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-noaddr -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
952 Turn off any specification of a language, so that subsequent files are
953 handled according to their file name suffixes (as they are if @option{-x}
954 has not been used at all).
956 @item -pass-exit-codes
957 @opindex pass-exit-codes
958 Normally the @command{gcc} program will exit with the code of 1 if any
959 phase of the compiler returns a non-success return code. If you specify
960 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
961 numerically highest error produced by any phase that returned an error
962 indication. The C, C++, and Fortran frontends return 4, if an internal
963 compiler error is encountered.
966 If you only want some of the stages of compilation, you can use
967 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
968 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
969 @command{gcc} is to stop. Note that some combinations (for example,
970 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
975 Compile or assemble the source files, but do not link. The linking
976 stage simply is not done. The ultimate output is in the form of an
977 object file for each source file.
979 By default, the object file name for a source file is made by replacing
980 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
982 Unrecognized input files, not requiring compilation or assembly, are
987 Stop after the stage of compilation proper; do not assemble. The output
988 is in the form of an assembler code file for each non-assembler input
991 By default, the assembler file name for a source file is made by
992 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
994 Input files that don't require compilation are ignored.
998 Stop after the preprocessing stage; do not run the compiler proper. The
999 output is in the form of preprocessed source code, which is sent to the
1002 Input files which don't require preprocessing are ignored.
1004 @cindex output file option
1007 Place output in file @var{file}. This applies regardless to whatever
1008 sort of output is being produced, whether it be an executable file,
1009 an object file, an assembler file or preprocessed C code.
1011 If @option{-o} is not specified, the default is to put an executable
1012 file in @file{a.out}, the object file for
1013 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1014 assembler file in @file{@var{source}.s}, a precompiled header file in
1015 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1020 Print (on standard error output) the commands executed to run the stages
1021 of compilation. Also print the version number of the compiler driver
1022 program and of the preprocessor and the compiler proper.
1026 Like @option{-v} except the commands are not executed and all command
1027 arguments are quoted. This is useful for shell scripts to capture the
1028 driver-generated command lines.
1032 Use pipes rather than temporary files for communication between the
1033 various stages of compilation. This fails to work on some systems where
1034 the assembler is unable to read from a pipe; but the GNU assembler has
1039 If you are compiling multiple source files, this option tells the driver
1040 to pass all the source files to the compiler at once (for those
1041 languages for which the compiler can handle this). This will allow
1042 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1043 language for which this is supported is C@. If you pass source files for
1044 multiple languages to the driver, using this option, the driver will invoke
1045 the compiler(s) that support IMA once each, passing each compiler all the
1046 source files appropriate for it. For those languages that do not support
1047 IMA this option will be ignored, and the compiler will be invoked once for
1048 each source file in that language. If you use this option in conjunction
1049 with @option{-save-temps}, the compiler will generate multiple
1051 (one for each source file), but only one (combined) @file{.o} or
1056 Print (on the standard output) a description of the command line options
1057 understood by @command{gcc}. If the @option{-v} option is also specified
1058 then @option{--help} will also be passed on to the various processes
1059 invoked by @command{gcc}, so that they can display the command line options
1060 they accept. If the @option{-Wextra} option is also specified then command
1061 line options which have no documentation associated with them will also
1065 @opindex target-help
1066 Print (on the standard output) a description of target specific command
1067 line options for each tool.
1071 Display the version number and copyrights of the invoked GCC@.
1073 @include @value{srcdir}/../libiberty/at-file.texi
1077 @section Compiling C++ Programs
1079 @cindex suffixes for C++ source
1080 @cindex C++ source file suffixes
1081 C++ source files conventionally use one of the suffixes @samp{.C},
1082 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1083 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1084 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1085 files with these names and compiles them as C++ programs even if you
1086 call the compiler the same way as for compiling C programs (usually
1087 with the name @command{gcc}).
1091 However, the use of @command{gcc} does not add the C++ library.
1092 @command{g++} is a program that calls GCC and treats @samp{.c},
1093 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1094 files unless @option{-x} is used, and automatically specifies linking
1095 against the C++ library. This program is also useful when
1096 precompiling a C header file with a @samp{.h} extension for use in C++
1097 compilations. On many systems, @command{g++} is also installed with
1098 the name @command{c++}.
1100 @cindex invoking @command{g++}
1101 When you compile C++ programs, you may specify many of the same
1102 command-line options that you use for compiling programs in any
1103 language; or command-line options meaningful for C and related
1104 languages; or options that are meaningful only for C++ programs.
1105 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1106 explanations of options for languages related to C@.
1107 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1108 explanations of options that are meaningful only for C++ programs.
1110 @node C Dialect Options
1111 @section Options Controlling C Dialect
1112 @cindex dialect options
1113 @cindex language dialect options
1114 @cindex options, dialect
1116 The following options control the dialect of C (or languages derived
1117 from C, such as C++, Objective-C and Objective-C++) that the compiler
1121 @cindex ANSI support
1125 In C mode, support all ISO C90 programs. In C++ mode,
1126 remove GNU extensions that conflict with ISO C++.
1128 This turns off certain features of GCC that are incompatible with ISO
1129 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1130 such as the @code{asm} and @code{typeof} keywords, and
1131 predefined macros such as @code{unix} and @code{vax} that identify the
1132 type of system you are using. It also enables the undesirable and
1133 rarely used ISO trigraph feature. For the C compiler,
1134 it disables recognition of C++ style @samp{//} comments as well as
1135 the @code{inline} keyword.
1137 The alternate keywords @code{__asm__}, @code{__extension__},
1138 @code{__inline__} and @code{__typeof__} continue to work despite
1139 @option{-ansi}. You would not want to use them in an ISO C program, of
1140 course, but it is useful to put them in header files that might be included
1141 in compilations done with @option{-ansi}. Alternate predefined macros
1142 such as @code{__unix__} and @code{__vax__} are also available, with or
1143 without @option{-ansi}.
1145 The @option{-ansi} option does not cause non-ISO programs to be
1146 rejected gratuitously. For that, @option{-pedantic} is required in
1147 addition to @option{-ansi}. @xref{Warning Options}.
1149 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1150 option is used. Some header files may notice this macro and refrain
1151 from declaring certain functions or defining certain macros that the
1152 ISO standard doesn't call for; this is to avoid interfering with any
1153 programs that might use these names for other things.
1155 Functions which would normally be built in but do not have semantics
1156 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1157 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1158 built-in functions provided by GCC}, for details of the functions
1163 Determine the language standard. This option is currently only
1164 supported when compiling C or C++. A value for this option must be
1165 provided; possible values are
1170 ISO C90 (same as @option{-ansi}).
1172 @item iso9899:199409
1173 ISO C90 as modified in amendment 1.
1179 ISO C99. Note that this standard is not yet fully supported; see
1180 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1181 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1184 Default, ISO C90 plus GNU extensions (including some C99 features).
1188 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1189 this will become the default. The name @samp{gnu9x} is deprecated.
1192 The 1998 ISO C++ standard plus amendments.
1195 The same as @option{-std=c++98} plus GNU extensions. This is the
1196 default for C++ code.
1199 Even when this option is not specified, you can still use some of the
1200 features of newer standards in so far as they do not conflict with
1201 previous C standards. For example, you may use @code{__restrict__} even
1202 when @option{-std=c99} is not specified.
1204 The @option{-std} options specifying some version of ISO C have the same
1205 effects as @option{-ansi}, except that features that were not in ISO C90
1206 but are in the specified version (for example, @samp{//} comments and
1207 the @code{inline} keyword in ISO C99) are not disabled.
1209 @xref{Standards,,Language Standards Supported by GCC}, for details of
1210 these standard versions.
1212 @item -aux-info @var{filename}
1214 Output to the given filename prototyped declarations for all functions
1215 declared and/or defined in a translation unit, including those in header
1216 files. This option is silently ignored in any language other than C@.
1218 Besides declarations, the file indicates, in comments, the origin of
1219 each declaration (source file and line), whether the declaration was
1220 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1221 @samp{O} for old, respectively, in the first character after the line
1222 number and the colon), and whether it came from a declaration or a
1223 definition (@samp{C} or @samp{F}, respectively, in the following
1224 character). In the case of function definitions, a K&R-style list of
1225 arguments followed by their declarations is also provided, inside
1226 comments, after the declaration.
1230 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1231 keyword, so that code can use these words as identifiers. You can use
1232 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1233 instead. @option{-ansi} implies @option{-fno-asm}.
1235 In C++, this switch only affects the @code{typeof} keyword, since
1236 @code{asm} and @code{inline} are standard keywords. You may want to
1237 use the @option{-fno-gnu-keywords} flag instead, which has the same
1238 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1239 switch only affects the @code{asm} and @code{typeof} keywords, since
1240 @code{inline} is a standard keyword in ISO C99.
1243 @itemx -fno-builtin-@var{function}
1244 @opindex fno-builtin
1245 @cindex built-in functions
1246 Don't recognize built-in functions that do not begin with
1247 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1248 functions provided by GCC}, for details of the functions affected,
1249 including those which are not built-in functions when @option{-ansi} or
1250 @option{-std} options for strict ISO C conformance are used because they
1251 do not have an ISO standard meaning.
1253 GCC normally generates special code to handle certain built-in functions
1254 more efficiently; for instance, calls to @code{alloca} may become single
1255 instructions that adjust the stack directly, and calls to @code{memcpy}
1256 may become inline copy loops. The resulting code is often both smaller
1257 and faster, but since the function calls no longer appear as such, you
1258 cannot set a breakpoint on those calls, nor can you change the behavior
1259 of the functions by linking with a different library. In addition,
1260 when a function is recognized as a built-in function, GCC may use
1261 information about that function to warn about problems with calls to
1262 that function, or to generate more efficient code, even if the
1263 resulting code still contains calls to that function. For example,
1264 warnings are given with @option{-Wformat} for bad calls to
1265 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1266 known not to modify global memory.
1268 With the @option{-fno-builtin-@var{function}} option
1269 only the built-in function @var{function} is
1270 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1271 function is named this is not built-in in this version of GCC, this
1272 option is ignored. There is no corresponding
1273 @option{-fbuiltin-@var{function}} option; if you wish to enable
1274 built-in functions selectively when using @option{-fno-builtin} or
1275 @option{-ffreestanding}, you may define macros such as:
1278 #define abs(n) __builtin_abs ((n))
1279 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1284 @cindex hosted environment
1286 Assert that compilation takes place in a hosted environment. This implies
1287 @option{-fbuiltin}. A hosted environment is one in which the
1288 entire standard library is available, and in which @code{main} has a return
1289 type of @code{int}. Examples are nearly everything except a kernel.
1290 This is equivalent to @option{-fno-freestanding}.
1292 @item -ffreestanding
1293 @opindex ffreestanding
1294 @cindex hosted environment
1296 Assert that compilation takes place in a freestanding environment. This
1297 implies @option{-fno-builtin}. A freestanding environment
1298 is one in which the standard library may not exist, and program startup may
1299 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1300 This is equivalent to @option{-fno-hosted}.
1302 @xref{Standards,,Language Standards Supported by GCC}, for details of
1303 freestanding and hosted environments.
1307 @cindex openmp parallel
1308 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1309 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1310 compiler generates parallel code according to the OpenMP Application
1311 Program Interface v2.5 @w{@uref{http://www.openmp.org/}}.
1313 @item -fms-extensions
1314 @opindex fms-extensions
1315 Accept some non-standard constructs used in Microsoft header files.
1317 Some cases of unnamed fields in structures and unions are only
1318 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1319 fields within structs/unions}, for details.
1323 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1324 options for strict ISO C conformance) implies @option{-trigraphs}.
1326 @item -no-integrated-cpp
1327 @opindex no-integrated-cpp
1328 Performs a compilation in two passes: preprocessing and compiling. This
1329 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1330 @option{-B} option. The user supplied compilation step can then add in
1331 an additional preprocessing step after normal preprocessing but before
1332 compiling. The default is to use the integrated cpp (internal cpp)
1334 The semantics of this option will change if "cc1", "cc1plus", and
1335 "cc1obj" are merged.
1337 @cindex traditional C language
1338 @cindex C language, traditional
1340 @itemx -traditional-cpp
1341 @opindex traditional-cpp
1342 @opindex traditional
1343 Formerly, these options caused GCC to attempt to emulate a pre-standard
1344 C compiler. They are now only supported with the @option{-E} switch.
1345 The preprocessor continues to support a pre-standard mode. See the GNU
1346 CPP manual for details.
1348 @item -fcond-mismatch
1349 @opindex fcond-mismatch
1350 Allow conditional expressions with mismatched types in the second and
1351 third arguments. The value of such an expression is void. This option
1352 is not supported for C++.
1354 @item -funsigned-char
1355 @opindex funsigned-char
1356 Let the type @code{char} be unsigned, like @code{unsigned char}.
1358 Each kind of machine has a default for what @code{char} should
1359 be. It is either like @code{unsigned char} by default or like
1360 @code{signed char} by default.
1362 Ideally, a portable program should always use @code{signed char} or
1363 @code{unsigned char} when it depends on the signedness of an object.
1364 But many programs have been written to use plain @code{char} and
1365 expect it to be signed, or expect it to be unsigned, depending on the
1366 machines they were written for. This option, and its inverse, let you
1367 make such a program work with the opposite default.
1369 The type @code{char} is always a distinct type from each of
1370 @code{signed char} or @code{unsigned char}, even though its behavior
1371 is always just like one of those two.
1374 @opindex fsigned-char
1375 Let the type @code{char} be signed, like @code{signed char}.
1377 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1378 the negative form of @option{-funsigned-char}. Likewise, the option
1379 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1381 @item -fsigned-bitfields
1382 @itemx -funsigned-bitfields
1383 @itemx -fno-signed-bitfields
1384 @itemx -fno-unsigned-bitfields
1385 @opindex fsigned-bitfields
1386 @opindex funsigned-bitfields
1387 @opindex fno-signed-bitfields
1388 @opindex fno-unsigned-bitfields
1389 These options control whether a bit-field is signed or unsigned, when the
1390 declaration does not use either @code{signed} or @code{unsigned}. By
1391 default, such a bit-field is signed, because this is consistent: the
1392 basic integer types such as @code{int} are signed types.
1395 @node C++ Dialect Options
1396 @section Options Controlling C++ Dialect
1398 @cindex compiler options, C++
1399 @cindex C++ options, command line
1400 @cindex options, C++
1401 This section describes the command-line options that are only meaningful
1402 for C++ programs; but you can also use most of the GNU compiler options
1403 regardless of what language your program is in. For example, you
1404 might compile a file @code{firstClass.C} like this:
1407 g++ -g -frepo -O -c firstClass.C
1411 In this example, only @option{-frepo} is an option meant
1412 only for C++ programs; you can use the other options with any
1413 language supported by GCC@.
