler
13496 @item GCC_EXEC_PREFIX
13497 @findex GCC_EXEC_PREFIX
13498 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13499 names of the subprograms executed by the compiler. No slash is added
13500 when this prefix is combined with the name of a subprogram, but you can
13501 specify a prefix that ends with a slash if you wish.
13503 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13504 an appropriate prefix to use based on the pathname it was invoked with.
13506 If GCC cannot find the subprogram using the specified prefix, it
13507 tries looking in the usual places for the subprogram.
13509 The default value of @env{GCC_EXEC_PREFIX} is
13510 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13511 of @code{prefix} when you ran the @file{configure} script.
13513 Other prefixes specified with @option{-B} take precedence over this prefix.
13515 This prefix is also used for finding files such as @file{crt0.o} that are
13518 In addition, the prefix is used in an unusual way in finding the
13519 directories to search for header files. For each of the standard
13520 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13521 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13522 replacing that beginning with the specified prefix to produce an
13523 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13524 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13525 These alternate directories are searched first; the standard directories
13528 @item COMPILER_PATH
13529 @findex COMPILER_PATH
13530 The value of @env{COMPILER_PATH} is a colon-separated list of
13531 directories, much like @env{PATH}. GCC tries the directories thus
13532 specified when searching for subprograms, if it can't find the
13533 subprograms using @env{GCC_EXEC_PREFIX}.
13536 @findex LIBRARY_PATH
13537 The value of @env{LIBRARY_PATH} is a colon-separated list of
13538 directories, much like @env{PATH}. When configured as a native compiler,
13539 GCC tries the directories thus specified when searching for special
13540 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13541 using GCC also uses these directories when searching for ordinary
13542 libraries for the @option{-l} option (but directories specified with
13543 @option{-L} come first).
13547 @cindex locale definition
13548 This variable is used to pass locale information to the compiler. One way in
13549 which this information is used is to determine the character set to be used
13550 when character literals, string literals and comments are parsed in C and C++.
13551 When the compiler is configured to allow multibyte characters,
13552 the following values for @env{LANG} are recognized:
13556 Recognize JIS characters.
13558 Recognize SJIS characters.
13560 Recognize EUCJP characters.
13563 If @env{LANG} is not defined, or if it has some other value, then the
13564 compiler will use mblen and mbtowc as defined by the default locale to
13565 recognize and translate multibyte characters.
13569 Some additional environments variables affect the behavior of the
13572 @include cppenv.texi
13576 @node Precompiled Headers
13577 @section Using Precompiled Headers
13578 @cindex precompiled headers
13579 @cindex speed of compilation
13581 Often large projects have many header files that are included in every
13582 source file. The time the compiler takes to process these header files
13583 over and over again can account for nearly all of the time required to
13584 build the project. To make builds faster, GCC allows users to
13585 `precompile' a header file; then, if builds can use the precompiled
13586 header file they will be much faster.
13588 To create a precompiled header file, simply compile it as you would any
13589 other file, if necessary using the @option{-x} option to make the driver
13590 treat it as a C or C++ header file. You will probably want to use a
13591 tool like @command{make} to keep the precompiled header up-to-date when
13592 the headers it contains change.
13594 A precompiled header file will be searched for when @code{#include} is
13595 seen in the compilation. As it searches for the included file
13596 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13597 compiler looks for a precompiled header in each directory just before it
13598 looks for the include file in that directory. The name searched for is
13599 the name specified in the @code{#include} with @samp{.gch} appended. If
13600 the precompiled header file can't be used, it is ignored.
13602 For instance, if you have @code{#include "all.h"}, and you have
13603 @file{all.h.gch} in the same directory as @file{all.h}, then the
13604 precompiled header file will be used if possible, and the original
13605 header will be used otherwise.
13607 Alternatively, you might decide to put the precompiled header file in a
13608 directory and use @option{-I} to ensure that directory is searched
13609 before (or instead of) the directory containing the original header.
13610 Then, if you want to check that the precompiled header file is always
13611 used, you can put a file of the same name as the original header in this
13612 directory containing an @code{#error} command.
