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 -fopenmp -fms-extensions @gol
170 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
171 -fallow-single-precision -fcond-mismatch @gol
172 -fsigned-bitfields -fsigned-char @gol
173 -funsigned-bitfields -funsigned-char}
175 @item C++ Language Options
176 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
177 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
178 -fconserve-space -ffriend-injection @gol
179 -fno-elide-constructors @gol
180 -fno-enforce-eh-specs @gol
181 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
182 -fno-implicit-templates @gol
183 -fno-implicit-inline-templates @gol
184 -fno-implement-inlines -fms-extensions @gol
185 -fno-nonansi-builtins -fno-operator-names @gol
186 -fno-optional-diags -fpermissive @gol
187 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
188 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
189 -fno-default-inline -fvisibility-inlines-hidden @gol
190 -Wabi -Wctor-dtor-privacy @gol
191 -Wnon-virtual-dtor -Wreorder @gol
192 -Weffc++ -Wno-deprecated -Wstrict-null-sentinel @gol
193 -Wno-non-template-friend -Wold-style-cast @gol
194 -Woverloaded-virtual -Wno-pmf-conversions @gol
197 @item Objective-C and Objective-C++ Language Options
198 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
199 Objective-C and Objective-C++ Dialects}.
201 -fconstant-string-class=@var{class-name} @gol
202 -fgnu-runtime -fnext-runtime @gol
203 -fno-nil-receivers @gol
204 -fobjc-call-cxx-cdtors @gol
205 -fobjc-direct-dispatch @gol
206 -fobjc-exceptions @gol
208 -freplace-objc-classes @gol
211 -Wassign-intercept @gol
212 -Wno-protocol -Wselector @gol
213 -Wstrict-selector-match @gol
214 -Wundeclared-selector}
216 @item Language Independent Options
217 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
218 @gccoptlist{-fmessage-length=@var{n} @gol
219 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}} @gol
220 -fdiagnostics-show-options
222 @item Warning Options
223 @xref{Warning Options,,Options to Request or Suppress Warnings}.
224 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
225 -w -Wextra -Wall -Waggregate-return -Walways-true -Wno-attributes @gol
226 -Wc++-compat -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
227 -Wconversion -Wno-deprecated-declarations @gol
228 -Wdisabled-optimization -Wno-div-by-zero -Wno-endif-labels @gol
229 -Werror -Werror-* -Werror-implicit-function-declaration @gol
230 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
231 -Wno-format-extra-args -Wformat-nonliteral @gol
232 -Wformat-security -Wformat-y2k @gol
233 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
234 -Wimport -Wno-import -Winit-self -Winline @gol
235 -Wno-int-to-pointer-cast @gol
236 -Wno-invalid-offsetof -Winvalid-pch @gol
237 -Wlarger-than-@var{len} -Wunsafe-loop-optimizations -Wlong-long @gol
238 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
239 -Wmissing-format-attribute -Wmissing-include-dirs @gol
240 -Wmissing-noreturn @gol
241 -Wno-multichar -Wnonnull -Wno-overflow @gol
242 -Woverlength-strings -Wpacked -Wpadded @gol
243 -Wparentheses -Wpointer-arith -Wno-pointer-to-int-cast @gol
244 -Wredundant-decls @gol
245 -Wreturn-type -Wsequence-point -Wshadow @gol
246 -Wsign-compare -Wstack-protector @gol
247 -Wstrict-aliasing -Wstrict-aliasing=2 @gol
248 -Wstring-literal-comparison @gol
249 -Wswitch -Wswitch-default -Wswitch-enum @gol
250 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
251 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
252 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
253 -Wunused-value -Wunused-variable -Wvariadic-macros @gol
254 -Wvolatile-register-var -Wwrite-strings}
256 @item C-only Warning Options
257 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
258 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
259 -Wstrict-prototypes -Wtraditional @gol
260 -Wdeclaration-after-statement -Wpointer-sign}
262 @item Debugging Options
263 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
264 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
265 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
266 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
267 -fdump-ipa-all -fdump-ipa-cgraph @gol
269 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
270 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
271 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
272 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
274 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
275 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
276 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
277 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
278 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
279 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
280 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
281 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
282 -fdump-tree-nrv -fdump-tree-vect @gol
283 -fdump-tree-sink @gol
284 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
285 -fdump-tree-salias @gol
286 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
288 -ftree-vectorizer-verbose=@var{n} @gol
289 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
290 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
291 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
292 -fmem-report -fprofile-arcs @gol
293 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
294 -ftest-coverage -ftime-report -fvar-tracking @gol
295 -g -g@var{level} -gcoff -gdwarf-2 @gol
296 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
297 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
298 -print-multi-directory -print-multi-lib @gol
299 -print-prog-name=@var{program} -print-search-dirs -Q @gol
302 @item Optimization Options
303 @xref{Optimize Options,,Options that Control Optimization}.
304 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
305 -falign-labels=@var{n} -falign-loops=@var{n} @gol
306 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
307 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
308 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
309 -fcaller-saves -fcprop-registers -fcse-follow-jumps @gol
310 -fcse-skip-blocks -fcx-limited-range -fdata-sections @gol
311 -fdelayed-branch -fdelete-null-pointer-checks -fearly-inlining @gol
312 -fexpensive-optimizations -ffast-math -ffloat-store @gol
313 -fforce-addr -ffunction-sections @gol
314 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
315 -fcrossjumping -fif-conversion -fif-conversion2 @gol
316 -finline-functions -finline-functions-called-once @gol
317 -finline-limit=@var{n} -fkeep-inline-functions @gol
318 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
319 -fmodulo-sched -fno-branch-count-reg @gol
320 -fno-default-inline -fno-defer-pop -fmove-loop-invariants @gol
321 -fno-function-cse -fno-guess-branch-probability @gol
322 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
323 -funsafe-math-optimizations -funsafe-loop-optimizations -ffinite-math-only @gol
324 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
325 -fomit-frame-pointer -foptimize-register-move @gol
326 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
327 -fprofile-generate -fprofile-use @gol
328 -fregmove -frename-registers @gol
329 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
330 -frerun-cse-after-loop @gol
331 -frounding-math -frtl-abstract-sequences @gol
332 -fschedule-insns -fschedule-insns2 @gol
333 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
334 -fsched-spec-load-dangerous @gol
335 -fsched-stalled-insns=@var{n} -fsched-stalled-insns-dep=@var{n} @gol
336 -fsched2-use-superblocks @gol
337 -fsched2-use-traces -fsee -freschedule-modulo-scheduled-loops @gol
338 -fsection-anchors -fsignaling-nans -fsingle-precision-constant @gol
339 -fstack-protector -fstack-protector-all @gol
340 -fstrict-aliasing -ftracer -fthread-jumps @gol
341 -funroll-all-loops -funroll-loops -fpeel-loops @gol
342 -fsplit-ivs-in-unroller -funswitch-loops @gol
343 -fvariable-expansion-in-unroller @gol
344 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
345 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
346 -ftree-dominator-opts -ftree-dse -ftree-copyrename -ftree-sink @gol
347 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
348 -ftree-vect-loop-version -ftree-salias -fipa-pta -fweb @gol
349 -ftree-copy-prop -ftree-store-ccp -ftree-store-copy-prop -fwhole-program @gol
350 --param @var{name}=@var{value}
351 -O -O0 -O1 -O2 -O3 -Os}
353 @item Preprocessor Options
354 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
355 @gccoptlist{-A@var{question}=@var{answer} @gol
356 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
357 -C -dD -dI -dM -dN @gol
358 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
359 -idirafter @var{dir} @gol
360 -include @var{file} -imacros @var{file} @gol
361 -iprefix @var{file} -iwithprefix @var{dir} @gol
362 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
363 -imultilib @var{dir} -isysroot @var{dir} @gol
364 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
365 -P -fworking-directory -remap @gol
366 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
367 -Xpreprocessor @var{option}}
369 @item Assembler Option
370 @xref{Assembler Options,,Passing Options to the Assembler}.
371 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
374 @xref{Link Options,,Options for Linking}.
375 @gccoptlist{@var{object-file-name} -l@var{library} @gol
376 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
377 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
378 -Wl,@var{option} -Xlinker @var{option} @gol
381 @item Directory Options
382 @xref{Directory Options,,Options for Directory Search}.
383 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
384 -specs=@var{file} -I- --sysroot=@var{dir}}
387 @c I wrote this xref this way to avoid overfull hbox. -- rms
388 @xref{Target Options}.
389 @gccoptlist{-V @var{version} -b @var{machine}}
391 @item Machine Dependent Options
392 @xref{Submodel Options,,Hardware Models and Configurations}.
393 @c This list is ordered alphanumerically by subsection name.
394 @c Try and put the significant identifier (CPU or system) first,
395 @c so users have a clue at guessing where the ones they want will be.
398 @gccoptlist{-EB -EL @gol
399 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
400 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
403 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
404 -mabi=@var{name} @gol
405 -mapcs-stack-check -mno-apcs-stack-check @gol
406 -mapcs-float -mno-apcs-float @gol
407 -mapcs-reentrant -mno-apcs-reentrant @gol
408 -msched-prolog -mno-sched-prolog @gol
409 -mlittle-endian -mbig-endian -mwords-little-endian @gol
410 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
411 -mthumb-interwork -mno-thumb-interwork @gol
412 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
413 -mstructure-size-boundary=@var{n} @gol
414 -mabort-on-noreturn @gol
415 -mlong-calls -mno-long-calls @gol
416 -msingle-pic-base -mno-single-pic-base @gol
417 -mpic-register=@var{reg} @gol
418 -mnop-fun-dllimport @gol
419 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
420 -mpoke-function-name @gol
422 -mtpcs-frame -mtpcs-leaf-frame @gol
423 -mcaller-super-interworking -mcallee-super-interworking @gol
427 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
428 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
430 @emph{Blackfin Options}
431 @gccoptlist{-momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
432 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
433 -mlow-64k -mno-low64k -mid-shared-library @gol
434 -mno-id-shared-library -mshared-library-id=@var{n} @gol
435 -mlong-calls -mno-long-calls}
438 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
439 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
440 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
441 -mstack-align -mdata-align -mconst-align @gol
442 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
443 -melf -maout -melinux -mlinux -sim -sim2 @gol
444 -mmul-bug-workaround -mno-mul-bug-workaround}
447 @gccoptlist{-mmac -mpush-args}
449 @emph{Darwin Options}
450 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
451 -arch_only -bind_at_load -bundle -bundle_loader @gol
452 -client_name -compatibility_version -current_version @gol
454 -dependency-file -dylib_file -dylinker_install_name @gol
455 -dynamic -dynamiclib -exported_symbols_list @gol
456 -filelist -flat_namespace -force_cpusubtype_ALL @gol
457 -force_flat_namespace -headerpad_max_install_names @gol
458 -image_base -init -install_name -keep_private_externs @gol
459 -multi_module -multiply_defined -multiply_defined_unused @gol
460 -noall_load -no_dead_strip_inits_and_terms @gol
461 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
462 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
463 -private_bundle -read_only_relocs -sectalign @gol
464 -sectobjectsymbols -whyload -seg1addr @gol
465 -sectcreate -sectobjectsymbols -sectorder @gol
466 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
467 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
468 -segprot -segs_read_only_addr -segs_read_write_addr @gol
469 -single_module -static -sub_library -sub_umbrella @gol
470 -twolevel_namespace -umbrella -undefined @gol
471 -unexported_symbols_list -weak_reference_mismatches @gol
472 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
475 @emph{DEC Alpha Options}
476 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
477 -mieee -mieee-with-inexact -mieee-conformant @gol
478 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
479 -mtrap-precision=@var{mode} -mbuild-constants @gol
480 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
481 -mbwx -mmax -mfix -mcix @gol
482 -mfloat-vax -mfloat-ieee @gol
483 -mexplicit-relocs -msmall-data -mlarge-data @gol
484 -msmall-text -mlarge-text @gol
485 -mmemory-latency=@var{time}}
487 @emph{DEC Alpha/VMS Options}
488 @gccoptlist{-mvms-return-codes}
491 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
492 -mhard-float -msoft-float @gol
493 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
494 -mdouble -mno-double @gol
495 -mmedia -mno-media -mmuladd -mno-muladd @gol
496 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
497 -mlinked-fp -mlong-calls -malign-labels @gol
498 -mlibrary-pic -macc-4 -macc-8 @gol
499 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
500 -moptimize-membar -mno-optimize-membar @gol
501 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
502 -mvliw-branch -mno-vliw-branch @gol
503 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
504 -mno-nested-cond-exec -mtomcat-stats @gol
508 @emph{GNU/Linux Options}
509 @gccoptlist{-muclibc}
511 @emph{H8/300 Options}
512 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
515 @gccoptlist{-march=@var{architecture-type} @gol
516 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
517 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
518 -mfixed-range=@var{register-range} @gol
519 -mjump-in-delay -mlinker-opt -mlong-calls @gol
520 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
521 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
522 -mno-jump-in-delay -mno-long-load-store @gol
523 -mno-portable-runtime -mno-soft-float @gol
524 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
525 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
526 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
527 -munix=@var{unix-std} -nolibdld -static -threads}
529 @emph{i386 and x86-64 Options}
530 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
531 -mfpmath=@var{unit} @gol
532 -masm=@var{dialect} -mno-fancy-math-387 @gol
533 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
534 -mno-wide-multiply -mrtd -malign-double @gol
535 -mpreferred-stack-boundary=@var{num} @gol
536 -mmmx -msse -msse2 -msse3 -m3dnow @gol
537 -mthreads -mno-align-stringops -minline-all-stringops @gol
538 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
539 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
541 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
542 -mcmodel=@var{code-model} @gol
543 -m32 -m64 -mlarge-data-threshold=@var{num}}
546 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
547 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
548 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
549 -minline-float-divide-max-throughput @gol
550 -minline-int-divide-min-latency @gol
551 -minline-int-divide-max-throughput @gol
552 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
553 -mno-dwarf2-asm -mearly-stop-bits @gol
554 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
555 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64 @gol
556 -mno-sched-br-data-spec -msched-ar-data-spec -mno-sched-control-spec @gol
557 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
558 -msched-ldc -mno-sched-control-ldc -mno-sched-spec-verbose @gol
559 -mno-sched-prefer-non-data-spec-insns @gol
560 -mno-sched-prefer-non-control-spec-insns @gol
561 -mno-sched-count-spec-in-critical-path}
563 @emph{M32R/D Options}
564 @gccoptlist{-m32r2 -m32rx -m32r @gol
566 -malign-loops -mno-align-loops @gol
567 -missue-rate=@var{number} @gol
568 -mbranch-cost=@var{number} @gol
569 -mmodel=@var{code-size-model-type} @gol
570 -msdata=@var{sdata-type} @gol
571 -mno-flush-func -mflush-func=@var{name} @gol
572 -mno-flush-trap -mflush-trap=@var{number} @gol
576 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
578 @emph{M680x0 Options}
579 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
580 -m68060 -mcpu32 -m5200 -mcfv4e -m68881 -mbitfield @gol
581 -mc68000 -mc68020 @gol
582 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
583 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
584 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
586 @emph{M68hc1x Options}
587 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
588 -mauto-incdec -minmax -mlong-calls -mshort @gol
589 -msoft-reg-count=@var{count}}
592 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
593 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
594 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
595 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
596 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
599 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
600 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
601 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
602 -mshared -mno-shared -mxgot -mno-xgot -mgp32 -mgp64 @gol
603 -mfp32 -mfp64 -mhard-float -msoft-float @gol
604 -msingle-float -mdouble-float -mdsp -mpaired-single -mips3d @gol
605 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
606 -G@var{num} -membedded-data -mno-embedded-data @gol
607 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
608 -msplit-addresses -mno-split-addresses @gol
609 -mexplicit-relocs -mno-explicit-relocs @gol
610 -mcheck-zero-division -mno-check-zero-division @gol
611 -mdivide-traps -mdivide-breaks @gol
612 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
613 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
614 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
615 -mfix-vr4120 -mno-fix-vr4120 -mfix-vr4130 @gol
616 -mfix-sb1 -mno-fix-sb1 @gol
617 -mflush-func=@var{func} -mno-flush-func @gol
618 -mbranch-likely -mno-branch-likely @gol
619 -mfp-exceptions -mno-fp-exceptions @gol
620 -mvr4130-align -mno-vr4130-align}
623 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
624 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
625 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
626 -mno-base-addresses -msingle-exit -mno-single-exit}
628 @emph{MN10300 Options}
629 @gccoptlist{-mmult-bug -mno-mult-bug @gol
630 -mam33 -mno-am33 @gol
631 -mam33-2 -mno-am33-2 @gol
632 -mreturn-pointer-on-d0 @gol
636 @gccoptlist{-mno-crt0 -mbacc -msim @gol
637 -march=@var{cpu-type} }
639 @emph{PDP-11 Options}
640 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
641 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
642 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
643 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
644 -mbranch-expensive -mbranch-cheap @gol
645 -msplit -mno-split -munix-asm -mdec-asm}
647 @emph{PowerPC Options}
648 See RS/6000 and PowerPC Options.
650 @emph{RS/6000 and PowerPC Options}
651 @gccoptlist{-mcpu=@var{cpu-type} @gol
652 -mtune=@var{cpu-type} @gol
653 -mpower -mno-power -mpower2 -mno-power2 @gol
654 -mpowerpc -mpowerpc64 -mno-powerpc @gol
655 -maltivec -mno-altivec @gol
656 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
657 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
658 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mfprnd -mno-fprnd @gol
659 -mnew-mnemonics -mold-mnemonics @gol
660 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
661 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
662 -malign-power -malign-natural @gol
663 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
664 -mstring -mno-string -mupdate -mno-update @gol
665 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
666 -mstrict-align -mno-strict-align -mrelocatable @gol
667 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
668 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
669 -mdynamic-no-pic -maltivec -mswdiv @gol
670 -mprioritize-restricted-insns=@var{priority} @gol
671 -msched-costly-dep=@var{dependence_type} @gol
672 -minsert-sched-nops=@var{scheme} @gol
673 -mcall-sysv -mcall-netbsd @gol
674 -maix-struct-return -msvr4-struct-return @gol
675 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
676 -misel -mno-isel @gol
677 -misel=yes -misel=no @gol
679 -mspe=yes -mspe=no @gol
680 -mvrsave -mno-vrsave @gol
681 -mmulhw -mno-mulhw @gol
682 -mdlmzb -mno-dlmzb @gol
683 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
684 -mprototype -mno-prototype @gol
685 -msim -mmvme -mads -myellowknife -memb -msdata @gol
686 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
688 @emph{S/390 and zSeries Options}
689 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
690 -mhard-float -msoft-float -mlong-double-64 -mlong-double-128 @gol
691 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
692 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
693 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
694 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
695 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
698 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
699 -m4-nofpu -m4-single-only -m4-single -m4 @gol
700 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
701 -m5-64media -m5-64media-nofpu @gol
702 -m5-32media -m5-32media-nofpu @gol
703 -m5-compact -m5-compact-nofpu @gol
704 -mb -ml -mdalign -mrelax @gol
705 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
706 -mieee -misize -mpadstruct -mspace @gol
707 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
708 -mdivsi3_libfunc=@var{name} @gol
709 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
713 @gccoptlist{-mcpu=@var{cpu-type} @gol
714 -mtune=@var{cpu-type} @gol
715 -mcmodel=@var{code-model} @gol
716 -m32 -m64 -mapp-regs -mno-app-regs @gol
717 -mfaster-structs -mno-faster-structs @gol
718 -mfpu -mno-fpu -mhard-float -msoft-float @gol
719 -mhard-quad-float -msoft-quad-float @gol
720 -mimpure-text -mno-impure-text -mlittle-endian @gol
721 -mstack-bias -mno-stack-bias @gol
722 -munaligned-doubles -mno-unaligned-doubles @gol
723 -mv8plus -mno-v8plus -mvis -mno-vis
724 -threads -pthreads -pthread}
726 @emph{System V Options}
727 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
729 @emph{TMS320C3x/C4x Options}
730 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
731 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
732 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
733 -mparallel-insns -mparallel-mpy -mpreserve-float}
736 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
737 -mprolog-function -mno-prolog-function -mspace @gol
738 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
739 -mapp-regs -mno-app-regs @gol
740 -mdisable-callt -mno-disable-callt @gol
746 @gccoptlist{-mg -mgnu -munix}
748 @emph{x86-64 Options}
749 See i386 and x86-64 Options.
751 @emph{Xstormy16 Options}
754 @emph{Xtensa Options}
755 @gccoptlist{-mconst16 -mno-const16 @gol
756 -mfused-madd -mno-fused-madd @gol
757 -mtext-section-literals -mno-text-section-literals @gol
758 -mtarget-align -mno-target-align @gol
759 -mlongcalls -mno-longcalls}
761 @emph{zSeries Options}
762 See S/390 and zSeries Options.
764 @item Code Generation Options
765 @xref{Code Gen Options,,Options for Code Generation Conventions}.
766 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
767 -ffixed-@var{reg} -fexceptions @gol
768 -fnon-call-exceptions -funwind-tables @gol
769 -fasynchronous-unwind-tables @gol
770 -finhibit-size-directive -finstrument-functions @gol
771 -fno-common -fno-ident @gol
772 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
773 -fno-jump-tables @gol
774 -freg-struct-return -fshort-enums @gol
775 -fshort-double -fshort-wchar @gol
776 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
777 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
778 -fargument-alias -fargument-noalias @gol
779 -fargument-noalias-global -fargument-noalias-anything
780 -fleading-underscore -ftls-model=@var{model} @gol
781 -ftrapv -fwrapv -fbounds-check @gol
786 * Overall Options:: Controlling the kind of output:
787 an executable, object files, assembler files,
788 or preprocessed source.
789 * C Dialect Options:: Controlling the variant of C language compiled.
790 * C++ Dialect Options:: Variations on C++.
791 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
793 * Language Independent Options:: Controlling how diagnostics should be
795 * Warning Options:: How picky should the compiler be?
796 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
797 * Optimize Options:: How much optimization?
798 * Preprocessor Options:: Controlling header files and macro definitions.
799 Also, getting dependency information for Make.
800 * Assembler Options:: Passing options to the assembler.
801 * Link Options:: Specifying libraries and so on.
802 * Directory Options:: Where to find header files and libraries.
803 Where to find the compiler executable files.
804 * Spec Files:: How to pass switches to sub-processes.
805 * Target Options:: Running a cross-compiler, or an old version of GCC.
808 @node Overall Options
809 @section Options Controlling the Kind of Output
811 Compilation can involve up to four stages: preprocessing, compilation
812 proper, assembly and linking, always in that order. GCC is capable of
813 preprocessing and compiling several files either into several
814 assembler input files, or into one assembler input file; then each
815 assembler input file produces an object file, and linking combines all
816 the object files (those newly compiled, and those specified as input)
817 into an executable file.
819 @cindex file name suffix
820 For any given input file, the file name suffix determines what kind of
825 C source code which must be preprocessed.
828 C source code which should not be preprocessed.
831 C++ source code which should not be preprocessed.
834 Objective-C source code. Note that you must link with the @file{libobjc}
835 library to make an Objective-C program work.
838 Objective-C source code which should not be preprocessed.
842 Objective-C++ source code. Note that you must link with the @file{libobjc}
843 library to make an Objective-C++ program work. Note that @samp{.M} refers
844 to a literal capital M@.
847 Objective-C++ source code which should not be preprocessed.
850 C, C++, Objective-C or Objective-C++ header file to be turned into a
855 @itemx @var{file}.cxx
856 @itemx @var{file}.cpp
857 @itemx @var{file}.CPP
858 @itemx @var{file}.c++
860 C++ source code which must be preprocessed. Note that in @samp{.cxx},
861 the last two letters must both be literally @samp{x}. Likewise,
862 @samp{.C} refers to a literal capital C@.
866 Objective-C++ source code which must be preprocessed.
869 Objective-C++ source code which should not be preprocessed.
873 C++ header file to be turned into a precompiled header.
876 @itemx @var{file}.for
877 @itemx @var{file}.FOR
878 Fixed form Fortran source code which should not be preprocessed.
881 @itemx @var{file}.fpp
882 @itemx @var{file}.FPP
883 Fixed form Fortran source code which must be preprocessed (with the traditional
887 @itemx @var{file}.f95
888 Free form Fortran source code which should not be preprocessed.
891 @itemx @var{file}.F95
892 Free form Fortran source code which must be preprocessed (with the
893 traditional preprocessor).
895 @c FIXME: Descriptions of Java file types.
902 Ada source code file which contains a library unit declaration (a
903 declaration of a package, subprogram, or generic, or a generic
904 instantiation), or a library unit renaming declaration (a package,
905 generic, or subprogram renaming declaration). Such files are also
908 @itemx @var{file}.adb
909 Ada source code file containing a library unit body (a subprogram or
910 package body). Such files are also called @dfn{bodies}.
912 @c GCC also knows about some suffixes for languages not yet included:
923 Assembler code which must be preprocessed.
926 An object file to be fed straight into linking.
927 Any file name with no recognized suffix is treated this way.
931 You can specify the input language explicitly with the @option{-x} option:
934 @item -x @var{language}
935 Specify explicitly the @var{language} for the following input files
936 (rather than letting the compiler choose a default based on the file
937 name suffix). This option applies to all following input files until
938 the next @option{-x} option. Possible values for @var{language} are:
940 c c-header c-cpp-output
941 c++ c++-header c++-cpp-output
942 objective-c objective-c-header objective-c-cpp-output
943 objective-c++ objective-c++-header objective-c++-cpp-output
944 assembler assembler-with-cpp
953 Turn off any specification of a language, so that subsequent files are
954 handled according to their file name suffixes (as they are if @option{-x}
955 has not been used at all).
957 @item -pass-exit-codes
958 @opindex pass-exit-codes
959 Normally the @command{gcc} program will exit with the code of 1 if any
960 phase of the compiler returns a non-success return code. If you specify
961 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
962 numerically highest error produced by any phase that returned an error
963 indication. The C, C++, and Fortran frontends return 4, if an internal
964 compiler error is encountered.
967 If you only want some of the stages of compilation, you can use
968 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
969 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
970 @command{gcc} is to stop. Note that some combinations (for example,
971 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
976 Compile or assemble the source files, but do not link. The linking
977 stage simply is not done. The ultimate output is in the form of an
978 object file for each source file.
980 By default, the object file name for a source file is made by replacing
981 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
983 Unrecognized input files, not requiring compilation or assembly, are
988 Stop after the stage of compilation proper; do not assemble. The output
989 is in the form of an assembler code file for each non-assembler input
992 By default, the assembler file name for a source file is made by
993 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
995 Input files that don't require compilation are ignored.
999 Stop after the preprocessing stage; do not run the compiler proper. The
1000 output is in the form of preprocessed source code, which is sent to the
1003 Input files which don't require preprocessing are ignored.
1005 @cindex output file option
1008 Place output in file @var{file}. This applies regardless to whatever
1009 sort of output is being produced, whether it be an executable file,
1010 an object file, an assembler file or preprocessed C code.
1012 If @option{-o} is not specified, the default is to put an executable
1013 file in @file{a.out}, the object file for
1014 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1015 assembler file in @file{@var{source}.s}, a precompiled header file in
1016 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1021 Print (on standard error output) the commands executed to run the stages
1022 of compilation. Also print the version number of the compiler driver
1023 program and of the preprocessor and the compiler proper.
1027 Like @option{-v} except the commands are not executed and all command
1028 arguments are quoted. This is useful for shell scripts to capture the
1029 driver-generated command lines.
1033 Use pipes rather than temporary files for communication between the
1034 various stages of compilation. This fails to work on some systems where
1035 the assembler is unable to read from a pipe; but the GNU assembler has
1040 If you are compiling multiple source files, this option tells the driver
1041 to pass all the source files to the compiler at once (for those
1042 languages for which the compiler can handle this). This will allow
1043 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1044 language for which this is supported is C@. If you pass source files for
1045 multiple languages to the driver, using this option, the driver will invoke
1046 the compiler(s) that support IMA once each, passing each compiler all the
1047 source files appropriate for it. For those languages that do not support
1048 IMA this option will be ignored, and the compiler will be invoked once for
1049 each source file in that language. If you use this option in conjunction
1050 with @option{-save-temps}, the compiler will generate multiple
1052 (one for each source file), but only one (combined) @file{.o} or
1057 Print (on the standard output) a description of the command line options
1058 understood by @command{gcc}. If the @option{-v} option is also specified
1059 then @option{--help} will also be passed on to the various processes
1060 invoked by @command{gcc}, so that they can display the command line options
1061 they accept. If the @option{-Wextra} option is also specified then command
1062 line options which have no documentation associated with them will also
1066 @opindex target-help
1067 Print (on the standard output) a description of target specific command
1068 line options for each tool.
1072 Display the version number and copyrights of the invoked GCC@.
1074 @include @value{srcdir}/../libiberty/at-file.texi
1078 @section Compiling C++ Programs
1080 @cindex suffixes for C++ source
1081 @cindex C++ source file suffixes
1082 C++ source files conventionally use one of the suffixes @samp{.C},
1083 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1084 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1085 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1086 files with these names and compiles them as C++ programs even if you
1087 call the compiler the same way as for compiling C programs (usually
1088 with the name @command{gcc}).
1092 However, the use of @command{gcc} does not add the C++ library.
1093 @command{g++} is a program that calls GCC and treats @samp{.c},
1094 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1095 files unless @option{-x} is used, and automatically specifies linking
1096 against the C++ library. This program is also useful when
1097 precompiling a C header file with a @samp{.h} extension for use in C++
1098 compilations. On many systems, @command{g++} is also installed with
1099 the name @command{c++}.
1101 @cindex invoking @command{g++}
1102 When you compile C++ programs, you may specify many of the same
1103 command-line options that you use for compiling programs in any
1104 language; or command-line options meaningful for C and related
1105 languages; or options that are meaningful only for C++ programs.
1106 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1107 explanations of options for languages related to C@.
1108 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1109 explanations of options that are meaningful only for C++ programs.
1111 @node C Dialect Options
1112 @section Options Controlling C Dialect
1113 @cindex dialect options
1114 @cindex language dialect options
1115 @cindex options, dialect
1117 The following options control the dialect of C (or languages derived
1118 from C, such as C++, Objective-C and Objective-C++) that the compiler
1122 @cindex ANSI support
1126 In C mode, support all ISO C90 programs. In C++ mode,
1127 remove GNU extensions that conflict with ISO C++.
1129 This turns off certain features of GCC that are incompatible with ISO
1130 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1131 such as the @code{asm} and @code{typeof} keywords, and
1132 predefined macros such as @code{unix} and @code{vax} that identify the
1133 type of system you are using. It also enables the undesirable and
1134 rarely used ISO trigraph feature. For the C compiler,
1135 it disables recognition of C++ style @samp{//} comments as well as
1136 the @code{inline} keyword.
1138 The alternate keywords @code{__asm__}, @code{__extension__},
1139 @code{__inline__} and @code{__typeof__} continue to work despite
1140 @option{-ansi}. You would not want to use them in an ISO C program, of
1141 course, but it is useful to put them in header files that might be included
1142 in compilations done with @option{-ansi}. Alternate predefined macros
1143 such as @code{__unix__} and @code{__vax__} are also available, with or
1144 without @option{-ansi}.
1146 The @option{-ansi} option does not cause non-ISO programs to be
1147 rejected gratuitously. For that, @option{-pedantic} is required in
1148 addition to @option{-ansi}. @xref{Warning Options}.
1150 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1151 option is used. Some header files may notice this macro and refrain
1152 from declaring certain functions or defining certain macros that the
1153 ISO standard doesn't call for; this is to avoid interfering with any
1154 programs that might use these names for other things.
1156 Functions which would normally be built in but do not have semantics
1157 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1158 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1159 built-in functions provided by GCC}, for details of the functions
1164 Determine the language standard. This option is currently only
1165 supported when compiling C or C++. A value for this option must be
1166 provided; possible values are
1171 ISO C90 (same as @option{-ansi}).
1173 @item iso9899:199409
1174 ISO C90 as modified in amendment 1.
1180 ISO C99. Note that this standard is not yet fully supported; see
1181 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1182 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1185 Default, ISO C90 plus GNU extensions (including some C99 features).
1189 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1190 this will become the default. The name @samp{gnu9x} is deprecated.
1193 The 1998 ISO C++ standard plus amendments.
1196 The same as @option{-std=c++98} plus GNU extensions. This is the
1197 default for C++ code.
1200 Even when this option is not specified, you can still use some of the
1201 features of newer standards in so far as they do not conflict with
1202 previous C standards. For example, you may use @code{__restrict__} even
1203 when @option{-std=c99} is not specified.
1205 The @option{-std} options specifying some version of ISO C have the same
1206 effects as @option{-ansi}, except that features that were not in ISO C90
1207 but are in the specified version (for example, @samp{//} comments and
1208 the @code{inline} keyword in ISO C99) are not disabled.
1210 @xref{Standards,,Language Standards Supported by GCC}, for details of
1211 these standard versions.
1213 @item -aux-info @var{filename}
1215 Output to the given filename prototyped declarations for all functions
1216 declared and/or defined in a translation unit, including those in header
1217 files. This option is silently ignored in any language other than C@.
1219 Besides declarations, the file indicates, in comments, the origin of
1220 each declaration (source file and line), whether the declaration was
1221 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1222 @samp{O} for old, respectively, in the first character after the line
1223 number and the colon), and whether it came from a declaration or a
1224 definition (@samp{C} or @samp{F}, respectively, in the following
1225 character). In the case of function definitions, a K&R-style list of
1226 arguments followed by their declarations is also provided, inside
1227 comments, after the declaration.
1231 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1232 keyword, so that code can use these words as identifiers. You can use
1233 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1234 instead. @option{-ansi} implies @option{-fno-asm}.
1236 In C++, this switch only affects the @code{typeof} keyword, since
1237 @code{asm} and @code{inline} are standard keywords. You may want to
1238 use the @option{-fno-gnu-keywords} flag instead, which has the same
1239 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1240 switch only affects the @code{asm} and @code{typeof} keywords, since
1241 @code{inline} is a standard keyword in ISO C99.
1244 @itemx -fno-builtin-@var{function}
1245 @opindex fno-builtin
1246 @cindex built-in functions
1247 Don't recognize built-in functions that do not begin with
1248 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1249 functions provided by GCC}, for details of the functions affected,
1250 including those which are not built-in functions when @option{-ansi} or
1251 @option{-std} options for strict ISO C conformance are used because they
1252 do not have an ISO standard meaning.
1254 GCC normally generates special code to handle certain built-in functions
1255 more efficiently; for instance, calls to @code{alloca} may become single
1256 instructions that adjust the stack directly, and calls to @code{memcpy}
1257 may become inline copy loops. The resulting code is often both smaller
1258 and faster, but since the function calls no longer appear as such, you
1259 cannot set a breakpoint on those calls, nor can you change the behavior
1260 of the functions by linking with a different library. In addition,
1261 when a function is recognized as a built-in function, GCC may use
1262 information about that function to warn about problems with calls to
1263 that function, or to generate more efficient code, even if the
1264 resulting code still contains calls to that function. For example,
1265 warnings are given with @option{-Wformat} for bad calls to
1266 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1267 known not to modify global memory.
1269 With the @option{-fno-builtin-@var{function}} option
1270 only the built-in function @var{function} is
1271 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1272 function is named this is not built-in in this version of GCC, this
1273 option is ignored. There is no corresponding
1274 @option{-fbuiltin-@var{function}} option; if you wish to enable
1275 built-in functions selectively when using @option{-fno-builtin} or
1276 @option{-ffreestanding}, you may define macros such as:
1279 #define abs(n) __builtin_abs ((n))
1280 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1285 @cindex hosted environment
1287 Assert that compilation takes place in a hosted environment. This implies
1288 @option{-fbuiltin}. A hosted environment is one in which the
1289 entire standard library is available, and in which @code{main} has a return
1290 type of @code{int}. Examples are nearly everything except a kernel.
1291 This is equivalent to @option{-fno-freestanding}.
1293 @item -ffreestanding
1294 @opindex ffreestanding
1295 @cindex hosted environment
1297 Assert that compilation takes place in a freestanding environment. This
1298 implies @option{-fno-builtin}. A freestanding environment
1299 is one in which the standard library may not exist, and program startup may
1300 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1301 This is equivalent to @option{-fno-hosted}.
1303 @xref{Standards,,Language Standards Supported by GCC}, for details of
1304 freestanding and hosted environments.
1308 @cindex openmp parallel
1309 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1310 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1311 compiler generates parallel code according to the OpenMP Application
1312 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1314 @item -fms-extensions
1315 @opindex fms-extensions
1316 Accept some non-standard constructs used in Microsoft header files.
1318 Some cases of unnamed fields in structures and unions are only
1319 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1320 fields within structs/unions}, for details.
1324 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1325 options for strict ISO C conformance) implies @option{-trigraphs}.
1327 @item -no-integrated-cpp
1328 @opindex no-integrated-cpp
1329 Performs a compilation in two passes: preprocessing and compiling. This
1330 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1331 @option{-B} option. The user supplied compilation step can then add in
1332 an additional preprocessing step after normal preprocessing but before
1333 compiling. The default is to use the integrated cpp (internal cpp)
1335 The semantics of this option will change if "cc1", "cc1plus", and
1336 "cc1obj" are merged.
1338 @cindex traditional C language
1339 @cindex C language, traditional
1341 @itemx -traditional-cpp
1342 @opindex traditional-cpp
1343 @opindex traditional
1344 Formerly, these options caused GCC to attempt to emulate a pre-standard
1345 C compiler. They are now only supported with the @option{-E} switch.
1346 The preprocessor continues to support a pre-standard mode. See the GNU
1347 CPP manual for details.
1349 @item -fcond-mismatch
1350 @opindex fcond-mismatch
1351 Allow conditional expressions with mismatched types in the second and
1352 third arguments. The value of such an expression is void. This option
1353 is not supported for C++.
1355 @item -funsigned-char
1356 @opindex funsigned-char
1357 Let the type @code{char} be unsigned, like @code{unsigned char}.
1359 Each kind of machine has a default for what @code{char} should
1360 be. It is either like @code{unsigned char} by default or like
1361 @code{signed char} by default.
1363 Ideally, a portable program should always use @code{signed char} or
1364 @code{unsigned char} when it depends on the signedness of an object.
1365 But many programs have been written to use plain @code{char} and
1366 expect it to be signed, or expect it to be unsigned, depending on the
1367 machines they were written for. This option, and its inverse, let you
1368 make such a program work with the opposite default.
1370 The type @code{char} is always a distinct type from each of
1371 @code{signed char} or @code{unsigned char}, even though its behavior
1372 is always just like one of those two.
1375 @opindex fsigned-char
1376 Let the type @code{char} be signed, like @code{signed char}.
1378 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1379 the negative form of @option{-funsigned-char}. Likewise, the option
1380 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1382 @item -fsigned-bitfields
1383 @itemx -funsigned-bitfields
1384 @itemx -fno-signed-bitfields
1385 @itemx -fno-unsigned-bitfields
1386 @opindex fsigned-bitfields
1387 @opindex funsigned-bitfields
1388 @opindex fno-signed-bitfields
1389 @opindex fno-unsigned-bitfields
1390 These options control whether a bit-field is signed or unsigned, when the
1391 declaration does not use either @code{signed} or @code{unsigned}. By
1392 default, such a bit-field is signed, because this is consistent: the
1393 basic integer types such as @code{int} are signed types.
1396 @node C++ Dialect Options
1397 @section Options Controlling C++ Dialect
1399 @cindex compiler options, C++
1400 @cindex C++ options, command line
1401 @cindex options, C++
1402 This section describes the command-line options that are only meaningful
1403 for C++ programs; but you can also use most of the GNU compiler options
1404 regardless of what language your program is in. For example, you
1405 might compile a file @code{firstClass.C} like this:
1408 g++ -g -frepo -O -c firstClass.C
1412 In this example, only @option{-frepo} is an option meant
1413 only for C++ programs; you can use the other options with any
1414 language supported by GCC@.
1416 Here is a list of options that are @emph{only} for compiling C++ programs:
1420 @item -fabi-version=@var{n}
1421 @opindex fabi-version
1422 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1423 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1424 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1425 the version that conforms most closely to the C++ ABI specification.
1426 Therefore, the ABI obtained using version 0 will change as ABI bugs
1429 The default is version 2.
1431 @item -fno-access-control
1432 @opindex fno-access-control
1433 Turn off all access checking. This switch is mainly useful for working
1434 around bugs in the access control code.
1438 Check that the pointer returned by @code{operator new} is non-null
1439 before attempting to modify the storage allocated. This check is
1440 normally unnecessary because the C++ standard specifies that
1441 @code{operator new} will only return @code{0} if it is declared
1442 @samp{throw()}, in which case the compiler will always check the
1443 return value even without this option. In all other cases, when
1444 @code{operator new} has a non-empty exception specification, memory
1445 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1446 @samp{new (nothrow)}.
1448 @item -fconserve-space
1449 @opindex fconserve-space
1450 Put uninitialized or runtime-initialized global variables into the
1451 common segment, as C does. This saves space in the executable at the
1452 cost of not diagnosing duplicate definitions. If you compile with this
1453 flag and your program mysteriously crashes after @code{main()} has
1454 completed, you may have an object that is being destroyed twice because
1455 two definitions were merged.
1457 This option is no longer useful on most targets, now that support has
1458 been added for putting variables into BSS without making them common.
1460 @item -ffriend-injection
1461 @opindex ffriend-injection
1462 Inject friend functions into the enclosing namespace, so that they are
1463 visible outside the scope of the class in which they are declared.
1464 Friend functions were documented to work this way in the old Annotated
1465 C++ Reference Manual, and versions of G++ before 4.1 always worked
1466 that way. However, in ISO C++ a friend function which is not declared
1467 in an enclosing scope can only be found using argument dependent
1468 lookup. This option causes friends to be injected as they were in
1471 This option is for compatibility, and may be removed in a future
1474 @item -fno-elide-constructors
1475 @opindex fno-elide-constructors
1476 The C++ standard allows an implementation to omit creating a temporary
1477 which is only used to initialize another object of the same type.
1478 Specifying this option disables that optimization, and forces G++ to
1479 call the copy constructor in all cases.
1481 @item -fno-enforce-eh-specs
1482 @opindex fno-enforce-eh-specs
1483 Don't generate code to check for violation of exception specifications
1484 at runtime. This option violates the C++ standard, but may be useful
1485 for reducing code size in production builds, much like defining
1486 @samp{NDEBUG}. This does not give user code permission to throw
1487 exceptions in violation of the exception specifications; the compiler
1488 will still optimize based on the specifications, so throwing an
1489 unexpected exception will result in undefined behavior.
1492 @itemx -fno-for-scope
1494 @opindex fno-for-scope
1495 If @option{-ffor-scope} is specified, the scope of variables declared in
1496 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1497 as specified by the C++ standard.
1498 If @option{-fno-for-scope} is specified, the scope of variables declared in
1499 a @i{for-init-statement} extends to the end of the enclosing scope,
1500 as was the case in old versions of G++, and other (traditional)
1501 implementations of C++.
1503 The default if neither flag is given to follow the standard,
1504 but to allow and give a warning for old-style code that would
1505 otherwise be invalid, or have different behavior.
1507 @item -fno-gnu-keywords
1508 @opindex fno-gnu-keywords
1509 Do not recognize @code{typeof} as a keyword, so that code can use this
1510 word as an identifier. You can use the keyword @code{__typeof__} instead.
1511 @option{-ansi} implies @option{-fno-gnu-keywords}.
1513 @item -fno-implicit-templates
1514 @opindex fno-implicit-templates
1515 Never emit code for non-inline templates which are instantiated
1516 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1517 @xref{Template Instantiation}, for more information.
1519 @item -fno-implicit-inline-templates
1520 @opindex fno-implicit-inline-templates
1521 Don't emit code for implicit instantiations of inline templates, either.
1522 The default is to handle inlines differently so that compiles with and
1523 without optimization will need the same set of explicit instantiations.
1525 @item -fno-implement-inlines
1526 @opindex fno-implement-inlines
1527 To save space, do not emit out-of-line copies of inline functions
1528 controlled by @samp{#pragma implementation}. This will cause linker
1529 errors if these functions are not inlined everywhere they are called.
1531 @item -fms-extensions
1532 @opindex fms-extensions
1533 Disable pedantic warnings about constructs used in MFC, such as implicit
1534 int and getting a pointer to member function via non-standard syntax.
1536 @item -fno-nonansi-builtins
1537 @opindex fno-nonansi-builtins
1538 Disable built-in declarations of functions that are not mandated by
1539 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1540 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1542 @item -fno-operator-names
1543 @opindex fno-operator-names
1544 Do not treat the operator name keywords @code{and}, @code{bitand},
1545 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1546 synonyms as keywords.
1548 @item -fno-optional-diags
1549 @opindex fno-optional-diags
1550 Disable diagnostics that the standard says a compiler does not need to
1551 issue. Currently, the only such diagnostic issued by G++ is the one for
1552 a name having multiple meanings within a class.
1555 @opindex fpermissive
1556 Downgrade some diagnostics about nonconformant code from errors to
1557 warnings. Thus, using @option{-fpermissive} will allow some
1558 nonconforming code to compile.
1562 Enable automatic template instantiation at link time. This option also
1563 implies @option{-fno-implicit-templates}. @xref{Template
1564 Instantiation}, for more information.
1568 Disable generation of information about every class with virtual
1569 functions for use by the C++ runtime type identification features
1570 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1571 of the language, you can save some space by using this flag. Note that
1572 exception handling uses the same information, but it will generate it as
1577 Emit statistics about front-end processing at the end of the compilation.
1578 This information is generally only useful to the G++ development team.
1580 @item -ftemplate-depth-@var{n}
1581 @opindex ftemplate-depth
1582 Set the maximum instantiation depth for template classes to @var{n}.
1583 A limit on the template instantiation depth is needed to detect
1584 endless recursions during template class instantiation. ANSI/ISO C++
1585 conforming programs must not rely on a maximum depth greater than 17.
1587 @item -fno-threadsafe-statics
1588 @opindex fno-threadsafe-statics
1589 Do not emit the extra code to use the routines specified in the C++
1590 ABI for thread-safe initialization of local statics. You can use this
1591 option to reduce code size slightly in code that doesn't need to be
1594 @item -fuse-cxa-atexit
1595 @opindex fuse-cxa-atexit
1596 Register destructors for objects with static storage duration with the
1597 @code{__cxa_atexit} function rather than the @code{atexit} function.
1598 This option is required for fully standards-compliant handling of static
1599 destructors, but will only work if your C library supports
1600 @code{__cxa_atexit}.
1602 @item -fno-use-cxa-get-exception-ptr
1603 @opindex fno-use-cxa-get-exception-ptr
1604 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1605 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1606 if the runtime routine is not available.
1608 @item -fvisibility-inlines-hidden
1609 @opindex fvisibility-inlines-hidden
1610 This switch declares that the user does not attempt to compare
1611 pointers to inline methods where the addresses of the two functions
1612 were taken in different shared objects.
1614 The effect of this is that GCC may, effectively, mark inline methods with
1615 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1616 appear in the export table of a DSO and do not require a PLT indirection
1617 when used within the DSO@. Enabling this option can have a dramatic effect
1618 on load and link times of a DSO as it massively reduces the size of the
1619 dynamic export table when the library makes heavy use of templates.
1621 The behaviour of this switch is not quite the same as marking the
1622 methods as hidden directly, because it does not affect static variables
1623 local to the function or cause the compiler to deduce that
1624 the function is defined in only one shared object.
1626 You may mark a method as having a visibility explicitly to negate the
1627 effect of the switch for that method. For example, if you do want to
1628 compare pointers to a particular inline method, you might mark it as
1629 having default visibility.
1633 Do not use weak symbol support, even if it is provided by the linker.
1634 By default, G++ will use weak symbols if they are available. This
1635 option exists only for testing, and should not be used by end-users;
1636 it will result in inferior code and has no benefits. This option may
1637 be removed in a future release of G++.
1641 Do not search for header files in the standard directories specific to
1642 C++, but do still search the other standard directories. (This option
1643 is used when building the C++ library.)
1646 In addition, these optimization, warning, and code generation options
1647 have meanings only for C++ programs:
1650 @item -fno-default-inline
1651 @opindex fno-default-inline
1652 Do not assume @samp{inline} for functions defined inside a class scope.
1653 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1654 functions will have linkage like inline functions; they just won't be
1657 @item -Wabi @r{(C++ only)}
1659 Warn when G++ generates code that is probably not compatible with the
1660 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1661 all such cases, there are probably some cases that are not warned about,
1662 even though G++ is generating incompatible code. There may also be
1663 cases where warnings are emitted even though the code that is generated
1666 You should rewrite your code to avoid these warnings if you are
1667 concerned about the fact that code generated by G++ may not be binary
1668 compatible with code generated by other compilers.
1670 The known incompatibilities at this point include:
1675 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1676 pack data into the same byte as a base class. For example:
1679 struct A @{ virtual void f(); int f1 : 1; @};
1680 struct B : public A @{ int f2 : 1; @};
1684 In this case, G++ will place @code{B::f2} into the same byte
1685 as@code{A::f1}; other compilers will not. You can avoid this problem
1686 by explicitly padding @code{A} so that its size is a multiple of the
1687 byte size on your platform; that will cause G++ and other compilers to
1688 layout @code{B} identically.
1691 Incorrect handling of tail-padding for virtual bases. G++ does not use
1692 tail padding when laying out virtual bases. For example:
1695 struct A @{ virtual void f(); char c1; @};
1696 struct B @{ B(); char c2; @};
1697 struct C : public A, public virtual B @{@};
1701 In this case, G++ will not place @code{B} into the tail-padding for
1702 @code{A}; other compilers will. You can avoid this problem by
1703 explicitly padding @code{A} so that its size is a multiple of its
1704 alignment (ignoring virtual base classes); that will cause G++ and other
1705 compilers to layout @code{C} identically.
1708 Incorrect handling of bit-fields with declared widths greater than that
1709 of their underlying types, when the bit-fields appear in a union. For
1713 union U @{ int i : 4096; @};
1717 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1718 union too small by the number of bits in an @code{int}.
1721 Empty classes can be placed at incorrect offsets. For example:
1731 struct C : public B, public A @{@};
1735 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1736 it should be placed at offset zero. G++ mistakenly believes that the
1737 @code{A} data member of @code{B} is already at offset zero.
1740 Names of template functions whose types involve @code{typename} or
1741 template template parameters can be mangled incorrectly.
1744 template <typename Q>
1745 void f(typename Q::X) @{@}
1747 template <template <typename> class Q>
1748 void f(typename Q<int>::X) @{@}
1752 Instantiations of these templates may be mangled incorrectly.
1756 @item -Wctor-dtor-privacy @r{(C++ only)}
1757 @opindex Wctor-dtor-privacy
1758 Warn when a class seems unusable because all the constructors or
1759 destructors in that class are private, and it has neither friends nor
1760 public static member functions.
1762 @item -Wnon-virtual-dtor @r{(C++ only)}
1763 @opindex Wnon-virtual-dtor
1764 Warn when a class appears to be polymorphic, thereby requiring a virtual
1765 destructor, yet it declares a non-virtual one. This warning is also
1766 enabled if -Weffc++ is specified.
1768 @item -Wreorder @r{(C++ only)}
1770 @cindex reordering, warning
1771 @cindex warning for reordering of member initializers
1772 Warn when the order of member initializers given in the code does not
1773 match the order in which they must be executed. For instance:
1779 A(): j (0), i (1) @{ @}
1783 The compiler will rearrange the member initializers for @samp{i}
1784 and @samp{j} to match the declaration order of the members, emitting
1785 a warning to that effect. This warning is enabled by @option{-Wall}.
1788 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1791 @item -Weffc++ @r{(C++ only)}
1793 Warn about violations of the following style guidelines from Scott Meyers'
1794 @cite{Effective C++} book:
1798 Item 11: Define a copy constructor and an assignment operator for classes
1799 with dynamically allocated memory.
1802 Item 12: Prefer initialization to assignment in constructors.
1805 Item 14: Make destructors virtual in base classes.
1808 Item 15: Have @code{operator=} return a reference to @code{*this}.
1811 Item 23: Don't try to return a reference when you must return an object.
1815 Also warn about violations of the following style guidelines from
1816 Scott Meyers' @cite{More Effective C++} book:
1820 Item 6: Distinguish between prefix and postfix forms of increment and
1821 decrement operators.
1824 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1828 When selecting this option, be aware that the standard library
1829 headers do not obey all of these guidelines; use @samp{grep -v}
1830 to filter out those warnings.
1832 @item -Wno-deprecated @r{(C++ only)}
1833 @opindex Wno-deprecated
1834 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1836 @item -Wstrict-null-sentinel @r{(C++ only)}
1837 @opindex Wstrict-null-sentinel
1838 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1839 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1840 to @code{__null}. Although it is a null pointer constant not a null pointer,
1841 it is guaranteed to of the same size as a pointer. But this use is
1842 not portable across different compilers.