1415 Here is a list of options that are @emph{only} for compiling C++ programs:
1419 @item -fabi-version=@var{n}
1420 @opindex fabi-version
1421 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1422 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1423 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1424 the version that conforms most closely to the C++ ABI specification.
1425 Therefore, the ABI obtained using version 0 will change as ABI bugs
1428 The default is version 2.
1430 @item -fno-access-control
1431 @opindex fno-access-control
1432 Turn off all access checking. This switch is mainly useful for working
1433 around bugs in the access control code.
1437 Check that the pointer returned by @code{operator new} is non-null
1438 before attempting to modify the storage allocated. This check is
1439 normally unnecessary because the C++ standard specifies that
1440 @code{operator new} will only return @code{0} if it is declared
1441 @samp{throw()}, in which case the compiler will always check the
1442 return value even without this option. In all other cases, when
1443 @code{operator new} has a non-empty exception specification, memory
1444 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1445 @samp{new (nothrow)}.
1447 @item -fconserve-space
1448 @opindex fconserve-space
1449 Put uninitialized or runtime-initialized global variables into the
1450 common segment, as C does. This saves space in the executable at the
1451 cost of not diagnosing duplicate definitions. If you compile with this
1452 flag and your program mysteriously crashes after @code{main()} has
1453 completed, you may have an object that is being destroyed twice because
1454 two definitions were merged.
1456 This option is no longer useful on most targets, now that support has
1457 been added for putting variables into BSS without making them common.
1459 @item -ffriend-injection
1460 @opindex ffriend-injection
1461 Inject friend functions into the enclosing namespace, so that they are
1462 visible outside the scope of the class in which they are declared.
1463 Friend functions were documented to work this way in the old Annotated
1464 C++ Reference Manual, and versions of G++ before 4.1 always worked
1465 that way. However, in ISO C++ a friend function which is not declared
1466 in an enclosing scope can only be found using argument dependent
1467 lookup. This option causes friends to be injected as they were in
1470 This option is for compatibility, and may be removed in a future
1473 @item -fno-elide-constructors
1474 @opindex fno-elide-constructors
1475 The C++ standard allows an implementation to omit creating a temporary
1476 which is only used to initialize another object of the same type.
1477 Specifying this option disables that optimization, and forces G++ to
1478 call the copy constructor in all cases.
1480 @item -fno-enforce-eh-specs
1481 @opindex fno-enforce-eh-specs
1482 Don't generate code to check for violation of exception specifications
1483 at runtime. This option violates the C++ standard, but may be useful
1484 for reducing code size in production builds, much like defining
1485 @samp{NDEBUG}. This does not give user code permission to throw
1486 exceptions in violation of the exception specifications; the compiler
1487 will still optimize based on the specifications, so throwing an
1488 unexpected exception will result in undefined behavior.
1491 @itemx -fno-for-scope
1493 @opindex fno-for-scope
1494 If @option{-ffor-scope} is specified, the scope of variables declared in
1495 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1496 as specified by the C++ standard.
1497 If @option{-fno-for-scope} is specified, the scope of variables declared in
1498 a @i{for-init-statement} extends to the end of the enclosing scope,
1499 as was the case in old versions of G++, and other (traditional)
1500 implementations of C++.
1502 The default if neither flag is given to follow the standard,
1503 but to allow and give a warning for old-style code that would
1504 otherwise be invalid, or have different behavior.
1506 @item -fno-gnu-keywords
1507 @opindex fno-gnu-keywords
1508 Do not recognize @code{typeof} as a keyword, so that code can use this
1509 word as an identifier. You can use the keyword @code{__typeof__} instead.
1510 @option{-ansi} implies @option{-fno-gnu-keywords}.
1512 @item -fno-implicit-templates
1513 @opindex fno-implicit-templates
1514 Never emit code for non-inline templates which are instantiated
1515 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1516 @xref{Template Instantiation}, for more information.
1518 @item -fno-implicit-inline-templates
1519 @opindex fno-implicit-inline-templates
1520 Don't emit code for implicit instantiations of inline templates, either.
1521 The default is to handle inlines differently so that compiles with and
1522 without optimization will need the same set of explicit instantiations.
1524 @item -fno-implement-inlines
1525 @opindex fno-implement-inlines
1526 To save space, do not emit out-of-line copies of inline functions
1527 controlled by @samp{#pragma implementation}. This will cause linker
1528 errors if these functions are not inlined everywhere they are called.
1530 @item -fms-extensions
1531 @opindex fms-extensions
1532 Disable pedantic warnings about constructs used in MFC, such as implicit
1533 int and getting a pointer to member function via non-standard syntax.
1535 @item -fno-nonansi-builtins
1536 @opindex fno-nonansi-builtins
1537 Disable built-in declarations of functions that are not mandated by
1538 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1539 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1541 @item -fno-operator-names
1542 @opindex fno-operator-names
1543 Do not treat the operator name keywords @code{and}, @code{bitand},
1544 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1545 synonyms as keywords.
1547 @item -fno-optional-diags
1548 @opindex fno-optional-diags
1549 Disable diagnostics that the standard says a compiler does not need to
1550 issue. Currently, the only such diagnostic issued by G++ is the one for
1551 a name having multiple meanings within a class.
1554 @opindex fpermissive
1555 Downgrade some diagnostics about nonconformant code from errors to
1556 warnings. Thus, using @option{-fpermissive} will allow some
1557 nonconforming code to compile.
1561 Enable automatic template instantiation at link time. This option also
1562 implies @option{-fno-implicit-templates}. @xref{Template
1563 Instantiation}, for more information.
1567 Disable generation of information about every class with virtual
1568 functions for use by the C++ runtime type identification features
1569 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1570 of the language, you can save some space by using this flag. Note that
1571 exception handling uses the same information, but it will generate it as
1572 needed. The @samp{dynamic_cast} operator can still be used for casts that
1573 do not require runtime type information, i.e. casts to @code{void *} or to
1574 unambiguous base classes.
1578 Emit statistics about front-end processing at the end of the compilation.
1579 This information is generally only useful to the G++ development team.
1581 @item -ftemplate-depth-@var{n}
1582 @opindex ftemplate-depth
1583 Set the maximum instantiation depth for template classes to @var{n}.
1584 A limit on the template instantiation depth is needed to detect
1585 endless recursions during template class instantiation. ANSI/ISO C++
1586 conforming programs must not rely on a maximum depth greater than 17.
1588 @item -fno-threadsafe-statics
1589 @opindex fno-threadsafe-statics
1590 Do not emit the extra code to use the routines specified in the C++
1591 ABI for thread-safe initialization of local statics. You can use this
1592 option to reduce code size slightly in code that doesn't need to be
1595 @item -fuse-cxa-atexit
1596 @opindex fuse-cxa-atexit
1597 Register destructors for objects with static storage duration with the
1598 @code{__cxa_atexit} function rather than the @code{atexit} function.
1599 This option is required for fully standards-compliant handling of static
1600 destructors, but will only work if your C library supports
1601 @code{__cxa_atexit}.
1603 @item -fno-use-cxa-get-exception-ptr
1604 @opindex fno-use-cxa-get-exception-ptr
1605 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1606 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1607 if the runtime routine is not available.
1609 @item -fvisibility-inlines-hidden
1610 @opindex fvisibility-inlines-hidden
1611 This switch declares that the user does not attempt to compare
1612 pointers to inline methods where the addresses of the two functions
1613 were taken in different shared objects.
1615 The effect of this is that GCC may, effectively, mark inline methods with
1616 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1617 appear in the export table of a DSO and do not require a PLT indirection
1618 when used within the DSO@. Enabling this option can have a dramatic effect
1619 on load and link times of a DSO as it massively reduces the size of the
1620 dynamic export table when the library makes heavy use of templates.
1622 The behaviour of this switch is not quite the same as marking the
1623 methods as hidden directly, because it does not affect static variables
1624 local to the function or cause the compiler to deduce that
1625 the function is defined in only one shared object.
1627 You may mark a method as having a visibility explicitly to negate the
1628 effect of the switch for that method. For example, if you do want to
1629 compare pointers to a particular inline method, you might mark it as
1630 having default visibility. Marking the enclosing class with explicit
1631 visibility will have no effect.
1633 Explicitly instantiated inline methods are unaffected by this option
1634 as their linkage might otherwise cross a shared library boundary.
1635 @xref{Template Instantiation}.
1639 Do not use weak symbol support, even if it is provided by the linker.
1640 By default, G++ will use weak symbols if they are available. This
1641 option exists only for testing, and should not be used by end-users;
1642 it will result in inferior code and has no benefits. This option may
1643 be removed in a future release of G++.
1647 Do not search for header files in the standard directories specific to
1648 C++, but do still search the other standard directories. (This option
1649 is used when building the C++ library.)
1652 In addition, these optimization, warning, and code generation options
1653 have meanings only for C++ programs:
1656 @item -fno-default-inline
1657 @opindex fno-default-inline
1658 Do not assume @samp{inline} for functions defined inside a class scope.
1659 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1660 functions will have linkage like inline functions; they just won't be
1663 @item -Wabi @r{(C++ only)}
1665 Warn when G++ generates code that is probably not compatible with the
1666 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1667 all such cases, there are probably some cases that are not warned about,
1668 even though G++ is generating incompatible code. There may also be
1669 cases where warnings are emitted even though the code that is generated
1672 You should rewrite your code to avoid these warnings if you are
1673 concerned about the fact that code generated by G++ may not be binary
1674 compatible with code generated by other compilers.
1676 The known incompatibilities at this point include:
1681 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1682 pack data into the same byte as a base class. For example:
1685 struct A @{ virtual void f(); int f1 : 1; @};
1686 struct B : public A @{ int f2 : 1; @};
1690 In this case, G++ will place @code{B::f2} into the same byte
1691 as@code{A::f1}; other compilers will not. You can avoid this problem
1692 by explicitly padding @code{A} so that its size is a multiple of the
1693 byte size on your platform; that will cause G++ and other compilers to
1694 layout @code{B} identically.
1697 Incorrect handling of tail-padding for virtual bases. G++ does not use
1698 tail padding when laying out virtual bases. For example:
1701 struct A @{ virtual void f(); char c1; @};
1702 struct B @{ B(); char c2; @};
1703 struct C : public A, public virtual B @{@};
1707 In this case, G++ will not place @code{B} into the tail-padding for
1708 @code{A}; other compilers will. You can avoid this problem by
1709 explicitly padding @code{A} so that its size is a multiple of its
1710 alignment (ignoring virtual base classes); that will cause G++ and other
1711 compilers to layout @code{C} identically.
1714 Incorrect handling of bit-fields with declared widths greater than that
1715 of their underlying types, when the bit-fields appear in a union. For
1719 union U @{ int i : 4096; @};
1723 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1724 union too small by the number of bits in an @code{int}.
1727 Empty classes can be placed at incorrect offsets. For example:
1737 struct C : public B, public A @{@};
1741 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1742 it should be placed at offset zero. G++ mistakenly believes that the
1743 @code{A} data member of @code{B} is already at offset zero.
1746 Names of template functions whose types involve @code{typename} or
1747 template template parameters can be mangled incorrectly.
1750 template <typename Q>
1751 void f(typename Q::X) @{@}
1753 template <template <typename> class Q>
1754 void f(typename Q<int>::X) @{@}
1758 Instantiations of these templates may be mangled incorrectly.
1762 @item -Wctor-dtor-privacy @r{(C++ only)}
1763 @opindex Wctor-dtor-privacy
1764 Warn when a class seems unusable because all the constructors or
1765 destructors in that class are private, and it has neither friends nor
1766 public static member functions.
1768 @item -Wnon-virtual-dtor @r{(C++ only)}
1769 @opindex Wnon-virtual-dtor
1770 Warn when a class appears to be polymorphic, thereby requiring a virtual
1771 destructor, yet it declares a non-virtual one. This warning is also
1772 enabled if -Weffc++ is specified.
1774 @item -Wreorder @r{(C++ only)}
1776 @cindex reordering, warning
1777 @cindex warning for reordering of member initializers
1778 Warn when the order of member initializers given in the code does not
1779 match the order in which they must be executed. For instance:
1785 A(): j (0), i (1) @{ @}
1789 The compiler will rearrange the member initializers for @samp{i}
1790 and @samp{j} to match the declaration order of the members, emitting
1791 a warning to that effect. This warning is enabled by @option{-Wall}.
1794 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1797 @item -Weffc++ @r{(C++ only)}
1799 Warn about violations of the following style guidelines from Scott Meyers'
1800 @cite{Effective C++} book:
1804 Item 11: Define a copy constructor and an assignment operator for classes
1805 with dynamically allocated memory.
1808 Item 12: Prefer initialization to assignment in constructors.
1811 Item 14: Make destructors virtual in base classes.
1814 Item 15: Have @code{operator=} return a reference to @code{*this}.
1817 Item 23: Don't try to return a reference when you must return an object.
1821 Also warn about violations of the following style guidelines from
1822 Scott Meyers' @cite{More Effective C++} book:
1826 Item 6: Distinguish between prefix and postfix forms of increment and
1827 decrement operators.
1830 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1834 When selecting this option, be aware that the standard library
1835 headers do not obey all of these guidelines; use @samp{grep -v}
1836 to filter out those warnings.
1838 @item -Wno-deprecated @r{(C++ only)}
1839 @opindex Wno-deprecated
1840 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1842 @item -Wstrict-null-sentinel @r{(C++ only)}
1843 @opindex Wstrict-null-sentinel
1844 Warn also about the use of an uncasted @code{NULL} as sentinel. When
1845 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
1846 to @code{__null}. Although it is a null pointer constant not a null pointer,
1847 it is guaranteed to of the same size as a pointer. But this use is
1848 not portable across different compilers.
1850 @item -Wno-non-template-friend @r{(C++ only)}
1851 @opindex Wno-non-template-friend
1852 Disable warnings when non-templatized friend functions are declared
1853 within a template. Since the advent of explicit template specification
1854 support in G++, if the name of the friend is an unqualified-id (i.e.,
1855 @samp{friend foo(int)}), the C++ language specification demands that the
1856 friend declare or define an ordinary, nontemplate function. (Section
1857 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1858 could be interpreted as a particular specialization of a templatized
1859 function. Because this non-conforming behavior is no longer the default
1860 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1861 check existing code for potential trouble spots and is on by default.
1862 This new compiler behavior can be turned off with
1863 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1864 but disables the helpful warning.
1866 @item -Wold-style-cast @r{(C++ only)}
1867 @opindex Wold-style-cast
1868 Warn if an old-style (C-style) cast to a non-void type is used within
1869 a C++ program. The new-style casts (@samp{dynamic_cast},
1870 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
1871 less vulnerable to unintended effects and much easier to search for.