13614 This also works with @option{-include}. So yet another way to use
13615 precompiled headers, good for projects not designed with precompiled
13616 header files in mind, is to simply take most of the header files used by
13617 a project, include them from another header file, precompile that header
13618 file, and @option{-include} the precompiled header. If the header files
13619 have guards against multiple inclusion, they will be skipped because
13620 they've already been included (in the precompiled header).
13622 If you need to precompile the same header file for different
13623 languages, targets, or compiler options, you can instead make a
13624 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13625 header in the directory, perhaps using @option{-o}. It doesn't matter
13626 what you call the files in the directory, every precompiled header in
13627 the directory will be considered. The first precompiled header
13628 encountered in the directory that is valid for this compilation will
13629 be used; they're searched in no particular order.
13631 There are many other possibilities, limited only by your imagination,
13632 good sense, and the constraints of your build system.
13634 A precompiled header file can be used only when these conditions apply:
13638 Only one precompiled header can be used in a particular compilation.
13641 A precompiled header can't be used once the first C token is seen. You
13642 can have preprocessor directives before a precompiled header; you can
13643 even include a precompiled header from inside another header, so long as
13644 there are no C tokens before the @code{#include}.
13647 The precompiled header file must be produced for the same language as
13648 the current compilation. You can't use a C precompiled header for a C++
13652 The precompiled header file must have been produced by the same compiler
13653 binary as the current compilation is using.
13656 Any macros defined before the precompiled header is included must
13657 either be defined in the same way as when the precompiled header was
13658 generated, or must not affect the precompiled header, which usually
13659 means that they don't appear in the precompiled header at all.
13661 The @option{-D} option is one way to define a macro before a
13662 precompiled header is included; using a @code{#define} can also do it.
13663 There are also some options that define macros implicitly, like
13664 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13667 @item If debugging information is output when using the precompiled
13668 header, using @option{-g} or similar, the same kind of debugging information
13669 must have been output when building the precompiled header. However,
13670 a precompiled header built using @option{-g} can be used in a compilation
13671 when no debugging information is being output.
13673 @item The same @option{-m} options must generally be used when building
13674 and using the precompiled header. @xref{Submodel Options},
13675 for any cases where this rule is relaxed.
13677 @item Each of the following options must be the same when building and using
13678 the precompiled header:
13680 @gccoptlist{-fexceptions -funit-at-a-time}
13683 Some other command-line options starting with @option{-f},
13684 @option{-p}, or @option{-O} must be defined in the same way as when
13685 the precompiled header was generated. At present, it's not clear
13686 which options are safe to change and which are not; the safest choice
13687 is to use exactly the same options when generating and using the
13688 precompiled header. The following are known to be safe:
13690 @gccoptlist{-fmessage-length= -fpreprocessed
13691 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13692 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13697 For all of these except the last, the compiler will automatically
13698 ignore the precompiled header if the conditions aren't met. If you
13699 find an option combination that doesn't work and doesn't cause the
13700 precompiled header to be ignored, please consider filing a bug report,
13703 If you do use differing options when generating and using the
13704 precompiled header, the actual behavior will be a mixture of the
13705 behavior for the options. For instance, if you use @option{-g} to
13706 generate the precompiled header but not when using it, you may or may
13707 not get debugging information for routines in the precompiled header.
13709 @node Running Protoize
13710 @section Running Protoize
13712 The program @code{protoize} is an optional part of GCC@. You can use
13713 it to add prototypes to a program, thus converting the program to ISO
13714 C in one respect. The companion program @code{unprotoize} does the
13715 reverse: it removes argument types from any prototypes that are found.
13717 When you run these programs, you must specify a set of source files as
13718 command line arguments. The conversion programs start out by compiling
13719 these files to see what functions they define. The information gathered
13720 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13722 After scanning comes actual conversion. The specified files are all
13723 eligible to be converted; any files they include (whether sources or
13724 just headers) are eligible as well.
13726 But not all the eligible files are converted. By default,
13727 @code{protoize} and @code{unprotoize} convert only source and header
13728 files in the current directory. You can specify additional directories
13729 whose files should be converted with the @option{-d @var{directory}}
13730 option. You can also specify particular files to exclude with the
13731 @option{-x @var{file}} option. A file is converted if it is eligible, its
13732 directory name matches one of the specified directory names, and its
13733 name within the directory has not been excluded.