1844 @item -Wno-non-template-friend @r{(C++ only)}
1845 @opindex Wno-non-template-friend
1846 Disable warnings when non-templatized friend functions are declared
1847 within a template. Since the advent of explicit template specification
1848 support in G++, if the name of the friend is an unqualified-id (i.e.,
1849 @samp{friend foo(int)}), the C++ language specification demands that the
1850 friend declare or define an ordinary, nontemplate function. (Section
1851 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1852 could be interpreted as a particular specialization of a templatized
1853 function. Because this non-conforming behavior is no longer the default
1854 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1855 check existing code for potential trouble spots and is on by default.
1856 This new compiler behavior can be turned off with
1857 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1858 but disables the helpful warning.
1860 @item -Wold-style-cast @r{(C++ only)}
1861 @opindex Wold-style-cast
1862 Warn if an old-style (C-style) cast to a non-void type is used within
1863 a C++ program. The new-style casts (@samp{dynamic_cast},
1864 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1865 less vulnerable to unintended effects and much easier to search for.
1867 @item -Woverloaded-virtual @r{(C++ only)}
1868 @opindex Woverloaded-virtual
1869 @cindex overloaded virtual fn, warning
1870 @cindex warning for overloaded virtual fn
1871 Warn when a function declaration hides virtual functions from a
1872 base class. For example, in:
1879 struct B: public A @{
1884 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1892 will fail to compile.
1894 @item -Wno-pmf-conversions @r{(C++ only)}
1895 @opindex Wno-pmf-conversions
1896 Disable the diagnostic for converting a bound pointer to member function
1899 @item -Wsign-promo @r{(C++ only)}
1900 @opindex Wsign-promo
1901 Warn when overload resolution chooses a promotion from unsigned or
1902 enumerated type to a signed type, over a conversion to an unsigned type of
1903 the same size. Previous versions of G++ would try to preserve
1904 unsignedness, but the standard mandates the current behavior.
1909 A& operator = (int);
1919 In this example, G++ will synthesize a default @samp{A& operator =
1920 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1923 @node Objective-C and Objective-C++ Dialect Options
1924 @section Options Controlling Objective-C and Objective-C++ Dialects
1926 @cindex compiler options, Objective-C and Objective-C++
1927 @cindex Objective-C and Objective-C++ options, command line
1928 @cindex options, Objective-C and Objective-C++
1929 (NOTE: This manual does not describe the Objective-C and Objective-C++
1930 languages themselves. See @xref{Standards,,Language Standards
1931 Supported by GCC}, for references.)
1933 This section describes the command-line options that are only meaningful
1934 for Objective-C and Objective-C++ programs, but you can also use most of
1935 the language-independent GNU compiler options.
1936 For example, you might compile a file @code{some_class.m} like this:
1939 gcc -g -fgnu-runtime -O -c some_class.m
1943 In this example, @option{-fgnu-runtime} is an option meant only for
1944 Objective-C and Objective-C++ programs; you can use the other options with
1945 any language supported by GCC@.
1947 Note that since Objective-C is an extension of the C language, Objective-C
1948 compilations may also use options specific to the C front-end (e.g.,
1949 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1950 C++-specific options (e.g., @option{-Wabi}).
1952 Here is a list of options that are @emph{only} for compiling Objective-C
1953 and Objective-C++ programs:
1956 @item -fconstant-string-class=@var{class-name}
1957 @opindex fconstant-string-class
1958 Use @var{class-name} as the name of the class to instantiate for each
1959 literal string specified with the syntax @code{@@"@dots{}"}. The default
1960 class name is @code{NXConstantString} if the GNU runtime is being used, and
1961 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1962 @option{-fconstant-cfstrings} option, if also present, will override the
1963 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1964 to be laid out as constant CoreFoundation strings.
1967 @opindex fgnu-runtime
1968 Generate object code compatible with the standard GNU Objective-C
1969 runtime. This is the default for most types of systems.
1971 @item -fnext-runtime
1972 @opindex fnext-runtime
1973 Generate output compatible with the NeXT runtime. This is the default
1974 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1975 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1978 @item -fno-nil-receivers
1979 @opindex fno-nil-receivers
1980 Assume that all Objective-C message dispatches (e.g.,
1981 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1982 is not @code{nil}. This allows for more efficient entry points in the runtime
1983 to be used. Currently, this option is only available in conjunction with
1984 the NeXT runtime on Mac OS X 10.3 and later.
1986 @item -fobjc-call-cxx-cdtors
1987 @opindex fobjc-call-cxx-cdtors
1988 For each Objective-C class, check if any of its instance variables is a
1989 C++ object with a non-trivial default constructor. If so, synthesize a
1990 special @code{- (id) .cxx_construct} instance method that will run
1991 non-trivial default constructors on any such instance variables, in order,
1992 and then return @code{self}. Similarly, check if any instance variable
1993 is a C++ object with a non-trivial destructor, and if so, synthesize a
1994 special @code{- (void) .cxx_destruct} method that will run
1995 all such default destructors, in reverse order.
1997 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
1998 thusly generated will only operate on instance variables declared in the
1999 current Objective-C class, and not those inherited from superclasses. It
2000 is the responsibility of the Objective-C runtime to invoke all such methods
2001 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2002 will be invoked by the runtime immediately after a new object
2003 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2004 be invoked immediately before the runtime deallocates an object instance.
2006 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2007 support for invoking the @code{- (id) .cxx_construct} and
2008 @code{- (void) .cxx_destruct} methods.
2010 @item -fobjc-direct-dispatch
2011 @opindex fobjc-direct-dispatch
2012 Allow fast jumps to the message dispatcher. On Darwin this is
2013 accomplished via the comm page.
2015 @item -fobjc-exceptions
2016 @opindex fobjc-exceptions
2017 Enable syntactic support for structured exception handling in Objective-C,
2018 similar to what is offered by C++ and Java. This option is
2019 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2028 @@catch (AnObjCClass *exc) @{
2035 @@catch (AnotherClass *exc) @{
2038 @@catch (id allOthers) @{
2048 The @code{@@throw} statement may appear anywhere in an Objective-C or
2049 Objective-C++ program; when used inside of a @code{@@catch} block, the
2050 @code{@@throw} may appear without an argument (as shown above), in which case
2051 the object caught by the @code{@@catch} will be rethrown.
2053 Note that only (pointers to) Objective-C objects may be thrown and
2054 caught using this scheme. When an object is thrown, it will be caught
2055 by the nearest @code{@@catch} clause capable of handling objects of that type,
2056 analogously to how @code{catch} blocks work in C++ and Java. A
2057 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2058 any and all Objective-C exceptions not caught by previous @code{@@catch}
2061 The @code{@@finally} clause, if present, will be executed upon exit from the
2062 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2063 regardless of whether any exceptions are thrown, caught or rethrown
2064 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2065 of the @code{finally} clause in Java.
2067 There are several caveats to using the new exception mechanism:
2071 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2072 idioms provided by the @code{NSException} class, the new
2073 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2074 systems, due to additional functionality needed in the (NeXT) Objective-C
2078 As mentioned above, the new exceptions do not support handling
2079 types other than Objective-C objects. Furthermore, when used from
2080 Objective-C++, the Objective-C exception model does not interoperate with C++
2081 exceptions at this time. This means you cannot @code{@@throw} an exception
2082 from Objective-C and @code{catch} it in C++, or vice versa
2083 (i.e., @code{throw @dots{} @@catch}).
2086 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2087 blocks for thread-safe execution:
2090 @@synchronized (ObjCClass *guard) @{
2095 Upon entering the @code{@@synchronized} block, a thread of execution shall
2096 first check whether a lock has been placed on the corresponding @code{guard}
2097 object by another thread. If it has, the current thread shall wait until
2098 the other thread relinquishes its lock. Once @code{guard} becomes available,
2099 the current thread will place its own lock on it, execute the code contained in
2100 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2101 making @code{guard} available to other threads).
2103 Unlike Java, Objective-C does not allow for entire methods to be marked
2104 @code{@@synchronized}. Note that throwing exceptions out of
2105 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2106 to be unlocked properly.
2110 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2112 @item -freplace-objc-classes
2113 @opindex freplace-objc-classes
2114 Emit a special marker instructing @command{ld(1)} not to statically link in
2115 the resulting object file, and allow @command{dyld(1)} to load it in at
2116 run time instead. This is used in conjunction with the Fix-and-Continue
2117 debugging mode, where the object file in question may be recompiled and
2118 dynamically reloaded in the course of program execution, without the need
2119 to restart the program itself. Currently, Fix-and-Continue functionality
2120 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2125 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2126 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2127 compile time) with static class references that get initialized at load time,
2128 which improves run-time performance. Specifying the @option{-fzero-link} flag
2129 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2130 to be retained. This is useful in Zero-Link debugging mode, since it allows
2131 for individual class implementations to be modified during program execution.
2135 Dump interface declarations for all classes seen in the source file to a
2136 file named @file{@var{sourcename}.decl}.
2138 @item -Wassign-intercept
2139 @opindex Wassign-intercept
2140 Warn whenever an Objective-C assignment is being intercepted by the
2144 @opindex Wno-protocol
2145 If a class is declared to implement a protocol, a warning is issued for
2146 every method in the protocol that is not implemented by the class. The
2147 default behavior is to issue a warning for every method not explicitly
2148 implemented in the class, even if a method implementation is inherited
2149 from the superclass. If you use the @option{-Wno-protocol} option, then
2150 methods inherited from the superclass are considered to be implemented,
2151 and no warning is issued for them.
2155 Warn if multiple methods of different types for the same selector are
2156 found during compilation. The check is performed on the list of methods
2157 in the final stage of compilation. Additionally, a check is performed
2158 for each selector appearing in a @code{@@selector(@dots{})}
2159 expression, and a corresponding method for that selector has been found
2160 during compilation. Because these checks scan the method table only at
2161 the end of compilation, these warnings are not produced if the final
2162 stage of compilation is not reached, for example because an error is
2163 found during compilation, or because the @option{-fsyntax-only} option is
2166 @item -Wstrict-selector-match
2167 @opindex Wstrict-selector-match
2168 Warn if multiple methods with differing argument and/or return types are
2169 found for a given selector when attempting to send a message using this
2170 selector to a receiver of type @code{id} or @code{Class}. When this flag
2171 is off (which is the default behavior), the compiler will omit such warnings
2172 if any differences found are confined to types which share the same size
2175 @item -Wundeclared-selector
2176 @opindex Wundeclared-selector
2177 Warn if a @code{@@selector(@dots{})} expression referring to an
2178 undeclared selector is found. A selector is considered undeclared if no
2179 method with that name has been declared before the
2180 @code{@@selector(@dots{})} expression, either explicitly in an
2181 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2182 an @code{@@implementation} section. This option always performs its
2183 checks as soon as a @code{@@selector(@dots{})} expression is found,
2184 while @option{-Wselector} only performs its checks in the final stage of
2185 compilation. This also enforces the coding style convention
2186 that methods and selectors must be declared before being used.
2188 @item -print-objc-runtime-info
2189 @opindex print-objc-runtime-info
2190 Generate C header describing the largest structure that is passed by
2195 @node Language Independent Options
2196 @section Options to Control Diagnostic Messages Formatting
2197 @cindex options to control diagnostics formatting
2198 @cindex diagnostic messages
2199 @cindex message formatting
2201 Traditionally, diagnostic messages have been formatted irrespective of
2202 the output device's aspect (e.g.@: its width, @dots{}). The options described
2203 below can be used to control the diagnostic messages formatting
2204 algorithm, e.g.@: how many characters per line, how often source location
2205 information should be reported. Right now, only the C++ front end can
2206 honor these options. However it is expected, in the near future, that
2207 the remaining front ends would be able to digest them correctly.
2210 @item -fmessage-length=@var{n}
2211 @opindex fmessage-length
2212 Try to format error messages so that they fit on lines of about @var{n}
2213 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2214 the front ends supported by GCC@. If @var{n} is zero, then no
2215 line-wrapping will be done; each error message will appear on a single
2218 @opindex fdiagnostics-show-location
2219 @item -fdiagnostics-show-location=once
2220 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2221 reporter to emit @emph{once} source location information; that is, in
2222 case the message is too long to fit on a single physical line and has to
2223 be wrapped, the source location won't be emitted (as prefix) again,
2224 over and over, in subsequent continuation lines. This is the default
2227 @item -fdiagnostics-show-location=every-line
2228 Only meaningful in line-wrapping mode. Instructs the diagnostic
2229 messages reporter to emit the same source location information (as
2230 prefix) for physical lines that result from the process of breaking
2231 a message which is too long to fit on a single line.
2233 @item -fdiagnostics-show-options
2234 @opindex fdiagnostics-show-options
2235 This option instructs the diagnostic machinery to add text to each
2236 diagnostic emitted, which indicates which command line option directly
2237 controls that diagnostic, when such an option is known to the
2238 diagnostic machinery.
2242 @node Warning Options
2243 @section Options to Request or Suppress Warnings
2244 @cindex options to control warnings
2245 @cindex warning messages
2246 @cindex messages, warning
2247 @cindex suppressing warnings
2249 Warnings are diagnostic messages that report constructions which
2250 are not inherently erroneous but which are risky or suggest there
2251 may have been an error.
2253 You can request many specific warnings with options beginning @samp{-W},
2254 for example @option{-Wimplicit} to request warnings on implicit
2255 declarations. Each of these specific warning options also has a
2256 negative form beginning @samp{-Wno-} to turn off warnings;
2257 for example, @option{-Wno-implicit}. This manual lists only one of the
2258 two forms, whichever is not the default.
2260 The following options control the amount and kinds of warnings produced
2261 by GCC; for further, language-specific options also refer to
2262 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2266 @cindex syntax checking
2268 @opindex fsyntax-only
2269 Check the code for syntax errors, but don't do anything beyond that.
2273 Issue all the warnings demanded by strict ISO C and ISO C++;
2274 reject all programs that use forbidden extensions, and some other
2275 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2276 version of the ISO C standard specified by any @option{-std} option used.
2278 Valid ISO C and ISO C++ programs should compile properly with or without
2279 this option (though a rare few will require @option{-ansi} or a
2280 @option{-std} option specifying the required version of ISO C)@. However,
2281 without this option, certain GNU extensions and traditional C and C++
2282 features are supported as well. With this option, they are rejected.
2284 @option{-pedantic} does not cause warning messages for use of the
2285 alternate keywords whose names begin and end with @samp{__}. Pedantic
2286 warnings are also disabled in the expression that follows
2287 @code{__extension__}. However, only system header files should use
2288 these escape routes; application programs should avoid them.
2289 @xref{Alternate Keywords}.
2291 Some users try to use @option{-pedantic} to check programs for strict ISO
2292 C conformance. They soon find that it does not do quite what they want:
2293 it finds some non-ISO practices, but not all---only those for which
2294 ISO C @emph{requires} a diagnostic, and some others for which
2295 diagnostics have been added.
2297 A feature to report any failure to conform to ISO C might be useful in
2298 some instances, but would require considerable additional work and would
2299 be quite different from @option{-pedantic}. We don't have plans to
2300 support such a feature in the near future.
2302 Where the standard specified with @option{-std} represents a GNU
2303 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2304 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2305 extended dialect is based. Warnings from @option{-pedantic} are given
2306 where they are required by the base standard. (It would not make sense
2307 for such warnings to be given only for features not in the specified GNU
2308 C dialect, since by definition the GNU dialects of C include all
2309 features the compiler supports with the given option, and there would be
2310 nothing to warn about.)
2312 @item -pedantic-errors
2313 @opindex pedantic-errors
2314 Like @option{-pedantic}, except that errors are produced rather than
2319 Inhibit all warning messages.
2323 Inhibit warning messages about the use of @samp{#import}.
2325 @item -Wchar-subscripts
2326 @opindex Wchar-subscripts
2327 Warn if an array subscript has type @code{char}. This is a common cause
2328 of error, as programmers often forget that this type is signed on some
2330 This warning is enabled by @option{-Wall}.
2334 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2335 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2336 This warning is enabled by @option{-Wall}.
2338 @item -Wfatal-errors
2339 @opindex Wfatal-errors
2340 This option causes the compiler to abort compilation on the first error
2341 occurred rather than trying to keep going and printing further error
2346 @opindex ffreestanding
2347 @opindex fno-builtin
2348 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2349 the arguments supplied have types appropriate to the format string
2350 specified, and that the conversions specified in the format string make
2351 sense. This includes standard functions, and others specified by format
2352 attributes (@pxref{Function Attributes}), in the @code{printf},
2353 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2354 not in the C standard) families (or other target-specific families).
2355 Which functions are checked without format attributes having been
2356 specified depends on the standard version selected, and such checks of
2357 functions without the attribute specified are disabled by
2358 @option{-ffreestanding} or @option{-fno-builtin}.
2360 The formats are checked against the format features supported by GNU
2361 libc version 2.2. These include all ISO C90 and C99 features, as well
2362 as features from the Single Unix Specification and some BSD and GNU
2363 extensions. Other library implementations may not support all these
2364 features; GCC does not support warning about features that go beyond a
2365 particular library's limitations. However, if @option{-pedantic} is used
2366 with @option{-Wformat}, warnings will be given about format features not
2367 in the selected standard version (but not for @code{strfmon} formats,
2368 since those are not in any version of the C standard). @xref{C Dialect
2369 Options,,Options Controlling C Dialect}.
2371 Since @option{-Wformat} also checks for null format arguments for
2372 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2374 @option{-Wformat} is included in @option{-Wall}. For more control over some
2375 aspects of format checking, the options @option{-Wformat-y2k},
2376 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2377 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2378 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2381 @opindex Wformat-y2k
2382 If @option{-Wformat} is specified, also warn about @code{strftime}
2383 formats which may yield only a two-digit year.
2385 @item -Wno-format-extra-args
2386 @opindex Wno-format-extra-args
2387 If @option{-Wformat} is specified, do not warn about excess arguments to a
2388 @code{printf} or @code{scanf} format function. The C standard specifies
2389 that such arguments are ignored.
2391 Where the unused arguments lie between used arguments that are
2392 specified with @samp{$} operand number specifications, normally
2393 warnings are still given, since the implementation could not know what
2394 type to pass to @code{va_arg} to skip the unused arguments. However,
2395 in the case of @code{scanf} formats, this option will suppress the
2396 warning if the unused arguments are all pointers, since the Single
2397 Unix Specification says that such unused arguments are allowed.
2399 @item -Wno-format-zero-length
2400 @opindex Wno-format-zero-length
2401 If @option{-Wformat} is specified, do not warn about zero-length formats.
2402 The C standard specifies that zero-length formats are allowed.
2404 @item -Wformat-nonliteral
2405 @opindex Wformat-nonliteral
2406 If @option{-Wformat} is specified, also warn if the format string is not a
2407 string literal and so cannot be checked, unless the format function
2408 takes its format arguments as a @code{va_list}.
2410 @item -Wformat-security
2411 @opindex Wformat-security
2412 If @option{-Wformat} is specified, also warn about uses of format
2413 functions that represent possible security problems. At present, this
2414 warns about calls to @code{printf} and @code{scanf} functions where the
2415 format string is not a string literal and there are no format arguments,
2416 as in @code{printf (foo);}. This may be a security hole if the format
2417 string came from untrusted input and contains @samp{%n}. (This is
2418 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2419 in future warnings may be added to @option{-Wformat-security} that are not
2420 included in @option{-Wformat-nonliteral}.)
2424 Enable @option{-Wformat} plus format checks not included in
2425 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2426 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2430 Warn about passing a null pointer for arguments marked as
2431 requiring a non-null value by the @code{nonnull} function attribute.
2433 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2434 can be disabled with the @option{-Wno-nonnull} option.
2436 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2438 Warn about uninitialized variables which are initialized with themselves.
2439 Note this option can only be used with the @option{-Wuninitialized} option,
2440 which in turn only works with @option{-O1} and above.
2442 For example, GCC will warn about @code{i} being uninitialized in the
2443 following snippet only when @option{-Winit-self} has been specified:
2454 @item -Wimplicit-int
2455 @opindex Wimplicit-int
2456 Warn when a declaration does not specify a type.
2457 This warning is enabled by @option{-Wall}.
2459 @item -Wimplicit-function-declaration
2460 @itemx -Werror-implicit-function-declaration
2461 @opindex Wimplicit-function-declaration
2462 @opindex Werror-implicit-function-declaration
2463 Give a warning (or error) whenever a function is used before being
2464 declared. The form @option{-Wno-error-implicit-function-declaration}
2466 This warning is enabled by @option{-Wall} (as a warning, not an error).
2470 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2471 This warning is enabled by @option{-Wall}.
2475 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2476 function with external linkage, returning int, taking either zero
2477 arguments, two, or three arguments of appropriate types.
2478 This warning is enabled by @option{-Wall}.
2480 @item -Wmissing-braces
2481 @opindex Wmissing-braces
2482 Warn if an aggregate or union initializer is not fully bracketed. In
2483 the following example, the initializer for @samp{a} is not fully
2484 bracketed, but that for @samp{b} is fully bracketed.
2487 int a[2][2] = @{ 0, 1, 2, 3 @};
2488 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2491 This warning is enabled by @option{-Wall}.
2493 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2494 @opindex Wmissing-include-dirs
2495 Warn if a user-supplied include directory does not exist.
2498 @opindex Wparentheses
2499 Warn if parentheses are omitted in certain contexts, such
2500 as when there is an assignment in a context where a truth value
2501 is expected, or when operators are nested whose precedence people
2502 often get confused about. Only the warning for an assignment used as
2503 a truth value is supported when compiling C++; the other warnings are
2504 only supported when compiling C@.
2506 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2507 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2508 interpretation from that of ordinary mathematical notation.
2510 Also warn about constructions where there may be confusion to which
2511 @code{if} statement an @code{else} branch belongs. Here is an example of
2526 In C, every @code{else} branch belongs to the innermost possible @code{if}
2527 statement, which in this example is @code{if (b)}. This is often not
2528 what the programmer expected, as illustrated in the above example by
2529 indentation the programmer chose. When there is the potential for this
2530 confusion, GCC will issue a warning when this flag is specified.
2531 To eliminate the warning, add explicit braces around the innermost
2532 @code{if} statement so there is no way the @code{else} could belong to
2533 the enclosing @code{if}. The resulting code would look like this:
2549 This warning is enabled by @option{-Wall}.
2551 @item -Wsequence-point
2552 @opindex Wsequence-point
2553 Warn about code that may have undefined semantics because of violations
2554 of sequence point rules in the C and C++ standards.
2556 The C and C++ standards defines the order in which expressions in a C/C++
2557 program are evaluated in terms of @dfn{sequence points}, which represent
2558 a partial ordering between the execution of parts of the program: those
2559 executed before the sequence point, and those executed after it. These
2560 occur after the evaluation of a full expression (one which is not part
2561 of a larger expression), after the evaluation of the first operand of a
2562 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2563 function is called (but after the evaluation of its arguments and the
2564 expression denoting the called function), and in certain other places.
2565 Other than as expressed by the sequence point rules, the order of
2566 evaluation of subexpressions of an expression is not specified. All
2567 these rules describe only a partial order rather than a total order,
2568 since, for example, if two functions are called within one expression
2569 with no sequence point between them, the order in which the functions
2570 are called is not specified. However, the standards committee have
2571 ruled that function calls do not overlap.
2573 It is not specified when between sequence points modifications to the
2574 values of objects take effect. Programs whose behavior depends on this
2575 have undefined behavior; the C and C++ standards specify that ``Between
2576 the previous and next sequence point an object shall have its stored
2577 value modified at most once by the evaluation of an expression.
2578 Furthermore, the prior value shall be read only to determine the value
2579 to be stored.''. If a program breaks these rules, the results on any
2580 particular implementation are entirely unpredictable.
2582 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2583 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2584 diagnosed by this option, and it may give an occasional false positive
2585 result, but in general it has been found fairly effective at detecting
2586 this sort of problem in programs.
2588 The standard is worded confusingly, therefore there is some debate
2589 over the precise meaning of the sequence point rules in subtle cases.
2590 Links to discussions of the problem, including proposed formal
2591 definitions, may be found on the GCC readings page, at
2592 @w{@uref{http://gcc.gnu.org/readings.html}}.
2594 This warning is enabled by @option{-Wall} for C and C++.
2597 @opindex Wreturn-type
2598 Warn whenever a function is defined with a return-type that defaults to
2599 @code{int}. Also warn about any @code{return} statement with no
2600 return-value in a function whose return-type is not @code{void}.
2602 For C, also warn if the return type of a function has a type qualifier
2603 such as @code{const}. Such a type qualifier has no effect, since the
2604 value returned by a function is not an lvalue. ISO C prohibits
2605 qualified @code{void} return types on function definitions, so such
2606 return types always receive a warning even without this option.
2608 For C++, a function without return type always produces a diagnostic
2609 message, even when @option{-Wno-return-type} is specified. The only
2610 exceptions are @samp{main} and functions defined in system headers.
2612 This warning is enabled by @option{-Wall}.
2616 Warn whenever a @code{switch} statement has an index of enumerated type
2617 and lacks a @code{case} for one or more of the named codes of that
2618 enumeration. (The presence of a @code{default} label prevents this
2619 warning.) @code{case} labels outside the enumeration range also
2620 provoke warnings when this option is used.
2621 This warning is enabled by @option{-Wall}.
2623 @item -Wswitch-default
2624 @opindex Wswitch-switch
2625 Warn whenever a @code{switch} statement does not have a @code{default}
2629 @opindex Wswitch-enum
2630 Warn whenever a @code{switch} statement has an index of enumerated type
2631 and lacks a @code{case} for one or more of the named codes of that
2632 enumeration. @code{case} labels outside the enumeration range also
2633 provoke warnings when this option is used.
2637 Warn if any trigraphs are encountered that might change the meaning of
2638 the program (trigraphs within comments are not warned about).
2639 This warning is enabled by @option{-Wall}.
2641 @item -Wunused-function
2642 @opindex Wunused-function
2643 Warn whenever a static function is declared but not defined or a
2644 non-inline static function is unused.
2645 This warning is enabled by @option{-Wall}.
2647 @item -Wunused-label
2648 @opindex Wunused-label
2649 Warn whenever a label is declared but not used.
2650 This warning is enabled by @option{-Wall}.
2652 To suppress this warning use the @samp{unused} attribute
2653 (@pxref{Variable Attributes}).
2655 @item -Wunused-parameter
2656 @opindex Wunused-parameter
2657 Warn whenever a function parameter is unused aside from its declaration.
2659 To suppress this warning use the @samp{unused} attribute
2660 (@pxref{Variable Attributes}).
2662 @item -Wunused-variable
2663 @opindex Wunused-variable
2664 Warn whenever a local variable or non-constant static variable is unused
2665 aside from its declaration
2666 This warning is enabled by @option{-Wall}.
2668 To suppress this warning use the @samp{unused} attribute
2669 (@pxref{Variable Attributes}).
2671 @item -Wunused-value
2672 @opindex Wunused-value
2673 Warn whenever a statement computes a result that is explicitly not used.
2674 This warning is enabled by @option{-Wall}.
2676 To suppress this warning cast the expression to @samp{void}.
2680 All the above @option{-Wunused} options combined.
2682 In order to get a warning about an unused function parameter, you must
2683 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2684 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2686 @item -Wuninitialized
2687 @opindex Wuninitialized
2688 Warn if an automatic variable is used without first being initialized or
2689 if a variable may be clobbered by a @code{setjmp} call.
2691 These warnings are possible only in optimizing compilation,
2692 because they require data flow information that is computed only
2693 when optimizing. If you do not specify @option{-O}, you will not get
2694 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2695 requiring @option{-O}.
2697 If you want to warn about code which uses the uninitialized value of the
2698 variable in its own initializer, use the @option{-Winit-self} option.
2700 These warnings occur for individual uninitialized or clobbered
2701 elements of structure, union or array variables as well as for
2702 variables which are uninitialized or clobbered as a whole. They do
2703 not occur for variables or elements declared @code{volatile}. Because
2704 these warnings depend on optimization, the exact variables or elements
2705 for which there are warnings will depend on the precise optimization
2706 options and version of GCC used.
2708 Note that there may be no warning about a variable that is used only
2709 to compute a value that itself is never used, because such
2710 computations may be deleted by data flow analysis before the warnings
2713 These warnings are made optional because GCC is not smart
2714 enough to see all the reasons why the code might be correct
2715 despite appearing to have an error. Here is one example of how
2736 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2737 always initialized, but GCC doesn't know this. Here is
2738 another common case:
2743 if (change_y) save_y = y, y = new_y;
2745 if (change_y) y = save_y;
2750 This has no bug because @code{save_y} is used only if it is set.
2752 @cindex @code{longjmp} warnings
2753 This option also warns when a non-volatile automatic variable might be
2754 changed by a call to @code{longjmp}. These warnings as well are possible
2755 only in optimizing compilation.
2757 The compiler sees only the calls to @code{setjmp}. It cannot know
2758 where @code{longjmp} will be called; in fact, a signal handler could
2759 call it at any point in the code. As a result, you may get a warning
2760 even when there is in fact no problem because @code{longjmp} cannot
2761 in fact be called at the place which would cause a problem.
2763 Some spurious warnings can be avoided if you declare all the functions
2764 you use that never return as @code{noreturn}. @xref{Function
2767 This warning is enabled by @option{-Wall}.
2769 @item -Wunknown-pragmas
2770 @opindex Wunknown-pragmas
2771 @cindex warning for unknown pragmas
2772 @cindex unknown pragmas, warning
2773 @cindex pragmas, warning of unknown
2774 Warn when a #pragma directive is encountered which is not understood by
2775 GCC@. If this command line option is used, warnings will even be issued
2776 for unknown pragmas in system header files. This is not the case if
2777 the warnings were only enabled by the @option{-Wall} command line option.
2780 @opindex Wno-pragmas
2782 Do not warn about misuses of pragmas, such as incorrect parameters,
2783 invalid syntax, or conflicts between pragmas. See also
2784 @samp{-Wunknown-pragmas}.
2786 @item -Wstrict-aliasing
2787 @opindex Wstrict-aliasing
2788 This option is only active when @option{-fstrict-aliasing} is active.
2789 It warns about code which might break the strict aliasing rules that the
2790 compiler is using for optimization. The warning does not catch all
2791 cases, but does attempt to catch the more common pitfalls. It is
2792 included in @option{-Wall}.
2794 @item -Wstrict-aliasing=2
2795 @opindex Wstrict-aliasing=2
2796 This option is only active when @option{-fstrict-aliasing} is active.
2797 It warns about code which might break the strict aliasing rules that the
2798 compiler is using for optimization. This warning catches more cases than
2799 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2800 cases that are safe.
2804 All of the above @samp{-W} options combined. This enables all the
2805 warnings about constructions that some users consider questionable, and
2806 that are easy to avoid (or modify to prevent the warning), even in
2807 conjunction with macros. This also enables some language-specific
2808 warnings described in @ref{C++ Dialect Options} and
2809 @ref{Objective-C and Objective-C++ Dialect Options}.
2812 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2813 Some of them warn about constructions that users generally do not
2814 consider questionable, but which occasionally you might wish to check
2815 for; others warn about constructions that are necessary or hard to avoid
2816 in some cases, and there is no simple way to modify the code to suppress
2823 (This option used to be called @option{-W}. The older name is still
2824 supported, but the newer name is more descriptive.) Print extra warning
2825 messages for these events:
2829 A function can return either with or without a value. (Falling
2830 off the end of the function body is considered returning without
2831 a value.) For example, this function would evoke such a
2845 An expression-statement or the left-hand side of a comma expression
2846 contains no side effects.
2847 To suppress the warning, cast the unused expression to void.
2848 For example, an expression such as @samp{x[i,j]} will cause a warning,
2849 but @samp{x[(void)i,j]} will not.
2852 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2855 Storage-class specifiers like @code{static} are not the first things in
2856 a declaration. According to the C Standard, this usage is obsolescent.
2859 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2863 A comparison between signed and unsigned values could produce an
2864 incorrect result when the signed value is converted to unsigned.
2865 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2868 An aggregate has an initializer which does not initialize all members.
2869 This warning can be independently controlled by
2870 @option{-Wmissing-field-initializers}.
2873 A function parameter is declared without a type specifier in K&R-style
2881 An empty body occurs in an @samp{if} or @samp{else} statement.
2884 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2885 @samp{>}, or @samp{>=}.
2888 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2891 Any of several floating-point events that often indicate errors, such as
2892 overflow, underflow, loss of precision, etc.
2894 @item @r{(C++ only)}
2895 An enumerator and a non-enumerator both appear in a conditional expression.
2897 @item @r{(C++ only)}
2898 A non-static reference or non-static @samp{const} member appears in a
2899 class without constructors.
2901 @item @r{(C++ only)}
2902 Ambiguous virtual bases.
2904 @item @r{(C++ only)}
2905 Subscripting an array which has been declared @samp{register}.
2907 @item @r{(C++ only)}
2908 Taking the address of a variable which has been declared @samp{register}.
2910 @item @r{(C++ only)}
2911 A base class is not initialized in a derived class' copy constructor.
2914 @item -Wno-div-by-zero
2915 @opindex Wno-div-by-zero
2916 @opindex Wdiv-by-zero
2917 Do not warn about compile-time integer division by zero. Floating point
2918 division by zero is not warned about, as it can be a legitimate way of
2919 obtaining infinities and NaNs.
2921 @item -Wsystem-headers
2922 @opindex Wsystem-headers
2923 @cindex warnings from system headers
2924 @cindex system headers, warnings from
2925 Print warning messages for constructs found in system header files.
2926 Warnings from system headers are normally suppressed, on the assumption
2927 that they usually do not indicate real problems and would only make the
2928 compiler output harder to read. Using this command line option tells
2929 GCC to emit warnings from system headers as if they occurred in user
2930 code. However, note that using @option{-Wall} in conjunction with this
2931 option will @emph{not} warn about unknown pragmas in system
2932 headers---for that, @option{-Wunknown-pragmas} must also be used.
2935 @opindex Wfloat-equal
2936 Warn if floating point values are used in equality comparisons.
2938 The idea behind this is that sometimes it is convenient (for the
2939 programmer) to consider floating-point values as approximations to
2940 infinitely precise real numbers. If you are doing this, then you need
2941 to compute (by analyzing the code, or in some other way) the maximum or
2942 likely maximum error that the computation introduces, and allow for it
2943 when performing comparisons (and when producing output, but that's a
2944 different problem). In particular, instead of testing for equality, you
2945 would check to see whether the two values have ranges that overlap; and
2946 this is done with the relational operators, so equality comparisons are
2949 @item -Wtraditional @r{(C only)}
2950 @opindex Wtraditional
2951 Warn about certain constructs that behave differently in traditional and
2952 ISO C@. Also warn about ISO C constructs that have no traditional C
2953 equivalent, and/or problematic constructs which should be avoided.
2957 Macro parameters that appear within string literals in the macro body.
2958 In traditional C macro replacement takes place within string literals,
2959 but does not in ISO C@.
2962 In traditional C, some preprocessor directives did not exist.
2963 Traditional preprocessors would only consider a line to be a directive
2964 if the @samp{#} appeared in column 1 on the line. Therefore
2965 @option{-Wtraditional} warns about directives that traditional C
2966 understands but would ignore because the @samp{#} does not appear as the
2967 first character on the line. It also suggests you hide directives like
2968 @samp{#pragma} not understood by traditional C by indenting them. Some
2969 traditional implementations would not recognize @samp{#elif}, so it
2970 suggests avoiding it altogether.
2973 A function-like macro that appears without arguments.
2976 The unary plus operator.
2979 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2980 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2981 constants.) Note, these suffixes appear in macros defined in the system
2982 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2983 Use of these macros in user code might normally lead to spurious
2984 warnings, however GCC's integrated preprocessor has enough context to
2985 avoid warning in these cases.
2988 A function declared external in one block and then used after the end of
2992 A @code{switch} statement has an operand of type @code{long}.
2995 A non-@code{static} function declaration follows a @code{static} one.
2996 This construct is not accepted by some traditional C compilers.
2999 The ISO type of an integer constant has a different width or
3000 signedness from its traditional type. This warning is only issued if
3001 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3002 typically represent bit patterns, are not warned about.
3005 Usage of ISO string concatenation is detected.
3008 Initialization of automatic aggregates.
3011 Identifier conflicts with labels. Traditional C lacks a separate
3012 namespace for labels.
3015 Initialization of unions. If the initializer is zero, the warning is
3016 omitted. This is done under the assumption that the zero initializer in
3017 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3018 initializer warnings and relies on default initialization to zero in the
3022 Conversions by prototypes between fixed/floating point values and vice
3023 versa. The absence of these prototypes when compiling with traditional
3024 C would cause serious problems. This is a subset of the possible
3025 conversion warnings, for the full set use @option{-Wconversion}.
3028 Use of ISO C style function definitions. This warning intentionally is
3029 @emph{not} issued for prototype declarations or variadic functions
3030 because these ISO C features will appear in your code when using
3031 libiberty's traditional C compatibility macros, @code{PARAMS} and
3032 @code{VPARAMS}. This warning is also bypassed for nested functions
3033 because that feature is already a GCC extension and thus not relevant to
3034 traditional C compatibility.
3037 @item -Wdeclaration-after-statement @r{(C only)}
3038 @opindex Wdeclaration-after-statement
3039 Warn when a declaration is found after a statement in a block. This
3040 construct, known from C++, was introduced with ISO C99 and is by default
3041 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3042 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3046 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3048 @item -Wno-endif-labels
3049 @opindex Wno-endif-labels
3050 @opindex Wendif-labels
3051 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3055 Warn whenever a local variable shadows another local variable, parameter or
3056 global variable or whenever a built-in function is shadowed.
3058 @item -Wlarger-than-@var{len}
3059 @opindex Wlarger-than
3060 Warn whenever an object of larger than @var{len} bytes is defined.
3062 @item -Wunsafe-loop-optimizations
3063 @opindex Wunsafe-loop-optimizations
3064 Warn if the loop cannot be optimized because the compiler could not
3065 assume anything on the bounds of the loop indices. With
3066 @option{-funsafe-loop-optimizations} warn if the compiler made
3069 @item -Wpointer-arith
3070 @opindex Wpointer-arith
3071 Warn about anything that depends on the ``size of'' a function type or
3072 of @code{void}. GNU C assigns these types a size of 1, for
3073 convenience in calculations with @code{void *} pointers and pointers
3076 @item -Wbad-function-cast @r{(C only)}
3077 @opindex Wbad-function-cast
3078 Warn whenever a function call is cast to a non-matching type.
3079 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3082 Warn about ISO C constructs that are outside of the common subset of
3083 ISO C and ISO C++, e.g.@: request for implicit conversion from
3084 @code{void *} to a pointer to non-@code{void} type.
3088 Warn whenever a pointer is cast so as to remove a type qualifier from
3089 the target type. For example, warn if a @code{const char *} is cast
3090 to an ordinary @code{char *}.
3093 @opindex Wcast-align
3094 Warn whenever a pointer is cast such that the required alignment of the
3095 target is increased. For example, warn if a @code{char *} is cast to
3096 an @code{int *} on machines where integers can only be accessed at
3097 two- or four-byte boundaries.
3099 @item -Wwrite-strings
3100 @opindex Wwrite-strings
3101 When compiling C, give string constants the type @code{const
3102 char[@var{length}]} so that
3103 copying the address of one into a non-@code{const} @code{char *}
3104 pointer will get a warning; when compiling C++, warn about the
3105 deprecated conversion from string literals to @code{char *}. This
3106 warning, by default, is enabled for C++ programs.
3107 These warnings will help you find at
3108 compile time code that can try to write into a string constant, but
3109 only if you have been very careful about using @code{const} in
3110 declarations and prototypes. Otherwise, it will just be a nuisance;
3111 this is why we did not make @option{-Wall} request these warnings.
3114 @opindex Wconversion
3115 Warn if a prototype causes a type conversion that is different from what
3116 would happen to the same argument in the absence of a prototype. This
3117 includes conversions of fixed point to floating and vice versa, and
3118 conversions changing the width or signedness of a fixed point argument
3119 except when the same as the default promotion.
3121 Also, warn if a negative integer constant expression is implicitly
3122 converted to an unsigned type. For example, warn about the assignment
3123 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3124 casts like @code{(unsigned) -1}.
3126 @item -Wsign-compare
3127 @opindex Wsign-compare
3128 @cindex warning for comparison of signed and unsigned values
3129 @cindex comparison of signed and unsigned values, warning
3130 @cindex signed and unsigned values, comparison warning
3131 Warn when a comparison between signed and unsigned values could produce
3132 an incorrect result when the signed value is converted to unsigned.
3133 This warning is also enabled by @option{-Wextra}; to get the other warnings
3134 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3136 @item -Waggregate-return
3137 @opindex Waggregate-return
3138 Warn if any functions that return structures or unions are defined or
3139 called. (In languages where you can return an array, this also elicits
3143 @opindex Walways-true
3144 Warn about comparisons which are always true such as testing if
3145 unsigned values are greater than or equal to zero. This warning is
3146 enabled by @option{-Wall}.
3148 @item -Wno-attributes
3149 @opindex Wno-attributes
3150 @opindex Wattributes
3151 Do not warn if an unexpected @code{__attribute__} is used, such as
3152 unrecognized attributes, function attributes applied to variables,
3153 etc. This will not stop errors for incorrect use of supported
3156 @item -Wstrict-prototypes @r{(C only)}
3157 @opindex Wstrict-prototypes
3158 Warn if a function is declared or defined without specifying the
3159 argument types. (An old-style function definition is permitted without
3160 a warning if preceded by a declaration which specifies the argument
3163 @item -Wold-style-definition @r{(C only)}
3164 @opindex Wold-style-definition
3165 Warn if an old-style function definition is used. A warning is given
3166 even if there is a previous prototype.
3168 @item -Wmissing-prototypes @r{(C only)}
3169 @opindex Wmissing-prototypes
3170 Warn if a global function is defined without a previous prototype
3171 declaration. This warning is issued even if the definition itself
3172 provides a prototype. The aim is to detect global functions that fail
3173 to be declared in header files.
3175 @item -Wmissing-declarations @r{(C only)}
3176 @opindex Wmissing-declarations
3177 Warn if a global function is defined without a previous declaration.
3178 Do so even if the definition itself provides a prototype.
3179 Use this option to detect global functions that are not declared in
3182 @item -Wmissing-field-initializers
3183 @opindex Wmissing-field-initializers
3186 Warn if a structure's initializer has some fields missing. For
3187 example, the following code would cause such a warning, because
3188 @code{x.h} is implicitly zero:
3191 struct s @{ int f, g, h; @};
3192 struct s x = @{ 3, 4 @};
3195 This option does not warn about designated initializers, so the following
3196 modification would not trigger a warning:
3199 struct s @{ int f, g, h; @};
3200 struct s x = @{ .f = 3, .g = 4 @};
3203 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3204 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3206 @item -Wmissing-noreturn
3207 @opindex Wmissing-noreturn
3208 Warn about functions which might be candidates for attribute @code{noreturn}.
3209 Note these are only possible candidates, not absolute ones. Care should
3210 be taken to manually verify functions actually do not ever return before
3211 adding the @code{noreturn} attribute, otherwise subtle code generation
3212 bugs could be introduced. You will not get a warning for @code{main} in
3213 hosted C environments.
3215 @item -Wmissing-format-attribute
3216 @opindex Wmissing-format-attribute
3218 Warn about function pointers which might be candidates for @code{format}
3219 attributes. Note these are only possible candidates, not absolute ones.
3220 GCC will guess that function pointers with @code{format} attributes that
3221 are used in assignment, initialization, parameter passing or return
3222 statements should have a corresponding @code{format} attribute in the
3223 resulting type. I.e.@: the left-hand side of the assignment or
3224 initialization, the type of the parameter variable, or the return type
3225 of the containing function respectively should also have a @code{format}
3226 attribute to avoid the warning.
3228 GCC will also warn about function definitions which might be
3229 candidates for @code{format} attributes. Again, these are only
3230 possible candidates. GCC will guess that @code{format} attributes
3231 might be appropriate for any function that calls a function like
3232 @code{vprintf} or @code{vscanf}, but this might not always be the
3233 case, and some functions for which @code{format} attributes are
3234 appropriate may not be detected.
3236 @item -Wno-multichar
3237 @opindex Wno-multichar
3239 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3240 Usually they indicate a typo in the user's code, as they have
3241 implementation-defined values, and should not be used in portable code.
3243 @item -Wnormalized=<none|id|nfc|nfkc>
3244 @opindex Wnormalized
3247 @cindex character set, input normalization
3248 In ISO C and ISO C++, two identifiers are different if they are
3249 different sequences of characters. However, sometimes when characters
3250 outside the basic ASCII character set are used, you can have two
3251 different character sequences that look the same. To avoid confusion,
3252 the ISO 10646 standard sets out some @dfn{normalization rules} which
3253 when applied ensure that two sequences that look the same are turned into
3254 the same sequence. GCC can warn you if you are using identifiers which
3255 have not been normalized; this option controls that warning.
3257 There are four levels of warning that GCC supports. The default is
3258 @option{-Wnormalized=nfc}, which warns about any identifier which is
3259 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3260 recommended form for most uses.
3262 Unfortunately, there are some characters which ISO C and ISO C++ allow
3263 in identifiers that when turned into NFC aren't allowable as
3264 identifiers. That is, there's no way to use these symbols in portable
3265 ISO C or C++ and have all your identifiers in NFC.
3266 @option{-Wnormalized=id} suppresses the warning for these characters.
3267 It is hoped that future versions of the standards involved will correct
3268 this, which is why this option is not the default.
3270 You can switch the warning off for all characters by writing
3271 @option{-Wnormalized=none}. You would only want to do this if you
3272 were using some other normalization scheme (like ``D''), because
3273 otherwise you can easily create bugs that are literally impossible to see.
3275 Some characters in ISO 10646 have distinct meanings but look identical
3276 in some fonts or display methodologies, especially once formatting has
3277 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3278 LETTER N'', will display just like a regular @code{n} which has been
3279 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3280 normalization scheme to convert all these into a standard form as
3281 well, and GCC will warn if your code is not in NFKC if you use
3282 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3283 about every identifier that contains the letter O because it might be
3284 confused with the digit 0, and so is not the default, but may be
3285 useful as a local coding convention if the programming environment is
3286 unable to be fixed to display these characters distinctly.
3288 @item -Wno-deprecated-declarations
3289 @opindex Wno-deprecated-declarations
3290 Do not warn about uses of functions, variables, and types marked as
3291 deprecated by using the @code{deprecated} attribute.
3292 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3293 @pxref{Type Attributes}.)
3296 @opindex Wno-overflow
3297 Do not warn about compile-time overflow in constant expressions.
3301 Warn if a structure is given the packed attribute, but the packed
3302 attribute has no effect on the layout or size of the structure.
3303 Such structures may be mis-aligned for little benefit. For
3304 instance, in this code, the variable @code{f.x} in @code{struct bar}
3305 will be misaligned even though @code{struct bar} does not itself
3306 have the packed attribute:
3313 @} __attribute__((packed));
3323 Warn if padding is included in a structure, either to align an element
3324 of the structure or to align the whole structure. Sometimes when this
3325 happens it is possible to rearrange the fields of the structure to
3326 reduce the padding and so make the structure smaller.
3328 @item -Wredundant-decls
3329 @opindex Wredundant-decls
3330 Warn if anything is declared more than once in the same scope, even in
3331 cases where multiple declaration is valid and changes nothing.
3333 @item -Wnested-externs @r{(C only)}
3334 @opindex Wnested-externs
3335 Warn if an @code{extern} declaration is encountered within a function.
3337 @item -Wunreachable-code
3338 @opindex Wunreachable-code
3339 Warn if the compiler detects that code will never be executed.
3341 This option is intended to warn when the compiler detects that at
3342 least a whole line of source code will never be executed, because
3343 some condition is never satisfied or because it is after a
3344 procedure that never returns.
3346 It is possible for this option to produce a warning even though there
3347 are circumstances under which part of the affected line can be executed,
3348 so care should be taken when removing apparently-unreachable code.
3350 For instance, when a function is inlined, a warning may mean that the
3351 line is unreachable in only one inlined copy of the function.
3353 This option is not made part of @option{-Wall} because in a debugging
3354 version of a program there is often substantial code which checks
3355 correct functioning of the program and is, hopefully, unreachable
3356 because the program does work. Another common use of unreachable
3357 code is to provide behavior which is selectable at compile-time.
3361 Warn if a function can not be inlined and it was declared as inline.
3362 Even with this option, the compiler will not warn about failures to
3363 inline functions declared in system headers.
3365 The compiler uses a variety of heuristics to determine whether or not
3366 to inline a function. For example, the compiler takes into account
3367 the size of the function being inlined and the amount of inlining
3368 that has already been done in the current function. Therefore,
3369 seemingly insignificant changes in the source program can cause the
3370 warnings produced by @option{-Winline} to appear or disappear.
3372 @item -Wno-invalid-offsetof @r{(C++ only)}
3373 @opindex Wno-invalid-offsetof
3374 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3375 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3376 to a non-POD type is undefined. In existing C++ implementations,
3377 however, @samp{offsetof} typically gives meaningful results even when
3378 applied to certain kinds of non-POD types. (Such as a simple
3379 @samp{struct} that fails to be a POD type only by virtue of having a
3380 constructor.) This flag is for users who are aware that they are
3381 writing nonportable code and who have deliberately chosen to ignore the
3384 The restrictions on @samp{offsetof} may be relaxed in a future version
3385 of the C++ standard.
3387 @item -Wno-int-to-pointer-cast @r{(C only)}
3388 @opindex Wno-int-to-pointer-cast
3389 Suppress warnings from casts to pointer type of an integer of a
3392 @item -Wno-pointer-to-int-cast @r{(C only)}
3393 @opindex Wno-pointer-to-int-cast
3394 Suppress warnings from casts from a pointer to an integer type of a
3398 @opindex Winvalid-pch
3399 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3400 the search path but can't be used.
3404 @opindex Wno-long-long
3405 Warn if @samp{long long} type is used. This is default. To inhibit
3406 the warning messages, use @option{-Wno-long-long}. Flags
3407 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3408 only when @option{-pedantic} flag is used.
3410 @item -Wvariadic-macros
3411 @opindex Wvariadic-macros
3412 @opindex Wno-variadic-macros
3413 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3414 alternate syntax when in pedantic ISO C99 mode. This is default.
3415 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3417 @item -Wvolatile-register-var
3418 @opindex Wvolatile-register-var
3419 @opindex Wno-volatile-register-var
3420 Warn if a register variable is declared volatile. The volatile
3421 modifier does not inhibit all optimizations that may eliminate reads
3422 and/or writes to register variables.
3424 @item -Wdisabled-optimization
3425 @opindex Wdisabled-optimization
3426 Warn if a requested optimization pass is disabled. This warning does
3427 not generally indicate that there is anything wrong with your code; it
3428 merely indicates that GCC's optimizers were unable to handle the code
3429 effectively. Often, the problem is that your code is too big or too
3430 complex; GCC will refuse to optimize programs when the optimization
3431 itself is likely to take inordinate amounts of time.
3433 @item -Wpointer-sign
3434 @opindex Wpointer-sign
3435 @opindex Wno-pointer-sign
3436 Warn for pointer argument passing or assignment with different signedness.
3437 This option is only supported for C and Objective-C@. It is implied by
3438 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3439 @option{-Wno-pointer-sign}.
3443 Make all warnings into errors.
3447 Make the specified warning into an errors. The specifier for a
3448 warning is appended, for example @option{-Werror=switch} turns the
3449 warnings controlled by @option{-Wswitch} into errors. This switch
3450 takes a negative form, to be used to negate @option{-Werror} for
3451 specific warnings, for example @option{-Wno-error=switch} makes
3452 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3453 is in effect. You can use the @option{-fdiagnostics-show-option}
3454 option to have each controllable warning amended with the option which
3455 controls it, to determine what to use with this option.
3457 Note that specifying @option{-Werror=}@var{foo} automatically implies
3458 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3461 @item -Wstack-protector
3462 @opindex Wstack-protector
3463 This option is only active when @option{-fstack-protector} is active. It
3464 warns about functions that will not be protected against stack smashing.
3466 @item -Wstring-literal-comparison
3467 @opindex Wstring-literal-comparison
3468 Warn about suspicious comparisons to string literal constants. In C,
3469 direct comparisons against the memory address of a string literal, such
3470 as @code{if (x == "abc")}, typically indicate a programmer error, and
3471 even when intentional, result in unspecified behavior and are not portable.
3472 Usually these warnings alert that the programmer intended to use
3473 @code{strcmp}. This warning is enabled by @option{-Wall}.
3475 @item -Woverlength-strings
3476 @opindex Woverlength-strings
3477 Warn about string constants which are longer than the ``minimum
3478 maximum'' length specified in the C standard. Modern compilers
3479 generally allow string constants which are much longer than the
3480 standard's minimum limit, but very portable programs should avoid
3481 using longer strings.