1873 @item -Woverloaded-virtual @r{(C++ only)}
1874 @opindex Woverloaded-virtual
1875 @cindex overloaded virtual fn, warning
1876 @cindex warning for overloaded virtual fn
1877 Warn when a function declaration hides virtual functions from a
1878 base class. For example, in:
1885 struct B: public A @{
1890 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1898 will fail to compile.
1900 @item -Wno-pmf-conversions @r{(C++ only)}
1901 @opindex Wno-pmf-conversions
1902 Disable the diagnostic for converting a bound pointer to member function
1905 @item -Wsign-promo @r{(C++ only)}
1906 @opindex Wsign-promo
1907 Warn when overload resolution chooses a promotion from unsigned or
1908 enumerated type to a signed type, over a conversion to an unsigned type of
1909 the same size. Previous versions of G++ would try to preserve
1910 unsignedness, but the standard mandates the current behavior.
1915 A& operator = (int);
1925 In this example, G++ will synthesize a default @samp{A& operator =
1926 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1929 @node Objective-C and Objective-C++ Dialect Options
1930 @section Options Controlling Objective-C and Objective-C++ Dialects
1932 @cindex compiler options, Objective-C and Objective-C++
1933 @cindex Objective-C and Objective-C++ options, command line
1934 @cindex options, Objective-C and Objective-C++
1935 (NOTE: This manual does not describe the Objective-C and Objective-C++
1936 languages themselves. See @xref{Standards,,Language Standards
1937 Supported by GCC}, for references.)
1939 This section describes the command-line options that are only meaningful
1940 for Objective-C and Objective-C++ programs, but you can also use most of
1941 the language-independent GNU compiler options.
1942 For example, you might compile a file @code{some_class.m} like this:
1945 gcc -g -fgnu-runtime -O -c some_class.m
1949 In this example, @option{-fgnu-runtime} is an option meant only for
1950 Objective-C and Objective-C++ programs; you can use the other options with
1951 any language supported by GCC@.
1953 Note that since Objective-C is an extension of the C language, Objective-C
1954 compilations may also use options specific to the C front-end (e.g.,
1955 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1956 C++-specific options (e.g., @option{-Wabi}).
1958 Here is a list of options that are @emph{only} for compiling Objective-C
1959 and Objective-C++ programs:
1962 @item -fconstant-string-class=@var{class-name}
1963 @opindex fconstant-string-class
1964 Use @var{class-name} as the name of the class to instantiate for each
1965 literal string specified with the syntax @code{@@"@dots{}"}. The default
1966 class name is @code{NXConstantString} if the GNU runtime is being used, and
1967 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1968 @option{-fconstant-cfstrings} option, if also present, will override the
1969 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1970 to be laid out as constant CoreFoundation strings.
1973 @opindex fgnu-runtime
1974 Generate object code compatible with the standard GNU Objective-C
1975 runtime. This is the default for most types of systems.
1977 @item -fnext-runtime
1978 @opindex fnext-runtime
1979 Generate output compatible with the NeXT runtime. This is the default
1980 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1981 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1984 @item -fno-nil-receivers
1985 @opindex fno-nil-receivers
1986 Assume that all Objective-C message dispatches (e.g.,
1987 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1988 is not @code{nil}. This allows for more efficient entry points in the runtime
1989 to be used. Currently, this option is only available in conjunction with
1990 the NeXT runtime on Mac OS X 10.3 and later.
1992 @item -fobjc-call-cxx-cdtors
1993 @opindex fobjc-call-cxx-cdtors
1994 For each Objective-C class, check if any of its instance variables is a
1995 C++ object with a non-trivial default constructor. If so, synthesize a
1996 special @code{- (id) .cxx_construct} instance method that will run
1997 non-trivial default constructors on any such instance variables, in order,
1998 and then return @code{self}. Similarly, check if any instance variable
1999 is a C++ object with a non-trivial destructor, and if so, synthesize a
2000 special @code{- (void) .cxx_destruct} method that will run
2001 all such default destructors, in reverse order.
2003 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2004 thusly generated will only operate on instance variables declared in the
2005 current Objective-C class, and not those inherited from superclasses. It
2006 is the responsibility of the Objective-C runtime to invoke all such methods
2007 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2008 will be invoked by the runtime immediately after a new object
2009 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2010 be invoked immediately before the runtime deallocates an object instance.
2012 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2013 support for invoking the @code{- (id) .cxx_construct} and
2014 @code{- (void) .cxx_destruct} methods.
2016 @item -fobjc-direct-dispatch
2017 @opindex fobjc-direct-dispatch
2018 Allow fast jumps to the message dispatcher. On Darwin this is
2019 accomplished via the comm page.
2021 @item -fobjc-exceptions
2022 @opindex fobjc-exceptions
2023 Enable syntactic support for structured exception handling in Objective-C,
2024 similar to what is offered by C++ and Java. This option is
2025 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2034 @@catch (AnObjCClass *exc) @{
2041 @@catch (AnotherClass *exc) @{
2044 @@catch (id allOthers) @{
2054 The @code{@@throw} statement may appear anywhere in an Objective-C or
2055 Objective-C++ program; when used inside of a @code{@@catch} block, the
2056 @code{@@throw} may appear without an argument (as shown above), in which case
2057 the object caught by the @code{@@catch} will be rethrown.
2059 Note that only (pointers to) Objective-C objects may be thrown and
2060 caught using this scheme. When an object is thrown, it will be caught
2061 by the nearest @code{@@catch} clause capable of handling objects of that type,
2062 analogously to how @code{catch} blocks work in C++ and Java. A
2063 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2064 any and all Objective-C exceptions not caught by previous @code{@@catch}
2067 The @code{@@finally} clause, if present, will be executed upon exit from the
2068 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2069 regardless of whether any exceptions are thrown, caught or rethrown
2070 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2071 of the @code{finally} clause in Java.
2073 There are several caveats to using the new exception mechanism:
2077 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2078 idioms provided by the @code{NSException} class, the new
2079 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2080 systems, due to additional functionality needed in the (NeXT) Objective-C
2084 As mentioned above, the new exceptions do not support handling
2085 types other than Objective-C objects. Furthermore, when used from
2086 Objective-C++, the Objective-C exception model does not interoperate with C++
2087 exceptions at this time. This means you cannot @code{@@throw} an exception
2088 from Objective-C and @code{catch} it in C++, or vice versa
2089 (i.e., @code{throw @dots{} @@catch}).
2092 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2093 blocks for thread-safe execution:
2096 @@synchronized (ObjCClass *guard) @{
2101 Upon entering the @code{@@synchronized} block, a thread of execution shall
2102 first check whether a lock has been placed on the corresponding @code{guard}
2103 object by another thread. If it has, the current thread shall wait until
2104 the other thread relinquishes its lock. Once @code{guard} becomes available,
2105 the current thread will place its own lock on it, execute the code contained in
2106 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2107 making @code{guard} available to other threads).
2109 Unlike Java, Objective-C does not allow for entire methods to be marked
2110 @code{@@synchronized}. Note that throwing exceptions out of
2111 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2112 to be unlocked properly.
2116 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2118 @item -freplace-objc-classes
2119 @opindex freplace-objc-classes
2120 Emit a special marker instructing @command{ld(1)} not to statically link in
2121 the resulting object file, and allow @command{dyld(1)} to load it in at
2122 run time instead. This is used in conjunction with the Fix-and-Continue
2123 debugging mode, where the object file in question may be recompiled and
2124 dynamically reloaded in the course of program execution, without the need
2125 to restart the program itself. Currently, Fix-and-Continue functionality
2126 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2131 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2132 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2133 compile time) with static class references that get initialized at load time,
2134 which improves run-time performance. Specifying the @option{-fzero-link} flag
2135 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2136 to be retained. This is useful in Zero-Link debugging mode, since it allows
2137 for individual class implementations to be modified during program execution.
2141 Dump interface declarations for all classes seen in the source file to a
2142 file named @file{@var{sourcename}.decl}.
2144 @item -Wassign-intercept
2145 @opindex Wassign-intercept
2146 Warn whenever an Objective-C assignment is being intercepted by the
2150 @opindex Wno-protocol
2151 If a class is declared to implement a protocol, a warning is issued for
2152 every method in the protocol that is not implemented by the class. The
2153 default behavior is to issue a warning for every method not explicitly
2154 implemented in the class, even if a method implementation is inherited
2155 from the superclass. If you use the @option{-Wno-protocol} option, then
2156 methods inherited from the superclass are considered to be implemented,
2157 and no warning is issued for them.
2161 Warn if multiple methods of different types for the same selector are
2162 found during compilation. The check is performed on the list of methods
2163 in the final stage of compilation. Additionally, a check is performed
2164 for each selector appearing in a @code{@@selector(@dots{})}
2165 expression, and a corresponding method for that selector has been found
2166 during compilation. Because these checks scan the method table only at
2167 the end of compilation, these warnings are not produced if the final
2168 stage of compilation is not reached, for example because an error is
2169 found during compilation, or because the @option{-fsyntax-only} option is
2172 @item -Wstrict-selector-match
2173 @opindex Wstrict-selector-match
2174 Warn if multiple methods with differing argument and/or return types are
2175 found for a given selector when attempting to send a message using this
2176 selector to a receiver of type @code{id} or @code{Class}. When this flag
2177 is off (which is the default behavior), the compiler will omit such warnings
2178 if any differences found are confined to types which share the same size
2181 @item -Wundeclared-selector
2182 @opindex Wundeclared-selector
2183 Warn if a @code{@@selector(@dots{})} expression referring to an
2184 undeclared selector is found. A selector is considered undeclared if no
2185 method with that name has been declared before the
2186 @code{@@selector(@dots{})} expression, either explicitly in an
2187 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2188 an @code{@@implementation} section. This option always performs its
2189 checks as soon as a @code{@@selector(@dots{})} expression is found,
2190 while @option{-Wselector} only performs its checks in the final stage of
2191 compilation. This also enforces the coding style convention
2192 that methods and selectors must be declared before being used.
2194 @item -print-objc-runtime-info
2195 @opindex print-objc-runtime-info
2196 Generate C header describing the largest structure that is passed by
2201 @node Language Independent Options
2202 @section Options to Control Diagnostic Messages Formatting
2203 @cindex options to control diagnostics formatting
2204 @cindex diagnostic messages
2205 @cindex message formatting
2207 Traditionally, diagnostic messages have been formatted irrespective of
2208 the output device's aspect (e.g.@: its width, @dots{}). The options described
2209 below can be used to control the diagnostic messages formatting
2210 algorithm, e.g.@: how many characters per line, how often source location
2211 information should be reported. Right now, only the C++ front end can
2212 honor these options. However it is expected, in the near future, that
2213 the remaining front ends would be able to digest them correctly.
2216 @item -fmessage-length=@var{n}
2217 @opindex fmessage-length
2218 Try to format error messages so that they fit on lines of about @var{n}
2219 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2220 the front ends supported by GCC@. If @var{n} is zero, then no
2221 line-wrapping will be done; each error message will appear on a single
2224 @opindex fdiagnostics-show-location
2225 @item -fdiagnostics-show-location=once
2226 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2227 reporter to emit @emph{once} source location information; that is, in
2228 case the message is too long to fit on a single physical line and has to
2229 be wrapped, the source location won't be emitted (as prefix) again,
2230 over and over, in subsequent continuation lines. This is the default
2233 @item -fdiagnostics-show-location=every-line
2234 Only meaningful in line-wrapping mode. Instructs the diagnostic
2235 messages reporter to emit the same source location information (as
2236 prefix) for physical lines that result from the process of breaking
2237 a message which is too long to fit on a single line.
2239 @item -fdiagnostics-show-options
2240 @opindex fdiagnostics-show-options
2241 This option instructs the diagnostic machinery to add text to each
2242 diagnostic emitted, which indicates which command line option directly
2243 controls that diagnostic, when such an option is known to the
2244 diagnostic machinery.
2248 @node Warning Options
2249 @section Options to Request or Suppress Warnings
2250 @cindex options to control warnings
2251 @cindex warning messages
2252 @cindex messages, warning
2253 @cindex suppressing warnings
2255 Warnings are diagnostic messages that report constructions which
2256 are not inherently erroneous but which are risky or suggest there
2257 may have been an error.
2259 You can request many specific warnings with options beginning @samp{-W},
2260 for example @option{-Wimplicit} to request warnings on implicit
2261 declarations. Each of these specific warning options also has a
2262 negative form beginning @samp{-Wno-} to turn off warnings;
2263 for example, @option{-Wno-implicit}. This manual lists only one of the
2264 two forms, whichever is not the default.
2266 The following options control the amount and kinds of warnings produced
2267 by GCC; for further, language-specific options also refer to
2268 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2272 @cindex syntax checking
2274 @opindex fsyntax-only
2275 Check the code for syntax errors, but don't do anything beyond that.
2279 Issue all the warnings demanded by strict ISO C and ISO C++;
2280 reject all programs that use forbidden extensions, and some other
2281 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2282 version of the ISO C standard specified by any @option{-std} option used.
2284 Valid ISO C and ISO C++ programs should compile properly with or without
2285 this option (though a rare few will require @option{-ansi} or a
2286 @option{-std} option specifying the required version of ISO C)@. However,
2287 without this option, certain GNU extensions and traditional C and C++
2288 features are supported as well. With this option, they are rejected.
2290 @option{-pedantic} does not cause warning messages for use of the
2291 alternate keywords whose names begin and end with @samp{__}. Pedantic
2292 warnings are also disabled in the expression that follows
2293 @code{__extension__}. However, only system header files should use
2294 these escape routes; application programs should avoid them.
2295 @xref{Alternate Keywords}.
2297 Some users try to use @option{-pedantic} to check programs for strict ISO
2298 C conformance. They soon find that it does not do quite what they want:
2299 it finds some non-ISO practices, but not all---only those for which
2300 ISO C @emph{requires} a diagnostic, and some others for which
2301 diagnostics have been added.
2303 A feature to report any failure to conform to ISO C might be useful in
2304 some instances, but would require considerable additional work and would
2305 be quite different from @option{-pedantic}. We don't have plans to
2306 support such a feature in the near future.
2308 Where the standard specified with @option{-std} represents a GNU
2309 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2310 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2311 extended dialect is based. Warnings from @option{-pedantic} are given
2312 where they are required by the base standard. (It would not make sense
2313 for such warnings to be given only for features not in the specified GNU
2314 C dialect, since by definition the GNU dialects of C include all
2315 features the compiler supports with the given option, and there would be
2316 nothing to warn about.)
2318 @item -pedantic-errors
2319 @opindex pedantic-errors
2320 Like @option{-pedantic}, except that errors are produced rather than
2325 Inhibit all warning messages.
2329 Inhibit warning messages about the use of @samp{#import}.
2331 @item -Wchar-subscripts
2332 @opindex Wchar-subscripts
2333 Warn if an array subscript has type @code{char}. This is a common cause
2334 of error, as programmers often forget that this type is signed on some
2336 This warning is enabled by @option{-Wall}.
2340 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2341 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2342 This warning is enabled by @option{-Wall}.