13735 Basic conversion with @code{protoize} consists of rewriting most
13736 function definitions and function declarations to specify the types of
13737 the arguments. The only ones not rewritten are those for varargs
13740 @code{protoize} optionally inserts prototype declarations at the
13741 beginning of the source file, to make them available for any calls that
13742 precede the function's definition. Or it can insert prototype
13743 declarations with block scope in the blocks where undeclared functions
13746 Basic conversion with @code{unprotoize} consists of rewriting most
13747 function declarations to remove any argument types, and rewriting
13748 function definitions to the old-style pre-ISO form.
13750 Both conversion programs print a warning for any function declaration or
13751 definition that they can't convert. You can suppress these warnings
13754 The output from @code{protoize} or @code{unprotoize} replaces the
13755 original source file. The original file is renamed to a name ending
13756 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13757 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13758 for DOS) file already exists, then the source file is simply discarded.
13760 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13761 scan the program and collect information about the functions it uses.
13762 So neither of these programs will work until GCC is installed.
13764 Here is a table of the options you can use with @code{protoize} and
13765 @code{unprotoize}. Each option works with both programs unless
13769 @item -B @var{directory}
13770 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13771 usual directory (normally @file{/usr/local/lib}). This file contains
13772 prototype information about standard system functions. This option
13773 applies only to @code{protoize}.
13775 @item -c @var{compilation-options}
13776 Use @var{compilation-options} as the options when running @command{gcc} to
13777 produce the @samp{.X} files. The special option @option{-aux-info} is
13778 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13780 Note that the compilation options must be given as a single argument to
13781 @code{protoize} or @code{unprotoize}. If you want to specify several
13782 @command{gcc} options, you must quote the entire set of compilation options
13783 to make them a single word in the shell.
13785 There are certain @command{gcc} arguments that you cannot use, because they
13786 would produce the wrong kind of output. These include @option{-g},
13787 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13788 the @var{compilation-options}, they are ignored.
13791 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13792 systems) instead of @samp{.c}. This is convenient if you are converting
13793 a C program to C++. This option applies only to @code{protoize}.
13796 Add explicit global declarations. This means inserting explicit
13797 declarations at the beginning of each source file for each function
13798 that is called in the file and was not declared. These declarations
13799 precede the first function definition that contains a call to an
13800 undeclared function. This option applies only to @code{protoize}.
13802 @item -i @var{string}
13803 Indent old-style parameter declarations with the string @var{string}.
13804 This option applies only to @code{protoize}.
13806 @code{unprotoize} converts prototyped function definitions to old-style
13807 function definitions, where the arguments are declared between the
13808 argument list and the initial @samp{@{}. By default, @code{unprotoize}
13809 uses five spaces as the indentation. If you want to indent with just
13810 one space instead, use @option{-i " "}.
13813 Keep the @samp{.X} files. Normally, they are deleted after conversion
13817 Add explicit local declarations. @code{protoize} with @option{-l} inserts
13818 a prototype declaration for each function in each block which calls the
13819 function without any declaration. This option applies only to
13823 Make no real changes. This mode just prints information about the conversions
13824 that would have been done without @option{-n}.
13827 Make no @samp{.save} files. The original files are simply deleted.
13828 Use this option with caution.
13830 @item -p @var{program}
13831 Use the program @var{program} as the compiler. Normally, the name
13832 @file{gcc} is used.
13835 Work quietly. Most warnings are suppressed.
13838 Print the version number, just like @option{-v} for @command{gcc}.
13841 If you need special compiler options to compile one of your program's
13842 source files, then you should generate that file's @samp{.X} file
13843 specially, by running @command{gcc} on that source file with the
13844 appropriate options and the option @option{-aux-info}. Then run
13845 @code{protoize} on the entire set of files. @code{protoize} will use
13846 the existing @samp{.X} file because it is newer than the source file.
13850 gcc -Dfoo=bar file1.c -aux-info file1.X
13855 You need to include the special files along with the rest in the
13856 @code{protoize} command, even though their @samp{.X} files already
13857 exist, because otherwise they won't get converted.
13859 @xref{Protoize Caveats}, for more information on how to use
13860 @code{protoize} successfully.