3483 The limit applies @emph{after} string constant concatenation, and does
3484 not count the trailing NUL@. In C89, the limit was 509 characters; in
3485 C99, it was raised to 4095. C++98 does not specify a normative
3486 minimum maximum, so we do not diagnose overlength strings in C++@.
3488 This option is implied by @option{-pedantic}, and can be disabled with
3489 @option{-Wno-overlength-strings}.
3492 @node Debugging Options
3493 @section Options for Debugging Your Program or GCC
3494 @cindex options, debugging
3495 @cindex debugging information options
3497 GCC has various special options that are used for debugging
3498 either your program or GCC:
3503 Produce debugging information in the operating system's native format
3504 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3507 On most systems that use stabs format, @option{-g} enables use of extra
3508 debugging information that only GDB can use; this extra information
3509 makes debugging work better in GDB but will probably make other debuggers
3511 refuse to read the program. If you want to control for certain whether
3512 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3513 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3515 GCC allows you to use @option{-g} with
3516 @option{-O}. The shortcuts taken by optimized code may occasionally
3517 produce surprising results: some variables you declared may not exist
3518 at all; flow of control may briefly move where you did not expect it;
3519 some statements may not be executed because they compute constant
3520 results or their values were already at hand; some statements may
3521 execute in different places because they were moved out of loops.
3523 Nevertheless it proves possible to debug optimized output. This makes
3524 it reasonable to use the optimizer for programs that might have bugs.
3526 The following options are useful when GCC is generated with the
3527 capability for more than one debugging format.
3531 Produce debugging information for use by GDB@. This means to use the
3532 most expressive format available (DWARF 2, stabs, or the native format
3533 if neither of those are supported), including GDB extensions if at all
3538 Produce debugging information in stabs format (if that is supported),
3539 without GDB extensions. This is the format used by DBX on most BSD
3540 systems. On MIPS, Alpha and System V Release 4 systems this option
3541 produces stabs debugging output which is not understood by DBX or SDB@.
3542 On System V Release 4 systems this option requires the GNU assembler.
3544 @item -feliminate-unused-debug-symbols
3545 @opindex feliminate-unused-debug-symbols
3546 Produce debugging information in stabs format (if that is supported),
3547 for only symbols that are actually used.
3549 @item -femit-class-debug-always
3550 Instead of emitting debugging information for a C++ class in only one
3551 object file, emit it in all object files using the class. This option
3552 should be used only with debuggers that are unable to handle the way GCC
3553 normally emits debugging information for classes because using this
3554 option will increase the size of debugging information by as much as a
3559 Produce debugging information in stabs format (if that is supported),
3560 using GNU extensions understood only by the GNU debugger (GDB)@. The
3561 use of these extensions is likely to make other debuggers crash or
3562 refuse to read the program.
3566 Produce debugging information in COFF format (if that is supported).
3567 This is the format used by SDB on most System V systems prior to
3572 Produce debugging information in XCOFF format (if that is supported).
3573 This is the format used by the DBX debugger on IBM RS/6000 systems.
3577 Produce debugging information in XCOFF format (if that is supported),
3578 using GNU extensions understood only by the GNU debugger (GDB)@. The
3579 use of these extensions is likely to make other debuggers crash or
3580 refuse to read the program, and may cause assemblers other than the GNU
3581 assembler (GAS) to fail with an error.
3585 Produce debugging information in DWARF version 2 format (if that is
3586 supported). This is the format used by DBX on IRIX 6. With this
3587 option, GCC uses features of DWARF version 3 when they are useful;
3588 version 3 is upward compatible with version 2, but may still cause
3589 problems for older debuggers.
3593 Produce debugging information in VMS debug format (if that is
3594 supported). This is the format used by DEBUG on VMS systems.
3597 @itemx -ggdb@var{level}
3598 @itemx -gstabs@var{level}
3599 @itemx -gcoff@var{level}
3600 @itemx -gxcoff@var{level}
3601 @itemx -gvms@var{level}
3602 Request debugging information and also use @var{level} to specify how
3603 much information. The default level is 2.
3605 Level 1 produces minimal information, enough for making backtraces in
3606 parts of the program that you don't plan to debug. This includes
3607 descriptions of functions and external variables, but no information
3608 about local variables and no line numbers.
3610 Level 3 includes extra information, such as all the macro definitions
3611 present in the program. Some debuggers support macro expansion when
3612 you use @option{-g3}.
3614 @option{-gdwarf-2} does not accept a concatenated debug level, because
3615 GCC used to support an option @option{-gdwarf} that meant to generate
3616 debug information in version 1 of the DWARF format (which is very
3617 different from version 2), and it would have been too confusing. That
3618 debug format is long obsolete, but the option cannot be changed now.
3619 Instead use an additional @option{-g@var{level}} option to change the
3620 debug level for DWARF2.
3622 @item -feliminate-dwarf2-dups
3623 @opindex feliminate-dwarf2-dups
3624 Compress DWARF2 debugging information by eliminating duplicated
3625 information about each symbol. This option only makes sense when
3626 generating DWARF2 debugging information with @option{-gdwarf-2}.
3628 @cindex @command{prof}
3631 Generate extra code to write profile information suitable for the
3632 analysis program @command{prof}. You must use this option when compiling
3633 the source files you want data about, and you must also use it when
3636 @cindex @command{gprof}
3639 Generate extra code to write profile information suitable for the
3640 analysis program @command{gprof}. You must use this option when compiling
3641 the source files you want data about, and you must also use it when
3646 Makes the compiler print out each function name as it is compiled, and
3647 print some statistics about each pass when it finishes.
3650 @opindex ftime-report
3651 Makes the compiler print some statistics about the time consumed by each
3652 pass when it finishes.
3655 @opindex fmem-report
3656 Makes the compiler print some statistics about permanent memory
3657 allocation when it finishes.
3659 @item -fprofile-arcs
3660 @opindex fprofile-arcs
3661 Add code so that program flow @dfn{arcs} are instrumented. During
3662 execution the program records how many times each branch and call is
3663 executed and how many times it is taken or returns. When the compiled
3664 program exits it saves this data to a file called
3665 @file{@var{auxname}.gcda} for each source file. The data may be used for
3666 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3667 test coverage analysis (@option{-ftest-coverage}). Each object file's
3668 @var{auxname} is generated from the name of the output file, if
3669 explicitly specified and it is not the final executable, otherwise it is
3670 the basename of the source file. In both cases any suffix is removed
3671 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3672 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3673 @xref{Cross-profiling}.
3675 @cindex @command{gcov}
3679 This option is used to compile and link code instrumented for coverage
3680 analysis. The option is a synonym for @option{-fprofile-arcs}
3681 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3682 linking). See the documentation for those options for more details.
3687 Compile the source files with @option{-fprofile-arcs} plus optimization
3688 and code generation options. For test coverage analysis, use the
3689 additional @option{-ftest-coverage} option. You do not need to profile
3690 every source file in a program.
3693 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3694 (the latter implies the former).
3697 Run the program on a representative workload to generate the arc profile
3698 information. This may be repeated any number of times. You can run
3699 concurrent instances of your program, and provided that the file system
3700 supports locking, the data files will be correctly updated. Also
3701 @code{fork} calls are detected and correctly handled (double counting
3705 For profile-directed optimizations, compile the source files again with
3706 the same optimization and code generation options plus
3707 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3708 Control Optimization}).
3711 For test coverage analysis, use @command{gcov} to produce human readable
3712 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3713 @command{gcov} documentation for further information.
3717 With @option{-fprofile-arcs}, for each function of your program GCC
3718 creates a program flow graph, then finds a spanning tree for the graph.
3719 Only arcs that are not on the spanning tree have to be instrumented: the
3720 compiler adds code to count the number of times that these arcs are
3721 executed. When an arc is the only exit or only entrance to a block, the
3722 instrumentation code can be added to the block; otherwise, a new basic
3723 block must be created to hold the instrumentation code.
3726 @item -ftest-coverage
3727 @opindex ftest-coverage
3728 Produce a notes file that the @command{gcov} code-coverage utility
3729 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3730 show program coverage. Each source file's note file is called
3731 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3732 above for a description of @var{auxname} and instructions on how to
3733 generate test coverage data. Coverage data will match the source files
3734 more closely, if you do not optimize.
3736 @item -d@var{letters}
3737 @item -fdump-rtl-@var{pass}
3739 Says to make debugging dumps during compilation at times specified by
3740 @var{letters}. This is used for debugging the RTL-based passes of the
3741 compiler. The file names for most of the dumps are made by appending a
3742 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3743 from the name of the output file, if explicitly specified and it is not
3744 an executable, otherwise it is the basename of the source file.
3746 Most debug dumps can be enabled either passing a letter to the @option{-d}
3747 option, or with a long @option{-fdump-rtl} switch; here are the possible
3748 letters for use in @var{letters} and @var{pass}, and their meanings:
3753 Annotate the assembler output with miscellaneous debugging information.
3756 @itemx -fdump-rtl-bbro
3758 @opindex fdump-rtl-bbro
3759 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3762 @itemx -fdump-rtl-combine
3764 @opindex fdump-rtl-combine
3765 Dump after instruction combination, to the file @file{@var{file}.129r.combine}.
3768 @itemx -fdump-rtl-ce1
3769 @itemx -fdump-rtl-ce2
3771 @opindex fdump-rtl-ce1
3772 @opindex fdump-rtl-ce2
3773 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3774 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3775 and @option{-fdump-rtl-ce2} enable dumping after the second if
3776 conversion, to the file @file{@var{file}.130r.ce2}.
3779 @itemx -fdump-rtl-btl
3780 @itemx -fdump-rtl-dbr
3782 @opindex fdump-rtl-btl
3783 @opindex fdump-rtl-dbr
3784 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3785 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3786 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3787 scheduling, to @file{@var{file}.36.dbr}.
3791 Dump all macro definitions, at the end of preprocessing, in addition to
3795 @itemx -fdump-rtl-ce3
3797 @opindex fdump-rtl-ce3
3798 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3801 @itemx -fdump-rtl-cfg
3802 @itemx -fdump-rtl-life
3804 @opindex fdump-rtl-cfg
3805 @opindex fdump-rtl-life
3806 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3807 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3808 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3809 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3812 @itemx -fdump-rtl-greg
3814 @opindex fdump-rtl-greg
3815 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3818 @itemx -fdump-rtl-gcse
3819 @itemx -fdump-rtl-bypass
3821 @opindex fdump-rtl-gcse
3822 @opindex fdump-rtl-bypass
3823 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3824 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3825 enable dumping after jump bypassing and control flow optimizations, to
3826 @file{@var{file}.115r.bypass}.
3829 @itemx -fdump-rtl-eh
3831 @opindex fdump-rtl-eh
3832 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3835 @itemx -fdump-rtl-sibling
3837 @opindex fdump-rtl-sibling
3838 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3841 @itemx -fdump-rtl-jump
3843 @opindex fdump-rtl-jump
3844 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3847 @itemx -fdump-rtl-stack
3849 @opindex fdump-rtl-stack
3850 Dump after conversion from registers to stack, to @file{@var{file}.152r.stack}.
3853 @itemx -fdump-rtl-lreg
3855 @opindex fdump-rtl-lreg
3856 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3859 @itemx -fdump-rtl-loop2
3861 @opindex fdump-rtl-loop2
3862 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3863 loop optimization pass, to @file{@var{file}.119r.loop2},
3864 @file{@var{file}.120r.loop2_init},
3865 @file{@var{file}.121r.loop2_invariant}, and
3866 @file{@var{file}.125r.loop2_done}.
3869 @itemx -fdump-rtl-sms
3871 @opindex fdump-rtl-sms
3872 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3875 @itemx -fdump-rtl-mach
3877 @opindex fdump-rtl-mach
3878 Dump after performing the machine dependent reorganization pass, to
3879 @file{@var{file}.155r.mach}.
3882 @itemx -fdump-rtl-rnreg
3884 @opindex fdump-rtl-rnreg
3885 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
3888 @itemx -fdump-rtl-regmove
3890 @opindex fdump-rtl-regmove
3891 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
3894 @itemx -fdump-rtl-postreload
3896 @opindex fdump-rtl-postreload
3897 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3900 @itemx -fdump-rtl-expand
3902 @opindex fdump-rtl-expand
3903 Dump after RTL generation, to @file{@var{file}.104r.expand}.
3906 @itemx -fdump-rtl-sched2
3908 @opindex fdump-rtl-sched2
3909 Dump after the second scheduling pass, to @file{@var{file}.150r.sched2}.
3912 @itemx -fdump-rtl-cse
3914 @opindex fdump-rtl-cse
3915 Dump after CSE (including the jump optimization that sometimes follows
3916 CSE), to @file{@var{file}.113r.cse}.
3919 @itemx -fdump-rtl-sched
3921 @opindex fdump-rtl-sched
3922 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3925 @itemx -fdump-rtl-cse2
3927 @opindex fdump-rtl-cse2
3928 Dump after the second CSE pass (including the jump optimization that
3929 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
3932 @itemx -fdump-rtl-tracer
3934 @opindex fdump-rtl-tracer
3935 Dump after running tracer, to @file{@var{file}.118r.tracer}.
3938 @itemx -fdump-rtl-vpt
3939 @itemx -fdump-rtl-vartrack
3941 @opindex fdump-rtl-vpt
3942 @opindex fdump-rtl-vartrack
3943 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3944 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3945 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3946 to @file{@var{file}.154r.vartrack}.
3949 @itemx -fdump-rtl-flow2
3951 @opindex fdump-rtl-flow2
3952 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
3955 @itemx -fdump-rtl-peephole2
3957 @opindex fdump-rtl-peephole2
3958 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
3961 @itemx -fdump-rtl-web
3963 @opindex fdump-rtl-web
3964 Dump after live range splitting, to @file{@var{file}.126r.web}.
3967 @itemx -fdump-rtl-all
3969 @opindex fdump-rtl-all
3970 Produce all the dumps listed above.
3974 Produce a core dump whenever an error occurs.
3978 Print statistics on memory usage, at the end of the run, to
3983 Annotate the assembler output with a comment indicating which
3984 pattern and alternative was used. The length of each instruction is
3989 Dump the RTL in the assembler output as a comment before each instruction.
3990 Also turns on @option{-dp} annotation.
3994 For each of the other indicated dump files (either with @option{-d} or
3995 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3996 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4000 Just generate RTL for a function instead of compiling it. Usually used
4001 with @samp{r} (@option{-fdump-rtl-expand}).
4005 Dump debugging information during parsing, to standard error.
4008 @item -fdump-unnumbered
4009 @opindex fdump-unnumbered
4010 When doing debugging dumps (see @option{-d} option above), suppress instruction
4011 numbers and line number note output. This makes it more feasible to
4012 use diff on debugging dumps for compiler invocations with different
4013 options, in particular with and without @option{-g}.
4015 @item -fdump-translation-unit @r{(C++ only)}
4016 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4017 @opindex fdump-translation-unit
4018 Dump a representation of the tree structure for the entire translation
4019 unit to a file. The file name is made by appending @file{.tu} to the
4020 source file name. If the @samp{-@var{options}} form is used, @var{options}
4021 controls the details of the dump as described for the
4022 @option{-fdump-tree} options.
4024 @item -fdump-class-hierarchy @r{(C++ only)}
4025 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4026 @opindex fdump-class-hierarchy
4027 Dump a representation of each class's hierarchy and virtual function
4028 table layout to a file. The file name is made by appending @file{.class}
4029 to the source file name. If the @samp{-@var{options}} form is used,
4030 @var{options} controls the details of the dump as described for the
4031 @option{-fdump-tree} options.
4033 @item -fdump-ipa-@var{switch}
4035 Control the dumping at various stages of inter-procedural analysis
4036 language tree to a file. The file name is generated by appending a switch
4037 specific suffix to the source file name. The following dumps are possible:
4041 Enables all inter-procedural analysis dumps; currently the only produced
4042 dump is the @samp{cgraph} dump.
4045 Dumps information about call-graph optimization, unused function removal,
4046 and inlining decisions.
4049 @item -fdump-tree-@var{switch}
4050 @itemx -fdump-tree-@var{switch}-@var{options}
4052 Control the dumping at various stages of processing the intermediate
4053 language tree to a file. The file name is generated by appending a switch
4054 specific suffix to the source file name. If the @samp{-@var{options}}
4055 form is used, @var{options} is a list of @samp{-} separated options that
4056 control the details of the dump. Not all options are applicable to all
4057 dumps, those which are not meaningful will be ignored. The following
4058 options are available
4062 Print the address of each node. Usually this is not meaningful as it
4063 changes according to the environment and source file. Its primary use
4064 is for tying up a dump file with a debug environment.
4066 Inhibit dumping of members of a scope or body of a function merely
4067 because that scope has been reached. Only dump such items when they
4068 are directly reachable by some other path. When dumping pretty-printed
4069 trees, this option inhibits dumping the bodies of control structures.
4071 Print a raw representation of the tree. By default, trees are
4072 pretty-printed into a C-like representation.
4074 Enable more detailed dumps (not honored by every dump option).
4076 Enable dumping various statistics about the pass (not honored by every dump
4079 Enable showing basic block boundaries (disabled in raw dumps).
4081 Enable showing virtual operands for every statement.
4083 Enable showing line numbers for statements.
4085 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4087 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4090 The following tree dumps are possible:
4094 Dump before any tree based optimization, to @file{@var{file}.original}.
4097 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4100 Dump after function inlining, to @file{@var{file}.inlined}.
4103 @opindex fdump-tree-gimple
4104 Dump each function before and after the gimplification pass to a file. The
4105 file name is made by appending @file{.gimple} to the source file name.
4108 @opindex fdump-tree-cfg
4109 Dump the control flow graph of each function to a file. The file name is
4110 made by appending @file{.cfg} to the source file name.
4113 @opindex fdump-tree-vcg
4114 Dump the control flow graph of each function to a file in VCG format. The
4115 file name is made by appending @file{.vcg} to the source file name. Note
4116 that if the file contains more than one function, the generated file cannot
4117 be used directly by VCG@. You will need to cut and paste each function's
4118 graph into its own separate file first.
4121 @opindex fdump-tree-ch
4122 Dump each function after copying loop headers. The file name is made by
4123 appending @file{.ch} to the source file name.
4126 @opindex fdump-tree-ssa
4127 Dump SSA related information to a file. The file name is made by appending
4128 @file{.ssa} to the source file name.
4131 @opindex fdump-tree-salias
4132 Dump structure aliasing variable information to a file. This file name
4133 is made by appending @file{.salias} to the source file name.
4136 @opindex fdump-tree-alias
4137 Dump aliasing information for each function. The file name is made by
4138 appending @file{.alias} to the source file name.
4141 @opindex fdump-tree-ccp
4142 Dump each function after CCP@. The file name is made by appending
4143 @file{.ccp} to the source file name.
4146 @opindex fdump-tree-storeccp
4147 Dump each function after STORE-CCP. The file name is made by appending
4148 @file{.storeccp} to the source file name.
4151 @opindex fdump-tree-pre
4152 Dump trees after partial redundancy elimination. The file name is made
4153 by appending @file{.pre} to the source file name.
4156 @opindex fdump-tree-fre
4157 Dump trees after full redundancy elimination. The file name is made
4158 by appending @file{.fre} to the source file name.
4161 @opindex fdump-tree-copyprop
4162 Dump trees after copy propagation. The file name is made
4163 by appending @file{.copyprop} to the source file name.
4165 @item store_copyprop
4166 @opindex fdump-tree-store_copyprop
4167 Dump trees after store copy-propagation. The file name is made
4168 by appending @file{.store_copyprop} to the source file name.
4171 @opindex fdump-tree-dce
4172 Dump each function after dead code elimination. The file name is made by
4173 appending @file{.dce} to the source file name.
4176 @opindex fdump-tree-mudflap
4177 Dump each function after adding mudflap instrumentation. The file name is
4178 made by appending @file{.mudflap} to the source file name.
4181 @opindex fdump-tree-sra
4182 Dump each function after performing scalar replacement of aggregates. The
4183 file name is made by appending @file{.sra} to the source file name.
4186 @opindex fdump-tree-sink
4187 Dump each function after performing code sinking. The file name is made
4188 by appending @file{.sink} to the source file name.
4191 @opindex fdump-tree-dom
4192 Dump each function after applying dominator tree optimizations. The file
4193 name is made by appending @file{.dom} to the source file name.
4196 @opindex fdump-tree-dse
4197 Dump each function after applying dead store elimination. The file
4198 name is made by appending @file{.dse} to the source file name.
4201 @opindex fdump-tree-phiopt
4202 Dump each function after optimizing PHI nodes into straightline code. The file
4203 name is made by appending @file{.phiopt} to the source file name.
4206 @opindex fdump-tree-forwprop
4207 Dump each function after forward propagating single use variables. The file
4208 name is made by appending @file{.forwprop} to the source file name.
4211 @opindex fdump-tree-copyrename
4212 Dump each function after applying the copy rename optimization. The file
4213 name is made by appending @file{.copyrename} to the source file name.
4216 @opindex fdump-tree-nrv
4217 Dump each function after applying the named return value optimization on
4218 generic trees. The file name is made by appending @file{.nrv} to the source
4222 @opindex fdump-tree-vect
4223 Dump each function after applying vectorization of loops. The file name is
4224 made by appending @file{.vect} to the source file name.
4227 @opindex fdump-tree-vrp
4228 Dump each function after Value Range Propagation (VRP). The file name
4229 is made by appending @file{.vrp} to the source file name.
4232 @opindex fdump-tree-all
4233 Enable all the available tree dumps with the flags provided in this option.
4236 @item -ftree-vectorizer-verbose=@var{n}
4237 @opindex ftree-vectorizer-verbose
4238 This option controls the amount of debugging output the vectorizer prints.
4239 This information is written to standard error, unless
4240 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4241 in which case it is output to the usual dump listing file, @file{.vect}.
4242 For @var{n}=0 no diagnostic information is reported.
4243 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4244 and the total number of loops that got vectorized.
4245 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4246 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4247 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4248 level that @option{-fdump-tree-vect-stats} uses.
4249 Higher verbosity levels mean either more information dumped for each
4250 reported loop, or same amount of information reported for more loops:
4251 If @var{n}=3, alignment related information is added to the reports.
4252 If @var{n}=4, data-references related information (e.g. memory dependences,
4253 memory access-patterns) is added to the reports.
4254 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4255 that did not pass the first analysis phase (i.e. may not be countable, or
4256 may have complicated control-flow).
4257 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4258 For @var{n}=7, all the information the vectorizer generates during its
4259 analysis and transformation is reported. This is the same verbosity level
4260 that @option{-fdump-tree-vect-details} uses.
4262 @item -frandom-seed=@var{string}
4263 @opindex frandom-string
4264 This option provides a seed that GCC uses when it would otherwise use
4265 random numbers. It is used to generate certain symbol names
4266 that have to be different in every compiled file. It is also used to
4267 place unique stamps in coverage data files and the object files that
4268 produce them. You can use the @option{-frandom-seed} option to produce
4269 reproducibly identical object files.
4271 The @var{string} should be different for every file you compile.
4273 @item -fsched-verbose=@var{n}
4274 @opindex fsched-verbose
4275 On targets that use instruction scheduling, this option controls the
4276 amount of debugging output the scheduler prints. This information is
4277 written to standard error, unless @option{-dS} or @option{-dR} is
4278 specified, in which case it is output to the usual dump
4279 listing file, @file{.sched} or @file{.sched2} respectively. However
4280 for @var{n} greater than nine, the output is always printed to standard
4283 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4284 same information as @option{-dRS}. For @var{n} greater than one, it
4285 also output basic block probabilities, detailed ready list information
4286 and unit/insn info. For @var{n} greater than two, it includes RTL
4287 at abort point, control-flow and regions info. And for @var{n} over
4288 four, @option{-fsched-verbose} also includes dependence info.
4292 Store the usual ``temporary'' intermediate files permanently; place them
4293 in the current directory and name them based on the source file. Thus,
4294 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4295 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4296 preprocessed @file{foo.i} output file even though the compiler now
4297 normally uses an integrated preprocessor.
4299 When used in combination with the @option{-x} command line option,
4300 @option{-save-temps} is sensible enough to avoid over writing an
4301 input source file with the same extension as an intermediate file.
4302 The corresponding intermediate file may be obtained by renaming the
4303 source file before using @option{-save-temps}.
4307 Report the CPU time taken by each subprocess in the compilation
4308 sequence. For C source files, this is the compiler proper and assembler
4309 (plus the linker if linking is done). The output looks like this:
4316 The first number on each line is the ``user time'', that is time spent
4317 executing the program itself. The second number is ``system time'',
4318 time spent executing operating system routines on behalf of the program.
4319 Both numbers are in seconds.
4321 @item -fvar-tracking
4322 @opindex fvar-tracking
4323 Run variable tracking pass. It computes where variables are stored at each
4324 position in code. Better debugging information is then generated
4325 (if the debugging information format supports this information).
4327 It is enabled by default when compiling with optimization (@option{-Os},
4328 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4329 the debug info format supports it.
4331 @item -print-file-name=@var{library}
4332 @opindex print-file-name
4333 Print the full absolute name of the library file @var{library} that
4334 would be used when linking---and don't do anything else. With this
4335 option, GCC does not compile or link anything; it just prints the
4338 @item -print-multi-directory
4339 @opindex print-multi-directory
4340 Print the directory name corresponding to the multilib selected by any
4341 other switches present in the command line. This directory is supposed
4342 to exist in @env{GCC_EXEC_PREFIX}.
4344 @item -print-multi-lib
4345 @opindex print-multi-lib
4346 Print the mapping from multilib directory names to compiler switches
4347 that enable them. The directory name is separated from the switches by
4348 @samp{;}, and each switch starts with an @samp{@@} instead of the
4349 @samp{-}, without spaces between multiple switches. This is supposed to
4350 ease shell-processing.
4352 @item -print-prog-name=@var{program}
4353 @opindex print-prog-name
4354 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4356 @item -print-libgcc-file-name
4357 @opindex print-libgcc-file-name
4358 Same as @option{-print-file-name=libgcc.a}.
4360 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4361 but you do want to link with @file{libgcc.a}. You can do
4364 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4367 @item -print-search-dirs
4368 @opindex print-search-dirs
4369 Print the name of the configured installation directory and a list of
4370 program and library directories @command{gcc} will search---and don't do anything else.
4372 This is useful when @command{gcc} prints the error message
4373 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4374 To resolve this you either need to put @file{cpp0} and the other compiler
4375 components where @command{gcc} expects to find them, or you can set the environment
4376 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4377 Don't forget the trailing @samp{/}.
4378 @xref{Environment Variables}.
4381 @opindex dumpmachine
4382 Print the compiler's target machine (for example,
4383 @samp{i686-pc-linux-gnu})---and don't do anything else.
4386 @opindex dumpversion
4387 Print the compiler version (for example, @samp{3.0})---and don't do
4392 Print the compiler's built-in specs---and don't do anything else. (This
4393 is used when GCC itself is being built.) @xref{Spec Files}.
4395 @item -feliminate-unused-debug-types
4396 @opindex feliminate-unused-debug-types
4397 Normally, when producing DWARF2 output, GCC will emit debugging
4398 information for all types declared in a compilation
4399 unit, regardless of whether or not they are actually used
4400 in that compilation unit. Sometimes this is useful, such as
4401 if, in the debugger, you want to cast a value to a type that is
4402 not actually used in your program (but is declared). More often,
4403 however, this results in a significant amount of wasted space.
4404 With this option, GCC will avoid producing debug symbol output
4405 for types that are nowhere used in the source file being compiled.
4408 @node Optimize Options
4409 @section Options That Control Optimization
4410 @cindex optimize options
4411 @cindex options, optimization
4413 These options control various sorts of optimizations.
4415 Without any optimization option, the compiler's goal is to reduce the
4416 cost of compilation and to make debugging produce the expected
4417 results. Statements are independent: if you stop the program with a
4418 breakpoint between statements, you can then assign a new value to any
4419 variable or change the program counter to any other statement in the
4420 function and get exactly the results you would expect from the source
4423 Turning on optimization flags makes the compiler attempt to improve
4424 the performance and/or code size at the expense of compilation time
4425 and possibly the ability to debug the program.
4427 The compiler performs optimization based on the knowledge it has of
4428 the program. Optimization levels @option{-O} and above, in
4429 particular, enable @emph{unit-at-a-time} mode, which allows the
4430 compiler to consider information gained from later functions in
4431 the file when compiling a function. Compiling multiple files at
4432 once to a single output file in @emph{unit-at-a-time} mode allows
4433 the compiler to use information gained from all of the files when
4434 compiling each of them.
4436 Not all optimizations are controlled directly by a flag. Only
4437 optimizations that have a flag are listed.
4444 Optimize. Optimizing compilation takes somewhat more time, and a lot
4445 more memory for a large function.
4447 With @option{-O}, the compiler tries to reduce code size and execution
4448 time, without performing any optimizations that take a great deal of
4451 @option{-O} turns on the following optimization flags:
4452 @gccoptlist{-fdefer-pop @gol
4453 -fdelayed-branch @gol
4454 -fguess-branch-probability @gol
4455 -fcprop-registers @gol
4456 -fif-conversion @gol
4457 -fif-conversion2 @gol
4460 -ftree-dominator-opts @gol
4465 -ftree-copyrename @gol
4468 -funit-at-a-time @gol
4471 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4472 where doing so does not interfere with debugging.
4476 Optimize even more. GCC performs nearly all supported optimizations
4477 that do not involve a space-speed tradeoff. The compiler does not
4478 perform loop unrolling or function inlining when you specify @option{-O2}.
4479 As compared to @option{-O}, this option increases both compilation time
4480 and the performance of the generated code.
4482 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4483 also turns on the following optimization flags:
4484 @gccoptlist{-fthread-jumps @gol
4486 -foptimize-sibling-calls @gol
4487 -fcse-follow-jumps -fcse-skip-blocks @gol
4488 -fgcse -fgcse-lm @gol
4489 -fexpensive-optimizations @gol
4490 -frerun-cse-after-loop @gol
4493 -fschedule-insns -fschedule-insns2 @gol
4494 -fsched-interblock -fsched-spec @gol
4496 -fstrict-aliasing @gol
4497 -fdelete-null-pointer-checks @gol
4498 -freorder-blocks -freorder-functions @gol
4499 -falign-functions -falign-jumps @gol
4500 -falign-loops -falign-labels @gol
4504 Please note the warning under @option{-fgcse} about
4505 invoking @option{-O2} on programs that use computed gotos.
4509 Optimize yet more. @option{-O3} turns on all optimizations specified by
4510 @option{-O2} and also turns on the @option{-finline-functions},
4511 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4515 Do not optimize. This is the default.
4519 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4520 do not typically increase code size. It also performs further
4521 optimizations designed to reduce code size.
4523 @option{-Os} disables the following optimization flags:
4524 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4525 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4526 -fprefetch-loop-arrays -ftree-vect-loop-version}
4528 If you use multiple @option{-O} options, with or without level numbers,
4529 the last such option is the one that is effective.
4532 Options of the form @option{-f@var{flag}} specify machine-independent
4533 flags. Most flags have both positive and negative forms; the negative
4534 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4535 below, only one of the forms is listed---the one you typically will
4536 use. You can figure out the other form by either removing @samp{no-}
4539 The following options control specific optimizations. They are either
4540 activated by @option{-O} options or are related to ones that are. You
4541 can use the following flags in the rare cases when ``fine-tuning'' of
4542 optimizations to be performed is desired.
4545 @item -fno-default-inline
4546 @opindex fno-default-inline
4547 Do not make member functions inline by default merely because they are
4548 defined inside the class scope (C++ only). Otherwise, when you specify
4549 @w{@option{-O}}, member functions defined inside class scope are compiled
4550 inline by default; i.e., you don't need to add @samp{inline} in front of
4551 the member function name.
4553 @item -fno-defer-pop
4554 @opindex fno-defer-pop
4555 Always pop the arguments to each function call as soon as that function
4556 returns. For machines which must pop arguments after a function call,
4557 the compiler normally lets arguments accumulate on the stack for several
4558 function calls and pops them all at once.
4560 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4564 Force memory operands to be copied into registers before doing
4565 arithmetic on them. This produces better code by making all memory
4566 references potential common subexpressions. When they are not common
4567 subexpressions, instruction combination should eliminate the separate
4568 register-load. This option is now a nop and will be removed in 4.2.
4571 @opindex fforce-addr
4572 Force memory address constants to be copied into registers before
4573 doing arithmetic on them.
4575 @item -fomit-frame-pointer
4576 @opindex fomit-frame-pointer
4577 Don't keep the frame pointer in a register for functions that
4578 don't need one. This avoids the instructions to save, set up and
4579 restore frame pointers; it also makes an extra register available
4580 in many functions. @strong{It also makes debugging impossible on
4583 On some machines, such as the VAX, this flag has no effect, because
4584 the standard calling sequence automatically handles the frame pointer
4585 and nothing is saved by pretending it doesn't exist. The
4586 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4587 whether a target machine supports this flag. @xref{Registers,,Register
4588 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4590 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4592 @item -foptimize-sibling-calls
4593 @opindex foptimize-sibling-calls
4594 Optimize sibling and tail recursive calls.
4596 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4600 Don't pay attention to the @code{inline} keyword. Normally this option
4601 is used to keep the compiler from expanding any functions inline.
4602 Note that if you are not optimizing, no functions can be expanded inline.
4604 @item -finline-functions
4605 @opindex finline-functions
4606 Integrate all simple functions into their callers. The compiler
4607 heuristically decides which functions are simple enough to be worth
4608 integrating in this way.
4610 If all calls to a given function are integrated, and the function is
4611 declared @code{static}, then the function is normally not output as
4612 assembler code in its own right.
4614 Enabled at level @option{-O3}.
4616 @item -finline-functions-called-once
4617 @opindex finline-functions-called-once
4618 Consider all @code{static} functions called once for inlining into their
4619 caller even if they are not marked @code{inline}. If a call to a given
4620 function is integrated, then the function is not output as assembler code
4623 Enabled if @option{-funit-at-a-time} is enabled.
4625 @item -fearly-inlining
4626 @opindex fearly-inlining
4627 Inline functions marked by @code{always_inline} and functions whose body seems
4628 smaller than the function call overhead early before doing
4629 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4630 makes profiling significantly cheaper and usually inlining faster on programs
4631 having large chains of nested wrapper functions.
4635 @item -finline-limit=@var{n}
4636 @opindex finline-limit
4637 By default, GCC limits the size of functions that can be inlined. This flag
4638 allows the control of this limit for functions that are explicitly marked as
4639 inline (i.e., marked with the inline keyword or defined within the class
4640 definition in c++). @var{n} is the size of functions that can be inlined in
4641 number of pseudo instructions (not counting parameter handling). The default
4642 value of @var{n} is 600.
4643 Increasing this value can result in more inlined code at
4644 the cost of compilation time and memory consumption. Decreasing usually makes
4645 the compilation faster and less code will be inlined (which presumably
4646 means slower programs). This option is particularly useful for programs that
4647 use inlining heavily such as those based on recursive templates with C++.
4649 Inlining is actually controlled by a number of parameters, which may be
4650 specified individually by using @option{--param @var{name}=@var{value}}.
4651 The @option{-finline-limit=@var{n}} option sets some of these parameters
4655 @item max-inline-insns-single
4656 is set to @var{n}/2.
4657 @item max-inline-insns-auto
4658 is set to @var{n}/2.
4659 @item min-inline-insns
4660 is set to 130 or @var{n}/4, whichever is smaller.
4661 @item max-inline-insns-rtl
4665 See below for a documentation of the individual
4666 parameters controlling inlining.
4668 @emph{Note:} pseudo instruction represents, in this particular context, an
4669 abstract measurement of function's size. In no way does it represent a count
4670 of assembly instructions and as such its exact meaning might change from one
4671 release to an another.
4673 @item -fkeep-inline-functions
4674 @opindex fkeep-inline-functions
4675 In C, emit @code{static} functions that are declared @code{inline}
4676 into the object file, even if the function has been inlined into all
4677 of its callers. This switch does not affect functions using the
4678 @code{extern inline} extension in GNU C@. In C++, emit any and all
4679 inline functions into the object file.
4681 @item -fkeep-static-consts
4682 @opindex fkeep-static-consts
4683 Emit variables declared @code{static const} when optimization isn't turned
4684 on, even if the variables aren't referenced.
4686 GCC enables this option by default. If you want to force the compiler to
4687 check if the variable was referenced, regardless of whether or not
4688 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4690 @item -fmerge-constants
4691 Attempt to merge identical constants (string constants and floating point
4692 constants) across compilation units.
4694 This option is the default for optimized compilation if the assembler and
4695 linker support it. Use @option{-fno-merge-constants} to inhibit this
4698 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4700 @item -fmerge-all-constants
4701 Attempt to merge identical constants and identical variables.
4703 This option implies @option{-fmerge-constants}. In addition to
4704 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4705 arrays or initialized constant variables with integral or floating point
4706 types. Languages like C or C++ require each non-automatic variable to
4707 have distinct location, so using this option will result in non-conforming
4710 @item -fmodulo-sched
4711 @opindex fmodulo-sched
4712 Perform swing modulo scheduling immediately before the first scheduling
4713 pass. This pass looks at innermost loops and reorders their
4714 instructions by overlapping different iterations.
4716 @item -fno-branch-count-reg
4717 @opindex fno-branch-count-reg
4718 Do not use ``decrement and branch'' instructions on a count register,
4719 but instead generate a sequence of instructions that decrement a
4720 register, compare it against zero, then branch based upon the result.
4721 This option is only meaningful on architectures that support such
4722 instructions, which include x86, PowerPC, IA-64 and S/390.
4724 The default is @option{-fbranch-count-reg}.
4726 @item -fno-function-cse
4727 @opindex fno-function-cse
4728 Do not put function addresses in registers; make each instruction that
4729 calls a constant function contain the function's address explicitly.
4731 This option results in less efficient code, but some strange hacks
4732 that alter the assembler output may be confused by the optimizations
4733 performed when this option is not used.
4735 The default is @option{-ffunction-cse}
4737 @item -fno-zero-initialized-in-bss
4738 @opindex fno-zero-initialized-in-bss
4739 If the target supports a BSS section, GCC by default puts variables that
4740 are initialized to zero into BSS@. This can save space in the resulting
4743 This option turns off this behavior because some programs explicitly
4744 rely on variables going to the data section. E.g., so that the
4745 resulting executable can find the beginning of that section and/or make
4746 assumptions based on that.
4748 The default is @option{-fzero-initialized-in-bss}.
4750 @item -fbounds-check
4751 @opindex fbounds-check
4752 For front-ends that support it, generate additional code to check that
4753 indices used to access arrays are within the declared range. This is
4754 currently only supported by the Java and Fortran front-ends, where
4755 this option defaults to true and false respectively.
4757 @item -fmudflap -fmudflapth -fmudflapir
4761 @cindex bounds checking
4763 For front-ends that support it (C and C++), instrument all risky
4764 pointer/array dereferencing operations, some standard library
4765 string/heap functions, and some other associated constructs with
4766 range/validity tests. Modules so instrumented should be immune to
4767 buffer overflows, invalid heap use, and some other classes of C/C++
4768 programming errors. The instrumentation relies on a separate runtime
4769 library (@file{libmudflap}), which will be linked into a program if
4770 @option{-fmudflap} is given at link time. Run-time behavior of the
4771 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4772 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4775 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4776 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4777 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4778 instrumentation should ignore pointer reads. This produces less
4779 instrumentation (and therefore faster execution) and still provides
4780 some protection against outright memory corrupting writes, but allows
4781 erroneously read data to propagate within a program.
4783 @item -fthread-jumps
4784 @opindex fthread-jumps
4785 Perform optimizations where we check to see if a jump branches to a
4786 location where another comparison subsumed by the first is found. If
4787 so, the first branch is redirected to either the destination of the
4788 second branch or a point immediately following it, depending on whether
4789 the condition is known to be true or false.
4791 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4793 @item -fcse-follow-jumps
4794 @opindex fcse-follow-jumps
4795 In common subexpression elimination, scan through jump instructions
4796 when the target of the jump is not reached by any other path. For
4797 example, when CSE encounters an @code{if} statement with an
4798 @code{else} clause, CSE will follow the jump when the condition
4801 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4803 @item -fcse-skip-blocks
4804 @opindex fcse-skip-blocks
4805 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4806 follow jumps which conditionally skip over blocks. When CSE
4807 encounters a simple @code{if} statement with no else clause,
4808 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4809 body of the @code{if}.
4811 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4813 @item -frerun-cse-after-loop
4814 @opindex frerun-cse-after-loop
4815 Re-run common subexpression elimination after loop optimizations has been
4818 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4822 Perform a global common subexpression elimination pass.
4823 This pass also performs global constant and copy propagation.
4825 @emph{Note:} When compiling a program using computed gotos, a GCC
4826 extension, you may get better runtime performance if you disable
4827 the global common subexpression elimination pass by adding
4828 @option{-fno-gcse} to the command line.
4830 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4834 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4835 attempt to move loads which are only killed by stores into themselves. This
4836 allows a loop containing a load/store sequence to be changed to a load outside
4837 the loop, and a copy/store within the loop.
4839 Enabled by default when gcse is enabled.
4843 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4844 global common subexpression elimination. This pass will attempt to move
4845 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4846 loops containing a load/store sequence can be changed to a load before
4847 the loop and a store after the loop.
4849 Not enabled at any optimization level.
4853 When @option{-fgcse-las} is enabled, the global common subexpression
4854 elimination pass eliminates redundant loads that come after stores to the
4855 same memory location (both partial and full redundancies).
4857 Not enabled at any optimization level.
4859 @item -fgcse-after-reload
4860 @opindex fgcse-after-reload
4861 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4862 pass is performed after reload. The purpose of this pass is to cleanup
4865 @item -funsafe-loop-optimizations
4866 @opindex funsafe-loop-optimizations
4867 If given, the loop optimizer will assume that loop indices do not
4868 overflow, and that the loops with nontrivial exit condition are not
4869 infinite. This enables a wider range of loop optimizations even if
4870 the loop optimizer itself cannot prove that these assumptions are valid.
4871 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4872 if it finds this kind of loop.
4874 @item -fcrossjumping
4875 @opindex crossjumping
4876 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4877 resulting code may or may not perform better than without cross-jumping.
4879 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4881 @item -fif-conversion
4882 @opindex if-conversion
4883 Attempt to transform conditional jumps into branch-less equivalents. This
4884 include use of conditional moves, min, max, set flags and abs instructions, and
4885 some tricks doable by standard arithmetics. The use of conditional execution
4886 on chips where it is available is controlled by @code{if-conversion2}.
4888 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4890 @item -fif-conversion2
4891 @opindex if-conversion2
4892 Use conditional execution (where available) to transform conditional jumps into
4893 branch-less equivalents.
4895 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4897 @item -fdelete-null-pointer-checks
4898 @opindex fdelete-null-pointer-checks
4899 Use global dataflow analysis to identify and eliminate useless checks
4900 for null pointers. The compiler assumes that dereferencing a null
4901 pointer would have halted the program. If a pointer is checked after
4902 it has already been dereferenced, it cannot be null.
4904 In some environments, this assumption is not true, and programs can
4905 safely dereference null pointers. Use
4906 @option{-fno-delete-null-pointer-checks} to disable this optimization
4907 for programs which depend on that behavior.
4909 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4911 @item -fexpensive-optimizations
4912 @opindex fexpensive-optimizations
4913 Perform a number of minor optimizations that are relatively expensive.
4915 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4917 @item -foptimize-register-move
4919 @opindex foptimize-register-move
4921 Attempt to reassign register numbers in move instructions and as
4922 operands of other simple instructions in order to maximize the amount of
4923 register tying. This is especially helpful on machines with two-operand
4926 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4929 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4931 @item -fdelayed-branch
4932 @opindex fdelayed-branch
4933 If supported for the target machine, attempt to reorder instructions
4934 to exploit instruction slots available after delayed branch
4937 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4939 @item -fschedule-insns
4940 @opindex fschedule-insns
4941 If supported for the target machine, attempt to reorder instructions to
4942 eliminate execution stalls due to required data being unavailable. This
4943 helps machines that have slow floating point or memory load instructions
4944 by allowing other instructions to be issued until the result of the load
4945 or floating point instruction is required.
4947 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4949 @item -fschedule-insns2
4950 @opindex fschedule-insns2
4951 Similar to @option{-fschedule-insns}, but requests an additional pass of
4952 instruction scheduling after register allocation has been done. This is
4953 especially useful on machines with a relatively small number of
4954 registers and where memory load instructions take more than one cycle.
4956 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4958 @item -fno-sched-interblock
4959 @opindex fno-sched-interblock
4960 Don't schedule instructions across basic blocks. This is normally
4961 enabled by default when scheduling before register allocation, i.e.@:
4962 with @option{-fschedule-insns} or at @option{-O2} or higher.
4964 @item -fno-sched-spec
4965 @opindex fno-sched-spec
4966 Don't allow speculative motion of non-load instructions. This is normally
4967 enabled by default when scheduling before register allocation, i.e.@:
4968 with @option{-fschedule-insns} or at @option{-O2} or higher.
4970 @item -fsched-spec-load
4971 @opindex fsched-spec-load
4972 Allow speculative motion of some load instructions. This only makes
4973 sense when scheduling before register allocation, i.e.@: with
4974 @option{-fschedule-insns} or at @option{-O2} or higher.
4976 @item -fsched-spec-load-dangerous
4977 @opindex fsched-spec-load-dangerous
4978 Allow speculative motion of more load instructions. This only makes
4979 sense when scheduling before register allocation, i.e.@: with
4980 @option{-fschedule-insns} or at @option{-O2} or higher.
4982 @item -fsched-stalled-insns=@var{n}
4983 @opindex fsched-stalled-insns
4984 Define how many insns (if any) can be moved prematurely from the queue
4985 of stalled insns into the ready list, during the second scheduling pass.
4987 @item -fsched-stalled-insns-dep=@var{n}
4988 @opindex fsched-stalled-insns-dep
4989 Define how many insn groups (cycles) will be examined for a dependency
4990 on a stalled insn that is candidate for premature removal from the queue
4991 of stalled insns. Has an effect only during the second scheduling pass,
4992 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4994 @item -fsched2-use-superblocks
4995 @opindex fsched2-use-superblocks
4996 When scheduling after register allocation, do use superblock scheduling
4997 algorithm. Superblock scheduling allows motion across basic block boundaries
4998 resulting on faster schedules. This option is experimental, as not all machine
4999 descriptions used by GCC model the CPU closely enough to avoid unreliable
5000 results from the algorithm.
5002 This only makes sense when scheduling after register allocation, i.e.@: with
5003 @option{-fschedule-insns2} or at @option{-O2} or higher.
5005 @item -fsched2-use-traces
5006 @opindex fsched2-use-traces
5007 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5008 allocation and additionally perform code duplication in order to increase the
5009 size of superblocks using tracer pass. See @option{-ftracer} for details on
5012 This mode should produce faster but significantly longer programs. Also
5013 without @option{-fbranch-probabilities} the traces constructed may not
5014 match the reality and hurt the performance. This only makes
5015 sense when scheduling after register allocation, i.e.@: with
5016 @option{-fschedule-insns2} or at @option{-O2} or higher.
5020 Eliminates redundant extension instructions and move the non redundant
5021 ones to optimal placement using LCM.
5023 @item -freschedule-modulo-scheduled-loops
5024 @opindex fscheduling-in-modulo-scheduled-loops
5025 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5026 we may want to prevent the later scheduling passes from changing its schedule, we use this
5027 option to control that.
5029 @item -fcaller-saves
5030 @opindex fcaller-saves
5031 Enable values to be allocated in registers that will be clobbered by
5032 function calls, by emitting extra instructions to save and restore the
5033 registers around such calls. Such allocation is done only when it
5034 seems to result in better code than would otherwise be produced.
5036 This option is always enabled by default on certain machines, usually
5037 those which have no call-preserved registers to use instead.
5039 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5042 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5043 enabled by default at @option{-O2} and @option{-O3}.
5046 Perform Full Redundancy Elimination (FRE) on trees. The difference
5047 between FRE and PRE is that FRE only considers expressions
5048 that are computed on all paths leading to the redundant computation.
5049 This analysis faster than PRE, though it exposes fewer redundancies.
5050 This flag is enabled by default at @option{-O} and higher.
5052 @item -ftree-copy-prop
5053 Perform copy propagation on trees. This pass eliminates unnecessary
5054 copy operations. This flag is enabled by default at @option{-O} and
5057 @item -ftree-store-copy-prop
5058 Perform copy propagation of memory loads and stores. This pass
5059 eliminates unnecessary copy operations in memory references
5060 (structures, global variables, arrays, etc). This flag is enabled by
5061 default at @option{-O2} and higher.
5064 Perform structural alias analysis on trees. This flag
5065 is enabled by default at @option{-O} and higher.
5068 Perform interprocedural pointer analysis.
5071 Perform forward store motion on trees. This flag is
5072 enabled by default at @option{-O} and higher.
5075 Perform sparse conditional constant propagation (CCP) on trees. This
5076 pass only operates on local scalar variables and is enabled by default
5077 at @option{-O} and higher.
5079 @item -ftree-store-ccp
5080 Perform sparse conditional constant propagation (CCP) on trees. This
5081 pass operates on both local scalar variables and memory stores and
5082 loads (global variables, structures, arrays, etc). This flag is
5083 enabled by default at @option{-O2} and higher.
5086 Perform dead code elimination (DCE) on trees. This flag is enabled by
5087 default at @option{-O} and higher.
5089 @item -ftree-dominator-opts
5090 Perform a variety of simple scalar cleanups (constant/copy
5091 propagation, redundancy elimination, range propagation and expression
5092 simplification) based on a dominator tree traversal. This also
5093 performs jump threading (to reduce jumps to jumps). This flag is
5094 enabled by default at @option{-O} and higher.
5097 Perform loop header copying on trees. This is beneficial since it increases
5098 effectiveness of code motion optimizations. It also saves one jump. This flag
5099 is enabled by default at @option{-O} and higher. It is not enabled
5100 for @option{-Os}, since it usually increases code size.
5102 @item -ftree-loop-optimize
5103 Perform loop optimizations on trees. This flag is enabled by default
5104 at @option{-O} and higher.
5106 @item -ftree-loop-linear
5107 Perform linear loop transformations on tree. This flag can improve cache
5108 performance and allow further loop optimizations to take place.
5110 @item -ftree-loop-im
5111 Perform loop invariant motion on trees. This pass moves only invariants that
5112 would be hard to handle at RTL level (function calls, operations that expand to
5113 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5114 operands of conditions that are invariant out of the loop, so that we can use
5115 just trivial invariantness analysis in loop unswitching. The pass also includes
5118 @item -ftree-loop-ivcanon
5119 Create a canonical counter for number of iterations in the loop for that
5120 determining number of iterations requires complicated analysis. Later
5121 optimizations then may determine the number easily. Useful especially
5122 in connection with unrolling.
5125 Perform induction variable optimizations (strength reduction, induction
5126 variable merging and induction variable elimination) on trees.
5129 Perform scalar replacement of aggregates. This pass replaces structure
5130 references with scalars to prevent committing structures to memory too
5131 early. This flag is enabled by default at @option{-O} and higher.
5133 @item -ftree-copyrename
5134 Perform copy renaming on trees. This pass attempts to rename compiler
5135 temporaries to other variables at copy locations, usually resulting in
5136 variable names which more closely resemble the original variables. This flag
5137 is enabled by default at @option{-O} and higher.
5140 Perform temporary expression replacement during the SSA->normal phase. Single
5141 use/single def temporaries are replaced at their use location with their
5142 defining expression. This results in non-GIMPLE code, but gives the expanders
5143 much more complex trees to work on resulting in better RTL generation. This is
5144 enabled by default at @option{-O} and higher.
5147 Perform live range splitting during the SSA->normal phase. Distinct live
5148 ranges of a variable are split into unique variables, allowing for better
5149 optimization later. This is enabled by default at @option{-O} and higher.
5151 @item -ftree-vectorize
5152 Perform loop vectorization on trees.
5154 @item -ftree-vect-loop-version
5155 @opindex ftree-vect-loop-version
5156 Perform loop versioning when doing loop vectorization on trees. When a loop
5157 appears to be vectorizable except that data alignment or data dependence cannot
5158 be determined at compile time then vectorized and non-vectorized versions of
5159 the loop are generated along with runtime checks for alignment or dependence
5160 to control which version is executed. This option is enabled by default
5161 except at level @option{-Os} where it is disabled.