2344 @item -Wfatal-errors
2345 @opindex Wfatal-errors
2346 This option causes the compiler to abort compilation on the first error
2347 occurred rather than trying to keep going and printing further error
2352 @opindex ffreestanding
2353 @opindex fno-builtin
2354 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2355 the arguments supplied have types appropriate to the format string
2356 specified, and that the conversions specified in the format string make
2357 sense. This includes standard functions, and others specified by format
2358 attributes (@pxref{Function Attributes}), in the @code{printf},
2359 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2360 not in the C standard) families (or other target-specific families).
2361 Which functions are checked without format attributes having been
2362 specified depends on the standard version selected, and such checks of
2363 functions without the attribute specified are disabled by
2364 @option{-ffreestanding} or @option{-fno-builtin}.
2366 The formats are checked against the format features supported by GNU
2367 libc version 2.2. These include all ISO C90 and C99 features, as well
2368 as features from the Single Unix Specification and some BSD and GNU
2369 extensions. Other library implementations may not support all these
2370 features; GCC does not support warning about features that go beyond a
2371 particular library's limitations. However, if @option{-pedantic} is used
2372 with @option{-Wformat}, warnings will be given about format features not
2373 in the selected standard version (but not for @code{strfmon} formats,
2374 since those are not in any version of the C standard). @xref{C Dialect
2375 Options,,Options Controlling C Dialect}.
2377 Since @option{-Wformat} also checks for null format arguments for
2378 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2380 @option{-Wformat} is included in @option{-Wall}. For more control over some
2381 aspects of format checking, the options @option{-Wformat-y2k},
2382 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2383 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2384 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2387 @opindex Wformat-y2k
2388 If @option{-Wformat} is specified, also warn about @code{strftime}
2389 formats which may yield only a two-digit year.
2391 @item -Wno-format-extra-args
2392 @opindex Wno-format-extra-args
2393 If @option{-Wformat} is specified, do not warn about excess arguments to a
2394 @code{printf} or @code{scanf} format function. The C standard specifies
2395 that such arguments are ignored.
2397 Where the unused arguments lie between used arguments that are
2398 specified with @samp{$} operand number specifications, normally
2399 warnings are still given, since the implementation could not know what
2400 type to pass to @code{va_arg} to skip the unused arguments. However,
2401 in the case of @code{scanf} formats, this option will suppress the
2402 warning if the unused arguments are all pointers, since the Single
2403 Unix Specification says that such unused arguments are allowed.
2405 @item -Wno-format-zero-length
2406 @opindex Wno-format-zero-length
2407 If @option{-Wformat} is specified, do not warn about zero-length formats.
2408 The C standard specifies that zero-length formats are allowed.
2410 @item -Wformat-nonliteral
2411 @opindex Wformat-nonliteral
2412 If @option{-Wformat} is specified, also warn if the format string is not a
2413 string literal and so cannot be checked, unless the format function
2414 takes its format arguments as a @code{va_list}.
2416 @item -Wformat-security
2417 @opindex Wformat-security
2418 If @option{-Wformat} is specified, also warn about uses of format
2419 functions that represent possible security problems. At present, this
2420 warns about calls to @code{printf} and @code{scanf} functions where the
2421 format string is not a string literal and there are no format arguments,
2422 as in @code{printf (foo);}. This may be a security hole if the format
2423 string came from untrusted input and contains @samp{%n}. (This is
2424 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2425 in future warnings may be added to @option{-Wformat-security} that are not
2426 included in @option{-Wformat-nonliteral}.)
2430 Enable @option{-Wformat} plus format checks not included in
2431 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2432 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2436 Warn about passing a null pointer for arguments marked as
2437 requiring a non-null value by the @code{nonnull} function attribute.
2439 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2440 can be disabled with the @option{-Wno-nonnull} option.
2442 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2444 Warn about uninitialized variables which are initialized with themselves.
2445 Note this option can only be used with the @option{-Wuninitialized} option,
2446 which in turn only works with @option{-O1} and above.
2448 For example, GCC will warn about @code{i} being uninitialized in the
2449 following snippet only when @option{-Winit-self} has been specified:
2460 @item -Wimplicit-int
2461 @opindex Wimplicit-int
2462 Warn when a declaration does not specify a type.
2463 This warning is enabled by @option{-Wall}.
2465 @item -Wimplicit-function-declaration
2466 @itemx -Werror-implicit-function-declaration
2467 @opindex Wimplicit-function-declaration
2468 @opindex Werror-implicit-function-declaration
2469 Give a warning (or error) whenever a function is used before being
2470 declared. The form @option{-Wno-error-implicit-function-declaration}
2472 This warning is enabled by @option{-Wall} (as a warning, not an error).
2476 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2477 This warning is enabled by @option{-Wall}.
2481 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2482 function with external linkage, returning int, taking either zero
2483 arguments, two, or three arguments of appropriate types.
2484 This warning is enabled by @option{-Wall}.
2486 @item -Wmissing-braces
2487 @opindex Wmissing-braces
2488 Warn if an aggregate or union initializer is not fully bracketed. In
2489 the following example, the initializer for @samp{a} is not fully
2490 bracketed, but that for @samp{b} is fully bracketed.
2493 int a[2][2] = @{ 0, 1, 2, 3 @};
2494 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2497 This warning is enabled by @option{-Wall}.
2499 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2500 @opindex Wmissing-include-dirs
2501 Warn if a user-supplied include directory does not exist.
2504 @opindex Wparentheses
2505 Warn if parentheses are omitted in certain contexts, such
2506 as when there is an assignment in a context where a truth value
2507 is expected, or when operators are nested whose precedence people
2508 often get confused about. Only the warning for an assignment used as
2509 a truth value is supported when compiling C++; the other warnings are
2510 only supported when compiling C@.
2512 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2513 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2514 interpretation from that of ordinary mathematical notation.
2516 Also warn about constructions where there may be confusion to which
2517 @code{if} statement an @code{else} branch belongs. Here is an example of
2532 In C, every @code{else} branch belongs to the innermost possible @code{if}
2533 statement, which in this example is @code{if (b)}. This is often not
2534 what the programmer expected, as illustrated in the above example by
2535 indentation the programmer chose. When there is the potential for this
2536 confusion, GCC will issue a warning when this flag is specified.
2537 To eliminate the warning, add explicit braces around the innermost
2538 @code{if} statement so there is no way the @code{else} could belong to
2539 the enclosing @code{if}. The resulting code would look like this:
2555 This warning is enabled by @option{-Wall}.
2557 @item -Wsequence-point
2558 @opindex Wsequence-point
2559 Warn about code that may have undefined semantics because of violations
2560 of sequence point rules in the C and C++ standards.
2562 The C and C++ standards defines the order in which expressions in a C/C++
2563 program are evaluated in terms of @dfn{sequence points}, which represent
2564 a partial ordering between the execution of parts of the program: those
2565 executed before the sequence point, and those executed after it. These
2566 occur after the evaluation of a full expression (one which is not part
2567 of a larger expression), after the evaluation of the first operand of a
2568 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2569 function is called (but after the evaluation of its arguments and the
2570 expression denoting the called function), and in certain other places.
2571 Other than as expressed by the sequence point rules, the order of
2572 evaluation of subexpressions of an expression is not specified. All
2573 these rules describe only a partial order rather than a total order,
2574 since, for example, if two functions are called within one expression
2575 with no sequence point between them, the order in which the functions
2576 are called is not specified. However, the standards committee have
2577 ruled that function calls do not overlap.
2579 It is not specified when between sequence points modifications to the
2580 values of objects take effect. Programs whose behavior depends on this
2581 have undefined behavior; the C and C++ standards specify that ``Between
2582 the previous and next sequence point an object shall have its stored
2583 value modified at most once by the evaluation of an expression.
2584 Furthermore, the prior value shall be read only to determine the value
2585 to be stored.''. If a program breaks these rules, the results on any
2586 particular implementation are entirely unpredictable.
2588 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2589 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2590 diagnosed by this option, and it may give an occasional false positive
2591 result, but in general it has been found fairly effective at detecting
2592 this sort of problem in programs.
2594 The standard is worded confusingly, therefore there is some debate
2595 over the precise meaning of the sequence point rules in subtle cases.
2596 Links to discussions of the problem, including proposed formal
2597 definitions, may be found on the GCC readings page, at
2598 @w{@uref{http://gcc.gnu.org/readings.html}}.
2600 This warning is enabled by @option{-Wall} for C and C++.
2603 @opindex Wreturn-type
2604 Warn whenever a function is defined with a return-type that defaults to
2605 @code{int}. Also warn about any @code{return} statement with no
2606 return-value in a function whose return-type is not @code{void}.
2608 For C, also warn if the return type of a function has a type qualifier
2609 such as @code{const}. Such a type qualifier has no effect, since the
2610 value returned by a function is not an lvalue. ISO C prohibits
2611 qualified @code{void} return types on function definitions, so such
2612 return types always receive a warning even without this option.
2614 For C++, a function without return type always produces a diagnostic
2615 message, even when @option{-Wno-return-type} is specified. The only
2616 exceptions are @samp{main} and functions defined in system headers.
2618 This warning is enabled by @option{-Wall}.
2622 Warn whenever a @code{switch} statement has an index of enumerated type
2623 and lacks a @code{case} for one or more of the named codes of that
2624 enumeration. (The presence of a @code{default} label prevents this
2625 warning.) @code{case} labels outside the enumeration range also
2626 provoke warnings when this option is used.
2627 This warning is enabled by @option{-Wall}.
2629 @item -Wswitch-default
2630 @opindex Wswitch-switch
2631 Warn whenever a @code{switch} statement does not have a @code{default}
2635 @opindex Wswitch-enum
2636 Warn whenever a @code{switch} statement has an index of enumerated type
2637 and lacks a @code{case} for one or more of the named codes of that
2638 enumeration. @code{case} labels outside the enumeration range also
2639 provoke warnings when this option is used.
2643 Warn if any trigraphs are encountered that might change the meaning of
2644 the program (trigraphs within comments are not warned about).
2645 This warning is enabled by @option{-Wall}.
2647 @item -Wunused-function
2648 @opindex Wunused-function
2649 Warn whenever a static function is declared but not defined or a
2650 non-inline static function is unused.
2651 This warning is enabled by @option{-Wall}.
2653 @item -Wunused-label
2654 @opindex Wunused-label
2655 Warn whenever a label is declared but not used.
2656 This warning is enabled by @option{-Wall}.
2658 To suppress this warning use the @samp{unused} attribute
2659 (@pxref{Variable Attributes}).
2661 @item -Wunused-parameter
2662 @opindex Wunused-parameter
2663 Warn whenever a function parameter is unused aside from its declaration.
2665 To suppress this warning use the @samp{unused} attribute
2666 (@pxref{Variable Attributes}).
2668 @item -Wunused-variable
2669 @opindex Wunused-variable
2670 Warn whenever a local variable or non-constant static variable is unused
2671 aside from its declaration.
2672 This warning is enabled by @option{-Wall}.
2674 To suppress this warning use the @samp{unused} attribute
2675 (@pxref{Variable Attributes}).
2677 @item -Wunused-value
2678 @opindex Wunused-value
2679 Warn whenever a statement computes a result that is explicitly not used.
2680 This warning is enabled by @option{-Wall}.
2682 To suppress this warning cast the expression to @samp{void}.
2686 All the above @option{-Wunused} options combined.
2688 In order to get a warning about an unused function parameter, you must
2689 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2690 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2692 @item -Wuninitialized
2693 @opindex Wuninitialized
2694 Warn if an automatic variable is used without first being initialized or
2695 if a variable may be clobbered by a @code{setjmp} call.
2697 These warnings are possible only in optimizing compilation,
2698 because they require data flow information that is computed only
2699 when optimizing. If you do not specify @option{-O}, you will not get
2700 these warnings. Instead, GCC will issue a warning about @option{-Wuninitialized}
2701 requiring @option{-O}.
2703 If you want to warn about code which uses the uninitialized value of the
2704 variable in its own initializer, use the @option{-Winit-self} option.
2706 These warnings occur for individual uninitialized or clobbered
2707 elements of structure, union or array variables as well as for
2708 variables which are uninitialized or clobbered as a whole. They do
2709 not occur for variables or elements declared @code{volatile}. Because
2710 these warnings depend on optimization, the exact variables or elements
2711 for which there are warnings will depend on the precise optimization
2712 options and version of GCC used.
2714 Note that there may be no warning about a variable that is used only
2715 to compute a value that itself is never used, because such
2716 computations may be deleted by data flow analysis before the warnings
2719 These warnings are made optional because GCC is not smart
2720 enough to see all the reasons why the code might be correct
2721 despite appearing to have an error. Here is one example of how
2742 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2743 always initialized, but GCC doesn't know this. Here is
2744 another common case:
2749 if (change_y) save_y = y, y = new_y;
2751 if (change_y) y = save_y;
2756 This has no bug because @code{save_y} is used only if it is set.
2758 @cindex @code{longjmp} warnings
2759 This option also warns when a non-volatile automatic variable might be
2760 changed by a call to @code{longjmp}. These warnings as well are possible
2761 only in optimizing compilation.
2763 The compiler sees only the calls to @code{setjmp}. It cannot know
2764 where @code{longjmp} will be called; in fact, a signal handler could
2765 call it at any point in the code. As a result, you may get a warning
2766 even when there is in fact no problem because @code{longjmp} cannot
2767 in fact be called at the place which would cause a problem.
2769 Some spurious warnings can be avoided if you declare all the functions
2770 you use that never return as @code{noreturn}. @xref{Function
2773 This warning is enabled by @option{-Wall}.
2775 @item -Wunknown-pragmas
2776 @opindex Wunknown-pragmas
2777 @cindex warning for unknown pragmas
2778 @cindex unknown pragmas, warning
2779 @cindex pragmas, warning of unknown
2780 Warn when a #pragma directive is encountered which is not understood by
2781 GCC@. If this command line option is used, warnings will even be issued
2782 for unknown pragmas in system header files. This is not the case if
2783 the warnings were only enabled by the @option{-Wall} command line option.
2786 @opindex Wno-pragmas
2788 Do not warn about misuses of pragmas, such as incorrect parameters,
2789 invalid syntax, or conflicts between pragmas. See also
2790 @samp{-Wunknown-pragmas}.
2792 @item -Wstrict-aliasing
2793 @opindex Wstrict-aliasing
2794 This option is only active when @option{-fstrict-aliasing} is active.
2795 It warns about code which might break the strict aliasing rules that the
2796 compiler is using for optimization. The warning does not catch all
2797 cases, but does attempt to catch the more common pitfalls. It is
2798 included in @option{-Wall}.
2800 @item -Wstrict-aliasing=2
2801 @opindex Wstrict-aliasing=2
2802 This option is only active when @option{-fstrict-aliasing} is active.
2803 It warns about code which might break the strict aliasing rules that the
2804 compiler is using for optimization. This warning catches more cases than
2805 @option{-Wstrict-aliasing}, but it will also give a warning for some ambiguous
2806 cases that are safe.
2810 All of the above @samp{-W} options combined. This enables all the
2811 warnings about constructions that some users consider questionable, and
2812 that are easy to avoid (or modify to prevent the warning), even in
2813 conjunction with macros. This also enables some language-specific
2814 warnings described in @ref{C++ Dialect Options} and
2815 @ref{Objective-C and Objective-C++ Dialect Options}.
2818 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2819 Some of them warn about constructions that users generally do not
2820 consider questionable, but which occasionally you might wish to check
2821 for; others warn about constructions that are necessary or hard to avoid
2822 in some cases, and there is no simple way to modify the code to suppress
2829 (This option used to be called @option{-W}. The older name is still
2830 supported, but the newer name is more descriptive.) Print extra warning
2831 messages for these events:
2835 A function can return either with or without a value. (Falling
2836 off the end of the function body is considered returning without
2837 a value.) For example, this function would evoke such a
2851 An expression-statement or the left-hand side of a comma expression
2852 contains no side effects.
2853 To suppress the warning, cast the unused expression to void.
2854 For example, an expression such as @samp{x[i,j]} will cause a warning,
2855 but @samp{x[(void)i,j]} will not.
2858 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2861 Storage-class specifiers like @code{static} are not the first things in
2862 a declaration. According to the C Standard, this usage is obsolescent.
2865 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2869 A comparison between signed and unsigned values could produce an
2870 incorrect result when the signed value is converted to unsigned.
2871 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2874 An aggregate has an initializer which does not initialize all members.
2875 This warning can be independently controlled by
2876 @option{-Wmissing-field-initializers}.
2879 A function parameter is declared without a type specifier in K&R-style
2887 An empty body occurs in an @samp{if} or @samp{else} statement.
2890 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2891 @samp{>}, or @samp{>=}.
2894 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2897 Any of several floating-point events that often indicate errors, such as
2898 overflow, underflow, loss of precision, etc.
2900 @item @r{(C++ only)}
2901 An enumerator and a non-enumerator both appear in a conditional expression.
2903 @item @r{(C++ only)}
2904 A non-static reference or non-static @samp{const} member appears in a
2905 class without constructors.
2907 @item @r{(C++ only)}
2908 Ambiguous virtual bases.
2910 @item @r{(C++ only)}
2911 Subscripting an array which has been declared @samp{register}.
2913 @item @r{(C++ only)}
2914 Taking the address of a variable which has been declared @samp{register}.
2916 @item @r{(C++ only)}
2917 A base class is not initialized in a derived class' copy constructor.
2920 @item -Wno-div-by-zero
2921 @opindex Wno-div-by-zero
2922 @opindex Wdiv-by-zero
2923 Do not warn about compile-time integer division by zero. Floating point
2924 division by zero is not warned about, as it can be a legitimate way of
2925 obtaining infinities and NaNs.
2927 @item -Wsystem-headers
2928 @opindex Wsystem-headers
2929 @cindex warnings from system headers
2930 @cindex system headers, warnings from
2931 Print warning messages for constructs found in system header files.
2932 Warnings from system headers are normally suppressed, on the assumption
2933 that they usually do not indicate real problems and would only make the
2934 compiler output harder to read. Using this command line option tells
2935 GCC to emit warnings from system headers as if they occurred in user
2936 code. However, note that using @option{-Wall} in conjunction with this
2937 option will @emph{not} warn about unknown pragmas in system
2938 headers---for that, @option{-Wunknown-pragmas} must also be used.
2941 @opindex Wfloat-equal
2942 Warn if floating point values are used in equality comparisons.
2944 The idea behind this is that sometimes it is convenient (for the
2945 programmer) to consider floating-point values as approximations to
2946 infinitely precise real numbers. If you are doing this, then you need
2947 to compute (by analyzing the code, or in some other way) the maximum or
2948 likely maximum error that the computation introduces, and allow for it
2949 when performing comparisons (and when producing output, but that's a
2950 different problem). In particular, instead of testing for equality, you
2951 would check to see whether the two values have ranges that overlap; and
2952 this is done with the relational operators, so equality comparisons are
2955 @item -Wtraditional @r{(C only)}
2956 @opindex Wtraditional
2957 Warn about certain constructs that behave differently in traditional and
2958 ISO C@. Also warn about ISO C constructs that have no traditional C
2959 equivalent, and/or problematic constructs which should be avoided.
2963 Macro parameters that appear within string literals in the macro body.
2964 In traditional C macro replacement takes place within string literals,
2965 but does not in ISO C@.
2968 In traditional C, some preprocessor directives did not exist.
2969 Traditional preprocessors would only consider a line to be a directive
2970 if the @samp{#} appeared in column 1 on the line. Therefore
2971 @option{-Wtraditional} warns about directives that traditional C
2972 understands but would ignore because the @samp{#} does not appear as the
2973 first character on the line. It also suggests you hide directives like
2974 @samp{#pragma} not understood by traditional C by indenting them. Some
2975 traditional implementations would not recognize @samp{#elif}, so it
2976 suggests avoiding it altogether.
2979 A function-like macro that appears without arguments.
2982 The unary plus operator.
2985 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2986 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2987 constants.) Note, these suffixes appear in macros defined in the system
2988 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2989 Use of these macros in user code might normally lead to spurious
2990 warnings, however GCC's integrated preprocessor has enough context to
2991 avoid warning in these cases.
2994 A function declared external in one block and then used after the end of
2998 A @code{switch} statement has an operand of type @code{long}.
3001 A non-@code{static} function declaration follows a @code{static} one.
3002 This construct is not accepted by some traditional C compilers.
3005 The ISO type of an integer constant has a different width or
3006 signedness from its traditional type. This warning is only issued if
3007 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3008 typically represent bit patterns, are not warned about.
3011 Usage of ISO string concatenation is detected.
3014 Initialization of automatic aggregates.
3017 Identifier conflicts with labels. Traditional C lacks a separate
3018 namespace for labels.
3021 Initialization of unions. If the initializer is zero, the warning is
3022 omitted. This is done under the assumption that the zero initializer in
3023 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3024 initializer warnings and relies on default initialization to zero in the
3028 Conversions by prototypes between fixed/floating point values and vice
3029 versa. The absence of these prototypes when compiling with traditional
3030 C would cause serious problems. This is a subset of the possible
3031 conversion warnings, for the full set use @option{-Wconversion}.
3034 Use of ISO C style function definitions. This warning intentionally is
3035 @emph{not} issued for prototype declarations or variadic functions
3036 because these ISO C features will appear in your code when using
3037 libiberty's traditional C compatibility macros, @code{PARAMS} and
3038 @code{VPARAMS}. This warning is also bypassed for nested functions
3039 because that feature is already a GCC extension and thus not relevant to
3040 traditional C compatibility.
3043 @item -Wdeclaration-after-statement @r{(C only)}
3044 @opindex Wdeclaration-after-statement
3045 Warn when a declaration is found after a statement in a block. This
3046 construct, known from C++, was introduced with ISO C99 and is by default
3047 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3048 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3052 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3054 @item -Wno-endif-labels
3055 @opindex Wno-endif-labels
3056 @opindex Wendif-labels
3057 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3061 Warn whenever a local variable shadows another local variable, parameter or
3062 global variable or whenever a built-in function is shadowed.
3064 @item -Wlarger-than-@var{len}
3065 @opindex Wlarger-than
3066 Warn whenever an object of larger than @var{len} bytes is defined.
3068 @item -Wunsafe-loop-optimizations
3069 @opindex Wunsafe-loop-optimizations
3070 Warn if the loop cannot be optimized because the compiler could not
3071 assume anything on the bounds of the loop indices. With
3072 @option{-funsafe-loop-optimizations} warn if the compiler made
3075 @item -Wpointer-arith
3076 @opindex Wpointer-arith
3077 Warn about anything that depends on the ``size of'' a function type or
3078 of @code{void}. GNU C assigns these types a size of 1, for
3079 convenience in calculations with @code{void *} pointers and pointers
3082 @item -Wbad-function-cast @r{(C only)}
3083 @opindex Wbad-function-cast
3084 Warn whenever a function call is cast to a non-matching type.
3085 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3088 Warn about ISO C constructs that are outside of the common subset of
3089 ISO C and ISO C++, e.g.@: request for implicit conversion from
3090 @code{void *} to a pointer to non-@code{void} type.
3094 Warn whenever a pointer is cast so as to remove a type qualifier from
3095 the target type. For example, warn if a @code{const char *} is cast
3096 to an ordinary @code{char *}.
3099 @opindex Wcast-align
3100 Warn whenever a pointer is cast such that the required alignment of the
3101 target is increased. For example, warn if a @code{char *} is cast to
3102 an @code{int *} on machines where integers can only be accessed at
3103 two- or four-byte boundaries.
3105 @item -Wwrite-strings
3106 @opindex Wwrite-strings
3107 When compiling C, give string constants the type @code{const
3108 char[@var{length}]} so that
3109 copying the address of one into a non-@code{const} @code{char *}
3110 pointer will get a warning; when compiling C++, warn about the
3111 deprecated conversion from string literals to @code{char *}. This
3112 warning, by default, is enabled for C++ programs.
3113 These warnings will help you find at
3114 compile time code that can try to write into a string constant, but
3115 only if you have been very careful about using @code{const} in
3116 declarations and prototypes. Otherwise, it will just be a nuisance;
3117 this is why we did not make @option{-Wall} request these warnings.
3120 @opindex Wconversion
3121 Warn if a prototype causes a type conversion that is different from what
3122 would happen to the same argument in the absence of a prototype. This
3123 includes conversions of fixed point to floating and vice versa, and
3124 conversions changing the width or signedness of a fixed point argument
3125 except when the same as the default promotion.
3127 Also, warn if a negative integer constant expression is implicitly
3128 converted to an unsigned type. For example, warn about the assignment
3129 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
3130 casts like @code{(unsigned) -1}.
3132 @item -Wsign-compare
3133 @opindex Wsign-compare
3134 @cindex warning for comparison of signed and unsigned values
3135 @cindex comparison of signed and unsigned values, warning
3136 @cindex signed and unsigned values, comparison warning
3137 Warn when a comparison between signed and unsigned values could produce
3138 an incorrect result when the signed value is converted to unsigned.
3139 This warning is also enabled by @option{-Wextra}; to get the other warnings
3140 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3142 @item -Waggregate-return
3143 @opindex Waggregate-return
3144 Warn if any functions that return structures or unions are defined or
3145 called. (In languages where you can return an array, this also elicits
3149 @opindex Walways-true
3150 Warn about comparisons which are always true such as testing if
3151 unsigned values are greater than or equal to zero. This warning is
3152 enabled by @option{-Wall}.
3154 @item -Wno-attributes
3155 @opindex Wno-attributes
3156 @opindex Wattributes
3157 Do not warn if an unexpected @code{__attribute__} is used, such as
3158 unrecognized attributes, function attributes applied to variables,
3159 etc. This will not stop errors for incorrect use of supported
3162 @item -Wstrict-prototypes @r{(C only)}
3163 @opindex Wstrict-prototypes
3164 Warn if a function is declared or defined without specifying the
3165 argument types. (An old-style function definition is permitted without
3166 a warning if preceded by a declaration which specifies the argument
3169 @item -Wold-style-definition @r{(C only)}
3170 @opindex Wold-style-definition
3171 Warn if an old-style function definition is used. A warning is given
3172 even if there is a previous prototype.
3174 @item -Wmissing-prototypes @r{(C only)}
3175 @opindex Wmissing-prototypes
3176 Warn if a global function is defined without a previous prototype
3177 declaration. This warning is issued even if the definition itself
3178 provides a prototype. The aim is to detect global functions that fail
3179 to be declared in header files.
3181 @item -Wmissing-declarations @r{(C only)}
3182 @opindex Wmissing-declarations
3183 Warn if a global function is defined without a previous declaration.
3184 Do so even if the definition itself provides a prototype.
3185 Use this option to detect global functions that are not declared in
3188 @item -Wmissing-field-initializers
3189 @opindex Wmissing-field-initializers
3192 Warn if a structure's initializer has some fields missing. For
3193 example, the following code would cause such a warning, because
3194 @code{x.h} is implicitly zero:
3197 struct s @{ int f, g, h; @};
3198 struct s x = @{ 3, 4 @};
3201 This option does not warn about designated initializers, so the following
3202 modification would not trigger a warning:
3205 struct s @{ int f, g, h; @};
3206 struct s x = @{ .f = 3, .g = 4 @};
3209 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3210 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3212 @item -Wmissing-noreturn
3213 @opindex Wmissing-noreturn
3214 Warn about functions which might be candidates for attribute @code{noreturn}.
3215 Note these are only possible candidates, not absolute ones. Care should
3216 be taken to manually verify functions actually do not ever return before
3217 adding the @code{noreturn} attribute, otherwise subtle code generation
3218 bugs could be introduced. You will not get a warning for @code{main} in
3219 hosted C environments.
3221 @item -Wmissing-format-attribute
3222 @opindex Wmissing-format-attribute
3224 Warn about function pointers which might be candidates for @code{format}
3225 attributes. Note these are only possible candidates, not absolute ones.
3226 GCC will guess that function pointers with @code{format} attributes that
3227 are used in assignment, initialization, parameter passing or return
3228 statements should have a corresponding @code{format} attribute in the
3229 resulting type. I.e.@: the left-hand side of the assignment or
3230 initialization, the type of the parameter variable, or the return type
3231 of the containing function respectively should also have a @code{format}
3232 attribute to avoid the warning.
3234 GCC will also warn about function definitions which might be
3235 candidates for @code{format} attributes. Again, these are only
3236 possible candidates. GCC will guess that @code{format} attributes
3237 might be appropriate for any function that calls a function like
3238 @code{vprintf} or @code{vscanf}, but this might not always be the
3239 case, and some functions for which @code{format} attributes are
3240 appropriate may not be detected.
3242 @item -Wno-multichar
3243 @opindex Wno-multichar
3245 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3246 Usually they indicate a typo in the user's code, as they have
3247 implementation-defined values, and should not be used in portable code.
3249 @item -Wnormalized=<none|id|nfc|nfkc>
3250 @opindex Wnormalized
3253 @cindex character set, input normalization
3254 In ISO C and ISO C++, two identifiers are different if they are
3255 different sequences of characters. However, sometimes when characters
3256 outside the basic ASCII character set are used, you can have two
3257 different character sequences that look the same. To avoid confusion,
3258 the ISO 10646 standard sets out some @dfn{normalization rules} which
3259 when applied ensure that two sequences that look the same are turned into
3260 the same sequence. GCC can warn you if you are using identifiers which
3261 have not been normalized; this option controls that warning.