5164 Perform Value Range Propagation on trees. This is similar to the
5165 constant propagation pass, but instead of values, ranges of values are
5166 propagated. This allows the optimizers to remove unnecessary range
5167 checks like array bound checks and null pointer checks. This is
5168 enabled by default at @option{-O2} and higher. Null pointer check
5169 elimination is only done if @option{-fdelete-null-pointer-checks} is
5174 Perform tail duplication to enlarge superblock size. This transformation
5175 simplifies the control flow of the function allowing other optimizations to do
5178 @item -funroll-loops
5179 @opindex funroll-loops
5180 Unroll loops whose number of iterations can be determined at compile
5181 time or upon entry to the loop. @option{-funroll-loops} implies
5182 @option{-frerun-cse-after-loop}. This option makes code larger,
5183 and may or may not make it run faster.
5185 @item -funroll-all-loops
5186 @opindex funroll-all-loops
5187 Unroll all loops, even if their number of iterations is uncertain when
5188 the loop is entered. This usually makes programs run more slowly.
5189 @option{-funroll-all-loops} implies the same options as
5190 @option{-funroll-loops},
5192 @item -fsplit-ivs-in-unroller
5193 @opindex -fsplit-ivs-in-unroller
5194 Enables expressing of values of induction variables in later iterations
5195 of the unrolled loop using the value in the first iteration. This breaks
5196 long dependency chains, thus improving efficiency of the scheduling passes.
5198 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5199 same effect. However in cases the loop body is more complicated than
5200 a single basic block, this is not reliable. It also does not work at all
5201 on some of the architectures due to restrictions in the CSE pass.
5203 This optimization is enabled by default.
5205 @item -fvariable-expansion-in-unroller
5206 @opindex -fvariable-expansion-in-unroller
5207 With this option, the compiler will create multiple copies of some
5208 local variables when unrolling a loop which can result in superior code.
5210 @item -fprefetch-loop-arrays
5211 @opindex fprefetch-loop-arrays
5212 If supported by the target machine, generate instructions to prefetch
5213 memory to improve the performance of loops that access large arrays.
5215 This option may generate better or worse code; results are highly
5216 dependent on the structure of loops within the source code.
5218 Disabled at level @option{-Os}.
5221 @itemx -fno-peephole2
5222 @opindex fno-peephole
5223 @opindex fno-peephole2
5224 Disable any machine-specific peephole optimizations. The difference
5225 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5226 are implemented in the compiler; some targets use one, some use the
5227 other, a few use both.
5229 @option{-fpeephole} is enabled by default.
5230 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5232 @item -fno-guess-branch-probability
5233 @opindex fno-guess-branch-probability
5234 Do not guess branch probabilities using heuristics.
5236 GCC will use heuristics to guess branch probabilities if they are
5237 not provided by profiling feedback (@option{-fprofile-arcs}). These
5238 heuristics are based on the control flow graph. If some branch probabilities
5239 are specified by @samp{__builtin_expect}, then the heuristics will be
5240 used to guess branch probabilities for the rest of the control flow graph,
5241 taking the @samp{__builtin_expect} info into account. The interactions
5242 between the heuristics and @samp{__builtin_expect} can be complex, and in
5243 some cases, it may be useful to disable the heuristics so that the effects
5244 of @samp{__builtin_expect} are easier to understand.
5246 The default is @option{-fguess-branch-probability} at levels
5247 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5249 @item -freorder-blocks
5250 @opindex freorder-blocks
5251 Reorder basic blocks in the compiled function in order to reduce number of
5252 taken branches and improve code locality.
5254 Enabled at levels @option{-O2}, @option{-O3}.
5256 @item -freorder-blocks-and-partition
5257 @opindex freorder-blocks-and-partition
5258 In addition to reordering basic blocks in the compiled function, in order
5259 to reduce number of taken branches, partitions hot and cold basic blocks
5260 into separate sections of the assembly and .o files, to improve
5261 paging and cache locality performance.
5263 This optimization is automatically turned off in the presence of
5264 exception handling, for linkonce sections, for functions with a user-defined
5265 section attribute and on any architecture that does not support named
5268 @item -freorder-functions
5269 @opindex freorder-functions
5270 Reorder functions in the object file in order to
5271 improve code locality. This is implemented by using special
5272 subsections @code{.text.hot} for most frequently executed functions and
5273 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5274 the linker so object file format must support named sections and linker must
5275 place them in a reasonable way.
5277 Also profile feedback must be available in to make this option effective. See
5278 @option{-fprofile-arcs} for details.
5280 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5282 @item -fstrict-aliasing
5283 @opindex fstrict-aliasing
5284 Allows the compiler to assume the strictest aliasing rules applicable to
5285 the language being compiled. For C (and C++), this activates
5286 optimizations based on the type of expressions. In particular, an
5287 object of one type is assumed never to reside at the same address as an
5288 object of a different type, unless the types are almost the same. For
5289 example, an @code{unsigned int} can alias an @code{int}, but not a
5290 @code{void*} or a @code{double}. A character type may alias any other
5293 Pay special attention to code like this:
5306 The practice of reading from a different union member than the one most
5307 recently written to (called ``type-punning'') is common. Even with
5308 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5309 is accessed through the union type. So, the code above will work as
5310 expected. However, this code might not:
5321 Every language that wishes to perform language-specific alias analysis
5322 should define a function that computes, given an @code{tree}
5323 node, an alias set for the node. Nodes in different alias sets are not
5324 allowed to alias. For an example, see the C front-end function
5325 @code{c_get_alias_set}.
5327 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5329 @item -falign-functions
5330 @itemx -falign-functions=@var{n}
5331 @opindex falign-functions
5332 Align the start of functions to the next power-of-two greater than
5333 @var{n}, skipping up to @var{n} bytes. For instance,
5334 @option{-falign-functions=32} aligns functions to the next 32-byte
5335 boundary, but @option{-falign-functions=24} would align to the next
5336 32-byte boundary only if this can be done by skipping 23 bytes or less.
5338 @option{-fno-align-functions} and @option{-falign-functions=1} are
5339 equivalent and mean that functions will not be aligned.
5341 Some assemblers only support this flag when @var{n} is a power of two;
5342 in that case, it is rounded up.
5344 If @var{n} is not specified or is zero, use a machine-dependent default.
5346 Enabled at levels @option{-O2}, @option{-O3}.
5348 @item -falign-labels
5349 @itemx -falign-labels=@var{n}
5350 @opindex falign-labels
5351 Align all branch targets to a power-of-two boundary, skipping up to
5352 @var{n} bytes like @option{-falign-functions}. This option can easily
5353 make code slower, because it must insert dummy operations for when the
5354 branch target is reached in the usual flow of the code.
5356 @option{-fno-align-labels} and @option{-falign-labels=1} are
5357 equivalent and mean that labels will not be aligned.
5359 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5360 are greater than this value, then their values are used instead.
5362 If @var{n} is not specified or is zero, use a machine-dependent default
5363 which is very likely to be @samp{1}, meaning no alignment.
5365 Enabled at levels @option{-O2}, @option{-O3}.
5368 @itemx -falign-loops=@var{n}
5369 @opindex falign-loops
5370 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5371 like @option{-falign-functions}. The hope is that the loop will be
5372 executed many times, which will make up for any execution of the dummy
5375 @option{-fno-align-loops} and @option{-falign-loops=1} are
5376 equivalent and mean that loops will not be aligned.
5378 If @var{n} is not specified or is zero, use a machine-dependent default.
5380 Enabled at levels @option{-O2}, @option{-O3}.
5383 @itemx -falign-jumps=@var{n}
5384 @opindex falign-jumps
5385 Align branch targets to a power-of-two boundary, for branch targets
5386 where the targets can only be reached by jumping, skipping up to @var{n}
5387 bytes like @option{-falign-functions}. In this case, no dummy operations
5390 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5391 equivalent and mean that loops will not be aligned.
5393 If @var{n} is not specified or is zero, use a machine-dependent default.
5395 Enabled at levels @option{-O2}, @option{-O3}.
5397 @item -funit-at-a-time
5398 @opindex funit-at-a-time
5399 Parse the whole compilation unit before starting to produce code.
5400 This allows some extra optimizations to take place but consumes
5401 more memory (in general). There are some compatibility issues
5402 with @emph{unit-at-a-time} mode:
5405 enabling @emph{unit-at-a-time} mode may change the order
5406 in which functions, variables, and top-level @code{asm} statements
5407 are emitted, and will likely break code relying on some particular
5408 ordering. The majority of such top-level @code{asm} statements,
5409 though, can be replaced by @code{section} attributes. The
5410 @option{fno-toplevel-reorder} option may be used to keep the ordering
5411 used in the input file, at the cost of some optimizations.
5414 @emph{unit-at-a-time} mode removes unreferenced static variables
5415 and functions. This may result in undefined references
5416 when an @code{asm} statement refers directly to variables or functions
5417 that are otherwise unused. In that case either the variable/function
5418 shall be listed as an operand of the @code{asm} statement operand or,
5419 in the case of top-level @code{asm} statements the attribute @code{used}
5420 shall be used on the declaration.
5423 Static functions now can use non-standard passing conventions that
5424 may break @code{asm} statements calling functions directly. Again,
5425 attribute @code{used} will prevent this behavior.
5428 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5429 but this scheme may not be supported by future releases of GCC@.
5431 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5433 @item -fno-toplevel-reorder
5434 Do not reorder top-level functions, variables, and @code{asm}
5435 statements. Output them in the same order that they appear in the
5436 input file. When this option is used, unreferenced static variables
5437 will not be removed. This option is intended to support existing code
5438 which relies on a particular ordering. For new code, it is better to
5443 Constructs webs as commonly used for register allocation purposes and assign
5444 each web individual pseudo register. This allows the register allocation pass
5445 to operate on pseudos directly, but also strengthens several other optimization
5446 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5447 however, make debugging impossible, since variables will no longer stay in a
5450 Enabled by default with @option{-funroll-loops}.
5452 @item -fwhole-program
5453 @opindex fwhole-program
5454 Assume that the current compilation unit represents whole program being
5455 compiled. All public functions and variables with the exception of @code{main}
5456 and those merged by attribute @code{externally_visible} become static functions
5457 and in a affect gets more aggressively optimized by interprocedural optimizers.
5458 While this option is equivalent to proper use of @code{static} keyword for
5459 programs consisting of single file, in combination with option
5460 @option{--combine} this flag can be used to compile most of smaller scale C
5461 programs since the functions and variables become local for the whole combined
5462 compilation unit, not for the single source file itself.
5465 @item -fno-cprop-registers
5466 @opindex fno-cprop-registers
5467 After register allocation and post-register allocation instruction splitting,
5468 we perform a copy-propagation pass to try to reduce scheduling dependencies
5469 and occasionally eliminate the copy.
5471 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5473 @item -fprofile-generate
5474 @opindex fprofile-generate
5476 Enable options usually used for instrumenting application to produce
5477 profile useful for later recompilation with profile feedback based
5478 optimization. You must use @option{-fprofile-generate} both when
5479 compiling and when linking your program.
5481 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5484 @opindex fprofile-use
5485 Enable profile feedback directed optimizations, and optimizations
5486 generally profitable only with profile feedback available.
5488 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5489 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5493 The following options control compiler behavior regarding floating
5494 point arithmetic. These options trade off between speed and
5495 correctness. All must be specifically enabled.
5499 @opindex ffloat-store
5500 Do not store floating point variables in registers, and inhibit other
5501 options that might change whether a floating point value is taken from a
5504 @cindex floating point precision
5505 This option prevents undesirable excess precision on machines such as
5506 the 68000 where the floating registers (of the 68881) keep more
5507 precision than a @code{double} is supposed to have. Similarly for the
5508 x86 architecture. For most programs, the excess precision does only
5509 good, but a few programs rely on the precise definition of IEEE floating
5510 point. Use @option{-ffloat-store} for such programs, after modifying
5511 them to store all pertinent intermediate computations into variables.
5515 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5516 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5517 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5518 and @option{fcx-limited-range}.
5520 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5522 This option should never be turned on by any @option{-O} option since
5523 it can result in incorrect output for programs which depend on
5524 an exact implementation of IEEE or ISO rules/specifications for
5527 @item -fno-math-errno
5528 @opindex fno-math-errno
5529 Do not set ERRNO after calling math functions that are executed
5530 with a single instruction, e.g., sqrt. A program that relies on
5531 IEEE exceptions for math error handling may want to use this flag
5532 for speed while maintaining IEEE arithmetic compatibility.
5534 This option should never be turned on by any @option{-O} option since
5535 it can result in incorrect output for programs which depend on
5536 an exact implementation of IEEE or ISO rules/specifications for
5539 The default is @option{-fmath-errno}.
5541 On Darwin systems, the math library never sets @code{errno}. There is therefore
5542 no reason for the compiler to consider the possibility that it might,
5543 and @option{-fno-math-errno} is the default.
5545 @item -funsafe-math-optimizations
5546 @opindex funsafe-math-optimizations
5547 Allow optimizations for floating-point arithmetic that (a) assume
5548 that arguments and results are valid and (b) may violate IEEE or
5549 ANSI standards. When used at link-time, it may include libraries
5550 or startup files that change the default FPU control word or other
5551 similar optimizations.
5553 This option should never be turned on by any @option{-O} option since
5554 it can result in incorrect output for programs which depend on
5555 an exact implementation of IEEE or ISO rules/specifications for
5558 The default is @option{-fno-unsafe-math-optimizations}.
5560 @item -ffinite-math-only
5561 @opindex ffinite-math-only
5562 Allow optimizations for floating-point arithmetic that assume
5563 that arguments and results are not NaNs or +-Infs.
5565 This option should never be turned on by any @option{-O} option since
5566 it can result in incorrect output for programs which depend on
5567 an exact implementation of IEEE or ISO rules/specifications.
5569 The default is @option{-fno-finite-math-only}.
5571 @item -fno-trapping-math
5572 @opindex fno-trapping-math
5573 Compile code assuming that floating-point operations cannot generate
5574 user-visible traps. These traps include division by zero, overflow,
5575 underflow, inexact result and invalid operation. This option implies
5576 @option{-fno-signaling-nans}. Setting this option may allow faster
5577 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5579 This option should never be turned on by any @option{-O} option since
5580 it can result in incorrect output for programs which depend on
5581 an exact implementation of IEEE or ISO rules/specifications for
5584 The default is @option{-ftrapping-math}.
5586 @item -frounding-math
5587 @opindex frounding-math
5588 Disable transformations and optimizations that assume default floating
5589 point rounding behavior. This is round-to-zero for all floating point
5590 to integer conversions, and round-to-nearest for all other arithmetic
5591 truncations. This option should be specified for programs that change
5592 the FP rounding mode dynamically, or that may be executed with a
5593 non-default rounding mode. This option disables constant folding of
5594 floating point expressions at compile-time (which may be affected by
5595 rounding mode) and arithmetic transformations that are unsafe in the
5596 presence of sign-dependent rounding modes.
5598 The default is @option{-fno-rounding-math}.
5600 This option is experimental and does not currently guarantee to
5601 disable all GCC optimizations that are affected by rounding mode.
5602 Future versions of GCC may provide finer control of this setting
5603 using C99's @code{FENV_ACCESS} pragma. This command line option
5604 will be used to specify the default state for @code{FENV_ACCESS}.
5606 @item -frtl-abstract-sequences
5607 @opindex frtl-abstract-sequences
5608 It is a size optimization method. This option is to find identical
5609 sequences of code, which can be turned into pseudo-procedures and
5610 then replace all occurrences with calls to the newly created
5611 subroutine. It is kind of an opposite of @option{-finline-functions}.
5612 This optimization runs at RTL level.
5614 @item -fsignaling-nans
5615 @opindex fsignaling-nans
5616 Compile code assuming that IEEE signaling NaNs may generate user-visible
5617 traps during floating-point operations. Setting this option disables
5618 optimizations that may change the number of exceptions visible with
5619 signaling NaNs. This option implies @option{-ftrapping-math}.
5621 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5624 The default is @option{-fno-signaling-nans}.
5626 This option is experimental and does not currently guarantee to
5627 disable all GCC optimizations that affect signaling NaN behavior.
5629 @item -fsingle-precision-constant
5630 @opindex fsingle-precision-constant
5631 Treat floating point constant as single precision constant instead of
5632 implicitly converting it to double precision constant.
5634 @item -fcx-limited-range
5635 @itemx -fno-cx-limited-range
5636 @opindex fcx-limited-range
5637 @opindex fno-cx-limited-range
5638 When enabled, this option states that a range reduction step is not
5639 needed when performing complex division. The default is
5640 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5642 This option controls the default setting of the ISO C99
5643 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5648 The following options control optimizations that may improve
5649 performance, but are not enabled by any @option{-O} options. This
5650 section includes experimental options that may produce broken code.
5653 @item -fbranch-probabilities
5654 @opindex fbranch-probabilities
5655 After running a program compiled with @option{-fprofile-arcs}
5656 (@pxref{Debugging Options,, Options for Debugging Your Program or
5657 @command{gcc}}), you can compile it a second time using
5658 @option{-fbranch-probabilities}, to improve optimizations based on
5659 the number of times each branch was taken. When the program
5660 compiled with @option{-fprofile-arcs} exits it saves arc execution
5661 counts to a file called @file{@var{sourcename}.gcda} for each source
5662 file The information in this data file is very dependent on the
5663 structure of the generated code, so you must use the same source code
5664 and the same optimization options for both compilations.
5666 With @option{-fbranch-probabilities}, GCC puts a
5667 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5668 These can be used to improve optimization. Currently, they are only
5669 used in one place: in @file{reorg.c}, instead of guessing which path a
5670 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5671 exactly determine which path is taken more often.
5673 @item -fprofile-values
5674 @opindex fprofile-values
5675 If combined with @option{-fprofile-arcs}, it adds code so that some
5676 data about values of expressions in the program is gathered.
5678 With @option{-fbranch-probabilities}, it reads back the data gathered
5679 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5680 notes to instructions for their later usage in optimizations.
5682 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5686 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5687 a code to gather information about values of expressions.
5689 With @option{-fbranch-probabilities}, it reads back the data gathered
5690 and actually performs the optimizations based on them.
5691 Currently the optimizations include specialization of division operation
5692 using the knowledge about the value of the denominator.
5694 @item -frename-registers
5695 @opindex frename-registers
5696 Attempt to avoid false dependencies in scheduled code by making use
5697 of registers left over after register allocation. This optimization
5698 will most benefit processors with lots of registers. Depending on the
5699 debug information format adopted by the target, however, it can
5700 make debugging impossible, since variables will no longer stay in
5701 a ``home register''.
5703 Enabled by default with @option{-funroll-loops}.
5707 Perform tail duplication to enlarge superblock size. This transformation
5708 simplifies the control flow of the function allowing other optimizations to do
5711 Enabled with @option{-fprofile-use}.
5713 @item -funroll-loops
5714 @opindex funroll-loops
5715 Unroll loops whose number of iterations can be determined at compile time or
5716 upon entry to the loop. @option{-funroll-loops} implies
5717 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5718 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5719 small constant number of iterations). This option makes code larger, and may
5720 or may not make it run faster.
5722 Enabled with @option{-fprofile-use}.
5724 @item -funroll-all-loops
5725 @opindex funroll-all-loops
5726 Unroll all loops, even if their number of iterations is uncertain when
5727 the loop is entered. This usually makes programs run more slowly.
5728 @option{-funroll-all-loops} implies the same options as
5729 @option{-funroll-loops}.
5732 @opindex fpeel-loops
5733 Peels the loops for that there is enough information that they do not
5734 roll much (from profile feedback). It also turns on complete loop peeling
5735 (i.e.@: complete removal of loops with small constant number of iterations).
5737 Enabled with @option{-fprofile-use}.
5739 @item -fmove-loop-invariants
5740 @opindex fmove-loop-invariants
5741 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5742 at level @option{-O1}
5744 @item -funswitch-loops
5745 @opindex funswitch-loops
5746 Move branches with loop invariant conditions out of the loop, with duplicates
5747 of the loop on both branches (modified according to result of the condition).
5749 @item -ffunction-sections
5750 @itemx -fdata-sections
5751 @opindex ffunction-sections
5752 @opindex fdata-sections
5753 Place each function or data item into its own section in the output
5754 file if the target supports arbitrary sections. The name of the
5755 function or the name of the data item determines the section's name
5758 Use these options on systems where the linker can perform optimizations
5759 to improve locality of reference in the instruction space. Most systems
5760 using the ELF object format and SPARC processors running Solaris 2 have
5761 linkers with such optimizations. AIX may have these optimizations in
5764 Only use these options when there are significant benefits from doing
5765 so. When you specify these options, the assembler and linker will
5766 create larger object and executable files and will also be slower.
5767 You will not be able to use @code{gprof} on all systems if you
5768 specify this option and you may have problems with debugging if
5769 you specify both this option and @option{-g}.
5771 @item -fbranch-target-load-optimize
5772 @opindex fbranch-target-load-optimize
5773 Perform branch target register load optimization before prologue / epilogue
5775 The use of target registers can typically be exposed only during reload,
5776 thus hoisting loads out of loops and doing inter-block scheduling needs
5777 a separate optimization pass.
5779 @item -fbranch-target-load-optimize2
5780 @opindex fbranch-target-load-optimize2
5781 Perform branch target register load optimization after prologue / epilogue
5784 @item -fbtr-bb-exclusive
5785 @opindex fbtr-bb-exclusive
5786 When performing branch target register load optimization, don't reuse
5787 branch target registers in within any basic block.
5789 @item -fstack-protector
5790 Emit extra code to check for buffer overflows, such as stack smashing
5791 attacks. This is done by adding a guard variable to functions with
5792 vulnerable objects. This includes functions that call alloca, and
5793 functions with buffers larger than 8 bytes. The guards are initialized
5794 when a function is entered and then checked when the function exits.
5795 If a guard check fails, an error message is printed and the program exits.
5797 @item -fstack-protector-all
5798 Like @option{-fstack-protector} except that all functions are protected.
5800 @item -fsection-anchors
5801 @opindex fsection-anchors
5802 Try to reduce the number of symbolic address calculations by using
5803 shared ``anchor'' symbols to address nearby objects. This transformation
5804 can help to reduce the number of GOT entries and GOT accesses on some
5807 For example, the implementation of the following function @code{foo}:
5811 int foo (void) @{ return a + b + c; @}
5814 would usually calculate the addresses of all three variables, but if you
5815 compile it with @option{-fsection-anchors}, it will access the variables
5816 from a common anchor point instead. The effect is similar to the
5817 following pseudocode (which isn't valid C):
5822 register int *xr = &x;
5823 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5827 Not all targets support this option.
5829 @item --param @var{name}=@var{value}
5831 In some places, GCC uses various constants to control the amount of
5832 optimization that is done. For example, GCC will not inline functions
5833 that contain more that a certain number of instructions. You can
5834 control some of these constants on the command-line using the
5835 @option{--param} option.
5837 The names of specific parameters, and the meaning of the values, are
5838 tied to the internals of the compiler, and are subject to change
5839 without notice in future releases.
5841 In each case, the @var{value} is an integer. The allowable choices for
5842 @var{name} are given in the following table:
5845 @item salias-max-implicit-fields
5846 The maximum number of fields in a variable without direct
5847 structure accesses for which structure aliasing will consider trying
5848 to track each field. The default is 5
5850 @item salias-max-array-elements
5851 The maximum number of elements an array can have and its elements
5852 still be tracked individually by structure aliasing. The default is 4
5854 @item sra-max-structure-size
5855 The maximum structure size, in bytes, at which the scalar replacement
5856 of aggregates (SRA) optimization will perform block copies. The
5857 default value, 0, implies that GCC will select the most appropriate
5860 @item sra-field-structure-ratio
5861 The threshold ratio (as a percentage) between instantiated fields and
5862 the complete structure size. We say that if the ratio of the number
5863 of bytes in instantiated fields to the number of bytes in the complete
5864 structure exceeds this parameter, then block copies are not used. The
5867 @item max-crossjump-edges
5868 The maximum number of incoming edges to consider for crossjumping.
5869 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5870 the number of edges incoming to each block. Increasing values mean
5871 more aggressive optimization, making the compile time increase with
5872 probably small improvement in executable size.
5874 @item min-crossjump-insns
5875 The minimum number of instructions which must be matched at the end
5876 of two blocks before crossjumping will be performed on them. This
5877 value is ignored in the case where all instructions in the block being
5878 crossjumped from are matched. The default value is 5.
5880 @item max-grow-copy-bb-insns
5881 The maximum code size expansion factor when copying basic blocks
5882 instead of jumping. The expansion is relative to a jump instruction.
5883 The default value is 8.
5885 @item max-goto-duplication-insns
5886 The maximum number of instructions to duplicate to a block that jumps
5887 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5888 passes, GCC factors computed gotos early in the compilation process,
5889 and unfactors them as late as possible. Only computed jumps at the
5890 end of a basic blocks with no more than max-goto-duplication-insns are
5891 unfactored. The default value is 8.
5893 @item max-delay-slot-insn-search
5894 The maximum number of instructions to consider when looking for an
5895 instruction to fill a delay slot. If more than this arbitrary number of
5896 instructions is searched, the time savings from filling the delay slot
5897 will be minimal so stop searching. Increasing values mean more
5898 aggressive optimization, making the compile time increase with probably
5899 small improvement in executable run time.
5901 @item max-delay-slot-live-search
5902 When trying to fill delay slots, the maximum number of instructions to
5903 consider when searching for a block with valid live register
5904 information. Increasing this arbitrarily chosen value means more
5905 aggressive optimization, increasing the compile time. This parameter
5906 should be removed when the delay slot code is rewritten to maintain the
5909 @item max-gcse-memory
5910 The approximate maximum amount of memory that will be allocated in
5911 order to perform the global common subexpression elimination
5912 optimization. If more memory than specified is required, the
5913 optimization will not be done.
5915 @item max-gcse-passes
5916 The maximum number of passes of GCSE to run. The default is 1.
5918 @item max-pending-list-length
5919 The maximum number of pending dependencies scheduling will allow
5920 before flushing the current state and starting over. Large functions
5921 with few branches or calls can create excessively large lists which
5922 needlessly consume memory and resources.
5924 @item max-inline-insns-single
5925 Several parameters control the tree inliner used in gcc.
5926 This number sets the maximum number of instructions (counted in GCC's
5927 internal representation) in a single function that the tree inliner
5928 will consider for inlining. This only affects functions declared
5929 inline and methods implemented in a class declaration (C++).
5930 The default value is 450.
5932 @item max-inline-insns-auto
5933 When you use @option{-finline-functions} (included in @option{-O3}),
5934 a lot of functions that would otherwise not be considered for inlining
5935 by the compiler will be investigated. To those functions, a different
5936 (more restrictive) limit compared to functions declared inline can
5938 The default value is 90.
5940 @item large-function-insns
5941 The limit specifying really large functions. For functions larger than this
5942 limit after inlining inlining is constrained by
5943 @option{--param large-function-growth}. This parameter is useful primarily
5944 to avoid extreme compilation time caused by non-linear algorithms used by the
5946 This parameter is ignored when @option{-funit-at-a-time} is not used.
5947 The default value is 2700.
5949 @item large-function-growth
5950 Specifies maximal growth of large function caused by inlining in percents.
5951 This parameter is ignored when @option{-funit-at-a-time} is not used.
5952 The default value is 100 which limits large function growth to 2.0 times
5955 @item large-unit-insns
5956 The limit specifying large translation unit. Growth caused by inlining of
5957 units larger than this limit is limited by @option{--param inline-unit-growth}.
5958 For small units this might be too tight (consider unit consisting of function A
5959 that is inline and B that just calls A three time. If B is small relative to
5960 A, the growth of unit is 300\% and yet such inlining is very sane. For very
5961 large units consisting of small inlininable functions however the overall unit
5962 growth limit is needed to avoid exponential explosion of code size. Thus for
5963 smaller units, the size is increased to @option{--param large-unit-insns}
5964 before applying @option{--param inline-unit-growth}. The default is 10000
5966 @item inline-unit-growth
5967 Specifies maximal overall growth of the compilation unit caused by inlining.
5968 This parameter is ignored when @option{-funit-at-a-time} is not used.
5969 The default value is 50 which limits unit growth to 1.5 times the original
5972 @item max-inline-insns-recursive
5973 @itemx max-inline-insns-recursive-auto
5974 Specifies maximum number of instructions out-of-line copy of self recursive inline
5975 function can grow into by performing recursive inlining.
5977 For functions declared inline @option{--param max-inline-insns-recursive} is
5978 taken into acount. For function not declared inline, recursive inlining
5979 happens only when @option{-finline-functions} (included in @option{-O3}) is
5980 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5981 default value is 450.
5983 @item max-inline-recursive-depth
5984 @itemx max-inline-recursive-depth-auto
5985 Specifies maximum recursion depth used by the recursive inlining.
5987 For functions declared inline @option{--param max-inline-recursive-depth} is
5988 taken into acount. For function not declared inline, recursive inlining
5989 happens only when @option{-finline-functions} (included in @option{-O3}) is
5990 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5991 default value is 450.
5993 @item min-inline-recursive-probability
5994 Recursive inlining is profitable only for function having deep recursion
5995 in average and can hurt for function having little recursion depth by
5996 increasing the prologue size or complexity of function body to other
5999 When profile feedback is available (see @option{-fprofile-generate}) the actual
6000 recursion depth can be guessed from probability that function will recurse via
6001 given call expression. This parameter limits inlining only to call expression
6002 whose probability exceeds given threshold (in percents). The default value is
6005 @item inline-call-cost
6006 Specify cost of call instruction relative to simple arithmetics operations
6007 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6008 functions and at the same time increases size of leaf function that is believed to
6009 reduce function size by being inlined. In effect it increases amount of
6010 inlining for code having large abstraction penalty (many functions that just
6011 pass the arguments to other functions) and decrease inlining for code with low
6012 abstraction penalty. The default value is 16.
6014 @item max-unrolled-insns
6015 The maximum number of instructions that a loop should have if that loop
6016 is unrolled, and if the loop is unrolled, it determines how many times
6017 the loop code is unrolled.
6019 @item max-average-unrolled-insns
6020 The maximum number of instructions biased by probabilities of their execution
6021 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6022 it determines how many times the loop code is unrolled.
6024 @item max-unroll-times
6025 The maximum number of unrollings of a single loop.
6027 @item max-peeled-insns
6028 The maximum number of instructions that a loop should have if that loop
6029 is peeled, and if the loop is peeled, it determines how many times
6030 the loop code is peeled.
6032 @item max-peel-times
6033 The maximum number of peelings of a single loop.
6035 @item max-completely-peeled-insns
6036 The maximum number of insns of a completely peeled loop.
6038 @item max-completely-peel-times
6039 The maximum number of iterations of a loop to be suitable for complete peeling.
6041 @item max-unswitch-insns
6042 The maximum number of insns of an unswitched loop.
6044 @item max-unswitch-level
6045 The maximum number of branches unswitched in a single loop.
6048 The minimum cost of an expensive expression in the loop invariant motion.
6050 @item iv-consider-all-candidates-bound
6051 Bound on number of candidates for induction variables below that
6052 all candidates are considered for each use in induction variable
6053 optimizations. Only the most relevant candidates are considered
6054 if there are more candidates, to avoid quadratic time complexity.
6056 @item iv-max-considered-uses
6057 The induction variable optimizations give up on loops that contain more
6058 induction variable uses.
6060 @item iv-always-prune-cand-set-bound
6061 If number of candidates in the set is smaller than this value,
6062 we always try to remove unnecessary ivs from the set during its
6063 optimization when a new iv is added to the set.
6065 @item scev-max-expr-size
6066 Bound on size of expressions used in the scalar evolutions analyzer.
6067 Large expressions slow the analyzer.
6069 @item vect-max-version-checks
6070 The maximum number of runtime checks that can be performed when doing
6071 loop versioning in the vectorizer. See option ftree-vect-loop-version
6072 for more information.
6074 @item max-iterations-to-track
6076 The maximum number of iterations of a loop the brute force algorithm
6077 for analysis of # of iterations of the loop tries to evaluate.
6079 @item hot-bb-count-fraction
6080 Select fraction of the maximal count of repetitions of basic block in program
6081 given basic block needs to have to be considered hot.
6083 @item hot-bb-frequency-fraction
6084 Select fraction of the maximal frequency of executions of basic block in
6085 function given basic block needs to have to be considered hot
6087 @item max-predicted-iterations
6088 The maximum number of loop iterations we predict statically. This is useful
6089 in cases where function contain single loop with known bound and other loop
6090 with unknown. We predict the known number of iterations correctly, while
6091 the unknown number of iterations average to roughly 10. This means that the
6092 loop without bounds would appear artificially cold relative to the other one.
6094 @item tracer-dynamic-coverage
6095 @itemx tracer-dynamic-coverage-feedback
6097 This value is used to limit superblock formation once the given percentage of
6098 executed instructions is covered. This limits unnecessary code size
6101 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6102 feedback is available. The real profiles (as opposed to statically estimated
6103 ones) are much less balanced allowing the threshold to be larger value.
6105 @item tracer-max-code-growth
6106 Stop tail duplication once code growth has reached given percentage. This is
6107 rather hokey argument, as most of the duplicates will be eliminated later in
6108 cross jumping, so it may be set to much higher values than is the desired code
6111 @item tracer-min-branch-ratio
6113 Stop reverse growth when the reverse probability of best edge is less than this
6114 threshold (in percent).
6116 @item tracer-min-branch-ratio
6117 @itemx tracer-min-branch-ratio-feedback
6119 Stop forward growth if the best edge do have probability lower than this
6122 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6123 compilation for profile feedback and one for compilation without. The value
6124 for compilation with profile feedback needs to be more conservative (higher) in
6125 order to make tracer effective.
6127 @item max-cse-path-length
6129 Maximum number of basic blocks on path that cse considers. The default is 10.
6132 The maximum instructions CSE process before flushing. The default is 1000.
6134 @item global-var-threshold
6136 Counts the number of function calls (@var{n}) and the number of
6137 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
6138 single artificial variable will be created to represent all the
6139 call-clobbered variables at function call sites. This artificial
6140 variable will then be made to alias every call-clobbered variable.
6141 (done as @code{int * size_t} on the host machine; beware overflow).
6143 @item max-aliased-vops
6145 Maximum number of virtual operands allowed to represent aliases
6146 before triggering the alias grouping heuristic. Alias grouping
6147 reduces compile times and memory consumption needed for aliasing at
6148 the expense of precision loss in alias information.
6150 @item ggc-min-expand
6152 GCC uses a garbage collector to manage its own memory allocation. This
6153 parameter specifies the minimum percentage by which the garbage
6154 collector's heap should be allowed to expand between collections.
6155 Tuning this may improve compilation speed; it has no effect on code
6158 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6159 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6160 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6161 GCC is not able to calculate RAM on a particular platform, the lower
6162 bound of 30% is used. Setting this parameter and
6163 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6164 every opportunity. This is extremely slow, but can be useful for
6167 @item ggc-min-heapsize
6169 Minimum size of the garbage collector's heap before it begins bothering
6170 to collect garbage. The first collection occurs after the heap expands
6171 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6172 tuning this may improve compilation speed, and has no effect on code
6175 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6176 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6177 with a lower bound of 4096 (four megabytes) and an upper bound of
6178 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6179 particular platform, the lower bound is used. Setting this parameter
6180 very large effectively disables garbage collection. Setting this
6181 parameter and @option{ggc-min-expand} to zero causes a full collection
6182 to occur at every opportunity.
6184 @item max-reload-search-insns
6185 The maximum number of instruction reload should look backward for equivalent
6186 register. Increasing values mean more aggressive optimization, making the
6187 compile time increase with probably slightly better performance. The default
6190 @item max-cselib-memory-locations
6191 The maximum number of memory locations cselib should take into acount.
6192 Increasing values mean more aggressive optimization, making the compile time
6193 increase with probably slightly better performance. The default value is 500.
6195 @item max-flow-memory-locations
6196 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6197 The default value is 100.
6199 @item reorder-blocks-duplicate
6200 @itemx reorder-blocks-duplicate-feedback
6202 Used by basic block reordering pass to decide whether to use unconditional
6203 branch or duplicate the code on its destination. Code is duplicated when its
6204 estimated size is smaller than this value multiplied by the estimated size of
6205 unconditional jump in the hot spots of the program.
6207 The @option{reorder-block-duplicate-feedback} is used only when profile
6208 feedback is available and may be set to higher values than
6209 @option{reorder-block-duplicate} since information about the hot spots is more
6212 @item max-sched-ready-insns
6213 The maximum number of instructions ready to be issued the scheduler should
6214 consider at any given time during the first scheduling pass. Increasing
6215 values mean more thorough searches, making the compilation time increase
6216 with probably little benefit. The default value is 100.
6218 @item max-sched-region-blocks
6219 The maximum number of blocks in a region to be considered for
6220 interblock scheduling. The default value is 10.
6222 @item max-sched-region-insns
6223 The maximum number of insns in a region to be considered for
6224 interblock scheduling. The default value is 100.
6227 The minimum probability (in percents) of reaching a source block
6228 for interblock speculative scheduling. The default value is 40.
6230 @item max-sched-extend-regions-iters
6231 The maximum number of iterations through CFG to extend regions.
6232 0 - disable region extension,
6233 N - do at most N iterations.
6234 The default value is 0.
6236 @item max-sched-insn-conflict-delay
6237 The maximum conflict delay for an insn to be considered for speculative motion.
6238 The default value is 3.
6240 @item sched-spec-prob-cutoff
6241 The minimal probability of speculation success (in percents), so that
6242 speculative insn will be scheduled.
6243 The default value is 40.
6245 @item max-last-value-rtl
6247 The maximum size measured as number of RTLs that can be recorded in an expression
6248 in combiner for a pseudo register as last known value of that register. The default
6251 @item integer-share-limit
6252 Small integer constants can use a shared data structure, reducing the
6253 compiler's memory usage and increasing its speed. This sets the maximum
6254 value of a shared integer constant's. The default value is 256.
6256 @item min-virtual-mappings
6257 Specifies the minimum number of virtual mappings in the incremental
6258 SSA updater that should be registered to trigger the virtual mappings
6259 heuristic defined by virtual-mappings-ratio. The default value is
6262 @item virtual-mappings-ratio
6263 If the number of virtual mappings is virtual-mappings-ratio bigger
6264 than the number of virtual symbols to be updated, then the incremental
6265 SSA updater switches to a full update for those symbols. The default
6268 @item ssp-buffer-size
6269 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6270 protection when @option{-fstack-protection} is used.
6272 @item max-jump-thread-duplication-stmts
6273 Maximum number of statements allowed in a block that needs to be
6274 duplicated when threading jumps.
6276 @item max-fields-for-field-sensitive
6277 Maximum number of fields in a structure we will treat in
6278 a field sensitive manner during pointer analysis.
6283 @node Preprocessor Options
6284 @section Options Controlling the Preprocessor
6285 @cindex preprocessor options
6286 @cindex options, preprocessor
6288 These options control the C preprocessor, which is run on each C source
6289 file before actual compilation.
6291 If you use the @option{-E} option, nothing is done except preprocessing.
6292 Some of these options make sense only together with @option{-E} because
6293 they cause the preprocessor output to be unsuitable for actual
6298 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6299 and pass @var{option} directly through to the preprocessor. If
6300 @var{option} contains commas, it is split into multiple options at the
6301 commas. However, many options are modified, translated or interpreted
6302 by the compiler driver before being passed to the preprocessor, and
6303 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6304 interface is undocumented and subject to change, so whenever possible
6305 you should avoid using @option{-Wp} and let the driver handle the
6308 @item -Xpreprocessor @var{option}
6309 @opindex preprocessor
6310 Pass @var{option} as an option to the preprocessor. You can use this to
6311 supply system-specific preprocessor options which GCC does not know how to
6314 If you want to pass an option that takes an argument, you must use
6315 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6318 @include cppopts.texi
6320 @node Assembler Options
6321 @section Passing Options to the Assembler
6323 @c prevent bad page break with this line
6324 You can pass options to the assembler.
6327 @item -Wa,@var{option}
6329 Pass @var{option} as an option to the assembler. If @var{option}
6330 contains commas, it is split into multiple options at the commas.
6332 @item -Xassembler @var{option}
6334 Pass @var{option} as an option to the assembler. You can use this to
6335 supply system-specific assembler options which GCC does not know how to
6338 If you want to pass an option that takes an argument, you must use
6339 @option{-Xassembler} twice, once for the option and once for the argument.
6344 @section Options for Linking
6345 @cindex link options
6346 @cindex options, linking
6348 These options come into play when the compiler links object files into
6349 an executable output file. They are meaningless if the compiler is
6350 not doing a link step.
6354 @item @var{object-file-name}
6355 A file name that does not end in a special recognized suffix is
6356 considered to name an object file or library. (Object files are
6357 distinguished from libraries by the linker according to the file
6358 contents.) If linking is done, these object files are used as input
6367 If any of these options is used, then the linker is not run, and
6368 object file names should not be used as arguments. @xref{Overall
6372 @item -l@var{library}
6373 @itemx -l @var{library}
6375 Search the library named @var{library} when linking. (The second
6376 alternative with the library as a separate argument is only for
6377 POSIX compliance and is not recommended.)
6379 It makes a difference where in the command you write this option; the
6380 linker searches and processes libraries and object files in the order they
6381 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6382 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6383 to functions in @samp{z}, those functions may not be loaded.
6385 The linker searches a standard list of directories for the library,
6386 which is actually a file named @file{lib@var{library}.a}. The linker
6387 then uses this file as if it had been specified precisely by name.
6389 The directories searched include several standard system directories
6390 plus any that you specify with @option{-L}.
6392 Normally the files found this way are library files---archive files
6393 whose members are object files. The linker handles an archive file by
6394 scanning through it for members which define symbols that have so far
6395 been referenced but not defined. But if the file that is found is an
6396 ordinary object file, it is linked in the usual fashion. The only
6397 difference between using an @option{-l} option and specifying a file name
6398 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6399 and searches several directories.
6403 You need this special case of the @option{-l} option in order to
6404 link an Objective-C or Objective-C++ program.
6407 @opindex nostartfiles
6408 Do not use the standard system startup files when linking.
6409 The standard system libraries are used normally, unless @option{-nostdlib}
6410 or @option{-nodefaultlibs} is used.
6412 @item -nodefaultlibs
6413 @opindex nodefaultlibs
6414 Do not use the standard system libraries when linking.
6415 Only the libraries you specify will be passed to the linker.
6416 The standard startup files are used normally, unless @option{-nostartfiles}
6417 is used. The compiler may generate calls to @code{memcmp},
6418 @code{memset}, @code{memcpy} and @code{memmove}.
6419 These entries are usually resolved by entries in
6420 libc. These entry points should be supplied through some other
6421 mechanism when this option is specified.
6425 Do not use the standard system startup files or libraries when linking.
6426 No startup files and only the libraries you specify will be passed to
6427 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6428 @code{memcpy} and @code{memmove}.
6429 These entries are usually resolved by entries in
6430 libc. These entry points should be supplied through some other
6431 mechanism when this option is specified.
6433 @cindex @option{-lgcc}, use with @option{-nostdlib}
6434 @cindex @option{-nostdlib} and unresolved references
6435 @cindex unresolved references and @option{-nostdlib}
6436 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6437 @cindex @option{-nodefaultlibs} and unresolved references
6438 @cindex unresolved references and @option{-nodefaultlibs}
6439 One of the standard libraries bypassed by @option{-nostdlib} and
6440 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6441 that GCC uses to overcome shortcomings of particular machines, or special
6442 needs for some languages.
6443 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6444 Collection (GCC) Internals},
6445 for more discussion of @file{libgcc.a}.)
6446 In most cases, you need @file{libgcc.a} even when you want to avoid
6447 other standard libraries. In other words, when you specify @option{-nostdlib}
6448 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6449 This ensures that you have no unresolved references to internal GCC
6450 library subroutines. (For example, @samp{__main}, used to ensure C++
6451 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6452 GNU Compiler Collection (GCC) Internals}.)
6456 Produce a position independent executable on targets which support it.
6457 For predictable results, you must also specify the same set of options
6458 that were used to generate code (@option{-fpie}, @option{-fPIE},
6459 or model suboptions) when you specify this option.
6463 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6464 that support it. This instructs the linker to add all symbols, not
6465 only used ones, to the dynamic symbol table. This option is needed
6466 for some uses of @code{dlopen} or to allow obtaining backtraces
6467 from within a program.
6471 Remove all symbol table and relocation information from the executable.
6475 On systems that support dynamic linking, this prevents linking with the shared
6476 libraries. On other systems, this option has no effect.
6480 Produce a shared object which can then be linked with other objects to
6481 form an executable. Not all systems support this option. For predictable
6482 results, you must also specify the same set of options that were used to
6483 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6484 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6485 needs to build supplementary stub code for constructors to work. On
6486 multi-libbed systems, @samp{gcc -shared} must select the correct support
6487 libraries to link against. Failing to supply the correct flags may lead
6488 to subtle defects. Supplying them in cases where they are not necessary
6491 @item -shared-libgcc
6492 @itemx -static-libgcc
6493 @opindex shared-libgcc
6494 @opindex static-libgcc
6495 On systems that provide @file{libgcc} as a shared library, these options
6496 force the use of either the shared or static version respectively.
6497 If no shared version of @file{libgcc} was built when the compiler was
6498 configured, these options have no effect.
6500 There are several situations in which an application should use the
6501 shared @file{libgcc} instead of the static version. The most common
6502 of these is when the application wishes to throw and catch exceptions
6503 across different shared libraries. In that case, each of the libraries
6504 as well as the application itself should use the shared @file{libgcc}.
6506 Therefore, the G++ and GCJ drivers automatically add
6507 @option{-shared-libgcc} whenever you build a shared library or a main
6508 executable, because C++ and Java programs typically use exceptions, so
6509 this is the right thing to do.
6511 If, instead, you use the GCC driver to create shared libraries, you may
6512 find that they will not always be linked with the shared @file{libgcc}.
6513 If GCC finds, at its configuration time, that you have a non-GNU linker
6514 or a GNU linker that does not support option @option{--eh-frame-hdr},
6515 it will link the shared version of @file{libgcc} into shared libraries
6516 by default. Otherwise, it will take advantage of the linker and optimize
6517 away the linking with the shared version of @file{libgcc}, linking with
6518 the static version of libgcc by default. This allows exceptions to
6519 propagate through such shared libraries, without incurring relocation
6520 costs at library load time.
6522 However, if a library or main executable is supposed to throw or catch
6523 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6524 for the languages used in the program, or using the option
6525 @option{-shared-libgcc}, such that it is linked with the shared
6530 Bind references to global symbols when building a shared object. Warn
6531 about any unresolved references (unless overridden by the link editor
6532 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6535 @item -Xlinker @var{option}
6537 Pass @var{option} as an option to the linker. You can use this to
6538 supply system-specific linker options which GCC does not know how to
6541 If you want to pass an option that takes an argument, you must use
6542 @option{-Xlinker} twice, once for the option and once for the argument.
6543 For example, to pass @option{-assert definitions}, you must write
6544 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6545 @option{-Xlinker "-assert definitions"}, because this passes the entire
6546 string as a single argument, which is not what the linker expects.
6548 @item -Wl,@var{option}
6550 Pass @var{option} as an option to the linker. If @var{option} contains
6551 commas, it is split into multiple options at the commas.
6553 @item -u @var{symbol}
6555 Pretend the symbol @var{symbol} is undefined, to force linking of
6556 library modules to define it. You can use @option{-u} multiple times with
6557 different symbols to force loading of additional library modules.
6560 @node Directory Options
6561 @section Options for Directory Search
6562 @cindex directory options
6563 @cindex options, directory search
6566 These options specify directories to search for header files, for
6567 libraries and for parts of the compiler:
6572 Add the directory @var{dir} to the head of the list of directories to be
6573 searched for header files. This can be used to override a system header
6574 file, substituting your own version, since these directories are
6575 searched before the system header file directories. However, you should
6576 not use this option to add directories that contain vendor-supplied
6577 system header files (use @option{-isystem} for that). If you use more than
6578 one @option{-I} option, the directories are scanned in left-to-right
6579 order; the standard system directories come after.
6581 If a standard system include directory, or a directory specified with
6582 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6583 option will be ignored. The directory will still be searched but as a
6584 system directory at its normal position in the system include chain.
6585 This is to ensure that GCC's procedure to fix buggy system headers and
6586 the ordering for the include_next directive are not inadvertently changed.
6587 If you really need to change the search order for system directories,
6588 use the @option{-nostdinc} and/or @option{-isystem} options.
6590 @item -iquote@var{dir}
6592 Add the directory @var{dir} to the head of the list of directories to
6593 be searched for header files only for the case of @samp{#include
6594 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6595 otherwise just like @option{-I}.
6599 Add directory @var{dir} to the list of directories to be searched
6602 @item -B@var{prefix}
6604 This option specifies where to find the executables, libraries,
6605 include files, and data files of the compiler itself.
6607 The compiler driver program runs one or more of the subprograms
6608 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6609 @var{prefix} as a prefix for each program it tries to run, both with and
6610 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6612 For each subprogram to be run, the compiler driver first tries the
6613 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6614 was not specified, the driver tries two standard prefixes, which are
6615 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6616 those results in a file name that is found, the unmodified program
6617 name is searched for using the directories specified in your
6618 @env{PATH} environment variable.
6620 The compiler will check to see if the path provided by the @option{-B}
6621 refers to a directory, and if necessary it will add a directory
6622 separator character at the end of the path.
6624 @option{-B} prefixes that effectively specify directory names also apply
6625 to libraries in the linker, because the compiler translates these
6626 options into @option{-L} options for the linker. They also apply to
6627 includes files in the preprocessor, because the compiler translates these
6628 options into @option{-isystem} options for the preprocessor. In this case,
6629 the compiler appends @samp{include} to the prefix.
6631 The run-time support file @file{libgcc.a} can also be searched for using
6632 the @option{-B} prefix, if needed. If it is not found there, the two
6633 standard prefixes above are tried, and that is all. The file is left
6634 out of the link if it is not found by those means.
6636 Another way to specify a prefix much like the @option{-B} prefix is to use
6637 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6640 As a special kludge, if the path provided by @option{-B} is
6641 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6642 9, then it will be replaced by @file{[dir/]include}. This is to help
6643 with boot-strapping the compiler.
6645 @item -specs=@var{file}
6647 Process @var{file} after the compiler reads in the standard @file{specs}
6648 file, in order to override the defaults that the @file{gcc} driver
6649 program uses when determining what switches to pass to @file{cc1},
6650 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6651 @option{-specs=@var{file}} can be specified on the command line, and they
6652 are processed in order, from left to right.
6654 @item --sysroot=@var{dir}
6656 Use @var{dir} as the logical root directory for headers and libraries.
6657 For example, if the compiler would normally search for headers in
6658 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6659 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6661 If you use both this option and the @option{-isysroot} option, then
6662 the @option{--sysroot} option will apply to libraries, but the
6663 @option{-isysroot} option will apply to header files.
6665 The GNU linker (beginning with version 2.16) has the necessary support
6666 for this option. If your linker does not support this option, the
6667 header file aspect of @option{--sysroot} will still work, but the
6668 library aspect will not.
6672 This option has been deprecated. Please use @option{-iquote} instead for
6673 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6674 Any directories you specify with @option{-I} options before the @option{-I-}
6675 option are searched only for the case of @samp{#include "@var{file}"};
6676 they are not searched for @samp{#include <@var{file}>}.
6678 If additional directories are specified with @option{-I} options after
6679 the @option{-I-}, these directories are searched for all @samp{#include}
6680 directives. (Ordinarily @emph{all} @option{-I} directories are used
6683 In addition, the @option{-I-} option inhibits the use of the current
6684 directory (where the current input file came from) as the first search
6685 directory for @samp{#include "@var{file}"}. There is no way to
6686 override this effect of @option{-I-}. With @option{-I.} you can specify
6687 searching the directory which was current when the compiler was
6688 invoked. That is not exactly the same as what the preprocessor does
6689 by default, but it is often satisfactory.
6691 @option{-I-} does not inhibit the use of the standard system directories
6692 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6699 @section Specifying subprocesses and the switches to pass to them
6702 @command{gcc} is a driver program. It performs its job by invoking a
6703 sequence of other programs to do the work of compiling, assembling and
6704 linking. GCC interprets its command-line parameters and uses these to
6705 deduce which programs it should invoke, and which command-line options
6706 it ought to place on their command lines. This behavior is controlled
6707 by @dfn{spec strings}. In most cases there is one spec string for each
6708 program that GCC can invoke, but a few programs have multiple spec
6709 strings to control their behavior. The spec strings built into GCC can
6710 be overridden by using the @option{-specs=} command-line switch to specify
6713 @dfn{Spec files} are plaintext files that are used to construct spec
6714 strings. They consist of a sequence of directives separated by blank
6715 lines. The type of directive is determined by the first non-whitespace
6716 character on the line and it can be one of the following:
6719 @item %@var{command}
6720 Issues a @var{command} to the spec file processor. The commands that can
6724 @item %include <@var{file}>
6726 Search for @var{file} and insert its text at the current point in the
6729 @item %include_noerr <@var{file}>
6730 @cindex %include_noerr
6731 Just like @samp{%include}, but do not generate an error message if the include
6732 file cannot be found.