3263 There are four levels of warning that GCC supports. The default is
3264 @option{-Wnormalized=nfc}, which warns about any identifier which is
3265 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
3266 recommended form for most uses.
3268 Unfortunately, there are some characters which ISO C and ISO C++ allow
3269 in identifiers that when turned into NFC aren't allowable as
3270 identifiers. That is, there's no way to use these symbols in portable
3271 ISO C or C++ and have all your identifiers in NFC.
3272 @option{-Wnormalized=id} suppresses the warning for these characters.
3273 It is hoped that future versions of the standards involved will correct
3274 this, which is why this option is not the default.
3276 You can switch the warning off for all characters by writing
3277 @option{-Wnormalized=none}. You would only want to do this if you
3278 were using some other normalization scheme (like ``D''), because
3279 otherwise you can easily create bugs that are literally impossible to see.
3281 Some characters in ISO 10646 have distinct meanings but look identical
3282 in some fonts or display methodologies, especially once formatting has
3283 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
3284 LETTER N'', will display just like a regular @code{n} which has been
3285 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
3286 normalization scheme to convert all these into a standard form as
3287 well, and GCC will warn if your code is not in NFKC if you use
3288 @option{-Wnormalized=nfkc}. This warning is comparable to warning
3289 about every identifier that contains the letter O because it might be
3290 confused with the digit 0, and so is not the default, but may be
3291 useful as a local coding convention if the programming environment is
3292 unable to be fixed to display these characters distinctly.
3294 @item -Wno-deprecated-declarations
3295 @opindex Wno-deprecated-declarations
3296 Do not warn about uses of functions (@pxref{Function Attributes}),
3297 variables (@pxref{Variable Attributes}), and types (@pxref{Type
3298 Attributes}) marked as deprecated by using the @code{deprecated}
3302 @opindex Wno-overflow
3303 Do not warn about compile-time overflow in constant expressions.
3307 Warn if a structure is given the packed attribute, but the packed
3308 attribute has no effect on the layout or size of the structure.
3309 Such structures may be mis-aligned for little benefit. For
3310 instance, in this code, the variable @code{f.x} in @code{struct bar}
3311 will be misaligned even though @code{struct bar} does not itself
3312 have the packed attribute:
3319 @} __attribute__((packed));
3329 Warn if padding is included in a structure, either to align an element
3330 of the structure or to align the whole structure. Sometimes when this
3331 happens it is possible to rearrange the fields of the structure to
3332 reduce the padding and so make the structure smaller.
3334 @item -Wredundant-decls
3335 @opindex Wredundant-decls
3336 Warn if anything is declared more than once in the same scope, even in
3337 cases where multiple declaration is valid and changes nothing.
3339 @item -Wnested-externs @r{(C only)}
3340 @opindex Wnested-externs
3341 Warn if an @code{extern} declaration is encountered within a function.
3343 @item -Wunreachable-code
3344 @opindex Wunreachable-code
3345 Warn if the compiler detects that code will never be executed.
3347 This option is intended to warn when the compiler detects that at
3348 least a whole line of source code will never be executed, because
3349 some condition is never satisfied or because it is after a
3350 procedure that never returns.
3352 It is possible for this option to produce a warning even though there
3353 are circumstances under which part of the affected line can be executed,
3354 so care should be taken when removing apparently-unreachable code.
3356 For instance, when a function is inlined, a warning may mean that the
3357 line is unreachable in only one inlined copy of the function.
3359 This option is not made part of @option{-Wall} because in a debugging
3360 version of a program there is often substantial code which checks
3361 correct functioning of the program and is, hopefully, unreachable
3362 because the program does work. Another common use of unreachable
3363 code is to provide behavior which is selectable at compile-time.
3367 Warn if a function can not be inlined and it was declared as inline.
3368 Even with this option, the compiler will not warn about failures to
3369 inline functions declared in system headers.
3371 The compiler uses a variety of heuristics to determine whether or not
3372 to inline a function. For example, the compiler takes into account
3373 the size of the function being inlined and the amount of inlining
3374 that has already been done in the current function. Therefore,
3375 seemingly insignificant changes in the source program can cause the
3376 warnings produced by @option{-Winline} to appear or disappear.
3378 @item -Wno-invalid-offsetof @r{(C++ only)}
3379 @opindex Wno-invalid-offsetof
3380 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3381 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3382 to a non-POD type is undefined. In existing C++ implementations,
3383 however, @samp{offsetof} typically gives meaningful results even when
3384 applied to certain kinds of non-POD types. (Such as a simple
3385 @samp{struct} that fails to be a POD type only by virtue of having a
3386 constructor.) This flag is for users who are aware that they are
3387 writing nonportable code and who have deliberately chosen to ignore the
3390 The restrictions on @samp{offsetof} may be relaxed in a future version
3391 of the C++ standard.
3393 @item -Wno-int-to-pointer-cast @r{(C only)}
3394 @opindex Wno-int-to-pointer-cast
3395 Suppress warnings from casts to pointer type of an integer of a
3398 @item -Wno-pointer-to-int-cast @r{(C only)}
3399 @opindex Wno-pointer-to-int-cast
3400 Suppress warnings from casts from a pointer to an integer type of a
3404 @opindex Winvalid-pch
3405 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3406 the search path but can't be used.
3410 @opindex Wno-long-long
3411 Warn if @samp{long long} type is used. This is default. To inhibit
3412 the warning messages, use @option{-Wno-long-long}. Flags
3413 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3414 only when @option{-pedantic} flag is used.
3416 @item -Wvariadic-macros
3417 @opindex Wvariadic-macros
3418 @opindex Wno-variadic-macros
3419 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3420 alternate syntax when in pedantic ISO C99 mode. This is default.
3421 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3423 @item -Wvolatile-register-var
3424 @opindex Wvolatile-register-var
3425 @opindex Wno-volatile-register-var
3426 Warn if a register variable is declared volatile. The volatile
3427 modifier does not inhibit all optimizations that may eliminate reads
3428 and/or writes to register variables.
3430 @item -Wdisabled-optimization
3431 @opindex Wdisabled-optimization
3432 Warn if a requested optimization pass is disabled. This warning does
3433 not generally indicate that there is anything wrong with your code; it
3434 merely indicates that GCC's optimizers were unable to handle the code
3435 effectively. Often, the problem is that your code is too big or too
3436 complex; GCC will refuse to optimize programs when the optimization
3437 itself is likely to take inordinate amounts of time.
3439 @item -Wpointer-sign
3440 @opindex Wpointer-sign
3441 @opindex Wno-pointer-sign
3442 Warn for pointer argument passing or assignment with different signedness.
3443 This option is only supported for C and Objective-C@. It is implied by
3444 @option{-Wall} and by @option{-pedantic}, which can be disabled with
3445 @option{-Wno-pointer-sign}.
3449 Make all warnings into errors.
3453 Make the specified warning into an errors. The specifier for a
3454 warning is appended, for example @option{-Werror=switch} turns the
3455 warnings controlled by @option{-Wswitch} into errors. This switch
3456 takes a negative form, to be used to negate @option{-Werror} for
3457 specific warnings, for example @option{-Wno-error=switch} makes
3458 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
3459 is in effect. You can use the @option{-fdiagnostics-show-option}
3460 option to have each controllable warning amended with the option which
3461 controls it, to determine what to use with this option.
3463 Note that specifying @option{-Werror=}@var{foo} automatically implies
3464 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
3467 @item -Wstack-protector
3468 @opindex Wstack-protector
3469 This option is only active when @option{-fstack-protector} is active. It
3470 warns about functions that will not be protected against stack smashing.
3472 @item -Wstring-literal-comparison
3473 @opindex Wstring-literal-comparison
3474 Warn about suspicious comparisons to string literal constants. In C,
3475 direct comparisons against the memory address of a string literal, such
3476 as @code{if (x == "abc")}, typically indicate a programmer error, and
3477 even when intentional, result in unspecified behavior and are not portable.
3478 Usually these warnings alert that the programmer intended to use
3479 @code{strcmp}. This warning is enabled by @option{-Wall}.
3481 @item -Woverlength-strings
3482 @opindex Woverlength-strings
3483 Warn about string constants which are longer than the ``minimum
3484 maximum'' length specified in the C standard. Modern compilers
3485 generally allow string constants which are much longer than the
3486 standard's minimum limit, but very portable programs should avoid
3487 using longer strings.
3489 The limit applies @emph{after} string constant concatenation, and does
3490 not count the trailing NUL@. In C89, the limit was 509 characters; in
3491 C99, it was raised to 4095. C++98 does not specify a normative
3492 minimum maximum, so we do not diagnose overlength strings in C++@.
3494 This option is implied by @option{-pedantic}, and can be disabled with
3495 @option{-Wno-overlength-strings}.
3498 @node Debugging Options
3499 @section Options for Debugging Your Program or GCC
3500 @cindex options, debugging
3501 @cindex debugging information options
3503 GCC has various special options that are used for debugging
3504 either your program or GCC:
3509 Produce debugging information in the operating system's native format
3510 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3513 On most systems that use stabs format, @option{-g} enables use of extra
3514 debugging information that only GDB can use; this extra information
3515 makes debugging work better in GDB but will probably make other debuggers
3517 refuse to read the program. If you want to control for certain whether
3518 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3519 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3521 GCC allows you to use @option{-g} with
3522 @option{-O}. The shortcuts taken by optimized code may occasionally
3523 produce surprising results: some variables you declared may not exist
3524 at all; flow of control may briefly move where you did not expect it;
3525 some statements may not be executed because they compute constant
3526 results or their values were already at hand; some statements may
3527 execute in different places because they were moved out of loops.
3529 Nevertheless it proves possible to debug optimized output. This makes
3530 it reasonable to use the optimizer for programs that might have bugs.
3532 The following options are useful when GCC is generated with the
3533 capability for more than one debugging format.
3537 Produce debugging information for use by GDB@. This means to use the
3538 most expressive format available (DWARF 2, stabs, or the native format
3539 if neither of those are supported), including GDB extensions if at all
3544 Produce debugging information in stabs format (if that is supported),
3545 without GDB extensions. This is the format used by DBX on most BSD
3546 systems. On MIPS, Alpha and System V Release 4 systems this option
3547 produces stabs debugging output which is not understood by DBX or SDB@.
3548 On System V Release 4 systems this option requires the GNU assembler.
3550 @item -feliminate-unused-debug-symbols
3551 @opindex feliminate-unused-debug-symbols
3552 Produce debugging information in stabs format (if that is supported),
3553 for only symbols that are actually used.
3555 @item -femit-class-debug-always
3556 Instead of emitting debugging information for a C++ class in only one
3557 object file, emit it in all object files using the class. This option
3558 should be used only with debuggers that are unable to handle the way GCC
3559 normally emits debugging information for classes because using this
3560 option will increase the size of debugging information by as much as a
3565 Produce debugging information in stabs format (if that is supported),
3566 using GNU extensions understood only by the GNU debugger (GDB)@. The
3567 use of these extensions is likely to make other debuggers crash or
3568 refuse to read the program.
3572 Produce debugging information in COFF format (if that is supported).
3573 This is the format used by SDB on most System V systems prior to
3578 Produce debugging information in XCOFF format (if that is supported).
3579 This is the format used by the DBX debugger on IBM RS/6000 systems.
3583 Produce debugging information in XCOFF format (if that is supported),
3584 using GNU extensions understood only by the GNU debugger (GDB)@. The
3585 use of these extensions is likely to make other debuggers crash or
3586 refuse to read the program, and may cause assemblers other than the GNU
3587 assembler (GAS) to fail with an error.
3591 Produce debugging information in DWARF version 2 format (if that is
3592 supported). This is the format used by DBX on IRIX 6. With this
3593 option, GCC uses features of DWARF version 3 when they are useful;
3594 version 3 is upward compatible with version 2, but may still cause
3595 problems for older debuggers.
3599 Produce debugging information in VMS debug format (if that is
3600 supported). This is the format used by DEBUG on VMS systems.
3603 @itemx -ggdb@var{level}
3604 @itemx -gstabs@var{level}
3605 @itemx -gcoff@var{level}
3606 @itemx -gxcoff@var{level}
3607 @itemx -gvms@var{level}
3608 Request debugging information and also use @var{level} to specify how
3609 much information. The default level is 2.
3611 Level 1 produces minimal information, enough for making backtraces in
3612 parts of the program that you don't plan to debug. This includes
3613 descriptions of functions and external variables, but no information
3614 about local variables and no line numbers.
3616 Level 3 includes extra information, such as all the macro definitions
3617 present in the program. Some debuggers support macro expansion when
3618 you use @option{-g3}.
3620 @option{-gdwarf-2} does not accept a concatenated debug level, because
3621 GCC used to support an option @option{-gdwarf} that meant to generate
3622 debug information in version 1 of the DWARF format (which is very
3623 different from version 2), and it would have been too confusing. That
3624 debug format is long obsolete, but the option cannot be changed now.
3625 Instead use an additional @option{-g@var{level}} option to change the
3626 debug level for DWARF2.
3628 @item -feliminate-dwarf2-dups
3629 @opindex feliminate-dwarf2-dups
3630 Compress DWARF2 debugging information by eliminating duplicated
3631 information about each symbol. This option only makes sense when
3632 generating DWARF2 debugging information with @option{-gdwarf-2}.
3634 @cindex @command{prof}
3637 Generate extra code to write profile information suitable for the
3638 analysis program @command{prof}. You must use this option when compiling
3639 the source files you want data about, and you must also use it when
3642 @cindex @command{gprof}
3645 Generate extra code to write profile information suitable for the
3646 analysis program @command{gprof}. You must use this option when compiling
3647 the source files you want data about, and you must also use it when
3652 Makes the compiler print out each function name as it is compiled, and
3653 print some statistics about each pass when it finishes.
3656 @opindex ftime-report
3657 Makes the compiler print some statistics about the time consumed by each
3658 pass when it finishes.
3661 @opindex fmem-report
3662 Makes the compiler print some statistics about permanent memory
3663 allocation when it finishes.
3665 @item -fprofile-arcs
3666 @opindex fprofile-arcs
3667 Add code so that program flow @dfn{arcs} are instrumented. During
3668 execution the program records how many times each branch and call is
3669 executed and how many times it is taken or returns. When the compiled
3670 program exits it saves this data to a file called
3671 @file{@var{auxname}.gcda} for each source file. The data may be used for
3672 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3673 test coverage analysis (@option{-ftest-coverage}). Each object file's
3674 @var{auxname} is generated from the name of the output file, if
3675 explicitly specified and it is not the final executable, otherwise it is
3676 the basename of the source file. In both cases any suffix is removed
3677 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3678 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3679 @xref{Cross-profiling}.
3681 @cindex @command{gcov}
3685 This option is used to compile and link code instrumented for coverage
3686 analysis. The option is a synonym for @option{-fprofile-arcs}
3687 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
3688 linking). See the documentation for those options for more details.