6734 @item %rename @var{old_name} @var{new_name}
6736 Rename the spec string @var{old_name} to @var{new_name}.
6740 @item *[@var{spec_name}]:
6741 This tells the compiler to create, override or delete the named spec
6742 string. All lines after this directive up to the next directive or
6743 blank line are considered to be the text for the spec string. If this
6744 results in an empty string then the spec will be deleted. (Or, if the
6745 spec did not exist, then nothing will happened.) Otherwise, if the spec
6746 does not currently exist a new spec will be created. If the spec does
6747 exist then its contents will be overridden by the text of this
6748 directive, unless the first character of that text is the @samp{+}
6749 character, in which case the text will be appended to the spec.
6751 @item [@var{suffix}]:
6752 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6753 and up to the next directive or blank line are considered to make up the
6754 spec string for the indicated suffix. When the compiler encounters an
6755 input file with the named suffix, it will processes the spec string in
6756 order to work out how to compile that file. For example:
6763 This says that any input file whose name ends in @samp{.ZZ} should be
6764 passed to the program @samp{z-compile}, which should be invoked with the
6765 command-line switch @option{-input} and with the result of performing the
6766 @samp{%i} substitution. (See below.)
6768 As an alternative to providing a spec string, the text that follows a
6769 suffix directive can be one of the following:
6772 @item @@@var{language}
6773 This says that the suffix is an alias for a known @var{language}. This is
6774 similar to using the @option{-x} command-line switch to GCC to specify a
6775 language explicitly. For example:
6782 Says that .ZZ files are, in fact, C++ source files.
6785 This causes an error messages saying:
6788 @var{name} compiler not installed on this system.
6792 GCC already has an extensive list of suffixes built into it.
6793 This directive will add an entry to the end of the list of suffixes, but
6794 since the list is searched from the end backwards, it is effectively
6795 possible to override earlier entries using this technique.
6799 GCC has the following spec strings built into it. Spec files can
6800 override these strings or create their own. Note that individual
6801 targets can also add their own spec strings to this list.
6804 asm Options to pass to the assembler
6805 asm_final Options to pass to the assembler post-processor
6806 cpp Options to pass to the C preprocessor
6807 cc1 Options to pass to the C compiler
6808 cc1plus Options to pass to the C++ compiler
6809 endfile Object files to include at the end of the link
6810 link Options to pass to the linker
6811 lib Libraries to include on the command line to the linker
6812 libgcc Decides which GCC support library to pass to the linker
6813 linker Sets the name of the linker
6814 predefines Defines to be passed to the C preprocessor
6815 signed_char Defines to pass to CPP to say whether @code{char} is signed
6817 startfile Object files to include at the start of the link
6820 Here is a small example of a spec file:
6826 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6829 This example renames the spec called @samp{lib} to @samp{old_lib} and
6830 then overrides the previous definition of @samp{lib} with a new one.
6831 The new definition adds in some extra command-line options before
6832 including the text of the old definition.
6834 @dfn{Spec strings} are a list of command-line options to be passed to their
6835 corresponding program. In addition, the spec strings can contain
6836 @samp{%}-prefixed sequences to substitute variable text or to
6837 conditionally insert text into the command line. Using these constructs
6838 it is possible to generate quite complex command lines.
6840 Here is a table of all defined @samp{%}-sequences for spec
6841 strings. Note that spaces are not generated automatically around the
6842 results of expanding these sequences. Therefore you can concatenate them
6843 together or combine them with constant text in a single argument.
6847 Substitute one @samp{%} into the program name or argument.
6850 Substitute the name of the input file being processed.
6853 Substitute the basename of the input file being processed.
6854 This is the substring up to (and not including) the last period
6855 and not including the directory.
6858 This is the same as @samp{%b}, but include the file suffix (text after
6862 Marks the argument containing or following the @samp{%d} as a
6863 temporary file name, so that that file will be deleted if GCC exits
6864 successfully. Unlike @samp{%g}, this contributes no text to the
6867 @item %g@var{suffix}
6868 Substitute a file name that has suffix @var{suffix} and is chosen
6869 once per compilation, and mark the argument in the same way as
6870 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6871 name is now chosen in a way that is hard to predict even when previously
6872 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6873 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6874 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6875 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6876 was simply substituted with a file name chosen once per compilation,
6877 without regard to any appended suffix (which was therefore treated
6878 just like ordinary text), making such attacks more likely to succeed.
6880 @item %u@var{suffix}
6881 Like @samp{%g}, but generates a new temporary file name even if
6882 @samp{%u@var{suffix}} was already seen.
6884 @item %U@var{suffix}
6885 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6886 new one if there is no such last file name. In the absence of any
6887 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6888 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6889 would involve the generation of two distinct file names, one
6890 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6891 simply substituted with a file name chosen for the previous @samp{%u},
6892 without regard to any appended suffix.
6894 @item %j@var{suffix}
6895 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6896 writable, and if save-temps is off; otherwise, substitute the name
6897 of a temporary file, just like @samp{%u}. This temporary file is not
6898 meant for communication between processes, but rather as a junk
6901 @item %|@var{suffix}
6902 @itemx %m@var{suffix}
6903 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6904 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6905 all. These are the two most common ways to instruct a program that it
6906 should read from standard input or write to standard output. If you
6907 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6908 construct: see for example @file{f/lang-specs.h}.
6910 @item %.@var{SUFFIX}
6911 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6912 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6913 terminated by the next space or %.
6916 Marks the argument containing or following the @samp{%w} as the
6917 designated output file of this compilation. This puts the argument
6918 into the sequence of arguments that @samp{%o} will substitute later.
6921 Substitutes the names of all the output files, with spaces
6922 automatically placed around them. You should write spaces
6923 around the @samp{%o} as well or the results are undefined.
6924 @samp{%o} is for use in the specs for running the linker.
6925 Input files whose names have no recognized suffix are not compiled
6926 at all, but they are included among the output files, so they will
6930 Substitutes the suffix for object files. Note that this is
6931 handled specially when it immediately follows @samp{%g, %u, or %U},
6932 because of the need for those to form complete file names. The
6933 handling is such that @samp{%O} is treated exactly as if it had already
6934 been substituted, except that @samp{%g, %u, and %U} do not currently
6935 support additional @var{suffix} characters following @samp{%O} as they would
6936 following, for example, @samp{.o}.
6939 Substitutes the standard macro predefinitions for the
6940 current target machine. Use this when running @code{cpp}.
6943 Like @samp{%p}, but puts @samp{__} before and after the name of each
6944 predefined macro, except for macros that start with @samp{__} or with
6945 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6949 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6950 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
6951 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6952 and @option{-imultilib} as necessary.
6955 Current argument is the name of a library or startup file of some sort.
6956 Search for that file in a standard list of directories and substitute
6957 the full name found.
6960 Print @var{str} as an error message. @var{str} is terminated by a newline.
6961 Use this when inconsistent options are detected.
6964 Substitute the contents of spec string @var{name} at this point.
6967 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6969 @item %x@{@var{option}@}
6970 Accumulate an option for @samp{%X}.
6973 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6977 Output the accumulated assembler options specified by @option{-Wa}.
6980 Output the accumulated preprocessor options specified by @option{-Wp}.
6983 Process the @code{asm} spec. This is used to compute the
6984 switches to be passed to the assembler.
6987 Process the @code{asm_final} spec. This is a spec string for
6988 passing switches to an assembler post-processor, if such a program is
6992 Process the @code{link} spec. This is the spec for computing the
6993 command line passed to the linker. Typically it will make use of the
6994 @samp{%L %G %S %D and %E} sequences.
6997 Dump out a @option{-L} option for each directory that GCC believes might
6998 contain startup files. If the target supports multilibs then the
6999 current multilib directory will be prepended to each of these paths.
7002 Process the @code{lib} spec. This is a spec string for deciding which
7003 libraries should be included on the command line to the linker.
7006 Process the @code{libgcc} spec. This is a spec string for deciding
7007 which GCC support library should be included on the command line to the linker.
7010 Process the @code{startfile} spec. This is a spec for deciding which
7011 object files should be the first ones passed to the linker. Typically
7012 this might be a file named @file{crt0.o}.
7015 Process the @code{endfile} spec. This is a spec string that specifies
7016 the last object files that will be passed to the linker.
7019 Process the @code{cpp} spec. This is used to construct the arguments
7020 to be passed to the C preprocessor.
7023 Process the @code{cc1} spec. This is used to construct the options to be
7024 passed to the actual C compiler (@samp{cc1}).
7027 Process the @code{cc1plus} spec. This is used to construct the options to be
7028 passed to the actual C++ compiler (@samp{cc1plus}).
7031 Substitute the variable part of a matched option. See below.
7032 Note that each comma in the substituted string is replaced by
7036 Remove all occurrences of @code{-S} from the command line. Note---this
7037 command is position dependent. @samp{%} commands in the spec string
7038 before this one will see @code{-S}, @samp{%} commands in the spec string
7039 after this one will not.
7041 @item %:@var{function}(@var{args})
7042 Call the named function @var{function}, passing it @var{args}.
7043 @var{args} is first processed as a nested spec string, then split
7044 into an argument vector in the usual fashion. The function returns
7045 a string which is processed as if it had appeared literally as part
7046 of the current spec.
7048 The following built-in spec functions are provided:
7051 @item @code{if-exists}
7052 The @code{if-exists} spec function takes one argument, an absolute
7053 pathname to a file. If the file exists, @code{if-exists} returns the
7054 pathname. Here is a small example of its usage:
7058 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7061 @item @code{if-exists-else}
7062 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7063 spec function, except that it takes two arguments. The first argument is
7064 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7065 returns the pathname. If it does not exist, it returns the second argument.
7066 This way, @code{if-exists-else} can be used to select one file or another,
7067 based on the existence of the first. Here is a small example of its usage:
7071 crt0%O%s %:if-exists(crti%O%s) \
7072 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7075 @item @code{replace-outfile}
7076 The @code{replace-outfile} spec function takes two arguments. It looks for the
7077 first argument in the outfiles array and replaces it with the second argument. Here
7078 is a small example of its usage:
7081 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7087 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7088 If that switch was not specified, this substitutes nothing. Note that
7089 the leading dash is omitted when specifying this option, and it is
7090 automatically inserted if the substitution is performed. Thus the spec
7091 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7092 and would output the command line option @option{-foo}.
7094 @item %W@{@code{S}@}
7095 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7098 @item %@{@code{S}*@}
7099 Substitutes all the switches specified to GCC whose names start
7100 with @code{-S}, but which also take an argument. This is used for
7101 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7102 GCC considers @option{-o foo} as being
7103 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7104 text, including the space. Thus two arguments would be generated.
7106 @item %@{@code{S}*&@code{T}*@}
7107 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7108 (the order of @code{S} and @code{T} in the spec is not significant).
7109 There can be any number of ampersand-separated variables; for each the
7110 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7112 @item %@{@code{S}:@code{X}@}
7113 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7115 @item %@{!@code{S}:@code{X}@}
7116 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7118 @item %@{@code{S}*:@code{X}@}
7119 Substitutes @code{X} if one or more switches whose names start with
7120 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7121 once, no matter how many such switches appeared. However, if @code{%*}
7122 appears somewhere in @code{X}, then @code{X} will be substituted once
7123 for each matching switch, with the @code{%*} replaced by the part of
7124 that switch that matched the @code{*}.
7126 @item %@{.@code{S}:@code{X}@}
7127 Substitutes @code{X}, if processing a file with suffix @code{S}.
7129 @item %@{!.@code{S}:@code{X}@}
7130 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7132 @item %@{@code{S}|@code{P}:@code{X}@}
7133 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7134 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7135 although they have a stronger binding than the @samp{|}. If @code{%*}
7136 appears in @code{X}, all of the alternatives must be starred, and only
7137 the first matching alternative is substituted.
7139 For example, a spec string like this:
7142 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7145 will output the following command-line options from the following input
7146 command-line options:
7151 -d fred.c -foo -baz -boggle
7152 -d jim.d -bar -baz -boggle
7155 @item %@{S:X; T:Y; :D@}
7157 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7158 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7159 be as many clauses as you need. This may be combined with @code{.},
7160 @code{!}, @code{|}, and @code{*} as needed.
7165 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7166 construct may contain other nested @samp{%} constructs or spaces, or
7167 even newlines. They are processed as usual, as described above.
7168 Trailing white space in @code{X} is ignored. White space may also
7169 appear anywhere on the left side of the colon in these constructs,
7170 except between @code{.} or @code{*} and the corresponding word.
7172 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7173 handled specifically in these constructs. If another value of
7174 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7175 @option{-W} switch is found later in the command line, the earlier
7176 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7177 just one letter, which passes all matching options.
7179 The character @samp{|} at the beginning of the predicate text is used to
7180 indicate that a command should be piped to the following command, but
7181 only if @option{-pipe} is specified.
7183 It is built into GCC which switches take arguments and which do not.
7184 (You might think it would be useful to generalize this to allow each
7185 compiler's spec to say which switches take arguments. But this cannot
7186 be done in a consistent fashion. GCC cannot even decide which input
7187 files have been specified without knowing which switches take arguments,
7188 and it must know which input files to compile in order to tell which
7191 GCC also knows implicitly that arguments starting in @option{-l} are to be
7192 treated as compiler output files, and passed to the linker in their
7193 proper position among the other output files.
7195 @c man begin OPTIONS
7197 @node Target Options
7198 @section Specifying Target Machine and Compiler Version
7199 @cindex target options
7200 @cindex cross compiling
7201 @cindex specifying machine version
7202 @cindex specifying compiler version and target machine
7203 @cindex compiler version, specifying
7204 @cindex target machine, specifying
7206 The usual way to run GCC is to run the executable called @file{gcc}, or
7207 @file{<machine>-gcc} when cross-compiling, or
7208 @file{<machine>-gcc-<version>} to run a version other than the one that
7209 was installed last. Sometimes this is inconvenient, so GCC provides
7210 options that will switch to another cross-compiler or version.
7213 @item -b @var{machine}
7215 The argument @var{machine} specifies the target machine for compilation.
7217 The value to use for @var{machine} is the same as was specified as the
7218 machine type when configuring GCC as a cross-compiler. For
7219 example, if a cross-compiler was configured with @samp{configure
7220 arm-elf}, meaning to compile for an arm processor with elf binaries,
7221 then you would specify @option{-b arm-elf} to run that cross compiler.
7222 Because there are other options beginning with @option{-b}, the
7223 configuration must contain a hyphen.
7225 @item -V @var{version}
7227 The argument @var{version} specifies which version of GCC to run.
7228 This is useful when multiple versions are installed. For example,
7229 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7232 The @option{-V} and @option{-b} options work by running the
7233 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7234 use them if you can just run that directly.
7236 @node Submodel Options
7237 @section Hardware Models and Configurations
7238 @cindex submodel options
7239 @cindex specifying hardware config
7240 @cindex hardware models and configurations, specifying
7241 @cindex machine dependent options
7243 Earlier we discussed the standard option @option{-b} which chooses among
7244 different installed compilers for completely different target
7245 machines, such as VAX vs.@: 68000 vs.@: 80386.
7247 In addition, each of these target machine types can have its own
7248 special options, starting with @samp{-m}, to choose among various
7249 hardware models or configurations---for example, 68010 vs 68020,
7250 floating coprocessor or none. A single installed version of the
7251 compiler can compile for any model or configuration, according to the
7254 Some configurations of the compiler also support additional special
7255 options, usually for compatibility with other compilers on the same
7258 @c This list is ordered alphanumerically by subsection name.
7259 @c It should be the same order and spelling as these options are listed
7260 @c in Machine Dependent Options
7266 * Blackfin Options::
7270 * DEC Alpha Options::
7271 * DEC Alpha/VMS Options::
7273 * GNU/Linux Options::
7276 * i386 and x86-64 Options::
7289 * RS/6000 and PowerPC Options::
7290 * S/390 and zSeries Options::
7293 * System V Options::
7294 * TMS320C3x/C4x Options::
7298 * Xstormy16 Options::
7304 @subsection ARC Options
7307 These options are defined for ARC implementations:
7312 Compile code for little endian mode. This is the default.
7316 Compile code for big endian mode.
7319 @opindex mmangle-cpu
7320 Prepend the name of the cpu to all public symbol names.
7321 In multiple-processor systems, there are many ARC variants with different
7322 instruction and register set characteristics. This flag prevents code
7323 compiled for one cpu to be linked with code compiled for another.
7324 No facility exists for handling variants that are ``almost identical''.
7325 This is an all or nothing option.
7327 @item -mcpu=@var{cpu}
7329 Compile code for ARC variant @var{cpu}.
7330 Which variants are supported depend on the configuration.
7331 All variants support @option{-mcpu=base}, this is the default.
7333 @item -mtext=@var{text-section}
7334 @itemx -mdata=@var{data-section}
7335 @itemx -mrodata=@var{readonly-data-section}
7339 Put functions, data, and readonly data in @var{text-section},
7340 @var{data-section}, and @var{readonly-data-section} respectively
7341 by default. This can be overridden with the @code{section} attribute.
7342 @xref{Variable Attributes}.
7347 @subsection ARM Options
7350 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7354 @item -mabi=@var{name}
7356 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7357 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7360 @opindex mapcs-frame
7361 Generate a stack frame that is compliant with the ARM Procedure Call
7362 Standard for all functions, even if this is not strictly necessary for
7363 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7364 with this option will cause the stack frames not to be generated for
7365 leaf functions. The default is @option{-mno-apcs-frame}.
7369 This is a synonym for @option{-mapcs-frame}.
7372 @c not currently implemented
7373 @item -mapcs-stack-check
7374 @opindex mapcs-stack-check
7375 Generate code to check the amount of stack space available upon entry to
7376 every function (that actually uses some stack space). If there is
7377 insufficient space available then either the function
7378 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7379 called, depending upon the amount of stack space required. The run time
7380 system is required to provide these functions. The default is
7381 @option{-mno-apcs-stack-check}, since this produces smaller code.
7383 @c not currently implemented
7385 @opindex mapcs-float
7386 Pass floating point arguments using the float point registers. This is
7387 one of the variants of the APCS@. This option is recommended if the
7388 target hardware has a floating point unit or if a lot of floating point
7389 arithmetic is going to be performed by the code. The default is
7390 @option{-mno-apcs-float}, since integer only code is slightly increased in
7391 size if @option{-mapcs-float} is used.
7393 @c not currently implemented
7394 @item -mapcs-reentrant
7395 @opindex mapcs-reentrant
7396 Generate reentrant, position independent code. The default is
7397 @option{-mno-apcs-reentrant}.
7400 @item -mthumb-interwork
7401 @opindex mthumb-interwork
7402 Generate code which supports calling between the ARM and Thumb
7403 instruction sets. Without this option the two instruction sets cannot
7404 be reliably used inside one program. The default is
7405 @option{-mno-thumb-interwork}, since slightly larger code is generated
7406 when @option{-mthumb-interwork} is specified.
7408 @item -mno-sched-prolog
7409 @opindex mno-sched-prolog
7410 Prevent the reordering of instructions in the function prolog, or the
7411 merging of those instruction with the instructions in the function's
7412 body. This means that all functions will start with a recognizable set
7413 of instructions (or in fact one of a choice from a small set of
7414 different function prologues), and this information can be used to
7415 locate the start if functions inside an executable piece of code. The
7416 default is @option{-msched-prolog}.
7419 @opindex mhard-float
7420 Generate output containing floating point instructions. This is the
7424 @opindex msoft-float
7425 Generate output containing library calls for floating point.
7426 @strong{Warning:} the requisite libraries are not available for all ARM
7427 targets. Normally the facilities of the machine's usual C compiler are
7428 used, but this cannot be done directly in cross-compilation. You must make
7429 your own arrangements to provide suitable library functions for
7432 @option{-msoft-float} changes the calling convention in the output file;
7433 therefore, it is only useful if you compile @emph{all} of a program with
7434 this option. In particular, you need to compile @file{libgcc.a}, the
7435 library that comes with GCC, with @option{-msoft-float} in order for
7438 @item -mfloat-abi=@var{name}
7440 Specifies which ABI to use for floating point values. Permissible values
7441 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7443 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7444 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7445 of floating point instructions, but still uses the soft-float calling
7448 @item -mlittle-endian
7449 @opindex mlittle-endian
7450 Generate code for a processor running in little-endian mode. This is
7451 the default for all standard configurations.
7454 @opindex mbig-endian
7455 Generate code for a processor running in big-endian mode; the default is
7456 to compile code for a little-endian processor.
7458 @item -mwords-little-endian
7459 @opindex mwords-little-endian
7460 This option only applies when generating code for big-endian processors.
7461 Generate code for a little-endian word order but a big-endian byte
7462 order. That is, a byte order of the form @samp{32107654}. Note: this
7463 option should only be used if you require compatibility with code for
7464 big-endian ARM processors generated by versions of the compiler prior to
7467 @item -mcpu=@var{name}
7469 This specifies the name of the target ARM processor. GCC uses this name
7470 to determine what kind of instructions it can emit when generating
7471 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7472 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7473 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7474 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7475 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7476 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7477 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7478 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7479 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7480 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7481 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7482 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7483 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7484 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7487 @itemx -mtune=@var{name}
7489 This option is very similar to the @option{-mcpu=} option, except that
7490 instead of specifying the actual target processor type, and hence
7491 restricting which instructions can be used, it specifies that GCC should
7492 tune the performance of the code as if the target were of the type
7493 specified in this option, but still choosing the instructions that it
7494 will generate based on the cpu specified by a @option{-mcpu=} option.
7495 For some ARM implementations better performance can be obtained by using
7498 @item -march=@var{name}
7500 This specifies the name of the target ARM architecture. GCC uses this
7501 name to determine what kind of instructions it can emit when generating
7502 assembly code. This option can be used in conjunction with or instead
7503 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7504 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7505 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7506 @samp{iwmmxt}, @samp{ep9312}.
7508 @item -mfpu=@var{name}
7509 @itemx -mfpe=@var{number}
7510 @itemx -mfp=@var{number}
7514 This specifies what floating point hardware (or hardware emulation) is
7515 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7516 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7517 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7518 with older versions of GCC@.
7520 If @option{-msoft-float} is specified this specifies the format of
7521 floating point values.
7523 @item -mstructure-size-boundary=@var{n}
7524 @opindex mstructure-size-boundary
7525 The size of all structures and unions will be rounded up to a multiple
7526 of the number of bits set by this option. Permissible values are 8, 32
7527 and 64. The default value varies for different toolchains. For the COFF
7528 targeted toolchain the default value is 8. A value of 64 is only allowed
7529 if the underlying ABI supports it.
7531 Specifying the larger number can produce faster, more efficient code, but
7532 can also increase the size of the program. Different values are potentially
7533 incompatible. Code compiled with one value cannot necessarily expect to
7534 work with code or libraries compiled with another value, if they exchange
7535 information using structures or unions.
7537 @item -mabort-on-noreturn
7538 @opindex mabort-on-noreturn
7539 Generate a call to the function @code{abort} at the end of a
7540 @code{noreturn} function. It will be executed if the function tries to
7544 @itemx -mno-long-calls
7545 @opindex mlong-calls
7546 @opindex mno-long-calls
7547 Tells the compiler to perform function calls by first loading the
7548 address of the function into a register and then performing a subroutine
7549 call on this register. This switch is needed if the target function
7550 will lie outside of the 64 megabyte addressing range of the offset based
7551 version of subroutine call instruction.
7553 Even if this switch is enabled, not all function calls will be turned
7554 into long calls. The heuristic is that static functions, functions
7555 which have the @samp{short-call} attribute, functions that are inside
7556 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7557 definitions have already been compiled within the current compilation
7558 unit, will not be turned into long calls. The exception to this rule is
7559 that weak function definitions, functions with the @samp{long-call}
7560 attribute or the @samp{section} attribute, and functions that are within
7561 the scope of a @samp{#pragma long_calls} directive, will always be
7562 turned into long calls.
7564 This feature is not enabled by default. Specifying
7565 @option{-mno-long-calls} will restore the default behavior, as will
7566 placing the function calls within the scope of a @samp{#pragma
7567 long_calls_off} directive. Note these switches have no effect on how
7568 the compiler generates code to handle function calls via function
7571 @item -mnop-fun-dllimport
7572 @opindex mnop-fun-dllimport
7573 Disable support for the @code{dllimport} attribute.
7575 @item -msingle-pic-base
7576 @opindex msingle-pic-base
7577 Treat the register used for PIC addressing as read-only, rather than
7578 loading it in the prologue for each function. The run-time system is
7579 responsible for initializing this register with an appropriate value
7580 before execution begins.
7582 @item -mpic-register=@var{reg}
7583 @opindex mpic-register
7584 Specify the register to be used for PIC addressing. The default is R10
7585 unless stack-checking is enabled, when R9 is used.
7587 @item -mcirrus-fix-invalid-insns
7588 @opindex mcirrus-fix-invalid-insns
7589 @opindex mno-cirrus-fix-invalid-insns
7590 Insert NOPs into the instruction stream to in order to work around
7591 problems with invalid Maverick instruction combinations. This option
7592 is only valid if the @option{-mcpu=ep9312} option has been used to
7593 enable generation of instructions for the Cirrus Maverick floating
7594 point co-processor. This option is not enabled by default, since the
7595 problem is only present in older Maverick implementations. The default
7596 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7599 @item -mpoke-function-name
7600 @opindex mpoke-function-name
7601 Write the name of each function into the text section, directly
7602 preceding the function prologue. The generated code is similar to this:
7606 .ascii "arm_poke_function_name", 0
7609 .word 0xff000000 + (t1 - t0)
7610 arm_poke_function_name
7612 stmfd sp!, @{fp, ip, lr, pc@}
7616 When performing a stack backtrace, code can inspect the value of
7617 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7618 location @code{pc - 12} and the top 8 bits are set, then we know that
7619 there is a function name embedded immediately preceding this location
7620 and has length @code{((pc[-3]) & 0xff000000)}.
7624 Generate code for the 16-bit Thumb instruction set. The default is to
7625 use the 32-bit ARM instruction set.
7628 @opindex mtpcs-frame
7629 Generate a stack frame that is compliant with the Thumb Procedure Call
7630 Standard for all non-leaf functions. (A leaf function is one that does
7631 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7633 @item -mtpcs-leaf-frame
7634 @opindex mtpcs-leaf-frame
7635 Generate a stack frame that is compliant with the Thumb Procedure Call
7636 Standard for all leaf functions. (A leaf function is one that does
7637 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7639 @item -mcallee-super-interworking
7640 @opindex mcallee-super-interworking
7641 Gives all externally visible functions in the file being compiled an ARM
7642 instruction set header which switches to Thumb mode before executing the
7643 rest of the function. This allows these functions to be called from
7644 non-interworking code.
7646 @item -mcaller-super-interworking
7647 @opindex mcaller-super-interworking
7648 Allows calls via function pointers (including virtual functions) to
7649 execute correctly regardless of whether the target code has been
7650 compiled for interworking or not. There is a small overhead in the cost
7651 of executing a function pointer if this option is enabled.
7653 @item -mtp=@var{name}
7655 Specify the access model for the thread local storage pointer. The valid
7656 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7657 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7658 (supported in the arm6k architecture), and @option{auto}, which uses the
7659 best available method for the selected processor. The default setting is
7665 @subsection AVR Options
7668 These options are defined for AVR implementations:
7671 @item -mmcu=@var{mcu}
7673 Specify ATMEL AVR instruction set or MCU type.
7675 Instruction set avr1 is for the minimal AVR core, not supported by the C
7676 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7677 attiny11, attiny12, attiny15, attiny28).
7679 Instruction set avr2 (default) is for the classic AVR core with up to
7680 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7681 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7682 at90c8534, at90s8535).
7684 Instruction set avr3 is for the classic AVR core with up to 128K program
7685 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7687 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7688 memory space (MCU types: atmega8, atmega83, atmega85).
7690 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7691 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7692 atmega64, atmega128, at43usb355, at94k).
7696 Output instruction sizes to the asm file.
7698 @item -minit-stack=@var{N}
7699 @opindex minit-stack
7700 Specify the initial stack address, which may be a symbol or numeric value,
7701 @samp{__stack} is the default.
7703 @item -mno-interrupts
7704 @opindex mno-interrupts
7705 Generated code is not compatible with hardware interrupts.
7706 Code size will be smaller.
7708 @item -mcall-prologues
7709 @opindex mcall-prologues
7710 Functions prologues/epilogues expanded as call to appropriate
7711 subroutines. Code size will be smaller.
7713 @item -mno-tablejump
7714 @opindex mno-tablejump
7715 Do not generate tablejump insns which sometimes increase code size.
7718 @opindex mtiny-stack
7719 Change only the low 8 bits of the stack pointer.
7723 Assume int to be 8 bit integer. This affects the sizes of all types: A
7724 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7725 and long long will be 4 bytes. Please note that this option does not
7726 comply to the C standards, but it will provide you with smaller code
7730 @node Blackfin Options
7731 @subsection Blackfin Options
7732 @cindex Blackfin Options
7735 @item -momit-leaf-frame-pointer
7736 @opindex momit-leaf-frame-pointer
7737 Don't keep the frame pointer in a register for leaf functions. This
7738 avoids the instructions to save, set up and restore frame pointers and
7739 makes an extra register available in leaf functions. The option
7740 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7741 which might make debugging harder.
7743 @item -mspecld-anomaly
7744 @opindex mspecld-anomaly
7745 When enabled, the compiler will ensure that the generated code does not
7746 contain speculative loads after jump instructions. This option is enabled
7749 @item -mno-specld-anomaly
7750 @opindex mno-specld-anomaly
7751 Don't generate extra code to prevent speculative loads from occurring.
7753 @item -mcsync-anomaly
7754 @opindex mcsync-anomaly
7755 When enabled, the compiler will ensure that the generated code does not
7756 contain CSYNC or SSYNC instructions too soon after conditional branches.
7757 This option is enabled by default.
7759 @item -mno-csync-anomaly
7760 @opindex mno-csync-anomaly
7761 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7762 occurring too soon after a conditional branch.
7766 When enabled, the compiler is free to take advantage of the knowledge that
7767 the entire program fits into the low 64k of memory.
7770 @opindex mno-low-64k
7771 Assume that the program is arbitrarily large. This is the default.
7773 @item -mid-shared-library
7774 @opindex mid-shared-library
7775 Generate code that supports shared libraries via the library ID method.
7776 This allows for execute in place and shared libraries in an environment
7777 without virtual memory management. This option implies @option{-fPIC}.
7779 @item -mno-id-shared-library
7780 @opindex mno-id-shared-library
7781 Generate code that doesn't assume ID based shared libraries are being used.
7782 This is the default.
7784 @item -mshared-library-id=n
7785 @opindex mshared-library-id
7786 Specified the identification number of the ID based shared library being
7787 compiled. Specifying a value of 0 will generate more compact code, specifying
7788 other values will force the allocation of that number to the current
7789 library but is no more space or time efficient than omitting this option.
7792 @itemx -mno-long-calls
7793 @opindex mlong-calls
7794 @opindex mno-long-calls
7795 Tells the compiler to perform function calls by first loading the
7796 address of the function into a register and then performing a subroutine
7797 call on this register. This switch is needed if the target function
7798 will lie outside of the 24 bit addressing range of the offset based
7799 version of subroutine call instruction.
7801 This feature is not enabled by default. Specifying
7802 @option{-mno-long-calls} will restore the default behavior. Note these
7803 switches have no effect on how the compiler generates code to handle
7804 function calls via function pointers.
7808 @subsection CRIS Options
7809 @cindex CRIS Options
7811 These options are defined specifically for the CRIS ports.
7814 @item -march=@var{architecture-type}
7815 @itemx -mcpu=@var{architecture-type}
7818 Generate code for the specified architecture. The choices for
7819 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7820 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7821 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7824 @item -mtune=@var{architecture-type}
7826 Tune to @var{architecture-type} everything applicable about the generated
7827 code, except for the ABI and the set of available instructions. The
7828 choices for @var{architecture-type} are the same as for
7829 @option{-march=@var{architecture-type}}.
7831 @item -mmax-stack-frame=@var{n}
7832 @opindex mmax-stack-frame
7833 Warn when the stack frame of a function exceeds @var{n} bytes.
7835 @item -melinux-stacksize=@var{n}
7836 @opindex melinux-stacksize
7837 Only available with the @samp{cris-axis-aout} target. Arranges for
7838 indications in the program to the kernel loader that the stack of the
7839 program should be set to @var{n} bytes.
7845 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7846 @option{-march=v3} and @option{-march=v8} respectively.
7848 @item -mmul-bug-workaround
7849 @itemx -mno-mul-bug-workaround
7850 @opindex mmul-bug-workaround
7851 @opindex mno-mul-bug-workaround
7852 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7853 models where it applies. This option is active by default.
7857 Enable CRIS-specific verbose debug-related information in the assembly
7858 code. This option also has the effect to turn off the @samp{#NO_APP}
7859 formatted-code indicator to the assembler at the beginning of the
7864 Do not use condition-code results from previous instruction; always emit
7865 compare and test instructions before use of condition codes.
7867 @item -mno-side-effects
7868 @opindex mno-side-effects
7869 Do not emit instructions with side-effects in addressing modes other than
7873 @itemx -mno-stack-align
7875 @itemx -mno-data-align
7876 @itemx -mconst-align
7877 @itemx -mno-const-align
7878 @opindex mstack-align
7879 @opindex mno-stack-align
7880 @opindex mdata-align
7881 @opindex mno-data-align
7882 @opindex mconst-align
7883 @opindex mno-const-align
7884 These options (no-options) arranges (eliminate arrangements) for the
7885 stack-frame, individual data and constants to be aligned for the maximum
7886 single data access size for the chosen CPU model. The default is to
7887 arrange for 32-bit alignment. ABI details such as structure layout are
7888 not affected by these options.
7896 Similar to the stack- data- and const-align options above, these options
7897 arrange for stack-frame, writable data and constants to all be 32-bit,
7898 16-bit or 8-bit aligned. The default is 32-bit alignment.
7900 @item -mno-prologue-epilogue
7901 @itemx -mprologue-epilogue
7902 @opindex mno-prologue-epilogue
7903 @opindex mprologue-epilogue
7904 With @option{-mno-prologue-epilogue}, the normal function prologue and
7905 epilogue that sets up the stack-frame are omitted and no return
7906 instructions or return sequences are generated in the code. Use this
7907 option only together with visual inspection of the compiled code: no
7908 warnings or errors are generated when call-saved registers must be saved,
7909 or storage for local variable needs to be allocated.
7915 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7916 instruction sequences that load addresses for functions from the PLT part
7917 of the GOT rather than (traditional on other architectures) calls to the
7918 PLT@. The default is @option{-mgotplt}.
7922 Legacy no-op option only recognized with the cris-axis-aout target.
7926 Legacy no-op option only recognized with the cris-axis-elf and
7927 cris-axis-linux-gnu targets.
7931 Only recognized with the cris-axis-aout target, where it selects a
7932 GNU/linux-like multilib, include files and instruction set for
7937 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7941 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7942 to link with input-output functions from a simulator library. Code,
7943 initialized data and zero-initialized data are allocated consecutively.
7947 Like @option{-sim}, but pass linker options to locate initialized data at
7948 0x40000000 and zero-initialized data at 0x80000000.
7952 @subsection CRX Options
7955 These options are defined specifically for the CRX ports.
7961 Enable the use of multiply-accumulate instructions. Disabled by default.
7965 Push instructions will be used to pass outgoing arguments when functions
7966 are called. Enabled by default.
7969 @node Darwin Options
7970 @subsection Darwin Options
7971 @cindex Darwin options
7973 These options are defined for all architectures running the Darwin operating
7976 FSF GCC on Darwin does not create ``fat'' object files; it will create
7977 an object file for the single architecture that it was built to
7978 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7979 @option{-arch} options are used; it does so by running the compiler or
7980 linker multiple times and joining the results together with
7983 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7984 @samp{i686}) is determined by the flags that specify the ISA
7985 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7986 @option{-force_cpusubtype_ALL} option can be used to override this.
7988 The Darwin tools vary in their behavior when presented with an ISA
7989 mismatch. The assembler, @file{as}, will only permit instructions to
7990 be used that are valid for the subtype of the file it is generating,
7991 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7992 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7993 and print an error if asked to create a shared library with a less
7994 restrictive subtype than its input files (for instance, trying to put
7995 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7996 for executables, @file{ld}, will quietly give the executable the most
7997 restrictive subtype of any of its input files.
8002 Add the framework directory @var{dir} to the head of the list of
8003 directories to be searched for header files. These directories are
8004 interleaved with those specified by @option{-I} options and are
8005 scanned in a left-to-right order.
8007 A framework directory is a directory with frameworks in it. A
8008 framework is a directory with a @samp{"Headers"} and/or
8009 @samp{"PrivateHeaders"} directory contained directly in it that ends
8010 in @samp{".framework"}. The name of a framework is the name of this
8011 directory excluding the @samp{".framework"}. Headers associated with
8012 the framework are found in one of those two directories, with
8013 @samp{"Headers"} being searched first. A subframework is a framework
8014 directory that is in a framework's @samp{"Frameworks"} directory.
8015 Includes of subframework headers can only appear in a header of a
8016 framework that contains the subframework, or in a sibling subframework
8017 header. Two subframeworks are siblings if they occur in the same
8018 framework. A subframework should not have the same name as a
8019 framework, a warning will be issued if this is violated. Currently a
8020 subframework cannot have subframeworks, in the future, the mechanism
8021 may be extended to support this. The standard frameworks can be found
8022 in @samp{"/System/Library/Frameworks"} and
8023 @samp{"/Library/Frameworks"}. An example include looks like
8024 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8025 the name of the framework and header.h is found in the
8026 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8030 Emit debugging information for symbols that are used. For STABS
8031 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8032 This is by default ON@.
8036 Emit debugging information for all symbols and types.
8038 @item -mmacosx-version-min=@var{version}
8039 The earliest version of MacOS X that this executable will run on
8040 is @var{version}. Typical values of @var{version} include @code{10.1},
8041 @code{10.2}, and @code{10.3.9}.
8043 The default for this option is to make choices that seem to be most
8046 @item -mone-byte-bool
8047 @opindex -mone-byte-bool
8048 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8049 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8050 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8051 option has no effect on x86.
8053 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8054 to generate code that is not binary compatible with code generated
8055 without that switch. Using this switch may require recompiling all
8056 other modules in a program, including system libraries. Use this
8057 switch to conform to a non-default data model.
8059 @item -mfix-and-continue
8060 @itemx -ffix-and-continue
8061 @itemx -findirect-data
8062 @opindex mfix-and-continue
8063 @opindex ffix-and-continue
8064 @opindex findirect-data
8065 Generate code suitable for fast turn around development. Needed to
8066 enable gdb to dynamically load @code{.o} files into already running
8067 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8068 are provided for backwards compatibility.
8072 Loads all members of static archive libraries.
8073 See man ld(1) for more information.
8075 @item -arch_errors_fatal
8076 @opindex arch_errors_fatal
8077 Cause the errors having to do with files that have the wrong architecture
8081 @opindex bind_at_load
8082 Causes the output file to be marked such that the dynamic linker will
8083 bind all undefined references when the file is loaded or launched.
8087 Produce a Mach-o bundle format file.
8088 See man ld(1) for more information.
8090 @item -bundle_loader @var{executable}
8091 @opindex bundle_loader
8092 This option specifies the @var{executable} that will be loading the build
8093 output file being linked. See man ld(1) for more information.
8096 @opindex -dynamiclib
8097 When passed this option, GCC will produce a dynamic library instead of
8098 an executable when linking, using the Darwin @file{libtool} command.
8100 @item -force_cpusubtype_ALL
8101 @opindex -force_cpusubtype_ALL
8102 This causes GCC's output file to have the @var{ALL} subtype, instead of
8103 one controlled by the @option{-mcpu} or @option{-march} option.
8105 @item -allowable_client @var{client_name}
8107 @itemx -compatibility_version
8108 @itemx -current_version
8110 @itemx -dependency-file
8112 @itemx -dylinker_install_name
8114 @itemx -exported_symbols_list
8116 @itemx -flat_namespace
8117 @itemx -force_flat_namespace
8118 @itemx -headerpad_max_install_names
8121 @itemx -install_name
8122 @itemx -keep_private_externs
8123 @itemx -multi_module
8124 @itemx -multiply_defined
8125 @itemx -multiply_defined_unused
8127 @itemx -no_dead_strip_inits_and_terms
8128 @itemx -nofixprebinding
8131 @itemx -noseglinkedit
8132 @itemx -pagezero_size
8134 @itemx -prebind_all_twolevel_modules
8135 @itemx -private_bundle
8136 @itemx -read_only_relocs
8138 @itemx -sectobjectsymbols
8142 @itemx -sectobjectsymbols
8145 @itemx -segs_read_only_addr
8146 @itemx -segs_read_write_addr
8147 @itemx -seg_addr_table
8148 @itemx -seg_addr_table_filename
8151 @itemx -segs_read_only_addr
8152 @itemx -segs_read_write_addr
8153 @itemx -single_module
8156 @itemx -sub_umbrella
8157 @itemx -twolevel_namespace
8160 @itemx -unexported_symbols_list
8161 @itemx -weak_reference_mismatches
8164 @opindex allowable_client
8165 @opindex client_name
8166 @opindex compatibility_version
8167 @opindex current_version
8169 @opindex dependency-file
8171 @opindex dylinker_install_name
8173 @opindex exported_symbols_list
8175 @opindex flat_namespace
8176 @opindex force_flat_namespace
8177 @opindex headerpad_max_install_names
8180 @opindex install_name
8181 @opindex keep_private_externs
8182 @opindex multi_module
8183 @opindex multiply_defined
8184 @opindex multiply_defined_unused
8186 @opindex no_dead_strip_inits_and_terms
8187 @opindex nofixprebinding
8188 @opindex nomultidefs
8190 @opindex noseglinkedit
8191 @opindex pagezero_size
8193 @opindex prebind_all_twolevel_modules
8194 @opindex private_bundle
8195 @opindex read_only_relocs
8197 @opindex sectobjectsymbols
8201 @opindex sectobjectsymbols
8204 @opindex segs_read_only_addr
8205 @opindex segs_read_write_addr
8206 @opindex seg_addr_table
8207 @opindex seg_addr_table_filename
8208 @opindex seglinkedit
8210 @opindex segs_read_only_addr
8211 @opindex segs_read_write_addr
8212 @opindex single_module
8214 @opindex sub_library
8215 @opindex sub_umbrella
8216 @opindex twolevel_namespace
8219 @opindex unexported_symbols_list
8220 @opindex weak_reference_mismatches
8221 @opindex whatsloaded
8223 These options are passed to the Darwin linker. The Darwin linker man page
8224 describes them in detail.
8227 @node DEC Alpha Options
8228 @subsection DEC Alpha Options
8230 These @samp{-m} options are defined for the DEC Alpha implementations:
8233 @item -mno-soft-float
8235 @opindex mno-soft-float
8236 @opindex msoft-float
8237 Use (do not use) the hardware floating-point instructions for
8238 floating-point operations. When @option{-msoft-float} is specified,
8239 functions in @file{libgcc.a} will be used to perform floating-point
8240 operations. Unless they are replaced by routines that emulate the
8241 floating-point operations, or compiled in such a way as to call such
8242 emulations routines, these routines will issue floating-point
8243 operations. If you are compiling for an Alpha without floating-point
8244 operations, you must ensure that the library is built so as not to call
8247 Note that Alpha implementations without floating-point operations are
8248 required to have floating-point registers.
8253 @opindex mno-fp-regs
8254 Generate code that uses (does not use) the floating-point register set.
8255 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8256 register set is not used, floating point operands are passed in integer
8257 registers as if they were integers and floating-point results are passed
8258 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8259 so any function with a floating-point argument or return value called by code
8260 compiled with @option{-mno-fp-regs} must also be compiled with that
8263 A typical use of this option is building a kernel that does not use,
8264 and hence need not save and restore, any floating-point registers.
8268 The Alpha architecture implements floating-point hardware optimized for
8269 maximum performance. It is mostly compliant with the IEEE floating
8270 point standard. However, for full compliance, software assistance is
8271 required. This option generates code fully IEEE compliant code
8272 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8273 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8274 defined during compilation. The resulting code is less efficient but is
8275 able to correctly support denormalized numbers and exceptional IEEE
8276 values such as not-a-number and plus/minus infinity. Other Alpha
8277 compilers call this option @option{-ieee_with_no_inexact}.
8279 @item -mieee-with-inexact
8280 @opindex mieee-with-inexact
8281 This is like @option{-mieee} except the generated code also maintains
8282 the IEEE @var{inexact-flag}. Turning on this option causes the
8283 generated code to implement fully-compliant IEEE math. In addition to
8284 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8285 macro. On some Alpha implementations the resulting code may execute
8286 significantly slower than the code generated by default. Since there is
8287 very little code that depends on the @var{inexact-flag}, you should
8288 normally not specify this option. Other Alpha compilers call this
8289 option @option{-ieee_with_inexact}.
8291 @item -mfp-trap-mode=@var{trap-mode}
8292 @opindex mfp-trap-mode
8293 This option controls what floating-point related traps are enabled.
8294 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8295 The trap mode can be set to one of four values:
8299 This is the default (normal) setting. The only traps that are enabled
8300 are the ones that cannot be disabled in software (e.g., division by zero
8304 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8308 Like @samp{su}, but the instructions are marked to be safe for software
8309 completion (see Alpha architecture manual for details).
8312 Like @samp{su}, but inexact traps are enabled as well.
8315 @item -mfp-rounding-mode=@var{rounding-mode}
8316 @opindex mfp-rounding-mode
8317 Selects the IEEE rounding mode. Other Alpha compilers call this option
8318 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8323 Normal IEEE rounding mode. Floating point numbers are rounded towards
8324 the nearest machine number or towards the even machine number in case
8328 Round towards minus infinity.
8331 Chopped rounding mode. Floating point numbers are rounded towards zero.
8334 Dynamic rounding mode. A field in the floating point control register
8335 (@var{fpcr}, see Alpha architecture reference manual) controls the
8336 rounding mode in effect. The C library initializes this register for
8337 rounding towards plus infinity. Thus, unless your program modifies the
8338 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8341 @item -mtrap-precision=@var{trap-precision}
8342 @opindex mtrap-precision
8343 In the Alpha architecture, floating point traps are imprecise. This
8344 means without software assistance it is impossible to recover from a
8345 floating trap and program execution normally needs to be terminated.
8346 GCC can generate code that can assist operating system trap handlers
8347 in determining the exact location that caused a floating point trap.
8348 Depending on the requirements of an application, different levels of
8349 precisions can be selected:
8353 Program precision. This option is the default and means a trap handler
8354 can only identify which program caused a floating point exception.
8357 Function precision. The trap handler can determine the function that
8358 caused a floating point exception.
8361 Instruction precision. The trap handler can determine the exact
8362 instruction that caused a floating point exception.
8365 Other Alpha compilers provide the equivalent options called
8366 @option{-scope_safe} and @option{-resumption_safe}.
8368 @item -mieee-conformant
8369 @opindex mieee-conformant
8370 This option marks the generated code as IEEE conformant. You must not
8371 use this option unless you also specify @option{-mtrap-precision=i} and either
8372 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8373 is to emit the line @samp{.eflag 48} in the function prologue of the
8374 generated assembly file. Under DEC Unix, this has the effect that
8375 IEEE-conformant math library routines will be linked in.
8377 @item -mbuild-constants
8378 @opindex mbuild-constants
8379 Normally GCC examines a 32- or 64-bit integer constant to
8380 see if it can construct it from smaller constants in two or three
8381 instructions. If it cannot, it will output the constant as a literal and
8382 generate code to load it from the data segment at runtime.
8384 Use this option to require GCC to construct @emph{all} integer constants
8385 using code, even if it takes more instructions (the maximum is six).
8387 You would typically use this option to build a shared library dynamic
8388 loader. Itself a shared library, it must relocate itself in memory
8389 before it can find the variables and constants in its own data segment.
8395 Select whether to generate code to be assembled by the vendor-supplied
8396 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8414 Indicate whether GCC should generate code to use the optional BWX,
8415 CIX, FIX and MAX instruction sets. The default is to use the instruction
8416 sets supported by the CPU type specified via @option{-mcpu=} option or that
8417 of the CPU on which GCC was built if none was specified.
8422 @opindex mfloat-ieee
8423 Generate code that uses (does not use) VAX F and G floating point
8424 arithmetic instead of IEEE single and double precision.
8426 @item -mexplicit-relocs
8427 @itemx -mno-explicit-relocs
8428 @opindex mexplicit-relocs
8429 @opindex mno-explicit-relocs
8430 Older Alpha assemblers provided no way to generate symbol relocations
8431 except via assembler macros. Use of these macros does not allow
8432 optimal instruction scheduling. GNU binutils as of version 2.12
8433 supports a new syntax that allows the compiler to explicitly mark
8434 which relocations should apply to which instructions. This option
8435 is mostly useful for debugging, as GCC detects the capabilities of
8436 the assembler when it is built and sets the default accordingly.
8440 @opindex msmall-data
8441 @opindex mlarge-data
8442 When @option{-mexplicit-relocs} is in effect, static data is
8443 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8444 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8445 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8446 16-bit relocations off of the @code{$gp} register. This limits the
8447 size of the small data area to 64KB, but allows the variables to be
8448 directly accessed via a single instruction.
8450 The default is @option{-mlarge-data}. With this option the data area
8451 is limited to just below 2GB@. Programs that require more than 2GB of
8452 data must use @code{malloc} or @code{mmap} to allocate the data in the
8453 heap instead of in the program's data segment.
8455 When generating code for shared libraries, @option{-fpic} implies
8456 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8460 @opindex msmall-text
8461 @opindex mlarge-text
8462 When @option{-msmall-text} is used, the compiler assumes that the
8463 code of the entire program (or shared library) fits in 4MB, and is
8464 thus reachable with a branch instruction. When @option{-msmall-data}
8465 is used, the compiler can assume that all local symbols share the
8466 same @code{$gp} value, and thus reduce the number of instructions
8467 required for a function call from 4 to 1.
8469 The default is @option{-mlarge-text}.
8471 @item -mcpu=@var{cpu_type}
8473 Set the instruction set and instruction scheduling parameters for
8474 machine type @var{cpu_type}. You can specify either the @samp{EV}
8475 style name or the corresponding chip number. GCC supports scheduling
8476 parameters for the EV4, EV5 and EV6 family of processors and will
8477 choose the default values for the instruction set from the processor
8478 you specify. If you do not specify a processor type, GCC will default
8479 to the processor on which the compiler was built.
8481 Supported values for @var{cpu_type} are
8487 Schedules as an EV4 and has no instruction set extensions.
8491 Schedules as an EV5 and has no instruction set extensions.
8495 Schedules as an EV5 and supports the BWX extension.
8500 Schedules as an EV5 and supports the BWX and MAX extensions.
8504 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8508 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8511 @item -mtune=@var{cpu_type}
8513 Set only the instruction scheduling parameters for machine type
8514 @var{cpu_type}. The instruction set is not changed.
8516 @item -mmemory-latency=@var{time}
8517 @opindex mmemory-latency
8518 Sets the latency the scheduler should assume for typical memory
8519 references as seen by the application. This number is highly
8520 dependent on the memory access patterns used by the application
8521 and the size of the external cache on the machine.
8523 Valid options for @var{time} are
8527 A decimal number representing clock cycles.
8533 The compiler contains estimates of the number of clock cycles for
8534 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8535 (also called Dcache, Scache, and Bcache), as well as to main memory.
8536 Note that L3 is only valid for EV5.
8541 @node DEC Alpha/VMS Options
8542 @subsection DEC Alpha/VMS Options
8544 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8547 @item -mvms-return-codes
8548 @opindex mvms-return-codes
8549 Return VMS condition codes from main. The default is to return POSIX
8550 style condition (e.g.@ error) codes.
8554 @subsection FRV Options
8561 Only use the first 32 general purpose registers.
8566 Use all 64 general purpose registers.
8571 Use only the first 32 floating point registers.
8576 Use all 64 floating point registers
8579 @opindex mhard-float
8581 Use hardware instructions for floating point operations.
8584 @opindex msoft-float
8586 Use library routines for floating point operations.
8591 Dynamically allocate condition code registers.
8596 Do not try to dynamically allocate condition code registers, only
8597 use @code{icc0} and @code{fcc0}.
8602 Change ABI to use double word insns.
8607 Do not use double word instructions.
8612 Use floating point double instructions.
8617 Do not use floating point double instructions.
8622 Use media instructions.
8627 Do not use media instructions.
8632 Use multiply and add/subtract instructions.
8637 Do not use multiply and add/subtract instructions.
8642 Select the FDPIC ABI, that uses function descriptors to represent
8643 pointers to functions. Without any PIC/PIE-related options, it
8644 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8645 assumes GOT entries and small data are within a 12-bit range from the
8646 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8647 are computed with 32 bits.