3693 Compile the source files with @option{-fprofile-arcs} plus optimization
3694 and code generation options. For test coverage analysis, use the
3695 additional @option{-ftest-coverage} option. You do not need to profile
3696 every source file in a program.
3699 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3700 (the latter implies the former).
3703 Run the program on a representative workload to generate the arc profile
3704 information. This may be repeated any number of times. You can run
3705 concurrent instances of your program, and provided that the file system
3706 supports locking, the data files will be correctly updated. Also
3707 @code{fork} calls are detected and correctly handled (double counting
3711 For profile-directed optimizations, compile the source files again with
3712 the same optimization and code generation options plus
3713 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3714 Control Optimization}).
3717 For test coverage analysis, use @command{gcov} to produce human readable
3718 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3719 @command{gcov} documentation for further information.
3723 With @option{-fprofile-arcs}, for each function of your program GCC
3724 creates a program flow graph, then finds a spanning tree for the graph.
3725 Only arcs that are not on the spanning tree have to be instrumented: the
3726 compiler adds code to count the number of times that these arcs are
3727 executed. When an arc is the only exit or only entrance to a block, the
3728 instrumentation code can be added to the block; otherwise, a new basic
3729 block must be created to hold the instrumentation code.
3732 @item -ftest-coverage
3733 @opindex ftest-coverage
3734 Produce a notes file that the @command{gcov} code-coverage utility
3735 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3736 show program coverage. Each source file's note file is called
3737 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3738 above for a description of @var{auxname} and instructions on how to
3739 generate test coverage data. Coverage data will match the source files
3740 more closely, if you do not optimize.
3742 @item -d@var{letters}
3743 @item -fdump-rtl-@var{pass}
3745 Says to make debugging dumps during compilation at times specified by
3746 @var{letters}. This is used for debugging the RTL-based passes of the
3747 compiler. The file names for most of the dumps are made by appending a
3748 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3749 from the name of the output file, if explicitly specified and it is not
3750 an executable, otherwise it is the basename of the source file.
3752 Most debug dumps can be enabled either passing a letter to the @option{-d}
3753 option, or with a long @option{-fdump-rtl} switch; here are the possible
3754 letters for use in @var{letters} and @var{pass}, and their meanings:
3759 Annotate the assembler output with miscellaneous debugging information.
3762 @itemx -fdump-rtl-bbro
3764 @opindex fdump-rtl-bbro
3765 Dump after block reordering, to @file{@var{file}.148r.bbro}.
3768 @itemx -fdump-rtl-combine
3770 @opindex fdump-rtl-combine
3771 Dump after instruction combination, to the file @file{@var{file}.129r.combine}.
3774 @itemx -fdump-rtl-ce1
3775 @itemx -fdump-rtl-ce2
3777 @opindex fdump-rtl-ce1
3778 @opindex fdump-rtl-ce2
3779 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3780 first if conversion, to the file @file{@var{file}.117r.ce1}. @option{-dC}
3781 and @option{-fdump-rtl-ce2} enable dumping after the second if
3782 conversion, to the file @file{@var{file}.130r.ce2}.
3785 @itemx -fdump-rtl-btl
3786 @itemx -fdump-rtl-dbr
3788 @opindex fdump-rtl-btl
3789 @opindex fdump-rtl-dbr
3790 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3791 target load optimization, to @file{@var{file}.31.btl}. @option{-dd}
3792 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3793 scheduling, to @file{@var{file}.36.dbr}.
3797 Dump all macro definitions, at the end of preprocessing, in addition to
3801 @itemx -fdump-rtl-ce3
3803 @opindex fdump-rtl-ce3
3804 Dump after the third if conversion, to @file{@var{file}.146r.ce3}.
3807 @itemx -fdump-rtl-cfg
3808 @itemx -fdump-rtl-life
3810 @opindex fdump-rtl-cfg
3811 @opindex fdump-rtl-life
3812 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3813 and data flow analysis, to @file{@var{file}.116r.cfg}. @option{-df}
3814 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3815 to @file{@var{file}.128r.life1} and @file{@var{file}.135r.life2}.
3818 @itemx -fdump-rtl-greg
3820 @opindex fdump-rtl-greg
3821 Dump after global register allocation, to @file{@var{file}.139r.greg}.
3824 @itemx -fdump-rtl-gcse
3825 @itemx -fdump-rtl-bypass
3827 @opindex fdump-rtl-gcse
3828 @opindex fdump-rtl-bypass
3829 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3830 @file{@var{file}.114r.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3831 enable dumping after jump bypassing and control flow optimizations, to
3832 @file{@var{file}.115r.bypass}.
3835 @itemx -fdump-rtl-eh
3837 @opindex fdump-rtl-eh
3838 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3841 @itemx -fdump-rtl-sibling
3843 @opindex fdump-rtl-sibling
3844 Dump after sibling call optimizations, to @file{@var{file}.106r.sibling}.
3847 @itemx -fdump-rtl-jump
3849 @opindex fdump-rtl-jump
3850 Dump after the first jump optimization, to @file{@var{file}.112r.jump}.
3853 @itemx -fdump-rtl-stack
3855 @opindex fdump-rtl-stack
3856 Dump after conversion from registers to stack, to @file{@var{file}.152r.stack}.
3859 @itemx -fdump-rtl-lreg
3861 @opindex fdump-rtl-lreg
3862 Dump after local register allocation, to @file{@var{file}.138r.lreg}.
3865 @itemx -fdump-rtl-loop2
3867 @opindex fdump-rtl-loop2
3868 @option{-dL} and @option{-fdump-rtl-loop2} enable dumping after the
3869 loop optimization pass, to @file{@var{file}.119r.loop2},
3870 @file{@var{file}.120r.loop2_init},
3871 @file{@var{file}.121r.loop2_invariant}, and
3872 @file{@var{file}.125r.loop2_done}.
3875 @itemx -fdump-rtl-sms
3877 @opindex fdump-rtl-sms
3878 Dump after modulo scheduling, to @file{@var{file}.136r.sms}.
3881 @itemx -fdump-rtl-mach
3883 @opindex fdump-rtl-mach
3884 Dump after performing the machine dependent reorganization pass, to
3885 @file{@var{file}.155r.mach}.
3888 @itemx -fdump-rtl-rnreg
3890 @opindex fdump-rtl-rnreg
3891 Dump after register renumbering, to @file{@var{file}.147r.rnreg}.
3894 @itemx -fdump-rtl-regmove
3896 @opindex fdump-rtl-regmove
3897 Dump after the register move pass, to @file{@var{file}.132r.regmove}.
3900 @itemx -fdump-rtl-postreload
3902 @opindex fdump-rtl-postreload
3903 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3906 @itemx -fdump-rtl-expand
3908 @opindex fdump-rtl-expand
3909 Dump after RTL generation, to @file{@var{file}.104r.expand}.
3912 @itemx -fdump-rtl-sched2
3914 @opindex fdump-rtl-sched2
3915 Dump after the second scheduling pass, to @file{@var{file}.150r.sched2}.
3918 @itemx -fdump-rtl-cse
3920 @opindex fdump-rtl-cse
3921 Dump after CSE (including the jump optimization that sometimes follows
3922 CSE), to @file{@var{file}.113r.cse}.
3925 @itemx -fdump-rtl-sched
3927 @opindex fdump-rtl-sched
3928 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3931 @itemx -fdump-rtl-cse2
3933 @opindex fdump-rtl-cse2
3934 Dump after the second CSE pass (including the jump optimization that
3935 sometimes follows CSE), to @file{@var{file}.127r.cse2}.
3938 @itemx -fdump-rtl-tracer
3940 @opindex fdump-rtl-tracer
3941 Dump after running tracer, to @file{@var{file}.118r.tracer}.
3944 @itemx -fdump-rtl-vpt
3945 @itemx -fdump-rtl-vartrack
3947 @opindex fdump-rtl-vpt
3948 @opindex fdump-rtl-vartrack
3949 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3950 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3951 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3952 to @file{@var{file}.154r.vartrack}.
3955 @itemx -fdump-rtl-flow2
3957 @opindex fdump-rtl-flow2
3958 Dump after the second flow pass, to @file{@var{file}.142r.flow2}.
3961 @itemx -fdump-rtl-peephole2
3963 @opindex fdump-rtl-peephole2
3964 Dump after the peephole pass, to @file{@var{file}.145r.peephole2}.
3967 @itemx -fdump-rtl-web
3969 @opindex fdump-rtl-web
3970 Dump after live range splitting, to @file{@var{file}.126r.web}.
3973 @itemx -fdump-rtl-all
3975 @opindex fdump-rtl-all
3976 Produce all the dumps listed above.
3980 Produce a core dump whenever an error occurs.
3984 Print statistics on memory usage, at the end of the run, to
3989 Annotate the assembler output with a comment indicating which
3990 pattern and alternative was used. The length of each instruction is
3995 Dump the RTL in the assembler output as a comment before each instruction.
3996 Also turns on @option{-dp} annotation.
4000 For each of the other indicated dump files (either with @option{-d} or
4001 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
4002 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
4006 Just generate RTL for a function instead of compiling it. Usually used
4007 with @samp{r} (@option{-fdump-rtl-expand}).
4011 Dump debugging information during parsing, to standard error.
4015 @opindex fdump-noaddr
4016 When doing debugging dumps (see @option{-d} option above), suppress
4017 address output. This makes it more feasible to use diff on debugging
4018 dumps for compiler invocations with different compiler binaries and/or
4019 different text / bss / data / heap / stack / dso start locations.
4021 @item -fdump-unnumbered
4022 @opindex fdump-unnumbered
4023 When doing debugging dumps (see @option{-d} option above), suppress instruction
4024 numbers, line number note and address output. This makes it more feasible to
4025 use diff on debugging dumps for compiler invocations with different
4026 options, in particular with and without @option{-g}.
4028 @item -fdump-translation-unit @r{(C++ only)}
4029 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
4030 @opindex fdump-translation-unit
4031 Dump a representation of the tree structure for the entire translation
4032 unit to a file. The file name is made by appending @file{.tu} to the
4033 source file name. If the @samp{-@var{options}} form is used, @var{options}
4034 controls the details of the dump as described for the
4035 @option{-fdump-tree} options.
4037 @item -fdump-class-hierarchy @r{(C++ only)}
4038 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
4039 @opindex fdump-class-hierarchy
4040 Dump a representation of each class's hierarchy and virtual function
4041 table layout to a file. The file name is made by appending @file{.class}
4042 to the source file name. If the @samp{-@var{options}} form is used,
4043 @var{options} controls the details of the dump as described for the
4044 @option{-fdump-tree} options.
4046 @item -fdump-ipa-@var{switch}
4048 Control the dumping at various stages of inter-procedural analysis
4049 language tree to a file. The file name is generated by appending a switch
4050 specific suffix to the source file name. The following dumps are possible:
4054 Enables all inter-procedural analysis dumps; currently the only produced
4055 dump is the @samp{cgraph} dump.
4058 Dumps information about call-graph optimization, unused function removal,
4059 and inlining decisions.
4062 @item -fdump-tree-@var{switch}
4063 @itemx -fdump-tree-@var{switch}-@var{options}
4065 Control the dumping at various stages of processing the intermediate
4066 language tree to a file. The file name is generated by appending a switch
4067 specific suffix to the source file name. If the @samp{-@var{options}}
4068 form is used, @var{options} is a list of @samp{-} separated options that
4069 control the details of the dump. Not all options are applicable to all
4070 dumps, those which are not meaningful will be ignored. The following
4071 options are available
4075 Print the address of each node. Usually this is not meaningful as it
4076 changes according to the environment and source file. Its primary use
4077 is for tying up a dump file with a debug environment.
4079 Inhibit dumping of members of a scope or body of a function merely
4080 because that scope has been reached. Only dump such items when they
4081 are directly reachable by some other path. When dumping pretty-printed
4082 trees, this option inhibits dumping the bodies of control structures.
4084 Print a raw representation of the tree. By default, trees are
4085 pretty-printed into a C-like representation.
4087 Enable more detailed dumps (not honored by every dump option).
4089 Enable dumping various statistics about the pass (not honored by every dump
4092 Enable showing basic block boundaries (disabled in raw dumps).
4094 Enable showing virtual operands for every statement.
4096 Enable showing line numbers for statements.
4098 Enable showing the unique ID (@code{DECL_UID}) for each variable.
4100 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
4103 The following tree dumps are possible:
4107 Dump before any tree based optimization, to @file{@var{file}.original}.
4110 Dump after all tree based optimization, to @file{@var{file}.optimized}.
4113 Dump after function inlining, to @file{@var{file}.inlined}.
4116 @opindex fdump-tree-gimple
4117 Dump each function before and after the gimplification pass to a file. The
4118 file name is made by appending @file{.gimple} to the source file name.
4121 @opindex fdump-tree-cfg
4122 Dump the control flow graph of each function to a file. The file name is
4123 made by appending @file{.cfg} to the source file name.
4126 @opindex fdump-tree-vcg
4127 Dump the control flow graph of each function to a file in VCG format. The
4128 file name is made by appending @file{.vcg} to the source file name. Note
4129 that if the file contains more than one function, the generated file cannot
4130 be used directly by VCG@. You will need to cut and paste each function's
4131 graph into its own separate file first.
4134 @opindex fdump-tree-ch
4135 Dump each function after copying loop headers. The file name is made by
4136 appending @file{.ch} to the source file name.
4139 @opindex fdump-tree-ssa
4140 Dump SSA related information to a file. The file name is made by appending
4141 @file{.ssa} to the source file name.
4144 @opindex fdump-tree-salias
4145 Dump structure aliasing variable information to a file. This file name
4146 is made by appending @file{.salias} to the source file name.
4149 @opindex fdump-tree-alias
4150 Dump aliasing information for each function. The file name is made by
4151 appending @file{.alias} to the source file name.
4154 @opindex fdump-tree-ccp
4155 Dump each function after CCP@. The file name is made by appending
4156 @file{.ccp} to the source file name.
4159 @opindex fdump-tree-storeccp
4160 Dump each function after STORE-CCP. The file name is made by appending
4161 @file{.storeccp} to the source file name.
4164 @opindex fdump-tree-pre
4165 Dump trees after partial redundancy elimination. The file name is made
4166 by appending @file{.pre} to the source file name.
4169 @opindex fdump-tree-fre
4170 Dump trees after full redundancy elimination. The file name is made
4171 by appending @file{.fre} to the source file name.
4174 @opindex fdump-tree-copyprop
4175 Dump trees after copy propagation. The file name is made
4176 by appending @file{.copyprop} to the source file name.
4178 @item store_copyprop
4179 @opindex fdump-tree-store_copyprop
4180 Dump trees after store copy-propagation. The file name is made
4181 by appending @file{.store_copyprop} to the source file name.