8650 @opindex minline-plt
8652 Enable inlining of PLT entries in function calls to functions that are
8653 not known to bind locally. It has no effect without @option{-mfdpic}.
8654 It's enabled by default if optimizing for speed and compiling for
8655 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8656 optimization option such as @option{-O3} or above is present in the
8662 Assume a large TLS segment when generating thread-local code.
8667 Do not assume a large TLS segment when generating thread-local code.
8672 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8673 that is known to be in read-only sections. It's enabled by default,
8674 except for @option{-fpic} or @option{-fpie}: even though it may help
8675 make the global offset table smaller, it trades 1 instruction for 4.
8676 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8677 one of which may be shared by multiple symbols, and it avoids the need
8678 for a GOT entry for the referenced symbol, so it's more likely to be a
8679 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8681 @item -multilib-library-pic
8682 @opindex multilib-library-pic
8684 Link with the (library, not FD) pic libraries. It's implied by
8685 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8686 @option{-fpic} without @option{-mfdpic}. You should never have to use
8692 Follow the EABI requirement of always creating a frame pointer whenever
8693 a stack frame is allocated. This option is enabled by default and can
8694 be disabled with @option{-mno-linked-fp}.
8697 @opindex mlong-calls
8699 Use indirect addressing to call functions outside the current
8700 compilation unit. This allows the functions to be placed anywhere
8701 within the 32-bit address space.
8703 @item -malign-labels
8704 @opindex malign-labels
8706 Try to align labels to an 8-byte boundary by inserting nops into the
8707 previous packet. This option only has an effect when VLIW packing
8708 is enabled. It doesn't create new packets; it merely adds nops to
8712 @opindex mlibrary-pic
8714 Generate position-independent EABI code.
8719 Use only the first four media accumulator registers.
8724 Use all eight media accumulator registers.
8729 Pack VLIW instructions.
8734 Do not pack VLIW instructions.
8739 Do not mark ABI switches in e_flags.
8744 Enable the use of conditional-move instructions (default).
8746 This switch is mainly for debugging the compiler and will likely be removed
8747 in a future version.
8749 @item -mno-cond-move
8750 @opindex mno-cond-move
8752 Disable the use of conditional-move instructions.
8754 This switch is mainly for debugging the compiler and will likely be removed
8755 in a future version.
8760 Enable the use of conditional set instructions (default).
8762 This switch is mainly for debugging the compiler and will likely be removed
8763 in a future version.
8768 Disable the use of conditional set instructions.
8770 This switch is mainly for debugging the compiler and will likely be removed
8771 in a future version.
8776 Enable the use of conditional execution (default).
8778 This switch is mainly for debugging the compiler and will likely be removed
8779 in a future version.
8781 @item -mno-cond-exec
8782 @opindex mno-cond-exec
8784 Disable the use of conditional execution.
8786 This switch is mainly for debugging the compiler and will likely be removed
8787 in a future version.
8790 @opindex mvliw-branch
8792 Run a pass to pack branches into VLIW instructions (default).
8794 This switch is mainly for debugging the compiler and will likely be removed
8795 in a future version.
8797 @item -mno-vliw-branch
8798 @opindex mno-vliw-branch
8800 Do not run a pass to pack branches into VLIW instructions.
8802 This switch is mainly for debugging the compiler and will likely be removed
8803 in a future version.
8805 @item -mmulti-cond-exec
8806 @opindex mmulti-cond-exec
8808 Enable optimization of @code{&&} and @code{||} in conditional execution
8811 This switch is mainly for debugging the compiler and will likely be removed
8812 in a future version.
8814 @item -mno-multi-cond-exec
8815 @opindex mno-multi-cond-exec
8817 Disable optimization of @code{&&} and @code{||} in conditional execution.
8819 This switch is mainly for debugging the compiler and will likely be removed
8820 in a future version.
8822 @item -mnested-cond-exec
8823 @opindex mnested-cond-exec
8825 Enable nested conditional execution optimizations (default).
8827 This switch is mainly for debugging the compiler and will likely be removed
8828 in a future version.
8830 @item -mno-nested-cond-exec
8831 @opindex mno-nested-cond-exec
8833 Disable nested conditional execution optimizations.
8835 This switch is mainly for debugging the compiler and will likely be removed
8836 in a future version.
8838 @item -moptimize-membar
8839 @opindex moptimize-membar
8841 This switch removes redundant @code{membar} instructions from the
8842 compiler generated code. It is enabled by default.
8844 @item -mno-optimize-membar
8845 @opindex mno-optimize-membar
8847 This switch disables the automatic removal of redundant @code{membar}
8848 instructions from the generated code.
8850 @item -mtomcat-stats
8851 @opindex mtomcat-stats
8853 Cause gas to print out tomcat statistics.
8855 @item -mcpu=@var{cpu}
8858 Select the processor type for which to generate code. Possible values are
8859 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8860 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8864 @node GNU/Linux Options
8865 @subsection GNU/Linux Options
8867 These @samp{-m} options are defined for GNU/Linux targets:
8872 Use the GNU C library instead of uClibc. This is the default except
8873 on @samp{*-*-linux-*uclibc*} targets.
8877 Use uClibc instead of the GNU C library. This is the default on
8878 @samp{*-*-linux-*uclibc*} targets.
8881 @node H8/300 Options
8882 @subsection H8/300 Options
8884 These @samp{-m} options are defined for the H8/300 implementations:
8889 Shorten some address references at link time, when possible; uses the
8890 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8891 ld, Using ld}, for a fuller description.
8895 Generate code for the H8/300H@.
8899 Generate code for the H8S@.
8903 Generate code for the H8S and H8/300H in the normal mode. This switch
8904 must be used either with @option{-mh} or @option{-ms}.
8908 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8912 Make @code{int} data 32 bits by default.
8916 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8917 The default for the H8/300H and H8S is to align longs and floats on 4
8919 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8920 This option has no effect on the H8/300.
8924 @subsection HPPA Options
8925 @cindex HPPA Options
8927 These @samp{-m} options are defined for the HPPA family of computers:
8930 @item -march=@var{architecture-type}
8932 Generate code for the specified architecture. The choices for
8933 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8934 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8935 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8936 architecture option for your machine. Code compiled for lower numbered
8937 architectures will run on higher numbered architectures, but not the
8941 @itemx -mpa-risc-1-1
8942 @itemx -mpa-risc-2-0
8943 @opindex mpa-risc-1-0
8944 @opindex mpa-risc-1-1
8945 @opindex mpa-risc-2-0
8946 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8949 @opindex mbig-switch
8950 Generate code suitable for big switch tables. Use this option only if
8951 the assembler/linker complain about out of range branches within a switch
8954 @item -mjump-in-delay
8955 @opindex mjump-in-delay
8956 Fill delay slots of function calls with unconditional jump instructions
8957 by modifying the return pointer for the function call to be the target
8958 of the conditional jump.
8960 @item -mdisable-fpregs
8961 @opindex mdisable-fpregs
8962 Prevent floating point registers from being used in any manner. This is
8963 necessary for compiling kernels which perform lazy context switching of
8964 floating point registers. If you use this option and attempt to perform
8965 floating point operations, the compiler will abort.
8967 @item -mdisable-indexing
8968 @opindex mdisable-indexing
8969 Prevent the compiler from using indexing address modes. This avoids some
8970 rather obscure problems when compiling MIG generated code under MACH@.
8972 @item -mno-space-regs
8973 @opindex mno-space-regs
8974 Generate code that assumes the target has no space registers. This allows
8975 GCC to generate faster indirect calls and use unscaled index address modes.
8977 Such code is suitable for level 0 PA systems and kernels.
8979 @item -mfast-indirect-calls
8980 @opindex mfast-indirect-calls
8981 Generate code that assumes calls never cross space boundaries. This
8982 allows GCC to emit code which performs faster indirect calls.
8984 This option will not work in the presence of shared libraries or nested
8987 @item -mfixed-range=@var{register-range}
8988 @opindex mfixed-range
8989 Generate code treating the given register range as fixed registers.
8990 A fixed register is one that the register allocator can not use. This is
8991 useful when compiling kernel code. A register range is specified as
8992 two registers separated by a dash. Multiple register ranges can be
8993 specified separated by a comma.
8995 @item -mlong-load-store
8996 @opindex mlong-load-store
8997 Generate 3-instruction load and store sequences as sometimes required by
8998 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9001 @item -mportable-runtime
9002 @opindex mportable-runtime
9003 Use the portable calling conventions proposed by HP for ELF systems.
9007 Enable the use of assembler directives only GAS understands.
9009 @item -mschedule=@var{cpu-type}
9011 Schedule code according to the constraints for the machine type
9012 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9013 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9014 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9015 proper scheduling option for your machine. The default scheduling is
9019 @opindex mlinker-opt
9020 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9021 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9022 linkers in which they give bogus error messages when linking some programs.
9025 @opindex msoft-float
9026 Generate output containing library calls for floating point.
9027 @strong{Warning:} the requisite libraries are not available for all HPPA
9028 targets. Normally the facilities of the machine's usual C compiler are
9029 used, but this cannot be done directly in cross-compilation. You must make
9030 your own arrangements to provide suitable library functions for
9031 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9032 does provide software floating point support.
9034 @option{-msoft-float} changes the calling convention in the output file;
9035 therefore, it is only useful if you compile @emph{all} of a program with
9036 this option. In particular, you need to compile @file{libgcc.a}, the
9037 library that comes with GCC, with @option{-msoft-float} in order for
9042 Generate the predefine, @code{_SIO}, for server IO@. The default is
9043 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9044 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9045 options are available under HP-UX and HI-UX@.
9049 Use GNU ld specific options. This passes @option{-shared} to ld when
9050 building a shared library. It is the default when GCC is configured,
9051 explicitly or implicitly, with the GNU linker. This option does not
9052 have any affect on which ld is called, it only changes what parameters
9053 are passed to that ld. The ld that is called is determined by the
9054 @option{--with-ld} configure option, GCC's program search path, and
9055 finally by the user's @env{PATH}. The linker used by GCC can be printed
9056 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9057 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9061 Use HP ld specific options. This passes @option{-b} to ld when building
9062 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9063 links. It is the default when GCC is configured, explicitly or
9064 implicitly, with the HP linker. This option does not have any affect on
9065 which ld is called, it only changes what parameters are passed to that
9066 ld. The ld that is called is determined by the @option{--with-ld}
9067 configure option, GCC's program search path, and finally by the user's
9068 @env{PATH}. The linker used by GCC can be printed using @samp{which
9069 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9070 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9073 @opindex mno-long-calls
9074 Generate code that uses long call sequences. This ensures that a call
9075 is always able to reach linker generated stubs. The default is to generate
9076 long calls only when the distance from the call site to the beginning
9077 of the function or translation unit, as the case may be, exceeds a
9078 predefined limit set by the branch type being used. The limits for
9079 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9080 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9083 Distances are measured from the beginning of functions when using the
9084 @option{-ffunction-sections} option, or when using the @option{-mgas}
9085 and @option{-mno-portable-runtime} options together under HP-UX with
9088 It is normally not desirable to use this option as it will degrade
9089 performance. However, it may be useful in large applications,
9090 particularly when partial linking is used to build the application.
9092 The types of long calls used depends on the capabilities of the
9093 assembler and linker, and the type of code being generated. The
9094 impact on systems that support long absolute calls, and long pic
9095 symbol-difference or pc-relative calls should be relatively small.
9096 However, an indirect call is used on 32-bit ELF systems in pic code
9097 and it is quite long.
9099 @item -munix=@var{unix-std}
9101 Generate compiler predefines and select a startfile for the specified
9102 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9103 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9104 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9105 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9106 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9109 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9110 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9111 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9112 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9113 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9114 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9116 It is @emph{important} to note that this option changes the interfaces
9117 for various library routines. It also affects the operational behavior
9118 of the C library. Thus, @emph{extreme} care is needed in using this
9121 Library code that is intended to operate with more than one UNIX
9122 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9123 as appropriate. Most GNU software doesn't provide this capability.
9127 Suppress the generation of link options to search libdld.sl when the
9128 @option{-static} option is specified on HP-UX 10 and later.
9132 The HP-UX implementation of setlocale in libc has a dependency on
9133 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9134 when the @option{-static} option is specified, special link options
9135 are needed to resolve this dependency.
9137 On HP-UX 10 and later, the GCC driver adds the necessary options to
9138 link with libdld.sl when the @option{-static} option is specified.
9139 This causes the resulting binary to be dynamic. On the 64-bit port,
9140 the linkers generate dynamic binaries by default in any case. The
9141 @option{-nolibdld} option can be used to prevent the GCC driver from
9142 adding these link options.
9146 Add support for multithreading with the @dfn{dce thread} library
9147 under HP-UX@. This option sets flags for both the preprocessor and
9151 @node i386 and x86-64 Options
9152 @subsection Intel 386 and AMD x86-64 Options
9153 @cindex i386 Options
9154 @cindex x86-64 Options
9155 @cindex Intel 386 Options
9156 @cindex AMD x86-64 Options
9158 These @samp{-m} options are defined for the i386 and x86-64 family of
9162 @item -mtune=@var{cpu-type}
9164 Tune to @var{cpu-type} everything applicable about the generated code, except
9165 for the ABI and the set of available instructions. The choices for
9169 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9170 If you know the CPU on which your code will run, then you should use
9171 the corresponding @option{-mtune} option instead of
9172 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9173 of your application will have, then you should use this option.
9175 As new processors are deployed in the marketplace, the behavior of this
9176 option will change. Therefore, if you upgrade to a newer version of
9177 GCC, the code generated option will change to reflect the processors
9178 that were most common when that version of GCC was released.
9180 There is no @option{-march=generic} option because @option{-march}
9181 indicates the instruction set the compiler can use, and there is no
9182 generic instruction set applicable to all processors. In contrast,
9183 @option{-mtune} indicates the processor (or, in this case, collection of
9184 processors) for which the code is optimized.
9186 This selects the CPU to tune for at compilation time by determining
9187 the processor type of the compiling machine. Using @option{-mtune=native}
9188 will produce code optimized for the local machine under the constraints
9189 of the selected instruction set. Using @option{-march=native} will
9190 enable all instruction subsets supported by the local machine (hence
9191 the result might not run on different machines).
9193 Original Intel's i386 CPU@.
9195 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9197 Intel Pentium CPU with no MMX support.
9199 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9201 Intel PentiumPro CPU@.
9203 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9204 instruction set will be used, so the code will run on all i686 family chips.
9206 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9207 @item pentium3, pentium3m
9208 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9211 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9212 support. Used by Centrino notebooks.
9213 @item pentium4, pentium4m
9214 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9216 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9219 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9220 SSE2 and SSE3 instruction set support.
9222 AMD K6 CPU with MMX instruction set support.
9224 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9225 @item athlon, athlon-tbird
9226 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9228 @item athlon-4, athlon-xp, athlon-mp
9229 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9230 instruction set support.
9231 @item k8, opteron, athlon64, athlon-fx
9232 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9233 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9235 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9238 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9239 instruction set support.
9241 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9242 implemented for this chip.)
9244 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9245 implemented for this chip.)
9248 While picking a specific @var{cpu-type} will schedule things appropriately
9249 for that particular chip, the compiler will not generate any code that
9250 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9253 @item -march=@var{cpu-type}
9255 Generate instructions for the machine type @var{cpu-type}. The choices
9256 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9257 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9259 @item -mcpu=@var{cpu-type}
9261 A deprecated synonym for @option{-mtune}.
9270 @opindex mpentiumpro
9271 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9272 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9273 These synonyms are deprecated.
9275 @item -mfpmath=@var{unit}
9277 Generate floating point arithmetics for selected unit @var{unit}. The choices
9282 Use the standard 387 floating point coprocessor present majority of chips and
9283 emulated otherwise. Code compiled with this option will run almost everywhere.
9284 The temporary results are computed in 80bit precision instead of precision
9285 specified by the type resulting in slightly different results compared to most
9286 of other chips. See @option{-ffloat-store} for more detailed description.
9288 This is the default choice for i386 compiler.
9291 Use scalar floating point instructions present in the SSE instruction set.
9292 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9293 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9294 instruction set supports only single precision arithmetics, thus the double and
9295 extended precision arithmetics is still done using 387. Later version, present
9296 only in Pentium4 and the future AMD x86-64 chips supports double precision
9299 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9300 or @option{-msse2} switches to enable SSE extensions and make this option
9301 effective. For the x86-64 compiler, these extensions are enabled by default.
9303 The resulting code should be considerably faster in the majority of cases and avoid
9304 the numerical instability problems of 387 code, but may break some existing
9305 code that expects temporaries to be 80bit.
9307 This is the default choice for the x86-64 compiler.
9310 Attempt to utilize both instruction sets at once. This effectively double the
9311 amount of available registers and on chips with separate execution units for
9312 387 and SSE the execution resources too. Use this option with care, as it is
9313 still experimental, because the GCC register allocator does not model separate
9314 functional units well resulting in instable performance.
9317 @item -masm=@var{dialect}
9318 @opindex masm=@var{dialect}
9319 Output asm instructions using selected @var{dialect}. Supported
9320 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9321 not support @samp{intel}.
9326 @opindex mno-ieee-fp
9327 Control whether or not the compiler uses IEEE floating point
9328 comparisons. These handle correctly the case where the result of a
9329 comparison is unordered.
9332 @opindex msoft-float
9333 Generate output containing library calls for floating point.
9334 @strong{Warning:} the requisite libraries are not part of GCC@.
9335 Normally the facilities of the machine's usual C compiler are used, but
9336 this can't be done directly in cross-compilation. You must make your
9337 own arrangements to provide suitable library functions for
9340 On machines where a function returns floating point results in the 80387
9341 register stack, some floating point opcodes may be emitted even if
9342 @option{-msoft-float} is used.
9344 @item -mno-fp-ret-in-387
9345 @opindex mno-fp-ret-in-387
9346 Do not use the FPU registers for return values of functions.
9348 The usual calling convention has functions return values of types
9349 @code{float} and @code{double} in an FPU register, even if there
9350 is no FPU@. The idea is that the operating system should emulate
9353 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9354 in ordinary CPU registers instead.
9356 @item -mno-fancy-math-387
9357 @opindex mno-fancy-math-387
9358 Some 387 emulators do not support the @code{sin}, @code{cos} and
9359 @code{sqrt} instructions for the 387. Specify this option to avoid
9360 generating those instructions. This option is the default on FreeBSD,
9361 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9362 indicates that the target cpu will always have an FPU and so the
9363 instruction will not need emulation. As of revision 2.6.1, these
9364 instructions are not generated unless you also use the
9365 @option{-funsafe-math-optimizations} switch.
9367 @item -malign-double
9368 @itemx -mno-align-double
9369 @opindex malign-double
9370 @opindex mno-align-double
9371 Control whether GCC aligns @code{double}, @code{long double}, and
9372 @code{long long} variables on a two word boundary or a one word
9373 boundary. Aligning @code{double} variables on a two word boundary will
9374 produce code that runs somewhat faster on a @samp{Pentium} at the
9375 expense of more memory.
9377 @strong{Warning:} if you use the @option{-malign-double} switch,
9378 structures containing the above types will be aligned differently than
9379 the published application binary interface specifications for the 386
9380 and will not be binary compatible with structures in code compiled
9381 without that switch.
9383 @item -m96bit-long-double
9384 @itemx -m128bit-long-double
9385 @opindex m96bit-long-double
9386 @opindex m128bit-long-double
9387 These switches control the size of @code{long double} type. The i386
9388 application binary interface specifies the size to be 96 bits,
9389 so @option{-m96bit-long-double} is the default in 32 bit mode.
9391 Modern architectures (Pentium and newer) would prefer @code{long double}
9392 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9393 conforming to the ABI, this would not be possible. So specifying a
9394 @option{-m128bit-long-double} will align @code{long double}
9395 to a 16 byte boundary by padding the @code{long double} with an additional
9398 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9399 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9401 Notice that neither of these options enable any extra precision over the x87
9402 standard of 80 bits for a @code{long double}.
9404 @strong{Warning:} if you override the default value for your target ABI, the
9405 structures and arrays containing @code{long double} variables will change
9406 their size as well as function calling convention for function taking
9407 @code{long double} will be modified. Hence they will not be binary
9408 compatible with arrays or structures in code compiled without that switch.
9410 @item -mmlarge-data-threshold=@var{number}
9411 @opindex mlarge-data-threshold=@var{number}
9412 When @option{-mcmodel=medium} is specified, the data greater than
9413 @var{threshold} are placed in large data section. This value must be the
9414 same across all object linked into the binary and defaults to 65535.
9417 @itemx -mno-svr3-shlib
9418 @opindex msvr3-shlib
9419 @opindex mno-svr3-shlib
9420 Control whether GCC places uninitialized local variables into the
9421 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9422 into @code{bss}. These options are meaningful only on System V Release 3.
9426 Use a different function-calling convention, in which functions that
9427 take a fixed number of arguments return with the @code{ret} @var{num}
9428 instruction, which pops their arguments while returning. This saves one
9429 instruction in the caller since there is no need to pop the arguments
9432 You can specify that an individual function is called with this calling
9433 sequence with the function attribute @samp{stdcall}. You can also
9434 override the @option{-mrtd} option by using the function attribute
9435 @samp{cdecl}. @xref{Function Attributes}.
9437 @strong{Warning:} this calling convention is incompatible with the one
9438 normally used on Unix, so you cannot use it if you need to call
9439 libraries compiled with the Unix compiler.
9441 Also, you must provide function prototypes for all functions that
9442 take variable numbers of arguments (including @code{printf});
9443 otherwise incorrect code will be generated for calls to those
9446 In addition, seriously incorrect code will result if you call a
9447 function with too many arguments. (Normally, extra arguments are
9448 harmlessly ignored.)
9450 @item -mregparm=@var{num}
9452 Control how many registers are used to pass integer arguments. By
9453 default, no registers are used to pass arguments, and at most 3
9454 registers can be used. You can control this behavior for a specific
9455 function by using the function attribute @samp{regparm}.
9456 @xref{Function Attributes}.
9458 @strong{Warning:} if you use this switch, and
9459 @var{num} is nonzero, then you must build all modules with the same
9460 value, including any libraries. This includes the system libraries and
9464 @opindex msseregparm
9465 Use SSE register passing conventions for float and double arguments
9466 and return values. You can control this behavior for a specific
9467 function by using the function attribute @samp{sseregparm}.
9468 @xref{Function Attributes}.
9470 @strong{Warning:} if you use this switch then you must build all
9471 modules with the same value, including any libraries. This includes
9472 the system libraries and startup modules.
9474 @item -mstackrealign
9475 @opindex mstackrealign
9476 Realign the stack at entry. On the Intel x86, the
9477 @option{-mstackrealign} option will generate an alternate prologue and
9478 epilogue that realigns the runtime stack. This supports mixing legacy
9479 codes that keep a 4-byte aligned stack with modern codes that keep a
9480 16-byte stack for SSE compatibility. The alternate prologue and
9481 epilogue are slower and bigger than the regular ones, and the
9482 alternate prologue requires an extra scratch register; this lowers the
9483 number of registers available if used in conjunction with the
9484 @code{regparm} attribute. The @option{-mstackrealign} option is
9485 incompatible with the nested function prologue; this is considered a
9486 hard error. See also the attribute @code{force_align_arg_pointer},
9487 applicable to individual functions.
9489 @item -mpreferred-stack-boundary=@var{num}
9490 @opindex mpreferred-stack-boundary
9491 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9492 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9493 the default is 4 (16 bytes or 128 bits), except when optimizing for code
9494 size (@option{-Os}), in which case the default is the minimum correct
9495 alignment (4 bytes for x86, and 8 bytes for x86-64).
9497 On Pentium and PentiumPro, @code{double} and @code{long double} values
9498 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9499 suffer significant run time performance penalties. On Pentium III, the
9500 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
9501 penalties if it is not 16 byte aligned.
9503 To ensure proper alignment of this values on the stack, the stack boundary
9504 must be as aligned as that required by any value stored on the stack.
9505 Further, every function must be generated such that it keeps the stack
9506 aligned. Thus calling a function compiled with a higher preferred
9507 stack boundary from a function compiled with a lower preferred stack
9508 boundary will most likely misalign the stack. It is recommended that
9509 libraries that use callbacks always use the default setting.
9511 This extra alignment does consume extra stack space, and generally
9512 increases code size. Code that is sensitive to stack space usage, such
9513 as embedded systems and operating system kernels, may want to reduce the
9514 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9532 These switches enable or disable the use of instructions in the MMX,
9533 SSE, SSE2 or 3DNow! extended instruction sets. These extensions are
9534 also available as built-in functions: see @ref{X86 Built-in Functions},
9535 for details of the functions enabled and disabled by these switches.
9537 To have SSE/SSE2 instructions generated automatically from floating-point
9538 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9540 These options will enable GCC to use these extended instructions in
9541 generated code, even without @option{-mfpmath=sse}. Applications which
9542 perform runtime CPU detection must compile separate files for each
9543 supported architecture, using the appropriate flags. In particular,
9544 the file containing the CPU detection code should be compiled without
9548 @itemx -mno-push-args
9550 @opindex mno-push-args
9551 Use PUSH operations to store outgoing parameters. This method is shorter
9552 and usually equally fast as method using SUB/MOV operations and is enabled
9553 by default. In some cases disabling it may improve performance because of
9554 improved scheduling and reduced dependencies.
9556 @item -maccumulate-outgoing-args
9557 @opindex maccumulate-outgoing-args
9558 If enabled, the maximum amount of space required for outgoing arguments will be
9559 computed in the function prologue. This is faster on most modern CPUs
9560 because of reduced dependencies, improved scheduling and reduced stack usage
9561 when preferred stack boundary is not equal to 2. The drawback is a notable
9562 increase in code size. This switch implies @option{-mno-push-args}.
9566 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9567 on thread-safe exception handling must compile and link all code with the
9568 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9569 @option{-D_MT}; when linking, it links in a special thread helper library
9570 @option{-lmingwthrd} which cleans up per thread exception handling data.
9572 @item -mno-align-stringops
9573 @opindex mno-align-stringops
9574 Do not align destination of inlined string operations. This switch reduces
9575 code size and improves performance in case the destination is already aligned,
9576 but GCC doesn't know about it.
9578 @item -minline-all-stringops
9579 @opindex minline-all-stringops
9580 By default GCC inlines string operations only when destination is known to be
9581 aligned at least to 4 byte boundary. This enables more inlining, increase code
9582 size, but may improve performance of code that depends on fast memcpy, strlen
9583 and memset for short lengths.
9585 @item -momit-leaf-frame-pointer
9586 @opindex momit-leaf-frame-pointer
9587 Don't keep the frame pointer in a register for leaf functions. This
9588 avoids the instructions to save, set up and restore frame pointers and
9589 makes an extra register available in leaf functions. The option
9590 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9591 which might make debugging harder.
9593 @item -mtls-direct-seg-refs
9594 @itemx -mno-tls-direct-seg-refs
9595 @opindex mtls-direct-seg-refs
9596 Controls whether TLS variables may be accessed with offsets from the
9597 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9598 or whether the thread base pointer must be added. Whether or not this
9599 is legal depends on the operating system, and whether it maps the
9600 segment to cover the entire TLS area.
9602 For systems that use GNU libc, the default is on.
9605 These @samp{-m} switches are supported in addition to the above
9606 on AMD x86-64 processors in 64-bit environments.
9613 Generate code for a 32-bit or 64-bit environment.
9614 The 32-bit environment sets int, long and pointer to 32 bits and
9615 generates code that runs on any i386 system.
9616 The 64-bit environment sets int to 32 bits and long and pointer
9617 to 64 bits and generates code for AMD's x86-64 architecture.
9620 @opindex no-red-zone
9621 Do not use a so called red zone for x86-64 code. The red zone is mandated
9622 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9623 stack pointer that will not be modified by signal or interrupt handlers
9624 and therefore can be used for temporary data without adjusting the stack
9625 pointer. The flag @option{-mno-red-zone} disables this red zone.
9627 @item -mcmodel=small
9628 @opindex mcmodel=small
9629 Generate code for the small code model: the program and its symbols must
9630 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9631 Programs can be statically or dynamically linked. This is the default
9634 @item -mcmodel=kernel
9635 @opindex mcmodel=kernel
9636 Generate code for the kernel code model. The kernel runs in the
9637 negative 2 GB of the address space.
9638 This model has to be used for Linux kernel code.
9640 @item -mcmodel=medium
9641 @opindex mcmodel=medium
9642 Generate code for the medium model: The program is linked in the lower 2
9643 GB of the address space but symbols can be located anywhere in the
9644 address space. Programs can be statically or dynamically linked, but
9645 building of shared libraries are not supported with the medium model.
9647 @item -mcmodel=large
9648 @opindex mcmodel=large
9649 Generate code for the large model: This model makes no assumptions
9650 about addresses and sizes of sections. Currently GCC does not implement
9655 @subsection IA-64 Options
9656 @cindex IA-64 Options
9658 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9662 @opindex mbig-endian
9663 Generate code for a big endian target. This is the default for HP-UX@.
9665 @item -mlittle-endian
9666 @opindex mlittle-endian
9667 Generate code for a little endian target. This is the default for AIX5
9674 Generate (or don't) code for the GNU assembler. This is the default.
9675 @c Also, this is the default if the configure option @option{--with-gnu-as}
9682 Generate (or don't) code for the GNU linker. This is the default.
9683 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9688 Generate code that does not use a global pointer register. The result
9689 is not position independent code, and violates the IA-64 ABI@.
9691 @item -mvolatile-asm-stop
9692 @itemx -mno-volatile-asm-stop
9693 @opindex mvolatile-asm-stop
9694 @opindex mno-volatile-asm-stop
9695 Generate (or don't) a stop bit immediately before and after volatile asm
9698 @item -mregister-names
9699 @itemx -mno-register-names
9700 @opindex mregister-names
9701 @opindex mno-register-names
9702 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9703 the stacked registers. This may make assembler output more readable.
9709 Disable (or enable) optimizations that use the small data section. This may
9710 be useful for working around optimizer bugs.
9713 @opindex mconstant-gp
9714 Generate code that uses a single constant global pointer value. This is
9715 useful when compiling kernel code.
9719 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9720 This is useful when compiling firmware code.
9722 @item -minline-float-divide-min-latency
9723 @opindex minline-float-divide-min-latency
9724 Generate code for inline divides of floating point values
9725 using the minimum latency algorithm.
9727 @item -minline-float-divide-max-throughput
9728 @opindex minline-float-divide-max-throughput
9729 Generate code for inline divides of floating point values
9730 using the maximum throughput algorithm.
9732 @item -minline-int-divide-min-latency
9733 @opindex minline-int-divide-min-latency
9734 Generate code for inline divides of integer values
9735 using the minimum latency algorithm.
9737 @item -minline-int-divide-max-throughput
9738 @opindex minline-int-divide-max-throughput
9739 Generate code for inline divides of integer values
9740 using the maximum throughput algorithm.
9742 @item -minline-sqrt-min-latency
9743 @opindex minline-sqrt-min-latency
9744 Generate code for inline square roots
9745 using the minimum latency algorithm.
9747 @item -minline-sqrt-max-throughput
9748 @opindex minline-sqrt-max-throughput
9749 Generate code for inline square roots
9750 using the maximum throughput algorithm.
9752 @item -mno-dwarf2-asm
9754 @opindex mno-dwarf2-asm
9755 @opindex mdwarf2-asm
9756 Don't (or do) generate assembler code for the DWARF2 line number debugging
9757 info. This may be useful when not using the GNU assembler.
9759 @item -mearly-stop-bits
9760 @itemx -mno-early-stop-bits
9761 @opindex mearly-stop-bits
9762 @opindex mno-early-stop-bits
9763 Allow stop bits to be placed earlier than immediately preceding the
9764 instruction that triggered the stop bit. This can improve instruction
9765 scheduling, but does not always do so.
9767 @item -mfixed-range=@var{register-range}
9768 @opindex mfixed-range
9769 Generate code treating the given register range as fixed registers.
9770 A fixed register is one that the register allocator can not use. This is
9771 useful when compiling kernel code. A register range is specified as
9772 two registers separated by a dash. Multiple register ranges can be
9773 specified separated by a comma.
9775 @item -mtls-size=@var{tls-size}
9777 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9780 @item -mtune=@var{cpu-type}
9782 Tune the instruction scheduling for a particular CPU, Valid values are
9783 itanium, itanium1, merced, itanium2, and mckinley.
9789 Add support for multithreading using the POSIX threads library. This
9790 option sets flags for both the preprocessor and linker. It does
9791 not affect the thread safety of object code produced by the compiler or
9792 that of libraries supplied with it. These are HP-UX specific flags.
9798 Generate code for a 32-bit or 64-bit environment.
9799 The 32-bit environment sets int, long and pointer to 32 bits.
9800 The 64-bit environment sets int to 32 bits and long and pointer
9801 to 64 bits. These are HP-UX specific flags.
9803 @item -mno-sched-br-data-spec
9804 @itemx -msched-br-data-spec
9805 @opindex -mno-sched-br-data-spec
9806 @opindex -msched-br-data-spec
9807 (Dis/En)able data speculative scheduling before reload.
9808 This will result in generation of the ld.a instructions and
9809 the corresponding check instructions (ld.c / chk.a).
9810 The default is 'disable'.
9812 @item -msched-ar-data-spec
9813 @itemx -mno-sched-ar-data-spec
9814 @opindex -msched-ar-data-spec
9815 @opindex -mno-sched-ar-data-spec
9816 (En/Dis)able data speculative scheduling after reload.
9817 This will result in generation of the ld.a instructions and
9818 the corresponding check instructions (ld.c / chk.a).
9819 The default is 'enable'.
9821 @item -mno-sched-control-spec
9822 @itemx -msched-control-spec
9823 @opindex -mno-sched-control-spec
9824 @opindex -msched-control-spec
9825 (Dis/En)able control speculative scheduling. This feature is
9826 available only during region scheduling (i.e. before reload).
9827 This will result in generation of the ld.s instructions and
9828 the corresponding check instructions chk.s .
9829 The default is 'disable'.
9831 @item -msched-br-in-data-spec
9832 @itemx -mno-sched-br-in-data-spec
9833 @opindex -msched-br-in-data-spec
9834 @opindex -mno-sched-br-in-data-spec
9835 (En/Dis)able speculative scheduling of the instructions that
9836 are dependent on the data speculative loads before reload.
9837 This is effective only with @option{-msched-br-data-spec} enabled.
9838 The default is 'enable'.
9840 @item -msched-ar-in-data-spec
9841 @itemx -mno-sched-ar-in-data-spec
9842 @opindex -msched-ar-in-data-spec
9843 @opindex -mno-sched-ar-in-data-spec
9844 (En/Dis)able speculative scheduling of the instructions that
9845 are dependent on the data speculative loads after reload.
9846 This is effective only with @option{-msched-ar-data-spec} enabled.
9847 The default is 'enable'.
9849 @item -msched-in-control-spec
9850 @itemx -mno-sched-in-control-spec
9851 @opindex -msched-in-control-spec
9852 @opindex -mno-sched-in-control-spec
9853 (En/Dis)able speculative scheduling of the instructions that
9854 are dependent on the control speculative loads.
9855 This is effective only with @option{-msched-control-spec} enabled.
9856 The default is 'enable'.
9859 @itemx -mno-sched-ldc
9860 @opindex -msched-ldc
9861 @opindex -mno-sched-ldc
9862 (En/Dis)able use of simple data speculation checks ld.c .
9863 If disabled, only chk.a instructions will be emitted to check
9864 data speculative loads.
9865 The default is 'enable'.
9867 @item -mno-sched-control-ldc
9868 @itemx -msched-control-ldc
9869 @opindex -mno-sched-control-ldc
9870 @opindex -msched-control-ldc
9871 (Dis/En)able use of ld.c instructions to check control speculative loads.
9872 If enabled, in case of control speculative load with no speculatively
9873 scheduled dependent instructions this load will be emitted as ld.sa and
9874 ld.c will be used to check it.
9875 The default is 'disable'.
9877 @item -mno-sched-spec-verbose
9878 @itemx -msched-spec-verbose
9879 @opindex -mno-sched-spec-verbose
9880 @opindex -msched-spec-verbose
9881 (Dis/En)able printing of the information about speculative motions.
9883 @item -mno-sched-prefer-non-data-spec-insns
9884 @itemx -msched-prefer-non-data-spec-insns
9885 @opindex -mno-sched-prefer-non-data-spec-insns
9886 @opindex -msched-prefer-non-data-spec-insns
9887 If enabled, data speculative instructions will be chosen for schedule
9888 only if there are no other choices at the moment. This will make
9889 the use of the data speculation much more conservative.
9890 The default is 'disable'.
9892 @item -mno-sched-prefer-non-control-spec-insns
9893 @itemx -msched-prefer-non-control-spec-insns
9894 @opindex -mno-sched-prefer-non-control-spec-insns
9895 @opindex -msched-prefer-non-control-spec-insns
9896 If enabled, control speculative instructions will be chosen for schedule
9897 only if there are no other choices at the moment. This will make
9898 the use of the control speculation much more conservative.
9899 The default is 'disable'.
9901 @item -mno-sched-count-spec-in-critical-path
9902 @itemx -msched-count-spec-in-critical-path
9903 @opindex -mno-sched-count-spec-in-critical-path
9904 @opindex -msched-count-spec-in-critical-path
9905 If enabled, speculative dependencies will be considered during
9906 computation of the instructions priorities. This will make the use of the
9907 speculation a bit more conservative.
9908 The default is 'disable'.
9913 @subsection M32C Options
9914 @cindex M32C options
9917 @item -mcpu=@var{name}
9919 Select the CPU for which code is generated. @var{name} may be one of
9920 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9921 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9926 Specifies that the program will be run on the simulator. This causes
9927 an alternate runtime library to be linked in which supports, for
9928 example, file I/O. You must not use this option when generating
9929 programs that will run on real hardware; you must provide your own
9930 runtime library for whatever I/O functions are needed.
9932 @item -memregs=@var{number}
9934 Specifies the number of memory-based pseudo-registers GCC will use
9935 during code generation. These pseudo-registers will be used like real
9936 registers, so there is a tradeoff between GCC's ability to fit the
9937 code into available registers, and the performance penalty of using
9938 memory instead of registers. Note that all modules in a program must
9939 be compiled with the same value for this option. Because of that, you
9940 must not use this option with the default runtime libraries gcc
9945 @node M32R/D Options
9946 @subsection M32R/D Options
9947 @cindex M32R/D options
9949 These @option{-m} options are defined for Renesas M32R/D architectures:
9954 Generate code for the M32R/2@.
9958 Generate code for the M32R/X@.
9962 Generate code for the M32R@. This is the default.
9965 @opindex mmodel=small
9966 Assume all objects live in the lower 16MB of memory (so that their addresses
9967 can be loaded with the @code{ld24} instruction), and assume all subroutines
9968 are reachable with the @code{bl} instruction.
9969 This is the default.
9971 The addressability of a particular object can be set with the
9972 @code{model} attribute.
9974 @item -mmodel=medium
9975 @opindex mmodel=medium
9976 Assume objects may be anywhere in the 32-bit address space (the compiler
9977 will generate @code{seth/add3} instructions to load their addresses), and
9978 assume all subroutines are reachable with the @code{bl} instruction.
9981 @opindex mmodel=large
9982 Assume objects may be anywhere in the 32-bit address space (the compiler
9983 will generate @code{seth/add3} instructions to load their addresses), and
9984 assume subroutines may not be reachable with the @code{bl} instruction
9985 (the compiler will generate the much slower @code{seth/add3/jl}
9986 instruction sequence).
9989 @opindex msdata=none
9990 Disable use of the small data area. Variables will be put into
9991 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
9992 @code{section} attribute has been specified).
9993 This is the default.
9995 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
9996 Objects may be explicitly put in the small data area with the
9997 @code{section} attribute using one of these sections.
10000 @opindex msdata=sdata
10001 Put small global and static data in the small data area, but do not
10002 generate special code to reference them.
10005 @opindex msdata=use
10006 Put small global and static data in the small data area, and generate
10007 special instructions to reference them.
10011 @cindex smaller data references
10012 Put global and static objects less than or equal to @var{num} bytes
10013 into the small data or bss sections instead of the normal data or bss
10014 sections. The default value of @var{num} is 8.
10015 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10016 for this option to have any effect.
10018 All modules should be compiled with the same @option{-G @var{num}} value.
10019 Compiling with different values of @var{num} may or may not work; if it
10020 doesn't the linker will give an error message---incorrect code will not be
10025 Makes the M32R specific code in the compiler display some statistics
10026 that might help in debugging programs.
10028 @item -malign-loops
10029 @opindex malign-loops
10030 Align all loops to a 32-byte boundary.
10032 @item -mno-align-loops
10033 @opindex mno-align-loops
10034 Do not enforce a 32-byte alignment for loops. This is the default.
10036 @item -missue-rate=@var{number}
10037 @opindex missue-rate=@var{number}
10038 Issue @var{number} instructions per cycle. @var{number} can only be 1
10041 @item -mbranch-cost=@var{number}
10042 @opindex mbranch-cost=@var{number}
10043 @var{number} can only be 1 or 2. If it is 1 then branches will be
10044 preferred over conditional code, if it is 2, then the opposite will
10047 @item -mflush-trap=@var{number}
10048 @opindex mflush-trap=@var{number}
10049 Specifies the trap number to use to flush the cache. The default is
10050 12. Valid numbers are between 0 and 15 inclusive.
10052 @item -mno-flush-trap
10053 @opindex mno-flush-trap
10054 Specifies that the cache cannot be flushed by using a trap.
10056 @item -mflush-func=@var{name}
10057 @opindex mflush-func=@var{name}
10058 Specifies the name of the operating system function to call to flush
10059 the cache. The default is @emph{_flush_cache}, but a function call
10060 will only be used if a trap is not available.
10062 @item -mno-flush-func
10063 @opindex mno-flush-func
10064 Indicates that there is no OS function for flushing the cache.
10068 @node M680x0 Options
10069 @subsection M680x0 Options
10070 @cindex M680x0 options
10072 These are the @samp{-m} options defined for the 68000 series. The default
10073 values for these options depends on which style of 68000 was selected when
10074 the compiler was configured; the defaults for the most common choices are
10082 Generate output for a 68000. This is the default
10083 when the compiler is configured for 68000-based systems.
10085 Use this option for microcontrollers with a 68000 or EC000 core,
10086 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10092 Generate output for a 68020. This is the default
10093 when the compiler is configured for 68020-based systems.
10097 Generate output containing 68881 instructions for floating point.
10098 This is the default for most 68020 systems unless @option{--nfp} was
10099 specified when the compiler was configured.
10103 Generate output for a 68030. This is the default when the compiler is
10104 configured for 68030-based systems.
10108 Generate output for a 68040. This is the default when the compiler is
10109 configured for 68040-based systems.
10111 This option inhibits the use of 68881/68882 instructions that have to be
10112 emulated by software on the 68040. Use this option if your 68040 does not
10113 have code to emulate those instructions.
10117 Generate output for a 68060. This is the default when the compiler is
10118 configured for 68060-based systems.
10120 This option inhibits the use of 68020 and 68881/68882 instructions that
10121 have to be emulated by software on the 68060. Use this option if your 68060
10122 does not have code to emulate those instructions.
10126 Generate output for a CPU32. This is the default
10127 when the compiler is configured for CPU32-based systems.
10129 Use this option for microcontrollers with a
10130 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10131 68336, 68340, 68341, 68349 and 68360.
10135 Generate output for a 520X ``coldfire'' family cpu. This is the default
10136 when the compiler is configured for 520X-based systems.
10138 Use this option for microcontroller with a 5200 core, including
10139 the MCF5202, MCF5203, MCF5204 and MCF5202.
10143 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
10144 This includes use of hardware floating point instructions.
10148 Generate output for a 68040, without using any of the new instructions.
10149 This results in code which can run relatively efficiently on either a
10150 68020/68881 or a 68030 or a 68040. The generated code does use the
10151 68881 instructions that are emulated on the 68040.
10155 Generate output for a 68060, without using any of the new instructions.
10156 This results in code which can run relatively efficiently on either a
10157 68020/68881 or a 68030 or a 68040. The generated code does use the
10158 68881 instructions that are emulated on the 68060.
10161 @opindex msoft-float
10162 Generate output containing library calls for floating point.
10163 @strong{Warning:} the requisite libraries are not available for all m68k
10164 targets. Normally the facilities of the machine's usual C compiler are
10165 used, but this can't be done directly in cross-compilation. You must
10166 make your own arrangements to provide suitable library functions for
10167 cross-compilation. The embedded targets @samp{m68k-*-aout} and
10168 @samp{m68k-*-coff} do provide software floating point support.
10172 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10173 Additionally, parameters passed on the stack are also aligned to a
10174 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10177 @opindex mnobitfield
10178 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10179 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10183 Do use the bit-field instructions. The @option{-m68020} option implies
10184 @option{-mbitfield}. This is the default if you use a configuration
10185 designed for a 68020.
10189 Use a different function-calling convention, in which functions
10190 that take a fixed number of arguments return with the @code{rtd}
10191 instruction, which pops their arguments while returning. This
10192 saves one instruction in the caller since there is no need to pop
10193 the arguments there.
10195 This calling convention is incompatible with the one normally
10196 used on Unix, so you cannot use it if you need to call libraries
10197 compiled with the Unix compiler.
10199 Also, you must provide function prototypes for all functions that
10200 take variable numbers of arguments (including @code{printf});
10201 otherwise incorrect code will be generated for calls to those
10204 In addition, seriously incorrect code will result if you call a
10205 function with too many arguments. (Normally, extra arguments are
10206 harmlessly ignored.)
10208 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10209 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10212 @itemx -mno-align-int
10213 @opindex malign-int
10214 @opindex mno-align-int
10215 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10216 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10217 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10218 Aligning variables on 32-bit boundaries produces code that runs somewhat
10219 faster on processors with 32-bit busses at the expense of more memory.
10221 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10222 align structures containing the above types differently than
10223 most published application binary interface specifications for the m68k.
10227 Use the pc-relative addressing mode of the 68000 directly, instead of
10228 using a global offset table. At present, this option implies @option{-fpic},
10229 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10230 not presently supported with @option{-mpcrel}, though this could be supported for
10231 68020 and higher processors.
10233 @item -mno-strict-align
10234 @itemx -mstrict-align
10235 @opindex mno-strict-align
10236 @opindex mstrict-align
10237 Do not (do) assume that unaligned memory references will be handled by
10241 Generate code that allows the data segment to be located in a different
10242 area of memory from the text segment. This allows for execute in place in
10243 an environment without virtual memory management. This option implies
10246 @item -mno-sep-data
10247 Generate code that assumes that the data segment follows the text segment.
10248 This is the default.
10250 @item -mid-shared-library
10251 Generate code that supports shared libraries via the library ID method.
10252 This allows for execute in place and shared libraries in an environment
10253 without virtual memory management. This option implies @option{-fPIC}.
10255 @item -mno-id-shared-library
10256 Generate code that doesn't assume ID based shared libraries are being used.
10257 This is the default.
10259 @item -mshared-library-id=n
10260 Specified the identification number of the ID based shared library being
10261 compiled. Specifying a value of 0 will generate more compact code, specifying
10262 other values will force the allocation of that number to the current
10263 library but is no more space or time efficient than omitting this option.
10267 @node M68hc1x Options
10268 @subsection M68hc1x Options
10269 @cindex M68hc1x options
10271 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10272 microcontrollers. The default values for these options depends on
10273 which style of microcontroller was selected when the compiler was configured;
10274 the defaults for the most common choices are given below.
10281 Generate output for a 68HC11. This is the default
10282 when the compiler is configured for 68HC11-based systems.
10288 Generate output for a 68HC12. This is the default
10289 when the compiler is configured for 68HC12-based systems.
10295 Generate output for a 68HCS12.
10297 @item -mauto-incdec
10298 @opindex mauto-incdec
10299 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10306 Enable the use of 68HC12 min and max instructions.
10309 @itemx -mno-long-calls
10310 @opindex mlong-calls
10311 @opindex mno-long-calls
10312 Treat all calls as being far away (near). If calls are assumed to be
10313 far away, the compiler will use the @code{call} instruction to
10314 call a function and the @code{rtc} instruction for returning.
10318 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10320 @item -msoft-reg-count=@var{count}
10321 @opindex msoft-reg-count
10322 Specify the number of pseudo-soft registers which are used for the
10323 code generation. The maximum number is 32. Using more pseudo-soft
10324 register may or may not result in better code depending on the program.
10325 The default is 4 for 68HC11 and 2 for 68HC12.
10329 @node MCore Options
10330 @subsection MCore Options
10331 @cindex MCore options
10333 These are the @samp{-m} options defined for the Motorola M*Core
10339 @itemx -mno-hardlit
10341 @opindex mno-hardlit
10342 Inline constants into the code stream if it can be done in two
10343 instructions or less.
10349 Use the divide instruction. (Enabled by default).
10351 @item -mrelax-immediate
10352 @itemx -mno-relax-immediate
10353 @opindex mrelax-immediate
10354 @opindex mno-relax-immediate
10355 Allow arbitrary sized immediates in bit operations.
10357 @item -mwide-bitfields
10358 @itemx -mno-wide-bitfields
10359 @opindex mwide-bitfields
10360 @opindex mno-wide-bitfields
10361 Always treat bit-fields as int-sized.
10363 @item -m4byte-functions
10364 @itemx -mno-4byte-functions
10365 @opindex m4byte-functions
10366 @opindex mno-4byte-functions
10367 Force all functions to be aligned to a four byte boundary.
10369 @item -mcallgraph-data
10370 @itemx -mno-callgraph-data
10371 @opindex mcallgraph-data
10372 @opindex mno-callgraph-data
10373 Emit callgraph information.
10376 @itemx -mno-slow-bytes
10377 @opindex mslow-bytes
10378 @opindex mno-slow-bytes
10379 Prefer word access when reading byte quantities.
10381 @item -mlittle-endian
10382 @itemx -mbig-endian
10383 @opindex mlittle-endian
10384 @opindex mbig-endian
10385 Generate code for a little endian target.
10391 Generate code for the 210 processor.
10395 @subsection MIPS Options
10396 @cindex MIPS options
10402 Generate big-endian code.
10406 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10409 @item -march=@var{arch}
10411 Generate code that will run on @var{arch}, which can be the name of a
10412 generic MIPS ISA, or the name of a particular processor.
10414 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10415 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10416 The processor names are:
10417 @samp{4kc}, @samp{4km}, @samp{4kp},
10418 @samp{5kc}, @samp{5kf},
10420 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10423 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10424 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10425 @samp{rm7000}, @samp{rm9000},
10428 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10429 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10430 The special value @samp{from-abi} selects the
10431 most compatible architecture for the selected ABI (that is,
10432 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10434 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10435 (for example, @samp{-march=r2k}). Prefixes are optional, and
10436 @samp{vr} may be written @samp{r}.
10438 GCC defines two macros based on the value of this option. The first
10439 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10440 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10441 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10442 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10443 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10445 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10446 above. In other words, it will have the full prefix and will not
10447 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10448 the macro names the resolved architecture (either @samp{"mips1"} or
10449 @samp{"mips3"}). It names the default architecture when no
10450 @option{-march} option is given.
10452 @item -mtune=@var{arch}
10454 Optimize for @var{arch}. Among other things, this option controls
10455 the way instructions are scheduled, and the perceived cost of arithmetic
10456 operations. The list of @var{arch} values is the same as for
10459 When this option is not used, GCC will optimize for the processor
10460 specified by @option{-march}. By using @option{-march} and
10461 @option{-mtune} together, it is possible to generate code that will
10462 run on a family of processors, but optimize the code for one
10463 particular member of that family.
10465 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10466 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10467 @samp{-march} ones described above.
10471 Equivalent to @samp{-march=mips1}.
10475 Equivalent to @samp{-march=mips2}.
10479 Equivalent to @samp{-march=mips3}.
10483 Equivalent to @samp{-march=mips4}.
10487 Equivalent to @samp{-march=mips32}.
10491 Equivalent to @samp{-march=mips32r2}.
10495 Equivalent to @samp{-march=mips64}.
10500 @opindex mno-mips16
10501 Generate (do not generate) MIPS16 code. If GCC is targetting a
10502 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10514 Generate code for the given ABI@.
10516 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10517 generates 64-bit code when you select a 64-bit architecture, but you
10518 can use @option{-mgp32} to get 32-bit code instead.
10520 For information about the O64 ABI, see
10521 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10524 @itemx -mno-abicalls
10526 @opindex mno-abicalls
10527 Generate (do not generate) code that is suitable for SVR4-style
10528 dynamic objects. @option{-mabicalls} is the default for SVR4-based
10533 Generate (do not generate) code that is fully position-independent,
10534 and that can therefore be linked into shared libraries. This option
10535 only affects @option{-mabicalls}.
10537 All @option{-mabicalls} code has traditionally been position-independent,
10538 regardless of options like @option{-fPIC} and @option{-fpic}. However,
10539 as an extension, the GNU toolchain allows executables to use absolute
10540 accesses for locally-binding symbols. It can also use shorter GP
10541 initialization sequences and generate direct calls to locally-defined
10542 functions. This mode is selected by @option{-mno-shared}.
10544 @option{-mno-shared} depends on binutils 2.16 or higher and generates
10545 objects that can only be linked by the GNU linker. However, the option
10546 does not affect the ABI of the final executable; it only affects the ABI
10547 of relocatable objects. Using @option{-mno-shared} will generally make
10548 executables both smaller and quicker.
10550 @option{-mshared} is the default.
10556 Lift (do not lift) the usual restrictions on the size of the global
10559 GCC normally uses a single instruction to load values from the GOT@.
10560 While this is relatively efficient, it will only work if the GOT
10561 is smaller than about 64k. Anything larger will cause the linker
10562 to report an error such as:
10564 @cindex relocation truncated to fit (MIPS)
10566 relocation truncated to fit: R_MIPS_GOT16 foobar
10569 If this happens, you should recompile your code with @option{-mxgot}.
10570 It should then work with very large GOTs, although it will also be
10571 less efficient, since it will take three instructions to fetch the
10572 value of a global symbol.
10574 Note that some linkers can create multiple GOTs. If you have such a
10575 linker, you should only need to use @option{-mxgot} when a single object
10576 file accesses more than 64k's worth of GOT entries. Very few do.
10578 These options have no effect unless GCC is generating position
10583 Assume that general-purpose registers are 32 bits wide.
10587 Assume that general-purpose registers are 64 bits wide.
10591 Assume that floating-point registers are 32 bits wide.
10595 Assume that floating-point registers are 64 bits wide.
10598 @opindex mhard-float
10599 Use floating-point coprocessor instructions.
10602 @opindex msoft-float
10603 Do not use floating-point coprocessor instructions. Implement
10604 floating-point calculations using library calls instead.
10606 @item -msingle-float
10607 @opindex msingle-float
10608 Assume that the floating-point coprocessor only supports single-precision
10611 @itemx -mdouble-float
10612 @opindex mdouble-float
10613 Assume that the floating-point coprocessor supports double-precision
10614 operations. This is the default.
10620 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10622 @itemx -mpaired-single
10623 @itemx -mno-paired-single
10624 @opindex mpaired-single
10625 @opindex mno-paired-single
10626 Use (do not use) paired-single floating-point instructions.
10627 @xref{MIPS Paired-Single Support}. This option can only be used
10628 when generating 64-bit code and requires hardware floating-point
10629 support to be enabled.
10634 @opindex mno-mips3d
10635 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10636 The option @option{-mips3d} implies @option{-mpaired-single}.
10640 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10641 an explanation of the default and the way that the pointer size is
10646 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10648 The default size of @code{int}s, @code{long}s and pointers depends on
10649 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10650 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10651 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10652 or the same size as integer registers, whichever is smaller.
10658 Assume (do not assume) that all symbols have 32-bit values, regardless
10659 of the selected ABI@. This option is useful in combination with
10660 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10661 to generate shorter and faster references to symbolic addresses.
10665 @cindex smaller data references (MIPS)
10666 @cindex gp-relative references (MIPS)
10667 Put global and static items less than or equal to @var{num} bytes into
10668 the small data or bss section instead of the normal data or bss section.
10669 This allows the data to be accessed using a single instruction.
10671 All modules should be compiled with the same @option{-G @var{num}}
10674 @item -membedded-data
10675 @itemx -mno-embedded-data
10676 @opindex membedded-data
10677 @opindex mno-embedded-data
10678 Allocate variables to the read-only data section first if possible, then
10679 next in the small data section if possible, otherwise in data. This gives
10680 slightly slower code than the default, but reduces the amount of RAM required
10681 when executing, and thus may be preferred for some embedded systems.
10683 @item -muninit-const-in-rodata
10684 @itemx -mno-uninit-const-in-rodata
10685 @opindex muninit-const-in-rodata
10686 @opindex mno-uninit-const-in-rodata
10687 Put uninitialized @code{const} variables in the read-only data section.
10688 This option is only meaningful in conjunction with @option{-membedded-data}.
10690 @item -msplit-addresses
10691 @itemx -mno-split-addresses
10692 @opindex msplit-addresses
10693 @opindex mno-split-addresses
10694 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10695 relocation operators. This option has been superseded by
10696 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10698 @item -mexplicit-relocs
10699 @itemx -mno-explicit-relocs
10700 @opindex mexplicit-relocs
10701 @opindex mno-explicit-relocs
10702 Use (do not use) assembler relocation operators when dealing with symbolic
10703 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10704 is to use assembler macros instead.
10706 @option{-mexplicit-relocs} is the default if GCC was configured
10707 to use an assembler that supports relocation operators.
10709 @item -mcheck-zero-division
10710 @itemx -mno-check-zero-division
10711 @opindex mcheck-zero-division
10712 @opindex mno-check-zero-division
10713 Trap (do not trap) on integer division by zero. The default is
10714 @option{-mcheck-zero-division}.
10716 @item -mdivide-traps
10717 @itemx -mdivide-breaks
10718 @opindex mdivide-traps
10719 @opindex mdivide-breaks
10720 MIPS systems check for division by zero by generating either a
10721 conditional trap or a break instruction. Using traps results in
10722 smaller code, but is only supported on MIPS II and later. Also, some
10723 versions of the Linux kernel have a bug that prevents trap from
10724 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10725 allow conditional traps on architectures that support them and
10726 @option{-mdivide-breaks} to force the use of breaks.
10728 The default is usually @option{-mdivide-traps}, but this can be
10729 overridden at configure time using @option{--with-divide=breaks}.
10730 Divide-by-zero checks can be completely disabled using
10731 @option{-mno-check-zero-division}.
10736 @opindex mno-memcpy
10737 Force (do not force) the use of @code{memcpy()} for non-trivial block
10738 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10739 most constant-sized copies.
10742 @itemx -mno-long-calls
10743 @opindex mlong-calls
10744 @opindex mno-long-calls
10745 Disable (do not disable) use of the @code{jal} instruction. Calling
10746 functions using @code{jal} is more efficient but requires the caller
10747 and callee to be in the same 256 megabyte segment.
10749 This option has no effect on abicalls code. The default is
10750 @option{-mno-long-calls}.
10756 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10757 instructions, as provided by the R4650 ISA@.
10760 @itemx -mno-fused-madd
10761 @opindex mfused-madd
10762 @opindex mno-fused-madd
10763 Enable (disable) use of the floating point multiply-accumulate
10764 instructions, when they are available. The default is
10765 @option{-mfused-madd}.
10767 When multiply-accumulate instructions are used, the intermediate
10768 product is calculated to infinite precision and is not subject to
10769 the FCSR Flush to Zero bit. This may be undesirable in some
10774 Tell the MIPS assembler to not run its preprocessor over user
10775 assembler files (with a @samp{.s} suffix) when assembling them.
10778 @itemx -mno-fix-r4000
10779 @opindex mfix-r4000
10780 @opindex mno-fix-r4000
10781 Work around certain R4000 CPU errata:
10784 A double-word or a variable shift may give an incorrect result if executed
10785 immediately after starting an integer division.
10787 A double-word or a variable shift may give an incorrect result if executed
10788 while an integer multiplication is in progress.
10790 An integer division may give an incorrect result if started in a delay slot
10791 of a taken branch or a jump.
10795 @itemx -mno-fix-r4400
10796 @opindex mfix-r4400
10797 @opindex mno-fix-r4400
10798 Work around certain R4400 CPU errata:
10801 A double-word or a variable shift may give an incorrect result if executed
10802 immediately after starting an integer division.
10806 @itemx -mno-fix-vr4120
10807 @opindex mfix-vr4120
10808 Work around certain VR4120 errata:
10811 @code{dmultu} does not always produce the correct result.
10813 @code{div} and @code{ddiv} do not always produce the correct result if one
10814 of the operands is negative.
10816 The workarounds for the division errata rely on special functions in
10817 @file{libgcc.a}. At present, these functions are only provided by
10818 the @code{mips64vr*-elf} configurations.
10820 Other VR4120 errata require a nop to be inserted between certain pairs of
10821 instructions. These errata are handled by the assembler, not by GCC itself.
10824 @opindex mfix-vr4130
10825 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10826 workarounds are implemented by the assembler rather than by GCC,
10827 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10828 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10829 instructions are available instead.
10832 @itemx -mno-fix-sb1
10834 Work around certain SB-1 CPU core errata.
10835 (This flag currently works around the SB-1 revision 2
10836 ``F1'' and ``F2'' floating point errata.)
10838 @item -mflush-func=@var{func}
10839 @itemx -mno-flush-func
10840 @opindex mflush-func
10841 Specifies the function to call to flush the I and D caches, or to not
10842 call any such function. If called, the function must take the same
10843 arguments as the common @code{_flush_func()}, that is, the address of the
10844 memory range for which the cache is being flushed, the size of the
10845 memory range, and the number 3 (to flush both caches). The default
10846 depends on the target GCC was configured for, but commonly is either
10847 @samp{_flush_func} or @samp{__cpu_flush}.
10849 @item -mbranch-likely
10850 @itemx -mno-branch-likely
10851 @opindex mbranch-likely
10852 @opindex mno-branch-likely
10853 Enable or disable use of Branch Likely instructions, regardless of the
10854 default for the selected architecture. By default, Branch Likely
10855 instructions may be generated if they are supported by the selected
10856 architecture. An exception is for the MIPS32 and MIPS64 architectures
10857 and processors which implement those architectures; for those, Branch
10858 Likely instructions will not be generated by default because the MIPS32
10859 and MIPS64 architectures specifically deprecate their use.
10861 @item -mfp-exceptions
10862 @itemx -mno-fp-exceptions
10863 @opindex mfp-exceptions
10864 Specifies whether FP exceptions are enabled. This affects how we schedule
10865 FP instructions for some processors. The default is that FP exceptions are
10868 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10869 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10872 @item -mvr4130-align
10873 @itemx -mno-vr4130-align
10874 @opindex mvr4130-align
10875 The VR4130 pipeline is two-way superscalar, but can only issue two
10876 instructions together if the first one is 8-byte aligned. When this
10877 option is enabled, GCC will align pairs of instructions that it
10878 thinks should execute in parallel.
10880 This option only has an effect when optimizing for the VR4130.
10881 It normally makes code faster, but at the expense of making it bigger.
10882 It is enabled by default at optimization level @option{-O3}.
10886 @subsection MMIX Options
10887 @cindex MMIX Options
10889 These options are defined for the MMIX:
10893 @itemx -mno-libfuncs
10895 @opindex mno-libfuncs
10896 Specify that intrinsic library functions are being compiled, passing all
10897 values in registers, no matter the size.
10900 @itemx -mno-epsilon
10902 @opindex mno-epsilon
10903 Generate floating-point comparison instructions that compare with respect
10904 to the @code{rE} epsilon register.
10906 @item -mabi=mmixware
10908 @opindex mabi-mmixware
10910 Generate code that passes function parameters and return values that (in
10911 the called function) are seen as registers @code{$0} and up, as opposed to
10912 the GNU ABI which uses global registers @code{$231} and up.
10914 @item -mzero-extend
10915 @itemx -mno-zero-extend
10916 @opindex mzero-extend
10917 @opindex mno-zero-extend
10918 When reading data from memory in sizes shorter than 64 bits, use (do not
10919 use) zero-extending load instructions by default, rather than
10920 sign-extending ones.
10923 @itemx -mno-knuthdiv
10925 @opindex mno-knuthdiv
10926 Make the result of a division yielding a remainder have the same sign as
10927 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10928 remainder follows the sign of the dividend. Both methods are
10929 arithmetically valid, the latter being almost exclusively used.
10931 @item -mtoplevel-symbols
10932 @itemx -mno-toplevel-symbols
10933 @opindex mtoplevel-symbols
10934 @opindex mno-toplevel-symbols
10935 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10936 code can be used with the @code{PREFIX} assembly directive.
10940 Generate an executable in the ELF format, rather than the default
10941 @samp{mmo} format used by the @command{mmix} simulator.
10943 @item -mbranch-predict
10944 @itemx -mno-branch-predict
10945 @opindex mbranch-predict
10946 @opindex mno-branch-predict
10947 Use (do not use) the probable-branch instructions, when static branch
10948 prediction indicates a probable branch.
10950 @item -mbase-addresses
10951 @itemx -mno-base-addresses
10952 @opindex mbase-addresses
10953 @opindex mno-base-addresses
10954 Generate (do not generate) code that uses @emph{base addresses}. Using a
10955 base address automatically generates a request (handled by the assembler
10956 and the linker) for a constant to be set up in a global register. The
10957 register is used for one or more base address requests within the range 0
10958 to 255 from the value held in the register. The generally leads to short
10959 and fast code, but the number of different data items that can be
10960 addressed is limited. This means that a program that uses lots of static
10961 data may require @option{-mno-base-addresses}.
10963 @item -msingle-exit
10964 @itemx -mno-single-exit
10965 @opindex msingle-exit
10966 @opindex mno-single-exit
10967 Force (do not force) generated code to have a single exit point in each
10971 @node MN10300 Options
10972 @subsection MN10300 Options
10973 @cindex MN10300 options
10975 These @option{-m} options are defined for Matsushita MN10300 architectures:
10980 Generate code to avoid bugs in the multiply instructions for the MN10300
10981 processors. This is the default.
10983 @item -mno-mult-bug
10984 @opindex mno-mult-bug
10985 Do not generate code to avoid bugs in the multiply instructions for the
10986 MN10300 processors.
10990 Generate code which uses features specific to the AM33 processor.
10994 Do not generate code which uses features specific to the AM33 processor. This
10997 @item -mreturn-pointer-on-d0
10998 @opindex mreturn-pointer-on-d0
10999 When generating a function which returns a pointer, return the pointer
11000 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
11001 only in a0, and attempts to call such functions without a prototype
11002 would result in errors. Note that this option is on by default; use
11003 @option{-mno-return-pointer-on-d0} to disable it.
11007 Do not link in the C run-time initialization object file.
11011 Indicate to the linker that it should perform a relaxation optimization pass
11012 to shorten branches, calls and absolute memory addresses. This option only
11013 has an effect when used on the command line for the final link step.
11015 This option makes symbolic debugging impossible.
11019 @subsection MT Options
11022 These @option{-m} options are defined for Morpho MT architectures:
11026 @item -march=@var{cpu-type}
11028 Generate code that will run on @var{cpu-type}, which is the name of a system
11029 representing a certain processor type. Possible values for
11030 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11031 @samp{ms1-16-003} and @samp{ms2}.
11033 When this option is not used, the default is @option{-march=ms1-16-002}.
11037 Use byte loads and stores when generating code.
11041 Do not use byte loads and stores when generating code.
11045 Use simulator runtime
11049 Do not link in the C run-time initialization object file
11050 @file{crti.o}. Other run-time initialization and termination files
11051 such as @file{startup.o} and @file{exit.o} are still included on the
11052 linker command line.
11056 @node PDP-11 Options
11057 @subsection PDP-11 Options
11058 @cindex PDP-11 Options
11060 These options are defined for the PDP-11:
11065 Use hardware FPP floating point. This is the default. (FIS floating
11066 point on the PDP-11/40 is not supported.)
11069 @opindex msoft-float
11070 Do not use hardware floating point.
11074 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11078 Return floating-point results in memory. This is the default.
11082 Generate code for a PDP-11/40.
11086 Generate code for a PDP-11/45. This is the default.
11090 Generate code for a PDP-11/10.
11092 @item -mbcopy-builtin
11093 @opindex bcopy-builtin
11094 Use inline @code{movmemhi} patterns for copying memory. This is the
11099 Do not use inline @code{movmemhi} patterns for copying memory.
11105 Use 16-bit @code{int}. This is the default.
11111 Use 32-bit @code{int}.
11114 @itemx -mno-float32
11116 @opindex mno-float32
11117 Use 64-bit @code{float}. This is the default.
11120 @itemx -mno-float64
11122 @opindex mno-float64
11123 Use 32-bit @code{float}.
11127 Use @code{abshi2} pattern. This is the default.
11131 Do not use @code{abshi2} pattern.
11133 @item -mbranch-expensive
11134 @opindex mbranch-expensive
11135 Pretend that branches are expensive. This is for experimenting with
11136 code generation only.
11138 @item -mbranch-cheap
11139 @opindex mbranch-cheap
11140 Do not pretend that branches are expensive. This is the default.
11144 Generate code for a system with split I&D@.
11148 Generate code for a system without split I&D@. This is the default.
11152 Use Unix assembler syntax. This is the default when configured for
11153 @samp{pdp11-*-bsd}.
11157 Use DEC assembler syntax. This is the default when configured for any
11158 PDP-11 target other than @samp{pdp11-*-bsd}.
11161 @node PowerPC Options
11162 @subsection PowerPC Options
11163 @cindex PowerPC options
11165 These are listed under @xref{RS/6000 and PowerPC Options}.
11167 @node RS/6000 and PowerPC Options
11168 @subsection IBM RS/6000 and PowerPC Options
11169 @cindex RS/6000 and PowerPC Options
11170 @cindex IBM RS/6000 and PowerPC Options
11172 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11179 @itemx -mno-powerpc
11180 @itemx -mpowerpc-gpopt
11181 @itemx -mno-powerpc-gpopt
11182 @itemx -mpowerpc-gfxopt
11183 @itemx -mno-powerpc-gfxopt
11185 @itemx -mno-powerpc64
11189 @itemx -mno-popcntb
11195 @opindex mno-power2
11197 @opindex mno-powerpc
11198 @opindex mpowerpc-gpopt
11199 @opindex mno-powerpc-gpopt
11200 @opindex mpowerpc-gfxopt
11201 @opindex mno-powerpc-gfxopt
11202 @opindex mpowerpc64
11203 @opindex mno-powerpc64
11207 @opindex mno-popcntb
11210 GCC supports two related instruction set architectures for the
11211 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11212 instructions supported by the @samp{rios} chip set used in the original
11213 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11214 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11215 the IBM 4xx, 6xx, and follow-on microprocessors.
11217 Neither architecture is a subset of the other. However there is a
11218 large common subset of instructions supported by both. An MQ
11219 register is included in processors supporting the POWER architecture.
11221 You use these options to specify which instructions are available on the
11222 processor you are using. The default value of these options is
11223 determined when configuring GCC@. Specifying the
11224 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11225 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11226 rather than the options listed above.
11228 The @option{-mpower} option allows GCC to generate instructions that
11229 are found only in the POWER architecture and to use the MQ register.
11230 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11231 to generate instructions that are present in the POWER2 architecture but
11232 not the original POWER architecture.
11234 The @option{-mpowerpc} option allows GCC to generate instructions that
11235 are found only in the 32-bit subset of the PowerPC architecture.
11236 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11237 GCC to use the optional PowerPC architecture instructions in the
11238 General Purpose group, including floating-point square root. Specifying
11239 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11240 use the optional PowerPC architecture instructions in the Graphics
11241 group, including floating-point select.
11243 The @option{-mmfcrf} option allows GCC to generate the move from
11244 condition register field instruction implemented on the POWER4
11245 processor and other processors that support the PowerPC V2.01
11247 The @option{-mpopcntb} option allows GCC to generate the popcount and
11248 double precision FP reciprocal estimate instruction implemented on the
11249 POWER5 processor and other processors that support the PowerPC V2.02
11251 The @option{-mfprnd} option allows GCC to generate the FP round to
11252 integer instructions implemented on the POWER5+ processor and other
11253 processors that support the PowerPC V2.03 architecture.
11255 The @option{-mpowerpc64} option allows GCC to generate the additional
11256 64-bit instructions that are found in the full PowerPC64 architecture
11257 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11258 @option{-mno-powerpc64}.
11260 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11261 will use only the instructions in the common subset of both
11262 architectures plus some special AIX common-mode calls, and will not use
11263 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11264 permits GCC to use any instruction from either architecture and to
11265 allow use of the MQ register; specify this for the Motorola MPC601.
11267 @item -mnew-mnemonics
11268 @itemx -mold-mnemonics
11269 @opindex mnew-mnemonics
11270 @opindex mold-mnemonics
11271 Select which mnemonics to use in the generated assembler code. With
11272 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11273 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11274 assembler mnemonics defined for the POWER architecture. Instructions
11275 defined in only one architecture have only one mnemonic; GCC uses that
11276 mnemonic irrespective of which of these options is specified.
11278 GCC defaults to the mnemonics appropriate for the architecture in
11279 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11280 value of these option. Unless you are building a cross-compiler, you
11281 should normally not specify either @option{-mnew-mnemonics} or
11282 @option{-mold-mnemonics}, but should instead accept the default.
11284 @item -mcpu=@var{cpu_type}
11286 Set architecture type, register usage, choice of mnemonics, and
11287 instruction scheduling parameters for machine type @var{cpu_type}.
11288 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11289 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11290 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11291 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11292 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11293 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11294 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11295 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
11296 @samp{common}, @samp{powerpc}, @samp{powerpc64},
11297 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11299 @option{-mcpu=common} selects a completely generic processor. Code
11300 generated under this option will run on any POWER or PowerPC processor.
11301 GCC will use only the instructions in the common subset of both
11302 architectures, and will not use the MQ register. GCC assumes a generic
11303 processor model for scheduling purposes.
11305 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11306 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11307 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11308 types, with an appropriate, generic processor model assumed for
11309 scheduling purposes.
11311 The other options specify a specific processor. Code generated under
11312 those options will run best on that processor, and may not run at all on
11315 The @option{-mcpu} options automatically enable or disable the
11316 following options: @option{-maltivec}, @option{-mfprnd},
11317 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11318 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11319 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11320 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw}, @option{dlmzb}.
11321 The particular options
11322 set for any particular CPU will vary between compiler versions,
11323 depending on what setting seems to produce optimal code for that CPU;
11324 it doesn't necessarily reflect the actual hardware's capabilities. If
11325 you wish to set an individual option to a particular value, you may
11326 specify it after the @option{-mcpu} option, like @samp{-mcpu=970
11329 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11330 not enabled or disabled by the @option{-mcpu} option at present because
11331 AIX does not have full support for these options. You may still
11332 enable or disable them individually if you're sure it'll work in your
11335 @item -mtune=@var{cpu_type}
11337 Set the instruction scheduling parameters for machine type
11338 @var{cpu_type}, but do not set the architecture type, register usage, or
11339 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11340 values for @var{cpu_type} are used for @option{-mtune} as for
11341 @option{-mcpu}. If both are specified, the code generated will use the
11342 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11343 scheduling parameters set by @option{-mtune}.
11349 Generate code to compute division as reciprocal estimate and iterative
11350 refinement, creating opportunities for increased throughput. This
11351 feature requires: optional PowerPC Graphics instruction set for single
11352 precision and FRE instruction for double precision, assuming divides
11353 cannot generate user-visible traps, and the domain values not include
11354 Infinities, denormals or zero denominator.
11357 @itemx -mno-altivec
11359 @opindex mno-altivec
11360 Generate code that uses (does not use) AltiVec instructions, and also
11361 enable the use of built-in functions that allow more direct access to
11362 the AltiVec instruction set. You may also need to set
11363 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11369 @opindex mno-vrsave
11370 Generate VRSAVE instructions when generating AltiVec code.
11373 @opindex msecure-plt
11374 Generate code that allows ld and ld.so to build executables and shared
11375 libraries with non-exec .plt and .got sections. This is a PowerPC
11376 32-bit SYSV ABI option.
11380 Generate code that uses a BSS .plt section that ld.so fills in, and
11381 requires .plt and .got sections that are both writable and executable.
11382 This is a PowerPC 32-bit SYSV ABI option.
11388 This switch enables or disables the generation of ISEL instructions.
11390 @item -misel=@var{yes/no}
11391 This switch has been deprecated. Use @option{-misel} and
11392 @option{-mno-isel} instead.
11398 This switch enables or disables the generation of SPE simd
11401 @item -mspe=@var{yes/no}
11402 This option has been deprecated. Use @option{-mspe} and
11403 @option{-mno-spe} instead.
11405 @item -mfloat-gprs=@var{yes/single/double/no}
11406 @itemx -mfloat-gprs
11407 @opindex mfloat-gprs
11408 This switch enables or disables the generation of floating point
11409 operations on the general purpose registers for architectures that
11412 The argument @var{yes} or @var{single} enables the use of
11413 single-precision floating point operations.
11415 The argument @var{double} enables the use of single and
11416 double-precision floating point operations.
11418 The argument @var{no} disables floating point operations on the
11419 general purpose registers.
11421 This option is currently only available on the MPC854x.
11427 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11428 targets (including GNU/Linux). The 32-bit environment sets int, long
11429 and pointer to 32 bits and generates code that runs on any PowerPC
11430 variant. The 64-bit environment sets int to 32 bits and long and
11431 pointer to 64 bits, and generates code for PowerPC64, as for
11432 @option{-mpowerpc64}.
11435 @itemx -mno-fp-in-toc
11436 @itemx -mno-sum-in-toc
11437 @itemx -mminimal-toc
11439 @opindex mno-fp-in-toc
11440 @opindex mno-sum-in-toc
11441 @opindex mminimal-toc
11442 Modify generation of the TOC (Table Of Contents), which is created for
11443 every executable file. The @option{-mfull-toc} option is selected by
11444 default. In that case, GCC will allocate at least one TOC entry for
11445 each unique non-automatic variable reference in your program. GCC
11446 will also place floating-point constants in the TOC@. However, only
11447 16,384 entries are available in the TOC@.
11449 If you receive a linker error message that saying you have overflowed
11450 the available TOC space, you can reduce the amount of TOC space used
11451 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11452 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11453 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11454 generate code to calculate the sum of an address and a constant at
11455 run-time instead of putting that sum into the TOC@. You may specify one
11456 or both of these options. Each causes GCC to produce very slightly
11457 slower and larger code at the expense of conserving TOC space.
11459 If you still run out of space in the TOC even when you specify both of
11460 these options, specify @option{-mminimal-toc} instead. This option causes
11461 GCC to make only one TOC entry for every file. When you specify this
11462 option, GCC will produce code that is slower and larger but which
11463 uses extremely little TOC space. You may wish to use this option
11464 only on files that contain less frequently executed code.
11470 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11471 @code{long} type, and the infrastructure needed to support them.
11472 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11473 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11474 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11477 @itemx -mno-xl-compat
11478 @opindex mxl-compat
11479 @opindex mno-xl-compat
11480 Produce code that conforms more closely to IBM XL compiler semantics
11481 when using AIX-compatible ABI. Pass floating-point arguments to
11482 prototyped functions beyond the register save area (RSA) on the stack
11483 in addition to argument FPRs. Do not assume that most significant
11484 double in 128-bit long double value is properly rounded when comparing
11485 values and converting to double. Use XL symbol names for long double
11488 The AIX calling convention was extended but not initially documented to
11489 handle an obscure K&R C case of calling a function that takes the
11490 address of its arguments with fewer arguments than declared. IBM XL
11491 compilers access floating point arguments which do not fit in the
11492 RSA from the stack when a subroutine is compiled without
11493 optimization. Because always storing floating-point arguments on the
11494 stack is inefficient and rarely needed, this option is not enabled by
11495 default and only is necessary when calling subroutines compiled by IBM
11496 XL compilers without optimization.
11500 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11501 application written to use message passing with special startup code to
11502 enable the application to run. The system must have PE installed in the
11503 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11504 must be overridden with the @option{-specs=} option to specify the
11505 appropriate directory location. The Parallel Environment does not
11506 support threads, so the @option{-mpe} option and the @option{-pthread}
11507 option are incompatible.
11509 @item -malign-natural
11510 @itemx -malign-power
11511 @opindex malign-natural
11512 @opindex malign-power
11513 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11514 @option{-malign-natural} overrides the ABI-defined alignment of larger
11515 types, such as floating-point doubles, on their natural size-based boundary.
11516 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11517 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11519 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11523 @itemx -mhard-float
11524 @opindex msoft-float
11525 @opindex mhard-float
11526 Generate code that does not use (uses) the floating-point register set.
11527 Software floating point emulation is provided if you use the
11528 @option{-msoft-float} option, and pass the option to GCC when linking.
11531 @itemx -mno-multiple
11533 @opindex mno-multiple
11534 Generate code that uses (does not use) the load multiple word
11535 instructions and the store multiple word instructions. These
11536 instructions are generated by default on POWER systems, and not
11537 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11538 endian PowerPC systems, since those instructions do not work when the
11539 processor is in little endian mode. The exceptions are PPC740 and
11540 PPC750 which permit the instructions usage in little endian mode.
11545 @opindex mno-string
11546 Generate code that uses (does not use) the load string instructions
11547 and the store string word instructions to save multiple registers and
11548 do small block moves. These instructions are generated by default on
11549 POWER systems, and not generated on PowerPC systems. Do not use
11550 @option{-mstring} on little endian PowerPC systems, since those
11551 instructions do not work when the processor is in little endian mode.
11552 The exceptions are PPC740 and PPC750 which permit the instructions
11553 usage in little endian mode.
11558 @opindex mno-update
11559 Generate code that uses (does not use) the load or store instructions
11560 that update the base register to the address of the calculated memory
11561 location. These instructions are generated by default. If you use
11562 @option{-mno-update}, there is a small window between the time that the
11563 stack pointer is updated and the address of the previous frame is
11564 stored, which means code that walks the stack frame across interrupts or
11565 signals may get corrupted data.
11568 @itemx -mno-fused-madd
11569 @opindex mfused-madd
11570 @opindex mno-fused-madd
11571 Generate code that uses (does not use) the floating point multiply and
11572 accumulate instructions. These instructions are generated by default if
11573 hardware floating is used.
11579 Generate code that uses (does not use) the half-word multiply and
11580 multiply-accumulate instructions on the IBM 405 and 440 processors.
11581 These instructions are generated by default when targetting those
11588 Generate code that uses (does not use) the string-search @samp{dlmzb}
11589 instruction on the IBM 405 and 440 processors. This instruction is
11590 generated by default when targetting those processors.
11592 @item -mno-bit-align
11594 @opindex mno-bit-align
11595 @opindex mbit-align
11596 On System V.4 and embedded PowerPC systems do not (do) force structures
11597 and unions that contain bit-fields to be aligned to the base type of the
11600 For example, by default a structure containing nothing but 8
11601 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11602 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11603 the structure would be aligned to a 1 byte boundary and be one byte in
11606 @item -mno-strict-align
11607 @itemx -mstrict-align
11608 @opindex mno-strict-align
11609 @opindex mstrict-align
11610 On System V.4 and embedded PowerPC systems do not (do) assume that
11611 unaligned memory references will be handled by the system.
11613 @item -mrelocatable
11614 @itemx -mno-relocatable
11615 @opindex mrelocatable
11616 @opindex mno-relocatable
11617 On embedded PowerPC systems generate code that allows (does not allow)
11618 the program to be relocated to a different address at runtime. If you
11619 use @option{-mrelocatable} on any module, all objects linked together must
11620 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11622 @item -mrelocatable-lib
11623 @itemx -mno-relocatable-lib
11624 @opindex mrelocatable-lib
11625 @opindex mno-relocatable-lib
11626 On embedded PowerPC systems generate code that allows (does not allow)
11627 the program to be relocated to a different address at runtime. Modules
11628 compiled with @option{-mrelocatable-lib} can be linked with either modules
11629 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11630 with modules compiled with the @option{-mrelocatable} options.
11636 On System V.4 and embedded PowerPC systems do not (do) assume that
11637 register 2 contains a pointer to a global area pointing to the addresses
11638 used in the program.
11641 @itemx -mlittle-endian
11643 @opindex mlittle-endian
11644 On System V.4 and embedded PowerPC systems compile code for the
11645 processor in little endian mode. The @option{-mlittle-endian} option is
11646 the same as @option{-mlittle}.
11649 @itemx -mbig-endian
11651 @opindex mbig-endian
11652 On System V.4 and embedded PowerPC systems compile code for the
11653 processor in big endian mode. The @option{-mbig-endian} option is
11654 the same as @option{-mbig}.
11656 @item -mdynamic-no-pic
11657 @opindex mdynamic-no-pic
11658 On Darwin and Mac OS X systems, compile code so that it is not
11659 relocatable, but that its external references are relocatable. The
11660 resulting code is suitable for applications, but not shared
11663 @item -mprioritize-restricted-insns=@var{priority}
11664 @opindex mprioritize-restricted-insns
11665 This option controls the priority that is assigned to
11666 dispatch-slot restricted instructions during the second scheduling
11667 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11668 @var{no/highest/second-highest} priority to dispatch slot restricted
11671 @item -msched-costly-dep=@var{dependence_type}
11672 @opindex msched-costly-dep
11673 This option controls which dependences are considered costly
11674 by the target during instruction scheduling. The argument
11675 @var{dependence_type} takes one of the following values:
11676 @var{no}: no dependence is costly,
11677 @var{all}: all dependences are costly,
11678 @var{true_store_to_load}: a true dependence from store to load is costly,
11679 @var{store_to_load}: any dependence from store to load is costly,
11680 @var{number}: any dependence which latency >= @var{number} is costly.
11682 @item -minsert-sched-nops=@var{scheme}
11683 @opindex minsert-sched-nops
11684 This option controls which nop insertion scheme will be used during
11685 the second scheduling pass. The argument @var{scheme} takes one of the
11687 @var{no}: Don't insert nops.
11688 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11689 according to the scheduler's grouping.
11690 @var{regroup_exact}: Insert nops to force costly dependent insns into
11691 separate groups. Insert exactly as many nops as needed to force an insn
11692 to a new group, according to the estimated processor grouping.
11693 @var{number}: Insert nops to force costly dependent insns into
11694 separate groups. Insert @var{number} nops to force an insn to a new group.
11697 @opindex mcall-sysv
11698 On System V.4 and embedded PowerPC systems compile code using calling
11699 conventions that adheres to the March 1995 draft of the System V
11700 Application Binary Interface, PowerPC processor supplement. This is the
11701 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11703 @item -mcall-sysv-eabi
11704 @opindex mcall-sysv-eabi
11705 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11707 @item -mcall-sysv-noeabi
11708 @opindex mcall-sysv-noeabi
11709 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11711 @item -mcall-solaris
11712 @opindex mcall-solaris
11713 On System V.4 and embedded PowerPC systems compile code for the Solaris
11717 @opindex mcall-linux
11718 On System V.4 and embedded PowerPC systems compile code for the
11719 Linux-based GNU system.
11723 On System V.4 and embedded PowerPC systems compile code for the
11724 Hurd-based GNU system.
11726 @item -mcall-netbsd
11727 @opindex mcall-netbsd
11728 On System V.4 and embedded PowerPC systems compile code for the
11729 NetBSD operating system.
11731 @item -maix-struct-return
11732 @opindex maix-struct-return
11733 Return all structures in memory (as specified by the AIX ABI)@.
11735 @item -msvr4-struct-return
11736 @opindex msvr4-struct-return
11737 Return structures smaller than 8 bytes in registers (as specified by the
11740 @item -mabi=@var{abi-type}
11742 Extend the current ABI with a particular extension, or remove such extension.
11743 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11744 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
11748 Extend the current ABI with SPE ABI extensions. This does not change
11749 the default ABI, instead it adds the SPE ABI extensions to the current
11753 @opindex mabi=no-spe
11754 Disable Booke SPE ABI extensions for the current ABI@.
11756 @item -mabi=ibmlongdouble
11757 @opindex mabi=ibmlongdouble
11758 Change the current ABI to use IBM extended precision long double.
11759 This is a PowerPC 32-bit SYSV ABI option.
11761 @item -mabi=ieeelongdouble
11762 @opindex mabi=ieeelongdouble
11763 Change the current ABI to use IEEE extended precision long double.
11764 This is a PowerPC 32-bit Linux ABI option.
11767 @itemx -mno-prototype
11768 @opindex mprototype
11769 @opindex mno-prototype
11770 On System V.4 and embedded PowerPC systems assume that all calls to
11771 variable argument functions are properly prototyped. Otherwise, the
11772 compiler must insert an instruction before every non prototyped call to
11773 set or clear bit 6 of the condition code register (@var{CR}) to
11774 indicate whether floating point values were passed in the floating point
11775 registers in case the function takes a variable arguments. With
11776 @option{-mprototype}, only calls to prototyped variable argument functions
11777 will set or clear the bit.
11781 On embedded PowerPC systems, assume that the startup module is called
11782 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11783 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11788 On embedded PowerPC systems, assume that the startup module is called
11789 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11794 On embedded PowerPC systems, assume that the startup module is called
11795 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11798 @item -myellowknife
11799 @opindex myellowknife
11800 On embedded PowerPC systems, assume that the startup module is called
11801 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11806 On System V.4 and embedded PowerPC systems, specify that you are
11807 compiling for a VxWorks system.
11811 Specify that you are compiling for the WindISS simulation environment.
11815 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11816 header to indicate that @samp{eabi} extended relocations are used.
11822 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11823 Embedded Applications Binary Interface (eabi) which is a set of
11824 modifications to the System V.4 specifications. Selecting @option{-meabi}
11825 means that the stack is aligned to an 8 byte boundary, a function
11826 @code{__eabi} is called to from @code{main} to set up the eabi
11827 environment, and the @option{-msdata} option can use both @code{r2} and
11828 @code{r13} to point to two separate small data areas. Selecting
11829 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11830 do not call an initialization function from @code{main}, and the
11831 @option{-msdata} option will only use @code{r13} to point to a single
11832 small data area. The @option{-meabi} option is on by default if you
11833 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11836 @opindex msdata=eabi
11837 On System V.4 and embedded PowerPC systems, put small initialized
11838 @code{const} global and static data in the @samp{.sdata2} section, which
11839 is pointed to by register @code{r2}. Put small initialized
11840 non-@code{const} global and static data in the @samp{.sdata} section,
11841 which is pointed to by register @code{r13}. Put small uninitialized
11842 global and static data in the @samp{.sbss} section, which is adjacent to
11843 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11844 incompatible with the @option{-mrelocatable} option. The
11845 @option{-msdata=eabi} option also sets the @option{-memb} option.
11848 @opindex msdata=sysv
11849 On System V.4 and embedded PowerPC systems, put small global and static
11850 data in the @samp{.sdata} section, which is pointed to by register
11851 @code{r13}. Put small uninitialized global and static data in the
11852 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11853 The @option{-msdata=sysv} option is incompatible with the
11854 @option{-mrelocatable} option.
11856 @item -msdata=default
11858 @opindex msdata=default
11860 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11861 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11862 same as @option{-msdata=sysv}.
11865 @opindex msdata-data
11866 On System V.4 and embedded PowerPC systems, put small global
11867 data in the @samp{.sdata} section. Put small uninitialized global
11868 data in the @samp{.sbss} section. Do not use register @code{r13}
11869 to address small data however. This is the default behavior unless
11870 other @option{-msdata} options are used.
11874 @opindex msdata=none
11876 On embedded PowerPC systems, put all initialized global and static data
11877 in the @samp{.data} section, and all uninitialized data in the
11878 @samp{.bss} section.
11882 @cindex smaller data references (PowerPC)
11883 @cindex .sdata/.sdata2 references (PowerPC)
11884 On embedded PowerPC systems, put global and static items less than or
11885 equal to @var{num} bytes into the small data or bss sections instead of
11886 the normal data or bss section. By default, @var{num} is 8. The
11887 @option{-G @var{num}} switch is also passed to the linker.
11888 All modules should be compiled with the same @option{-G @var{num}} value.
11891 @itemx -mno-regnames
11893 @opindex mno-regnames
11894 On System V.4 and embedded PowerPC systems do (do not) emit register
11895 names in the assembly language output using symbolic forms.
11898 @itemx -mno-longcall
11900 @opindex mno-longcall
11901 Default to making all function calls indirectly, using a register, so
11902 that functions which reside further than 32 megabytes (33,554,432
11903 bytes) from the current location can be called. This setting can be
11904 overridden by the @code{shortcall} function attribute, or by
11905 @code{#pragma longcall(0)}.
11907 Some linkers are capable of detecting out-of-range calls and generating
11908 glue code on the fly. On these systems, long calls are unnecessary and
11909 generate slower code. As of this writing, the AIX linker can do this,
11910 as can the GNU linker for PowerPC/64. It is planned to add this feature
11911 to the GNU linker for 32-bit PowerPC systems as well.
11913 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11914 callee, L42'', plus a ``branch island'' (glue code). The two target
11915 addresses represent the callee and the ``branch island''. The
11916 Darwin/PPC linker will prefer the first address and generate a ``bl
11917 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11918 otherwise, the linker will generate ``bl L42'' to call the ``branch
11919 island''. The ``branch island'' is appended to the body of the
11920 calling function; it computes the full 32-bit address of the callee
11923 On Mach-O (Darwin) systems, this option directs the compiler emit to
11924 the glue for every direct call, and the Darwin linker decides whether
11925 to use or discard it.
11927 In the future, we may cause GCC to ignore all longcall specifications
11928 when the linker is known to generate glue.
11932 Adds support for multithreading with the @dfn{pthreads} library.
11933 This option sets flags for both the preprocessor and linker.
11937 @node S/390 and zSeries Options
11938 @subsection S/390 and zSeries Options
11939 @cindex S/390 and zSeries Options
11941 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
11945 @itemx -msoft-float
11946 @opindex mhard-float
11947 @opindex msoft-float
11948 Use (do not use) the hardware floating-point instructions and registers
11949 for floating-point operations. When @option{-msoft-float} is specified,
11950 functions in @file{libgcc.a} will be used to perform floating-point
11951 operations. When @option{-mhard-float} is specified, the compiler
11952 generates IEEE floating-point instructions. This is the default.
11954 @item -mlong-double-64
11955 @itemx -mlong-double-128
11956 @opindex mlong-double-64
11957 @opindex mlong-double-128
11958 These switches control the size of @code{long double} type. A size
11959 of 64bit makes the @code{long double} type equivalent to the @code{double}
11960 type. This is the default.
11963 @itemx -mno-backchain
11964 @opindex mbackchain
11965 @opindex mno-backchain
11966 Store (do not store) the address of the caller's frame as backchain pointer
11967 into the callee's stack frame.
11968 A backchain may be needed to allow debugging using tools that do not understand
11969 DWARF-2 call frame information.
11970 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
11971 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
11972 the backchain is placed into the topmost word of the 96/160 byte register
11975 In general, code compiled with @option{-mbackchain} is call-compatible with
11976 code compiled with @option{-mmo-backchain}; however, use of the backchain
11977 for debugging purposes usually requires that the whole binary is built with
11978 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
11979 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11980 to build a linux kernel use @option{-msoft-float}.
11982 The default is to not maintain the backchain.
11984 @item -mpacked-stack
11985 @item -mno-packed-stack
11986 @opindex mpacked-stack
11987 @opindex mno-packed-stack
11988 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
11989 specified, the compiler uses the all fields of the 96/160 byte register save
11990 area only for their default purpose; unused fields still take up stack space.
11991 When @option{-mpacked-stack} is specified, register save slots are densely
11992 packed at the top of the register save area; unused space is reused for other
11993 purposes, allowing for more efficient use of the available stack space.
11994 However, when @option{-mbackchain} is also in effect, the topmost word of
11995 the save area is always used to store the backchain, and the return address
11996 register is always saved two words below the backchain.
11998 As long as the stack frame backchain is not used, code generated with
11999 @option{-mpacked-stack} is call-compatible with code generated with
12000 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
12001 S/390 or zSeries generated code that uses the stack frame backchain at run
12002 time, not just for debugging purposes. Such code is not call-compatible
12003 with code compiled with @option{-mpacked-stack}. Also, note that the
12004 combination of @option{-mbackchain},
12005 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12006 to build a linux kernel use @option{-msoft-float}.
12008 The default is to not use the packed stack layout.
12011 @itemx -mno-small-exec
12012 @opindex msmall-exec
12013 @opindex mno-small-exec
12014 Generate (or do not generate) code using the @code{bras} instruction
12015 to do subroutine calls.
12016 This only works reliably if the total executable size does not
12017 exceed 64k. The default is to use the @code{basr} instruction instead,
12018 which does not have this limitation.
12024 When @option{-m31} is specified, generate code compliant to the
12025 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12026 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12027 particular to generate 64-bit instructions. For the @samp{s390}
12028 targets, the default is @option{-m31}, while the @samp{s390x}
12029 targets default to @option{-m64}.
12035 When @option{-mzarch} is specified, generate code using the
12036 instructions available on z/Architecture.
12037 When @option{-mesa} is specified, generate code using the
12038 instructions available on ESA/390. Note that @option{-mesa} is
12039 not possible with @option{-m64}.
12040 When generating code compliant to the GNU/Linux for S/390 ABI,
12041 the default is @option{-mesa}. When generating code compliant
12042 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12048 Generate (or do not generate) code using the @code{mvcle} instruction
12049 to perform block moves. When @option{-mno-mvcle} is specified,
12050 use a @code{mvc} loop instead. This is the default unless optimizing for
12057 Print (or do not print) additional debug information when compiling.
12058 The default is to not print debug information.
12060 @item -march=@var{cpu-type}
12062 Generate code that will run on @var{cpu-type}, which is the name of a system
12063 representing a certain processor type. Possible values for
12064 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12065 When generating code using the instructions available on z/Architecture,
12066 the default is @option{-march=z900}. Otherwise, the default is
12067 @option{-march=g5}.
12069 @item -mtune=@var{cpu-type}
12071 Tune to @var{cpu-type} everything applicable about the generated code,
12072 except for the ABI and the set of available instructions.
12073 The list of @var{cpu-type} values is the same as for @option{-march}.
12074 The default is the value used for @option{-march}.
12077 @itemx -mno-tpf-trace
12078 @opindex mtpf-trace
12079 @opindex mno-tpf-trace
12080 Generate code that adds (does not add) in TPF OS specific branches to trace
12081 routines in the operating system. This option is off by default, even
12082 when compiling for the TPF OS@.
12085 @itemx -mno-fused-madd
12086 @opindex mfused-madd
12087 @opindex mno-fused-madd
12088 Generate code that uses (does not use) the floating point multiply and
12089 accumulate instructions. These instructions are generated by default if
12090 hardware floating point is used.
12092 @item -mwarn-framesize=@var{framesize}
12093 @opindex mwarn-framesize
12094 Emit a warning if the current function exceeds the given frame size. Because
12095 this is a compile time check it doesn't need to be a real problem when the program
12096 runs. It is intended to identify functions which most probably cause
12097 a stack overflow. It is useful to be used in an environment with limited stack
12098 size e.g.@: the linux kernel.
12100 @item -mwarn-dynamicstack
12101 @opindex mwarn-dynamicstack
12102 Emit a warning if the function calls alloca or uses dynamically
12103 sized arrays. This is generally a bad idea with a limited stack size.
12105 @item -mstack-guard=@var{stack-guard}
12106 @item -mstack-size=@var{stack-size}
12107 @opindex mstack-guard
12108 @opindex mstack-size
12109 These arguments always have to be used in conjunction. If they are present the s390
12110 back end emits additional instructions in the function prologue which trigger a trap
12111 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12112 (remember that the stack on s390 grows downward). These options are intended to
12113 be used to help debugging stack overflow problems. The additionally emitted code
12114 causes only little overhead and hence can also be used in production like systems
12115 without greater performance degradation. The given values have to be exact
12116 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12118 In order to be efficient the extra code makes the assumption that the stack starts
12119 at an address aligned to the value given by @var{stack-size}.
12123 @subsection SH Options
12125 These @samp{-m} options are defined for the SH implementations:
12130 Generate code for the SH1.
12134 Generate code for the SH2.
12137 Generate code for the SH2e.
12141 Generate code for the SH3.
12145 Generate code for the SH3e.
12149 Generate code for the SH4 without a floating-point unit.
12151 @item -m4-single-only
12152 @opindex m4-single-only
12153 Generate code for the SH4 with a floating-point unit that only
12154 supports single-precision arithmetic.
12158 Generate code for the SH4 assuming the floating-point unit is in
12159 single-precision mode by default.
12163 Generate code for the SH4.
12167 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12168 floating-point unit is not used.
12170 @item -m4a-single-only
12171 @opindex m4a-single-only
12172 Generate code for the SH4a, in such a way that no double-precision
12173 floating point operations are used.
12176 @opindex m4a-single
12177 Generate code for the SH4a assuming the floating-point unit is in
12178 single-precision mode by default.
12182 Generate code for the SH4a.
12186 Same as @option{-m4a-nofpu}, except that it implicitly passes
12187 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12188 instructions at the moment.
12192 Compile code for the processor in big endian mode.
12196 Compile code for the processor in little endian mode.
12200 Align doubles at 64-bit boundaries. Note that this changes the calling
12201 conventions, and thus some functions from the standard C library will
12202 not work unless you recompile it first with @option{-mdalign}.
12206 Shorten some address references at link time, when possible; uses the
12207 linker option @option{-relax}.
12211 Use 32-bit offsets in @code{switch} tables. The default is to use
12216 Enable the use of the instruction @code{fmovd}.
12220 Comply with the calling conventions defined by Renesas.
12224 Comply with the calling conventions defined by Renesas.
12228 Comply with the calling conventions defined for GCC before the Renesas
12229 conventions were available. This option is the default for all
12230 targets of the SH toolchain except for @samp{sh-symbianelf}.
12233 @opindex mnomacsave
12234 Mark the @code{MAC} register as call-clobbered, even if
12235 @option{-mhitachi} is given.
12239 Increase IEEE-compliance of floating-point code.
12240 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12241 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12242 comparisons of NANs / infinities incurs extra overhead in every
12243 floating point comparison, therefore the default is set to
12244 @option{-ffinite-math-only}.
12248 Dump instruction size and location in the assembly code.
12251 @opindex mpadstruct
12252 This option is deprecated. It pads structures to multiple of 4 bytes,
12253 which is incompatible with the SH ABI@.
12257 Optimize for space instead of speed. Implied by @option{-Os}.
12260 @opindex mprefergot
12261 When generating position-independent code, emit function calls using
12262 the Global Offset Table instead of the Procedure Linkage Table.
12266 Generate a library function call to invalidate instruction cache
12267 entries, after fixing up a trampoline. This library function call
12268 doesn't assume it can write to the whole memory address space. This
12269 is the default when the target is @code{sh-*-linux*}.
12271 @item -multcost=@var{number}
12272 @opindex multcost=@var{number}
12273 Set the cost to assume for a multiply insn.
12275 @item -mdiv=@var{strategy}
12276 @opindex mdiv=@var{strategy}
12277 Set the division strategy to use for SHmedia code. @var{strategy} must be
12278 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12279 inv:call2, inv:fp .
12280 "fp" performs the operation in floating point. This has a very high latency,
12281 but needs only a few instructions, so it might be a good choice if
12282 your code has enough easily exploitable ILP to allow the compiler to
12283 schedule the floating point instructions together with other instructions.
12284 Division by zero causes a floating point exception.
12285 "inv" uses integer operations to calculate the inverse of the divisor,
12286 and then multiplies the dividend with the inverse. This strategy allows
12287 cse and hoisting of the inverse calculation. Division by zero calculates
12288 an unspecified result, but does not trap.
12289 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12290 have been found, or if the entire operation has been hoisted to the same
12291 place, the last stages of the inverse calculation are intertwined with the
12292 final multiply to reduce the overall latency, at the expense of using a few
12293 more instructions, and thus offering fewer scheduling opportunities with
12295 "call" calls a library function that usually implements the inv:minlat
12297 This gives high code density for m5-*media-nofpu compilations.
12298 "call2" uses a different entry point of the same library function, where it
12299 assumes that a pointer to a lookup table has already been set up, which
12300 exposes the pointer load to cse / code hoisting optimizations.
12301 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12302 code generation, but if the code stays unoptimized, revert to the "call",
12303 "call2", or "fp" strategies, respectively. Note that the
12304 potentially-trapping side effect of division by zero is carried by a
12305 separate instruction, so it is possible that all the integer instructions
12306 are hoisted out, but the marker for the side effect stays where it is.
12307 A recombination to fp operations or a call is not possible in that case.
12308 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12309 that the inverse calculation was nor separated from the multiply, they speed
12310 up division where the dividend fits into 20 bits (plus sign where applicable),
12311 by inserting a test to skip a number of operations in this case; this test
12312 slows down the case of larger dividends. inv20u assumes the case of a such
12313 a small dividend to be unlikely, and inv20l assumes it to be likely.
12315 @item -mdivsi3_libfunc=@var{name}
12316 @opindex mdivsi3_libfunc=@var{name}
12317 Set the name of the library function used for 32 bit signed division to
12318 @var{name}. This only affect the name used in the call and inv:call
12319 division strategies, and the compiler will still expect the same
12320 sets of input/output/clobbered registers as if this option was not present.
12322 @item -madjust-unroll
12323 @opindex madjust-unroll
12324 Throttle unrolling to avoid thrashing target registers.
12325 This option only has an effect if the gcc code base supports the
12326 TARGET_ADJUST_UNROLL_MAX target hook.
12328 @item -mindexed-addressing
12329 @opindex mindexed-addressing
12330 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12331 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12332 semantics for the indexed addressing mode. The architecture allows the
12333 implementation of processors with 64 bit MMU, which the OS could use to
12334 get 32 bit addressing, but since no current hardware implementation supports
12335 this or any other way to make the indexed addressing mode safe to use in
12336 the 32 bit ABI, the default is -mno-indexed-addressing.
12338 @item -mgettrcost=@var{number}
12339 @opindex mgettrcost=@var{number}
12340 Set the cost assumed for the gettr instruction to @var{number}.
12341 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12345 Assume pt* instructions won't trap. This will generally generate better
12346 scheduled code, but is unsafe on current hardware. The current architecture
12347 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12348 This has the unintentional effect of making it unsafe to schedule ptabs /
12349 ptrel before a branch, or hoist it out of a loop. For example,
12350 __do_global_ctors, a part of libgcc that runs constructors at program
12351 startup, calls functions in a list which is delimited by -1. With the
12352 -mpt-fixed option, the ptabs will be done before testing against -1.
12353 That means that all the constructors will be run a bit quicker, but when
12354 the loop comes to the end of the list, the program crashes because ptabs
12355 loads -1 into a target register. Since this option is unsafe for any
12356 hardware implementing the current architecture specification, the default
12357 is -mno-pt-fixed. Unless the user specifies a specific cost with
12358 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12359 this deters register allocation using target registers for storing
12362 @item -minvalid-symbols
12363 @opindex minvalid-symbols
12364 Assume symbols might be invalid. Ordinary function symbols generated by
12365 the compiler will always be valid to load with movi/shori/ptabs or
12366 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12367 to generate symbols that will cause ptabs / ptrel to trap.
12368 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12369 It will then prevent cross-basic-block cse, hoisting and most scheduling
12370 of symbol loads. The default is @option{-mno-invalid-symbols}.
12373 @node SPARC Options
12374 @subsection SPARC Options
12375 @cindex SPARC options
12377 These @samp{-m} options are supported on the SPARC:
12380 @item -mno-app-regs
12382 @opindex mno-app-regs
12384 Specify @option{-mapp-regs} to generate output using the global registers
12385 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12388 To be fully SVR4 ABI compliant at the cost of some performance loss,
12389 specify @option{-mno-app-regs}. You should compile libraries and system
12390 software with this option.
12393 @itemx -mhard-float
12395 @opindex mhard-float
12396 Generate output containing floating point instructions. This is the
12400 @itemx -msoft-float
12402 @opindex msoft-float
12403 Generate output containing library calls for floating point.
12404 @strong{Warning:} the requisite libraries are not available for all SPARC
12405 targets. Normally the facilities of the machine's usual C compiler are
12406 used, but this cannot be done directly in cross-compilation. You must make
12407 your own arrangements to provide suitable library functions for
12408 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12409 @samp{sparclite-*-*} do provide software floating point support.
12411 @option{-msoft-float} changes the calling convention in the output file;
12412 therefore, it is only useful if you compile @emph{all} of a program with
12413 this option. In particular, you need to compile @file{libgcc.a}, the
12414 library that comes with GCC, with @option{-msoft-float} in order for
12417 @item -mhard-quad-float
12418 @opindex mhard-quad-float
12419 Generate output containing quad-word (long double) floating point
12422 @item -msoft-quad-float
12423 @opindex msoft-quad-float
12424 Generate output containing library calls for quad-word (long double)
12425 floating point instructions. The functions called are those specified
12426 in the SPARC ABI@. This is the default.
12428 As of this writing, there are no SPARC implementations that have hardware
12429 support for the quad-word floating point instructions. They all invoke
12430 a trap handler for one of these instructions, and then the trap handler
12431 emulates the effect of the instruction. Because of the trap handler overhead,
12432 this is much slower than calling the ABI library routines. Thus the
12433 @option{-msoft-quad-float} option is the default.
12435 @item -mno-unaligned-doubles
12436 @itemx -munaligned-doubles
12437 @opindex mno-unaligned-doubles
12438 @opindex munaligned-doubles
12439 Assume that doubles have 8 byte alignment. This is the default.
12441 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12442 alignment only if they are contained in another type, or if they have an
12443 absolute address. Otherwise, it assumes they have 4 byte alignment.
12444 Specifying this option avoids some rare compatibility problems with code
12445 generated by other compilers. It is not the default because it results
12446 in a performance loss, especially for floating point code.
12448 @item -mno-faster-structs
12449 @itemx -mfaster-structs
12450 @opindex mno-faster-structs
12451 @opindex mfaster-structs
12452 With @option{-mfaster-structs}, the compiler assumes that structures
12453 should have 8 byte alignment. This enables the use of pairs of
12454 @code{ldd} and @code{std} instructions for copies in structure
12455 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12456 However, the use of this changed alignment directly violates the SPARC
12457 ABI@. Thus, it's intended only for use on targets where the developer
12458 acknowledges that their resulting code will not be directly in line with
12459 the rules of the ABI@.
12461 @item -mimpure-text
12462 @opindex mimpure-text
12463 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12464 the compiler to not pass @option{-z text} to the linker when linking a
12465 shared object. Using this option, you can link position-dependent
12466 code into a shared object.
12468 @option{-mimpure-text} suppresses the ``relocations remain against
12469 allocatable but non-writable sections'' linker error message.
12470 However, the necessary relocations will trigger copy-on-write, and the
12471 shared object is not actually shared across processes. Instead of
12472 using @option{-mimpure-text}, you should compile all source code with
12473 @option{-fpic} or @option{-fPIC}.
12475 This option is only available on SunOS and Solaris.
12477 @item -mcpu=@var{cpu_type}
12479 Set the instruction set, register set, and instruction scheduling parameters
12480 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12481 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12482 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12483 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12484 @samp{ultrasparc3}, and @samp{niagara}.
12486 Default instruction scheduling parameters are used for values that select
12487 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12488 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12490 Here is a list of each supported architecture and their supported
12495 v8: supersparc, hypersparc
12496 sparclite: f930, f934, sparclite86x
12498 v9: ultrasparc, ultrasparc3, niagara
12501 By default (unless configured otherwise), GCC generates code for the V7
12502 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12503 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12504 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12505 SPARCStation 1, 2, IPX etc.
12507 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12508 architecture. The only difference from V7 code is that the compiler emits
12509 the integer multiply and integer divide instructions which exist in SPARC-V8
12510 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12511 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12514 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12515 the SPARC architecture. This adds the integer multiply, integer divide step
12516 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12517 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12518 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12519 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12520 MB86934 chip, which is the more recent SPARClite with FPU@.
12522 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12523 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12524 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12525 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12526 optimizes it for the TEMIC SPARClet chip.
12528 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12529 architecture. This adds 64-bit integer and floating-point move instructions,
12530 3 additional floating-point condition code registers and conditional move
12531 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12532 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12533 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12534 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12535 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12536 Sun UltraSPARC T1 chips.
12538 @item -mtune=@var{cpu_type}
12540 Set the instruction scheduling parameters for machine type
12541 @var{cpu_type}, but do not set the instruction set or register set that the
12542 option @option{-mcpu=@var{cpu_type}} would.
12544 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12545 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12546 that select a particular cpu implementation. Those are @samp{cypress},
12547 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12548 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12549 @samp{ultrasparc3}, and @samp{niagara}.
12554 @opindex mno-v8plus
12555 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12556 difference from the V8 ABI is that the global and out registers are
12557 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12558 mode for all SPARC-V9 processors.
12564 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12565 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12568 These @samp{-m} options are supported in addition to the above
12569 on SPARC-V9 processors in 64-bit environments:
12572 @item -mlittle-endian
12573 @opindex mlittle-endian
12574 Generate code for a processor running in little-endian mode. It is only
12575 available for a few configurations and most notably not on Solaris and Linux.
12581 Generate code for a 32-bit or 64-bit environment.
12582 The 32-bit environment sets int, long and pointer to 32 bits.
12583 The 64-bit environment sets int to 32 bits and long and pointer
12586 @item -mcmodel=medlow
12587 @opindex mcmodel=medlow
12588 Generate code for the Medium/Low code model: 64-bit addresses, programs
12589 must be linked in the low 32 bits of memory. Programs can be statically
12590 or dynamically linked.
12592 @item -mcmodel=medmid
12593 @opindex mcmodel=medmid
12594 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12595 must be linked in the low 44 bits of memory, the text and data segments must
12596 be less than 2GB in size and the data segment must be located within 2GB of
12599 @item -mcmodel=medany
12600 @opindex mcmodel=medany
12601 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12602 may be linked anywhere in memory, the text and data segments must be less
12603 than 2GB in size and the data segment must be located within 2GB of the
12606 @item -mcmodel=embmedany
12607 @opindex mcmodel=embmedany
12608 Generate code for the Medium/Anywhere code model for embedded systems:
12609 64-bit addresses, the text and data segments must be less than 2GB in
12610 size, both starting anywhere in memory (determined at link time). The
12611 global register %g4 points to the base of the data segment. Programs
12612 are statically linked and PIC is not supported.
12615 @itemx -mno-stack-bias
12616 @opindex mstack-bias
12617 @opindex mno-stack-bias
12618 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12619 frame pointer if present, are offset by @minus{}2047 which must be added back
12620 when making stack frame references. This is the default in 64-bit mode.
12621 Otherwise, assume no such offset is present.
12624 These switches are supported in addition to the above on Solaris:
12629 Add support for multithreading using the Solaris threads library. This
12630 option sets flags for both the preprocessor and linker. This option does
12631 not affect the thread safety of object code produced by the compiler or
12632 that of libraries supplied with it.
12636 Add support for multithreading using the POSIX threads library. This
12637 option sets flags for both the preprocessor and linker. This option does
12638 not affect the thread safety of object code produced by the compiler or
12639 that of libraries supplied with it.
12643 This is a synonym for @option{-pthreads}.
12646 @node System V Options
12647 @subsection Options for System V
12649 These additional options are available on System V Release 4 for
12650 compatibility with other compilers on those systems:
12655 Create a shared object.
12656 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12660 Identify the versions of each tool used by the compiler, in a
12661 @code{.ident} assembler directive in the output.
12665 Refrain from adding @code{.ident} directives to the output file (this is
12668 @item -YP,@var{dirs}
12670 Search the directories @var{dirs}, and no others, for libraries
12671 specified with @option{-l}.
12673 @item -Ym,@var{dir}
12675 Look in the directory @var{dir} to find the M4 preprocessor.
12676 The assembler uses this option.
12677 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12678 @c the generic assembler that comes with Solaris takes just -Ym.
12681 @node TMS320C3x/C4x Options
12682 @subsection TMS320C3x/C4x Options
12683 @cindex TMS320C3x/C4x Options
12685 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12689 @item -mcpu=@var{cpu_type}
12691 Set the instruction set, register set, and instruction scheduling
12692 parameters for machine type @var{cpu_type}. Supported values for
12693 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12694 @samp{c44}. The default is @samp{c40} to generate code for the
12699 @itemx -msmall-memory
12701 @opindex mbig-memory
12703 @opindex msmall-memory
12705 Generates code for the big or small memory model. The small memory
12706 model assumed that all data fits into one 64K word page. At run-time
12707 the data page (DP) register must be set to point to the 64K page
12708 containing the .bss and .data program sections. The big memory model is
12709 the default and requires reloading of the DP register for every direct
12716 Allow (disallow) allocation of general integer operands into the block
12717 count register BK@.
12723 Enable (disable) generation of code using decrement and branch,
12724 DBcond(D), instructions. This is enabled by default for the C4x. To be
12725 on the safe side, this is disabled for the C3x, since the maximum
12726 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12727 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12728 that it can utilize the decrement and branch instruction, but will give
12729 up if there is more than one memory reference in the loop. Thus a loop
12730 where the loop counter is decremented can generate slightly more
12731 efficient code, in cases where the RPTB instruction cannot be utilized.
12733 @item -mdp-isr-reload
12735 @opindex mdp-isr-reload
12737 Force the DP register to be saved on entry to an interrupt service
12738 routine (ISR), reloaded to point to the data section, and restored on
12739 exit from the ISR@. This should not be required unless someone has
12740 violated the small memory model by modifying the DP register, say within
12747 For the C3x use the 24-bit MPYI instruction for integer multiplies
12748 instead of a library call to guarantee 32-bit results. Note that if one
12749 of the operands is a constant, then the multiplication will be performed
12750 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12751 then squaring operations are performed inline instead of a library call.
12754 @itemx -mno-fast-fix
12756 @opindex mno-fast-fix
12757 The C3x/C4x FIX instruction to convert a floating point value to an
12758 integer value chooses the nearest integer less than or equal to the
12759 floating point value rather than to the nearest integer. Thus if the
12760 floating point number is negative, the result will be incorrectly
12761 truncated an additional code is necessary to detect and correct this
12762 case. This option can be used to disable generation of the additional
12763 code required to correct the result.
12769 Enable (disable) generation of repeat block sequences using the RPTB
12770 instruction for zero overhead looping. The RPTB construct is only used
12771 for innermost loops that do not call functions or jump across the loop
12772 boundaries. There is no advantage having nested RPTB loops due to the
12773 overhead required to save and restore the RC, RS, and RE registers.
12774 This is enabled by default with @option{-O2}.
12776 @item -mrpts=@var{count}
12780 Enable (disable) the use of the single instruction repeat instruction
12781 RPTS@. If a repeat block contains a single instruction, and the loop
12782 count can be guaranteed to be less than the value @var{count}, GCC will
12783 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12784 then a RPTS will be emitted even if the loop count cannot be determined
12785 at compile time. Note that the repeated instruction following RPTS does
12786 not have to be reloaded from memory each iteration, thus freeing up the
12787 CPU buses for operands. However, since interrupts are blocked by this
12788 instruction, it is disabled by default.
12790 @item -mloop-unsigned
12791 @itemx -mno-loop-unsigned
12792 @opindex mloop-unsigned
12793 @opindex mno-loop-unsigned
12794 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12795 is @math{2^{31} + 1} since these instructions test if the iteration count is
12796 negative to terminate the loop. If the iteration count is unsigned
12797 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12798 exceeded. This switch allows an unsigned iteration count.
12802 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12803 with. This also enforces compatibility with the API employed by the TI
12804 C3x C compiler. For example, long doubles are passed as structures
12805 rather than in floating point registers.
12811 Generate code that uses registers (stack) for passing arguments to functions.
12812 By default, arguments are passed in registers where possible rather
12813 than by pushing arguments on to the stack.
12815 @item -mparallel-insns
12816 @itemx -mno-parallel-insns
12817 @opindex mparallel-insns
12818 @opindex mno-parallel-insns
12819 Allow the generation of parallel instructions. This is enabled by
12820 default with @option{-O2}.
12822 @item -mparallel-mpy
12823 @itemx -mno-parallel-mpy
12824 @opindex mparallel-mpy
12825 @opindex mno-parallel-mpy
12826 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12827 provided @option{-mparallel-insns} is also specified. These instructions have
12828 tight register constraints which can pessimize the code generation
12829 of large functions.
12834 @subsection V850 Options
12835 @cindex V850 Options
12837 These @samp{-m} options are defined for V850 implementations:
12841 @itemx -mno-long-calls
12842 @opindex mlong-calls
12843 @opindex mno-long-calls
12844 Treat all calls as being far away (near). If calls are assumed to be
12845 far away, the compiler will always load the functions address up into a
12846 register, and call indirect through the pointer.
12852 Do not optimize (do optimize) basic blocks that use the same index
12853 pointer 4 or more times to copy pointer into the @code{ep} register, and
12854 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12855 option is on by default if you optimize.
12857 @item -mno-prolog-function
12858 @itemx -mprolog-function
12859 @opindex mno-prolog-function
12860 @opindex mprolog-function
12861 Do not use (do use) external functions to save and restore registers
12862 at the prologue and epilogue of a function. The external functions
12863 are slower, but use less code space if more than one function saves
12864 the same number of registers. The @option{-mprolog-function} option
12865 is on by default if you optimize.
12869 Try to make the code as small as possible. At present, this just turns
12870 on the @option{-mep} and @option{-mprolog-function} options.
12872 @item -mtda=@var{n}
12874 Put static or global variables whose size is @var{n} bytes or less into
12875 the tiny data area that register @code{ep} points to. The tiny data
12876 area can hold up to 256 bytes in total (128 bytes for byte references).
12878 @item -msda=@var{n}
12880 Put static or global variables whose size is @var{n} bytes or less into
12881 the small data area that register @code{gp} points to. The small data
12882 area can hold up to 64 kilobytes.
12884 @item -mzda=@var{n}
12886 Put static or global variables whose size is @var{n} bytes or less into
12887 the first 32 kilobytes of memory.
12891 Specify that the target processor is the V850.
12894 @opindex mbig-switch
12895 Generate code suitable for big switch tables. Use this option only if
12896 the assembler/linker complain about out of range branches within a switch
12901 This option will cause r2 and r5 to be used in the code generated by
12902 the compiler. This setting is the default.
12904 @item -mno-app-regs
12905 @opindex mno-app-regs
12906 This option will cause r2 and r5 to be treated as fixed registers.
12910 Specify that the target processor is the V850E1. The preprocessor
12911 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
12912 this option is used.
12916 Specify that the target processor is the V850E@. The preprocessor
12917 constant @samp{__v850e__} will be defined if this option is used.
12919 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
12920 are defined then a default target processor will be chosen and the
12921 relevant @samp{__v850*__} preprocessor constant will be defined.
12923 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
12924 defined, regardless of which processor variant is the target.
12926 @item -mdisable-callt
12927 @opindex mdisable-callt
12928 This option will suppress generation of the CALLT instruction for the
12929 v850e and v850e1 flavors of the v850 architecture. The default is
12930 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
12935 @subsection VAX Options
12936 @cindex VAX options
12938 These @samp{-m} options are defined for the VAX:
12943 Do not output certain jump instructions (@code{aobleq} and so on)
12944 that the Unix assembler for the VAX cannot handle across long
12949 Do output those jump instructions, on the assumption that you
12950 will assemble with the GNU assembler.
12954 Output code for g-format floating point numbers instead of d-format.
12957 @node x86-64 Options
12958 @subsection x86-64 Options
12959 @cindex x86-64 options
12961 These are listed under @xref{i386 and x86-64 Options}.
12963 @node Xstormy16 Options
12964 @subsection Xstormy16 Options
12965 @cindex Xstormy16 Options
12967 These options are defined for Xstormy16:
12972 Choose startup files and linker script suitable for the simulator.
12975 @node Xtensa Options
12976 @subsection Xtensa Options
12977 @cindex Xtensa Options
12979 These options are supported for Xtensa targets:
12983 @itemx -mno-const16
12985 @opindex mno-const16
12986 Enable or disable use of @code{CONST16} instructions for loading
12987 constant values. The @code{CONST16} instruction is currently not a
12988 standard option from Tensilica. When enabled, @code{CONST16}
12989 instructions are always used in place of the standard @code{L32R}
12990 instructions. The use of @code{CONST16} is enabled by default only if
12991 the @code{L32R} instruction is not available.
12994 @itemx -mno-fused-madd
12995 @opindex mfused-madd
12996 @opindex mno-fused-madd
12997 Enable or disable use of fused multiply/add and multiply/subtract
12998 instructions in the floating-point option. This has no effect if the
12999 floating-point option is not also enabled. Disabling fused multiply/add
13000 and multiply/subtract instructions forces the compiler to use separate
13001 instructions for the multiply and add/subtract operations. This may be
13002 desirable in some cases where strict IEEE 754-compliant results are
13003 required: the fused multiply add/subtract instructions do not round the
13004 intermediate result, thereby producing results with @emph{more} bits of
13005 precision than specified by the IEEE standard. Disabling fused multiply
13006 add/subtract instructions also ensures that the program output is not
13007 sensitive to the compiler's ability to combine multiply and add/subtract
13010 @item -mtext-section-literals
13011 @itemx -mno-text-section-literals
13012 @opindex mtext-section-literals
13013 @opindex mno-text-section-literals
13014 Control the treatment of literal pools. The default is
13015 @option{-mno-text-section-literals}, which places literals in a separate
13016 section in the output file. This allows the literal pool to be placed
13017 in a data RAM/ROM, and it also allows the linker to combine literal
13018 pools from separate object files to remove redundant literals and
13019 improve code size. With @option{-mtext-section-literals}, the literals
13020 are interspersed in the text section in order to keep them as close as
13021 possible to their references. This may be necessary for large assembly
13024 @item -mtarget-align
13025 @itemx -mno-target-align
13026 @opindex mtarget-align
13027 @opindex mno-target-align
13028 When this option is enabled, GCC instructs the assembler to
13029 automatically align instructions to reduce branch penalties at the
13030 expense of some code density. The assembler attempts to widen density
13031 instructions to align branch targets and the instructions following call
13032 instructions. If there are not enough preceding safe density
13033 instructions to align a target, no widening will be performed. The
13034 default is @option{-mtarget-align}. These options do not affect the
13035 treatment of auto-aligned instructions like @code{LOOP}, which the
13036 assembler will always align, either by widening density instructions or
13037 by inserting no-op instructions.
13040 @itemx -mno-longcalls
13041 @opindex mlongcalls
13042 @opindex mno-longcalls
13043 When this option is enabled, GCC instructs the assembler to translate
13044 direct calls to indirect calls unless it can determine that the target
13045 of a direct call is in the range allowed by the call instruction. This
13046 translation typically occurs for calls to functions in other source
13047 files. Specifically, the assembler translates a direct @code{CALL}
13048 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13049 The default is @option{-mno-longcalls}. This option should be used in
13050 programs where the call target can potentially be out of range. This
13051 option is implemented in the assembler, not the compiler, so the
13052 assembly code generated by GCC will still show direct call
13053 instructions---look at the disassembled object code to see the actual
13054 instructions. Note that the assembler will use an indirect call for
13055 every cross-file call, not just those that really will be out of range.
13058 @node zSeries Options
13059 @subsection zSeries Options
13060 @cindex zSeries options
13062 These are listed under @xref{S/390 and zSeries Options}.
13064 @node Code Gen Options
13065 @section Options for Code Generation Conventions
13066 @cindex code generation conventions
13067 @cindex options, code generation
13068 @cindex run-time options
13070 These machine-independent options control the interface conventions
13071 used in code generation.
13073 Most of them have both positive and negative forms; the negative form
13074 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13075 one of the forms is listed---the one which is not the default. You
13076 can figure out the other form by either removing @samp{no-} or adding
13080 @item -fbounds-check
13081 @opindex fbounds-check
13082 For front-ends that support it, generate additional code to check that
13083 indices used to access arrays are within the declared range. This is
13084 currently only supported by the Java and Fortran 77 front-ends, where
13085 this option defaults to true and false respectively.
13089 This option generates traps for signed overflow on addition, subtraction,
13090 multiplication operations.
13094 This option instructs the compiler to assume that signed arithmetic
13095 overflow of addition, subtraction and multiplication wraps around
13096 using twos-complement representation. This flag enables some optimizations
13097 and disables others. This option is enabled by default for the Java
13098 front-end, as required by the Java language specification.
13101 @opindex fexceptions
13102 Enable exception handling. Generates extra code needed to propagate
13103 exceptions. For some targets, this implies GCC will generate frame
13104 unwind information for all functions, which can produce significant data
13105 size overhead, although it does not affect execution. If you do not
13106 specify this option, GCC will enable it by default for languages like
13107 C++ which normally require exception handling, and disable it for
13108 languages like C that do not normally require it. However, you may need
13109 to enable this option when compiling C code that needs to interoperate
13110 properly with exception handlers written in C++. You may also wish to
13111 disable this option if you are compiling older C++ programs that don't
13112 use exception handling.
13114 @item -fnon-call-exceptions
13115 @opindex fnon-call-exceptions
13116 Generate code that allows trapping instructions to throw exceptions.
13117 Note that this requires platform-specific runtime support that does
13118 not exist everywhere. Moreover, it only allows @emph{trapping}
13119 instructions to throw exceptions, i.e.@: memory references or floating
13120 point instructions. It does not allow exceptions to be thrown from
13121 arbitrary signal handlers such as @code{SIGALRM}.
13123 @item -funwind-tables
13124 @opindex funwind-tables
13125 Similar to @option{-fexceptions}, except that it will just generate any needed
13126 static data, but will not affect the generated code in any other way.
13127 You will normally not enable this option; instead, a language processor
13128 that needs this handling would enable it on your behalf.
13130 @item -fasynchronous-unwind-tables
13131 @opindex fasynchronous-unwind-tables
13132 Generate unwind table in dwarf2 format, if supported by target machine. The
13133 table is exact at each instruction boundary, so it can be used for stack
13134 unwinding from asynchronous events (such as debugger or garbage collector).
13136 @item -fpcc-struct-return
13137 @opindex fpcc-struct-return
13138 Return ``short'' @code{struct} and @code{union} values in memory like
13139 longer ones, rather than in registers. This convention is less
13140 efficient, but it has the advantage of allowing intercallability between
13141 GCC-compiled files and files compiled with other compilers, particularly
13142 the Portable C Compiler (pcc).
13144 The precise convention for returning structures in memory depends
13145 on the target configuration macros.
13147 Short structures and unions are those whose size and alignment match
13148 that of some integer type.
13150 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13151 switch is not binary compatible with code compiled with the
13152 @option{-freg-struct-return} switch.
13153 Use it to conform to a non-default application binary interface.
13155 @item -freg-struct-return
13156 @opindex freg-struct-return
13157 Return @code{struct} and @code{union} values in registers when possible.
13158 This is more efficient for small structures than
13159 @option{-fpcc-struct-return}.
13161 If you specify neither @option{-fpcc-struct-return} nor
13162 @option{-freg-struct-return}, GCC defaults to whichever convention is
13163 standard for the target. If there is no standard convention, GCC
13164 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13165 the principal compiler. In those cases, we can choose the standard, and
13166 we chose the more efficient register return alternative.
13168 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13169 switch is not binary compatible with code compiled with the
13170 @option{-fpcc-struct-return} switch.
13171 Use it to conform to a non-default application binary interface.
13173 @item -fshort-enums
13174 @opindex fshort-enums
13175 Allocate to an @code{enum} type only as many bytes as it needs for the
13176 declared range of possible values. Specifically, the @code{enum} type
13177 will be equivalent to the smallest integer type which has enough room.
13179 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13180 code that is not binary compatible with code generated without that switch.
13181 Use it to conform to a non-default application binary interface.
13183 @item -fshort-double
13184 @opindex fshort-double
13185 Use the same size for @code{double} as for @code{float}.
13187 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13188 code that is not binary compatible with code generated without that switch.
13189 Use it to conform to a non-default application binary interface.
13191 @item -fshort-wchar
13192 @opindex fshort-wchar
13193 Override the underlying type for @samp{wchar_t} to be @samp{short
13194 unsigned int} instead of the default for the target. This option is
13195 useful for building programs to run under WINE@.
13197 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13198 code that is not binary compatible with code generated without that switch.
13199 Use it to conform to a non-default application binary interface.
13202 @opindex fno-common
13203 In C, allocate even uninitialized global variables in the data section of the
13204 object file, rather than generating them as common blocks. This has the
13205 effect that if the same variable is declared (without @code{extern}) in
13206 two different compilations, you will get an error when you link them.
13207 The only reason this might be useful is if you wish to verify that the
13208 program will work on other systems which always work this way.
13212 Ignore the @samp{#ident} directive.
13214 @item -finhibit-size-directive
13215 @opindex finhibit-size-directive
13216 Don't output a @code{.size} assembler directive, or anything else that
13217 would cause trouble if the function is split in the middle, and the
13218 two halves are placed at locations far apart in memory. This option is
13219 used when compiling @file{crtstuff.c}; you should not need to use it
13222 @item -fverbose-asm
13223 @opindex fverbose-asm
13224 Put extra commentary information in the generated assembly code to
13225 make it more readable. This option is generally only of use to those
13226 who actually need to read the generated assembly code (perhaps while
13227 debugging the compiler itself).
13229 @option{-fno-verbose-asm}, the default, causes the
13230 extra information to be omitted and is useful when comparing two assembler
13235 @cindex global offset table
13237 Generate position-independent code (PIC) suitable for use in a shared
13238 library, if supported for the target machine. Such code accesses all
13239 constant addresses through a global offset table (GOT)@. The dynamic
13240 loader resolves the GOT entries when the program starts (the dynamic
13241 loader is not part of GCC; it is part of the operating system). If
13242 the GOT size for the linked executable exceeds a machine-specific
13243 maximum size, you get an error message from the linker indicating that
13244 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13245 instead. (These maximums are 8k on the SPARC and 32k
13246 on the m68k and RS/6000. The 386 has no such limit.)
13248 Position-independent code requires special support, and therefore works
13249 only on certain machines. For the 386, GCC supports PIC for System V
13250 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13251 position-independent.
13253 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13258 If supported for the target machine, emit position-independent code,
13259 suitable for dynamic linking and avoiding any limit on the size of the
13260 global offset table. This option makes a difference on the m68k,
13261 PowerPC and SPARC@.
13263 Position-independent code requires special support, and therefore works
13264 only on certain machines.
13266 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13273 These options are similar to @option{-fpic} and @option{-fPIC}, but
13274 generated position independent code can be only linked into executables.
13275 Usually these options are used when @option{-pie} GCC option will be
13276 used during linking.
13278 @item -fno-jump-tables
13279 @opindex fno-jump-tables
13280 Do not use jump tables for switch statements even where it would be
13281 more efficient than other code generation strategies. This option is
13282 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13283 building code which forms part of a dynamic linker and cannot
13284 reference the address of a jump table. On some targets, jump tables
13285 do not require a GOT and this option is not needed.
13287 @item -ffixed-@var{reg}
13289 Treat the register named @var{reg} as a fixed register; generated code
13290 should never refer to it (except perhaps as a stack pointer, frame
13291 pointer or in some other fixed role).
13293 @var{reg} must be the name of a register. The register names accepted
13294 are machine-specific and are defined in the @code{REGISTER_NAMES}
13295 macro in the machine description macro file.
13297 This flag does not have a negative form, because it specifies a
13300 @item -fcall-used-@var{reg}
13301 @opindex fcall-used
13302 Treat the register named @var{reg} as an allocable register that is
13303 clobbered by function calls. It may be allocated for temporaries or
13304 variables that do not live across a call. Functions compiled this way
13305 will not save and restore the register @var{reg}.
13307 It is an error to used this flag with the frame pointer or stack pointer.
13308 Use of this flag for other registers that have fixed pervasive roles in
13309 the machine's execution model will produce disastrous results.
13311 This flag does not have a negative form, because it specifies a
13314 @item -fcall-saved-@var{reg}
13315 @opindex fcall-saved
13316 Treat the register named @var{reg} as an allocable register saved by
13317 functions. It may be allocated even for temporaries or variables that
13318 live across a call. Functions compiled this way will save and restore
13319 the register @var{reg} if they use it.
13321 It is an error to used this flag with the frame pointer or stack pointer.
13322 Use of this flag for other registers that have fixed pervasive roles in
13323 the machine's execution model will produce disastrous results.
13325 A different sort of disaster will result from the use of this flag for
13326 a register in which function values may be returned.
13328 This flag does not have a negative form, because it specifies a
13331 @item -fpack-struct[=@var{n}]
13332 @opindex fpack-struct
13333 Without a value specified, pack all structure members together without
13334 holes. When a value is specified (which must be a small power of two), pack
13335 structure members according to this value, representing the maximum
13336 alignment (that is, objects with default alignment requirements larger than
13337 this will be output potentially unaligned at the next fitting location.
13339 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13340 code that is not binary compatible with code generated without that switch.
13341 Additionally, it makes the code suboptimal.
13342 Use it to conform to a non-default application binary interface.
13344 @item -finstrument-functions
13345 @opindex finstrument-functions
13346 Generate instrumentation calls for entry and exit to functions. Just
13347 after function entry and just before function exit, the following
13348 profiling functions will be called with the address of the current
13349 function and its call site. (On some platforms,
13350 @code{__builtin_return_address} does not work beyond the current
13351 function, so the call site information may not be available to the
13352 profiling functions otherwise.)
13355 void __cyg_profile_func_enter (void *this_fn,
13357 void __cyg_profile_func_exit (void *this_fn,
13361 The first argument is the address of the start of the current function,
13362 which may be looked up exactly in the symbol table.
13364 This instrumentation is also done for functions expanded inline in other
13365 functions. The profiling calls will indicate where, conceptually, the
13366 inline function is entered and exited. This means that addressable
13367 versions of such functions must be available. If all your uses of a
13368 function are expanded inline, this may mean an additional expansion of
13369 code size. If you use @samp{extern inline} in your C code, an
13370 addressable version of such functions must be provided. (This is
13371 normally the case anyways, but if you get lucky and the optimizer always
13372 expands the functions inline, you might have gotten away without
13373 providing static copies.)
13375 A function may be given the attribute @code{no_instrument_function}, in
13376 which case this instrumentation will not be done. This can be used, for
13377 example, for the profiling functions listed above, high-priority
13378 interrupt routines, and any functions from which the profiling functions
13379 cannot safely be called (perhaps signal handlers, if the profiling
13380 routines generate output or allocate memory).
13382 @item -fstack-check
13383 @opindex fstack-check
13384 Generate code to verify that you do not go beyond the boundary of the
13385 stack. You should specify this flag if you are running in an
13386 environment with multiple threads, but only rarely need to specify it in
13387 a single-threaded environment since stack overflow is automatically
13388 detected on nearly all systems if there is only one stack.
13390 Note that this switch does not actually cause checking to be done; the
13391 operating system must do that. The switch causes generation of code
13392 to ensure that the operating system sees the stack being extended.
13394 @item -fstack-limit-register=@var{reg}
13395 @itemx -fstack-limit-symbol=@var{sym}
13396 @itemx -fno-stack-limit
13397 @opindex fstack-limit-register
13398 @opindex fstack-limit-symbol
13399 @opindex fno-stack-limit
13400 Generate code to ensure that the stack does not grow beyond a certain value,
13401 either the value of a register or the address of a symbol. If the stack
13402 would grow beyond the value, a signal is raised. For most targets,
13403 the signal is raised before the stack overruns the boundary, so
13404 it is possible to catch the signal without taking special precautions.
13406 For instance, if the stack starts at absolute address @samp{0x80000000}
13407 and grows downwards, you can use the flags
13408 @option{-fstack-limit-symbol=__stack_limit} and
13409 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13410 of 128KB@. Note that this may only work with the GNU linker.
13412 @cindex aliasing of parameters
13413 @cindex parameters, aliased
13414 @item -fargument-alias
13415 @itemx -fargument-noalias
13416 @itemx -fargument-noalias-global
13417 @itemx -fargument-noalias-anything
13418 @opindex fargument-alias
13419 @opindex fargument-noalias
13420 @opindex fargument-noalias-global
13421 @opindex fargument-noalias-anything
13422 Specify the possible relationships among parameters and between
13423 parameters and global data.
13425 @option{-fargument-alias} specifies that arguments (parameters) may
13426 alias each other and may alias global storage.@*
13427 @option{-fargument-noalias} specifies that arguments do not alias
13428 each other, but may alias global storage.@*
13429 @option{-fargument-noalias-global} specifies that arguments do not
13430 alias each other and do not alias global storage.
13431 @option{-fargument-noalias-anything} specifies that arguments do not
13432 alias any other storage.
13434 Each language will automatically use whatever option is required by
13435 the language standard. You should not need to use these options yourself.
13437 @item -fleading-underscore
13438 @opindex fleading-underscore
13439 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13440 change the way C symbols are represented in the object file. One use
13441 is to help link with legacy assembly code.
13443 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13444 generate code that is not binary compatible with code generated without that
13445 switch. Use it to conform to a non-default application binary interface.
13446 Not all targets provide complete support for this switch.
13448 @item -ftls-model=@var{model}
13449 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13450 The @var{model} argument should be one of @code{global-dynamic},
13451 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13453 The default without @option{-fpic} is @code{initial-exec}; with
13454 @option{-fpic} the default is @code{global-dynamic}.
13456 @item -fvisibility=@var{default|internal|hidden|protected}
13457 @opindex fvisibility
13458 Set the default ELF image symbol visibility to the specified option---all
13459 symbols will be marked with this unless overridden within the code.
13460 Using this feature can very substantially improve linking and
13461 load times of shared object libraries, produce more optimized
13462 code, provide near-perfect API export and prevent symbol clashes.
13463 It is @strong{strongly} recommended that you use this in any shared objects
13466 Despite the nomenclature, @code{default} always means public ie;
13467 available to be linked against from outside the shared object.
13468 @code{protected} and @code{internal} are pretty useless in real-world
13469 usage so the only other commonly used option will be @code{hidden}.
13470 The default if @option{-fvisibility} isn't specified is
13471 @code{default}, i.e., make every
13472 symbol public---this causes the same behavior as previous versions of
13475 A good explanation of the benefits offered by ensuring ELF
13476 symbols have the correct visibility is given by ``How To Write
13477 Shared Libraries'' by Ulrich Drepper (which can be found at
13478 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13479 solution made possible by this option to marking things hidden when
13480 the default is public is to make the default hidden and mark things
13481 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13482 and @code{__attribute__ ((visibility("default")))} instead of
13483 @code{__declspec(dllexport)} you get almost identical semantics with
13484 identical syntax. This is a great boon to those working with
13485 cross-platform projects.
13487 For those adding visibility support to existing code, you may find
13488 @samp{#pragma GCC visibility} of use. This works by you enclosing
13489 the declarations you wish to set visibility for with (for example)
13490 @samp{#pragma GCC visibility push(hidden)} and
13491 @samp{#pragma GCC visibility pop}.
13492 Bear in mind that symbol visibility should be viewed @strong{as
13493 part of the API interface contract} and thus all new code should
13494 always specify visibility when it is not the default ie; declarations
13495 only for use within the local DSO should @strong{always} be marked explicitly
13496 as hidden as so to avoid PLT indirection overheads---making this
13497 abundantly clear also aids readability and self-documentation of the code.
13498 Note that due to ISO C++ specification requirements, operator new and
13499 operator delete must always be of default visibility.
13501 Be aware that headers from outside your project, in particular system
13502 headers and headers from any other library you use, may not be
13503 expecting to be compiled with visibility other than the default. You
13504 may need to explicitly say @samp{#pragma GCC visibility push(default)}
13505 before including any such headers.
13507 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
13508 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
13509 no modifications. However, this means that calls to @samp{extern}
13510 functions with no explicit visibility will use the PLT, so it is more
13511 effective to use @samp{__attribute ((visibility))} and/or
13512 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
13513 declarations should be treated as hidden.
13515 Note that @samp{-fvisibility} does affect C++ vague linkage
13516 entities. This means that, for instance, an exception class that will
13517 be thrown between DSOs must be explicitly marked with default
13518 visibility so that the @samp{type_info} nodes will be unified between
13521 An overview of these techniques, their benefits and how to use them
13522 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13528 @node Environment Variables
13529 @section Environment Variables Affecting GCC
13530 @cindex environment variables
13532 @c man begin ENVIRONMENT
13533 This section describes several environment variables that affect how GCC
13534 operates. Some of them work by specifying directories or prefixes to use
13535 when searching for various kinds of files. Some are used to specify other
13536 aspects of the compilation environment.
13538 Note that you can also specify places to search using options such as
13539 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13540 take precedence over places specified using environment variables, which
13541 in turn take precedence over those specified by the configuration of GCC@.
13542 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13543 GNU Compiler Collection (GCC) Internals}.
13548 @c @itemx LC_COLLATE
13550 @c @itemx LC_MONETARY
13551 @c @itemx LC_NUMERIC
13556 @c @findex LC_COLLATE
13557 @findex LC_MESSAGES
13558 @c @findex LC_MONETARY
13559 @c @findex LC_NUMERIC
13563 These environment variables control the way that GCC uses
13564 localization information that allow GCC to work with different
13565 national conventions. GCC inspects the locale categories
13566 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13567 so. These locale categories can be set to any value supported by your
13568 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13569 Kingdom encoded in UTF-8.
13571 The @env{LC_CTYPE} environment variable specifies character
13572 classification. GCC uses it to determine the character boundaries in
13573 a string; this is needed for some multibyte encodings that contain quote
13574 and escape characters that would otherwise be interpreted as a string
13577 The @env{LC_MESSAGES} environment variable specifies the language to
13578 use in diagnostic messages.
13580 If the @env{LC_ALL} environment variable is set, it overrides the value
13581 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13582 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13583 environment variable. If none of these variables are set, GCC
13584 defaults to traditional C English behavior.
13588 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13589 files. GCC uses temporary files to hold the output of one stage of
13590 compilation which is to be used as input to the next stage: for example,
13591 the output of the preprocessor, which is the input to the compiler
13594 @item GCC_EXEC_PREFIX
13595 @findex GCC_EXEC_PREFIX
13596 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13597 names of the subprograms executed by the compiler. No slash is added
13598 when this prefix is combined with the name of a subprogram, but you can
13599 specify a prefix that ends with a slash if you wish.
13601 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13602 an appropriate prefix to use based on the pathname it was invoked with.
13604 If GCC cannot find the subprogram using the specified prefix, it
13605 tries looking in the usual places for the subprogram.
13607 The default value of @env{GCC_EXEC_PREFIX} is
13608 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13609 of @code{prefix} when you ran the @file{configure} script.
13611 Other prefixes specified with @option{-B} take precedence over this prefix.
13613 This prefix is also used for finding files such as @file{crt0.o} that are
13616 In addition, the prefix is used in an unusual way in finding the
13617 directories to search for header files. For each of the standard
13618 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13619 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13620 replacing that beginning with the specified prefix to produce an
13621 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13622 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13623 These alternate directories are searched first; the standard directories
13626 @item COMPILER_PATH
13627 @findex COMPILER_PATH
13628 The value of @env{COMPILER_PATH} is a colon-separated list of
13629 directories, much like @env{PATH}. GCC tries the directories thus
13630 specified when searching for subprograms, if it can't find the
13631 subprograms using @env{GCC_EXEC_PREFIX}.
13634 @findex LIBRARY_PATH
13635 The value of @env{LIBRARY_PATH} is a colon-separated list of
13636 directories, much like @env{PATH}. When configured as a native compiler,
13637 GCC tries the directories thus specified when searching for special
13638 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13639 using GCC also uses these directories when searching for ordinary
13640 libraries for the @option{-l} option (but directories specified with
13641 @option{-L} come first).
13645 @cindex locale definition
13646 This variable is used to pass locale information to the compiler. One way in
13647 which this information is used is to determine the character set to be used
13648 when character literals, string literals and comments are parsed in C and C++.
13649 When the compiler is configured to allow multibyte characters,
13650 the following values for @env{LANG} are recognized:
13654 Recognize JIS characters.
13656 Recognize SJIS characters.
13658 Recognize EUCJP characters.
13661 If @env{LANG} is not defined, or if it has some other value, then the
13662 compiler will use mblen and mbtowc as defined by the default locale to
13663 recognize and translate multibyte characters.
13667 Some additional environments variables affect the behavior of the
13670 @include cppenv.texi
13674 @node Precompiled Headers
13675 @section Using Precompiled Headers
13676 @cindex precompiled headers
13677 @cindex speed of compilation
13679 Often large projects have many header files that are included in every
13680 source file. The time the compiler takes to process these header files
13681 over and over again can account for nearly all of the time required to
13682 build the project. To make builds faster, GCC allows users to
13683 `precompile' a header file; then, if builds can use the precompiled
13684 header file they will be much faster.
13686 To create a precompiled header file, simply compile it as you would any
13687 other file, if necessary using the @option{-x} option to make the driver
13688 treat it as a C or C++ header file. You will probably want to use a
13689 tool like @command{make} to keep the precompiled header up-to-date when
13690 the headers it contains change.
13692 A precompiled header file will be searched for when @code{#include} is
13693 seen in the compilation. As it searches for the included file
13694 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13695 compiler looks for a precompiled header in each directory just before it
13696 looks for the include file in that directory. The name searched for is
13697 the name specified in the @code{#include} with @samp{.gch} appended. If
13698 the precompiled header file can't be used, it is ignored.
13700 For instance, if you have @code{#include "all.h"}, and you have
13701 @file{all.h.gch} in the same directory as @file{all.h}, then the
13702 precompiled header file will be used if possible, and the original
13703 header will be used otherwise.
13705 Alternatively, you might decide to put the precompiled header file in a
13706 directory and use @option{-I} to ensure that directory is searched
13707 before (or instead of) the directory containing the original header.
13708 Then, if you want to check that the precompiled header file is always
13709 used, you can put a file of the same name as the original header in this
13710 directory containing an @code{#error} command.
13712 This also works with @option{-include}. So yet another way to use
13713 precompiled headers, good for projects not designed with precompiled
13714 header files in mind, is to simply take most of the header files used by
13715 a project, include them from another header file, precompile that header
13716 file, and @option{-include} the precompiled header. If the header files
13717 have guards against multiple inclusion, they will be skipped because
13718 they've already been included (in the precompiled header).
13720 If you need to precompile the same header file for different
13721 languages, targets, or compiler options, you can instead make a
13722 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13723 header in the directory, perhaps using @option{-o}. It doesn't matter
13724 what you call the files in the directory, every precompiled header in
13725 the directory will be considered. The first precompiled header
13726 encountered in the directory that is valid for this compilation will
13727 be used; they're searched in no particular order.
13729 There are many other possibilities, limited only by your imagination,
13730 good sense, and the constraints of your build system.
13732 A precompiled header file can be used only when these conditions apply:
13736 Only one precompiled header can be used in a particular compilation.
13739 A precompiled header can't be used once the first C token is seen. You
13740 can have preprocessor directives before a precompiled header; you can
13741 even include a precompiled header from inside another header, so long as
13742 there are no C tokens before the @code{#include}.
13745 The precompiled header file must be produced for the same language as
13746 the current compilation. You can't use a C precompiled header for a C++
13750 The precompiled header file must have been produced by the same compiler
13751 binary as the current compilation is using.
13754 Any macros defined before the precompiled header is included must
13755 either be defined in the same way as when the precompiled header was
13756 generated, or must not affect the precompiled header, which usually
13757 means that they don't appear in the precompiled header at all.
13759 The @option{-D} option is one way to define a macro before a
13760 precompiled header is included; using a @code{#define} can also do it.
13761 There are also some options that define macros implicitly, like
13762 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13765 @item If debugging information is output when using the precompiled
13766 header, using @option{-g} or similar, the same kind of debugging information
13767 must have been output when building the precompiled header. However,
13768 a precompiled header built using @option{-g} can be used in a compilation
13769 when no debugging information is being output.
13771 @item The same @option{-m} options must generally be used when building
13772 and using the precompiled header. @xref{Submodel Options},
13773 for any cases where this rule is relaxed.
13775 @item Each of the following options must be the same when building and using
13776 the precompiled header:
13778 @gccoptlist{-fexceptions -funit-at-a-time}
13781 Some other command-line options starting with @option{-f},
13782 @option{-p}, or @option{-O} must be defined in the same way as when
13783 the precompiled header was generated. At present, it's not clear
13784 which options are safe to change and which are not; the safest choice
13785 is to use exactly the same options when generating and using the
13786 precompiled header. The following are known to be safe:
13788 @gccoptlist{-fmessage-length= -fpreprocessed
13789 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13790 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13795 For all of these except the last, the compiler will automatically
13796 ignore the precompiled header if the conditions aren't met. If you
13797 find an option combination that doesn't work and doesn't cause the
13798 precompiled header to be ignored, please consider filing a bug report,
13801 If you do use differing options when generating and using the
13802 precompiled header, the actual behavior will be a mixture of the
13803 behavior for the options. For instance, if you use @option{-g} to
13804 generate the precompiled header but not when using it, you may or may
13805 not get debugging information for routines in the precompiled header.
13807 @node Running Protoize
13808 @section Running Protoize
13810 The program @code{protoize} is an optional part of GCC@. You can use
13811 it to add prototypes to a program, thus converting the program to ISO
13812 C in one respect. The companion program @code{unprotoize} does the
13813 reverse: it removes argument types from any prototypes that are found.
13815 When you run these programs, you must specify a set of source files as
13816 command line arguments. The conversion programs start out by compiling
13817 these files to see what functions they define. The information gathered
13818 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13820 After scanning comes actual conversion. The specified files are all
13821 eligible to be converted; any files they include (whether sources or
13822 just headers) are eligible as well.
13824 But not all the eligible files are converted. By default,
13825 @code{protoize} and @code{unprotoize} convert only source and header
13826 files in the current directory. You can specify additional directories
13827 whose files should be converted with the @option{-d @var{directory}}
13828 option. You can also specify particular files to exclude with the
13829 @option{-x @var{file}} option. A file is converted if it is eligible, its
13830 directory name matches one of the specified directory names, and its
13831 name within the directory has not been excluded.
13833 Basic conversion with @code{protoize} consists of rewriting most
13834 function definitions and function declarations to specify the types of
13835 the arguments. The only ones not rewritten are those for varargs
13838 @code{protoize} optionally inserts prototype declarations at the
13839 beginning of the source file, to make them available for any calls that
13840 precede the function's definition. Or it can insert prototype
13841 declarations with block scope in the blocks where undeclared functions
13844 Basic conversion with @code{unprotoize} consists of rewriting most
13845 function declarations to remove any argument types, and rewriting
13846 function definitions to the old-style pre-ISO form.
13848 Both conversion programs print a warning for any function declaration or
13849 definition that they can't convert. You can suppress these warnings
13852 The output from @code{protoize} or @code{unprotoize} replaces the
13853 original source file. The original file is renamed to a name ending
13854 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13855 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13856 for DOS) file already exists, then the source file is simply discarded.
13858 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13859 scan the program and collect information about the functions it uses.
13860 So neither of these programs will work until GCC is installed.
13862 Here is a table of the options you can use with @code{protoize} and
13863 @code{unprotoize}. Each option works with both programs unless
13867 @item -B @var{directory}
13868 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13869 usual directory (normally @file{/usr/local/lib}). This file contains
13870 prototype information about standard system functions. This option
13871 applies only to @code{protoize}.
13873 @item -c @var{compilation-options}
13874 Use @var{compilation-options} as the options when running @command{gcc} to
13875 produce the @samp{.X} files. The special option @option{-aux-info} is
13876 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13878 Note that the compilation options must be given as a single argument to
13879 @code{protoize} or @code{unprotoize}. If you want to specify several
13880 @command{gcc} options, you must quote the entire set of compilation options
13881 to make them a single word in the shell.
13883 There are certain @command{gcc} arguments that you cannot use, because they
13884 would produce the wrong kind of output. These include @option{-g},
13885 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13886 the @var{compilation-options}, they are ignored.
13889 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13890 systems) instead of @samp{.c}. This is convenient if you are converting
13891 a C program to C++. This option applies only to @code{protoize}.
13894 Add explicit global declarations. This means inserting explicit
13895 declarations at the beginning of each source file for each function
13896 that is called in the file and was not declared. These declarations
13897 precede the first function definition that contains a call to an
13898 undeclared function. This option applies only to @code{protoize}.
13900 @item -i @var{string}
13901 Indent old-style parameter declarations with the string @var{string}.
13902 This option applies only to @code{protoize}.
13904 @code{unprotoize} converts prototyped function definitions to old-style
13905 function definitions, where the arguments are declared between the
13906 argument list and the initial @samp{@{}. By default, @code{unprotoize}
13907 uses five spaces as the indentation. If you want to indent with just
13908 one space instead, use @option{-i " "}.
13911 Keep the @samp{.X} files. Normally, they are deleted after conversion
13915 Add explicit local declarations. @code{protoize} with @option{-l} inserts
13916 a prototype declaration for each function in each block which calls the
13917 function without any declaration. This option applies only to
13921 Make no real changes. This mode just prints information about the conversions
13922 that would have been done without @option{-n}.
13925 Make no @samp{.save} files. The original files are simply deleted.
13926 Use this option with caution.
13928 @item -p @var{program}
13929 Use the program @var{program} as the compiler. Normally, the name
13930 @file{gcc} is used.
13933 Work quietly. Most warnings are suppressed.
13936 Print the version number, just like @option{-v} for @command{gcc}.
13939 If you need special compiler options to compile one of your program's
13940 source files, then you should generate that file's @samp{.X} file
13941 specially, by running @command{gcc} on that source file with the
13942 appropriate options and the option @option{-aux-info}. Then run
13943 @code{protoize} on the entire set of files. @code{protoize} will use
13944 the existing @samp{.X} file because it is newer than the source file.
13948 gcc -Dfoo=bar file1.c -aux-info file1.X
13953 You need to include the special files along with the rest in the
13954 @code{protoize} command, even though their @samp{.X} files already
13955 exist, because otherwise they won't get converted.
13957 @xref{Protoize Caveats}, for more information on how to use
13958 @code{protoize} successfully.