4184 @opindex fdump-tree-dce
4185 Dump each function after dead code elimination. The file name is made by
4186 appending @file{.dce} to the source file name.
4189 @opindex fdump-tree-mudflap
4190 Dump each function after adding mudflap instrumentation. The file name is
4191 made by appending @file{.mudflap} to the source file name.
4194 @opindex fdump-tree-sra
4195 Dump each function after performing scalar replacement of aggregates. The
4196 file name is made by appending @file{.sra} to the source file name.
4199 @opindex fdump-tree-sink
4200 Dump each function after performing code sinking. The file name is made
4201 by appending @file{.sink} to the source file name.
4204 @opindex fdump-tree-dom
4205 Dump each function after applying dominator tree optimizations. The file
4206 name is made by appending @file{.dom} to the source file name.
4209 @opindex fdump-tree-dse
4210 Dump each function after applying dead store elimination. The file
4211 name is made by appending @file{.dse} to the source file name.
4214 @opindex fdump-tree-phiopt
4215 Dump each function after optimizing PHI nodes into straightline code. The file
4216 name is made by appending @file{.phiopt} to the source file name.
4219 @opindex fdump-tree-forwprop
4220 Dump each function after forward propagating single use variables. The file
4221 name is made by appending @file{.forwprop} to the source file name.
4224 @opindex fdump-tree-copyrename
4225 Dump each function after applying the copy rename optimization. The file
4226 name is made by appending @file{.copyrename} to the source file name.
4229 @opindex fdump-tree-nrv
4230 Dump each function after applying the named return value optimization on
4231 generic trees. The file name is made by appending @file{.nrv} to the source
4235 @opindex fdump-tree-vect
4236 Dump each function after applying vectorization of loops. The file name is
4237 made by appending @file{.vect} to the source file name.
4240 @opindex fdump-tree-vrp
4241 Dump each function after Value Range Propagation (VRP). The file name
4242 is made by appending @file{.vrp} to the source file name.
4245 @opindex fdump-tree-all
4246 Enable all the available tree dumps with the flags provided in this option.
4249 @item -ftree-vectorizer-verbose=@var{n}
4250 @opindex ftree-vectorizer-verbose
4251 This option controls the amount of debugging output the vectorizer prints.
4252 This information is written to standard error, unless
4253 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
4254 in which case it is output to the usual dump listing file, @file{.vect}.
4255 For @var{n}=0 no diagnostic information is reported.
4256 If @var{n}=1 the vectorizer reports each loop that got vectorized,
4257 and the total number of loops that got vectorized.
4258 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
4259 the first analysis phase (vect_analyze_loop_form) - i.e. countable,
4260 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
4261 level that @option{-fdump-tree-vect-stats} uses.
4262 Higher verbosity levels mean either more information dumped for each
4263 reported loop, or same amount of information reported for more loops:
4264 If @var{n}=3, alignment related information is added to the reports.
4265 If @var{n}=4, data-references related information (e.g. memory dependences,
4266 memory access-patterns) is added to the reports.
4267 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
4268 that did not pass the first analysis phase (i.e. may not be countable, or
4269 may have complicated control-flow).
4270 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
4271 For @var{n}=7, all the information the vectorizer generates during its
4272 analysis and transformation is reported. This is the same verbosity level
4273 that @option{-fdump-tree-vect-details} uses.
4275 @item -frandom-seed=@var{string}
4276 @opindex frandom-string
4277 This option provides a seed that GCC uses when it would otherwise use
4278 random numbers. It is used to generate certain symbol names
4279 that have to be different in every compiled file. It is also used to
4280 place unique stamps in coverage data files and the object files that
4281 produce them. You can use the @option{-frandom-seed} option to produce
4282 reproducibly identical object files.
4284 The @var{string} should be different for every file you compile.
4286 @item -fsched-verbose=@var{n}
4287 @opindex fsched-verbose
4288 On targets that use instruction scheduling, this option controls the
4289 amount of debugging output the scheduler prints. This information is
4290 written to standard error, unless @option{-dS} or @option{-dR} is
4291 specified, in which case it is output to the usual dump
4292 listing file, @file{.sched} or @file{.sched2} respectively. However
4293 for @var{n} greater than nine, the output is always printed to standard
4296 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
4297 same information as @option{-dRS}. For @var{n} greater than one, it
4298 also output basic block probabilities, detailed ready list information
4299 and unit/insn info. For @var{n} greater than two, it includes RTL
4300 at abort point, control-flow and regions info. And for @var{n} over
4301 four, @option{-fsched-verbose} also includes dependence info.
4305 Store the usual ``temporary'' intermediate files permanently; place them
4306 in the current directory and name them based on the source file. Thus,
4307 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
4308 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
4309 preprocessed @file{foo.i} output file even though the compiler now
4310 normally uses an integrated preprocessor.
4312 When used in combination with the @option{-x} command line option,
4313 @option{-save-temps} is sensible enough to avoid over writing an
4314 input source file with the same extension as an intermediate file.
4315 The corresponding intermediate file may be obtained by renaming the
4316 source file before using @option{-save-temps}.
4320 Report the CPU time taken by each subprocess in the compilation
4321 sequence. For C source files, this is the compiler proper and assembler
4322 (plus the linker if linking is done). The output looks like this:
4329 The first number on each line is the ``user time'', that is time spent
4330 executing the program itself. The second number is ``system time'',
4331 time spent executing operating system routines on behalf of the program.
4332 Both numbers are in seconds.
4334 @item -fvar-tracking
4335 @opindex fvar-tracking
4336 Run variable tracking pass. It computes where variables are stored at each
4337 position in code. Better debugging information is then generated
4338 (if the debugging information format supports this information).
4340 It is enabled by default when compiling with optimization (@option{-Os},
4341 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
4342 the debug info format supports it.
4344 @item -print-file-name=@var{library}
4345 @opindex print-file-name
4346 Print the full absolute name of the library file @var{library} that
4347 would be used when linking---and don't do anything else. With this
4348 option, GCC does not compile or link anything; it just prints the
4351 @item -print-multi-directory
4352 @opindex print-multi-directory
4353 Print the directory name corresponding to the multilib selected by any
4354 other switches present in the command line. This directory is supposed
4355 to exist in @env{GCC_EXEC_PREFIX}.
4357 @item -print-multi-lib
4358 @opindex print-multi-lib
4359 Print the mapping from multilib directory names to compiler switches
4360 that enable them. The directory name is separated from the switches by
4361 @samp{;}, and each switch starts with an @samp{@@} instead of the
4362 @samp{-}, without spaces between multiple switches. This is supposed to
4363 ease shell-processing.
4365 @item -print-prog-name=@var{program}
4366 @opindex print-prog-name
4367 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
4369 @item -print-libgcc-file-name
4370 @opindex print-libgcc-file-name
4371 Same as @option{-print-file-name=libgcc.a}.
4373 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
4374 but you do want to link with @file{libgcc.a}. You can do
4377 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
4380 @item -print-search-dirs
4381 @opindex print-search-dirs
4382 Print the name of the configured installation directory and a list of
4383 program and library directories @command{gcc} will search---and don't do anything else.
4385 This is useful when @command{gcc} prints the error message
4386 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4387 To resolve this you either need to put @file{cpp0} and the other compiler
4388 components where @command{gcc} expects to find them, or you can set the environment
4389 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4390 Don't forget the trailing @samp{/}.
4391 @xref{Environment Variables}.
4394 @opindex dumpmachine
4395 Print the compiler's target machine (for example,
4396 @samp{i686-pc-linux-gnu})---and don't do anything else.
4399 @opindex dumpversion
4400 Print the compiler version (for example, @samp{3.0})---and don't do
4405 Print the compiler's built-in specs---and don't do anything else. (This
4406 is used when GCC itself is being built.) @xref{Spec Files}.
4408 @item -feliminate-unused-debug-types
4409 @opindex feliminate-unused-debug-types
4410 Normally, when producing DWARF2 output, GCC will emit debugging
4411 information for all types declared in a compilation
4412 unit, regardless of whether or not they are actually used
4413 in that compilation unit. Sometimes this is useful, such as
4414 if, in the debugger, you want to cast a value to a type that is
4415 not actually used in your program (but is declared). More often,
4416 however, this results in a significant amount of wasted space.
4417 With this option, GCC will avoid producing debug symbol output
4418 for types that are nowhere used in the source file being compiled.
4421 @node Optimize Options
4422 @section Options That Control Optimization
4423 @cindex optimize options
4424 @cindex options, optimization
4426 These options control various sorts of optimizations.
4428 Without any optimization option, the compiler's goal is to reduce the
4429 cost of compilation and to make debugging produce the expected
4430 results. Statements are independent: if you stop the program with a
4431 breakpoint between statements, you can then assign a new value to any
4432 variable or change the program counter to any other statement in the
4433 function and get exactly the results you would expect from the source
4436 Turning on optimization flags makes the compiler attempt to improve
4437 the performance and/or code size at the expense of compilation time
4438 and possibly the ability to debug the program.
4440 The compiler performs optimization based on the knowledge it has of
4441 the program. Optimization levels @option{-O} and above, in
4442 particular, enable @emph{unit-at-a-time} mode, which allows the
4443 compiler to consider information gained from later functions in
4444 the file when compiling a function. Compiling multiple files at
4445 once to a single output file in @emph{unit-at-a-time} mode allows
4446 the compiler to use information gained from all of the files when
4447 compiling each of them.
4449 Not all optimizations are controlled directly by a flag. Only
4450 optimizations that have a flag are listed.
4457 Optimize. Optimizing compilation takes somewhat more time, and a lot
4458 more memory for a large function.
4460 With @option{-O}, the compiler tries to reduce code size and execution
4461 time, without performing any optimizations that take a great deal of
4464 @option{-O} turns on the following optimization flags:
4465 @gccoptlist{-fdefer-pop @gol
4466 -fdelayed-branch @gol
4467 -fguess-branch-probability @gol
4468 -fcprop-registers @gol
4469 -fif-conversion @gol
4470 -fif-conversion2 @gol
4473 -ftree-dominator-opts @gol
4478 -ftree-copyrename @gol
4481 -funit-at-a-time @gol
4484 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4485 where doing so does not interfere with debugging.
4489 Optimize even more. GCC performs nearly all supported optimizations
4490 that do not involve a space-speed tradeoff. The compiler does not
4491 perform loop unrolling or function inlining when you specify @option{-O2}.
4492 As compared to @option{-O}, this option increases both compilation time
4493 and the performance of the generated code.
4495 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4496 also turns on the following optimization flags:
4497 @gccoptlist{-fthread-jumps @gol
4499 -foptimize-sibling-calls @gol
4500 -fcse-follow-jumps -fcse-skip-blocks @gol
4501 -fgcse -fgcse-lm @gol
4502 -fexpensive-optimizations @gol
4503 -frerun-cse-after-loop @gol
4506 -fschedule-insns -fschedule-insns2 @gol
4507 -fsched-interblock -fsched-spec @gol
4509 -fstrict-aliasing @gol
4510 -fdelete-null-pointer-checks @gol
4511 -freorder-blocks -freorder-functions @gol
4512 -falign-functions -falign-jumps @gol
4513 -falign-loops -falign-labels @gol
4517 Please note the warning under @option{-fgcse} about
4518 invoking @option{-O2} on programs that use computed gotos.
4522 Optimize yet more. @option{-O3} turns on all optimizations specified by
4523 @option{-O2} and also turns on the @option{-finline-functions},
4524 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4528 Do not optimize. This is the default.
4532 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4533 do not typically increase code size. It also performs further
4534 optimizations designed to reduce code size.
4536 @option{-Os} disables the following optimization flags:
4537 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4538 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
4539 -fprefetch-loop-arrays -ftree-vect-loop-version}
4541 If you use multiple @option{-O} options, with or without level numbers,
4542 the last such option is the one that is effective.
4545 Options of the form @option{-f@var{flag}} specify machine-independent
4546 flags. Most flags have both positive and negative forms; the negative
4547 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4548 below, only one of the forms is listed---the one you typically will
4549 use. You can figure out the other form by either removing @samp{no-}
4552 The following options control specific optimizations. They are either
4553 activated by @option{-O} options or are related to ones that are. You
4554 can use the following flags in the rare cases when ``fine-tuning'' of
4555 optimizations to be performed is desired.
4558 @item -fno-default-inline
4559 @opindex fno-default-inline
4560 Do not make member functions inline by default merely because they are
4561 defined inside the class scope (C++ only). Otherwise, when you specify
4562 @w{@option{-O}}, member functions defined inside class scope are compiled
4563 inline by default; i.e., you don't need to add @samp{inline} in front of
4564 the member function name.
4566 @item -fno-defer-pop
4567 @opindex fno-defer-pop
4568 Always pop the arguments to each function call as soon as that function
4569 returns. For machines which must pop arguments after a function call,
4570 the compiler normally lets arguments accumulate on the stack for several
4571 function calls and pops them all at once.
4573 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4577 Force memory operands to be copied into registers before doing
4578 arithmetic on them. This produces better code by making all memory
4579 references potential common subexpressions. When they are not common
4580 subexpressions, instruction combination should eliminate the separate
4581 register-load. This option is now a nop and will be removed in 4.2.
4584 @opindex fforce-addr
4585 Force memory address constants to be copied into registers before
4586 doing arithmetic on them.
4588 @item -fomit-frame-pointer
4589 @opindex fomit-frame-pointer
4590 Don't keep the frame pointer in a register for functions that
4591 don't need one. This avoids the instructions to save, set up and
4592 restore frame pointers; it also makes an extra register available
4593 in many functions. @strong{It also makes debugging impossible on
4596 On some machines, such as the VAX, this flag has no effect, because
4597 the standard calling sequence automatically handles the frame pointer
4598 and nothing is saved by pretending it doesn't exist. The
4599 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4600 whether a target machine supports this flag. @xref{Registers,,Register
4601 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4603 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4605 @item -foptimize-sibling-calls
4606 @opindex foptimize-sibling-calls
4607 Optimize sibling and tail recursive calls.
4609 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4613 Don't pay attention to the @code{inline} keyword. Normally this option
4614 is used to keep the compiler from expanding any functions inline.
4615 Note that if you are not optimizing, no functions can be expanded inline.
4617 @item -finline-functions
4618 @opindex finline-functions
4619 Integrate all simple functions into their callers. The compiler
4620 heuristically decides which functions are simple enough to be worth
4621 integrating in this way.
4623 If all calls to a given function are integrated, and the function is
4624 declared @code{static}, then the function is normally not output as
4625 assembler code in its own right.
4627 Enabled at level @option{-O3}.
4629 @item -finline-functions-called-once
4630 @opindex finline-functions-called-once
4631 Consider all @code{static} functions called once for inlining into their
4632 caller even if they are not marked @code{inline}. If a call to a given
4633 function is integrated, then the function is not output as assembler code
4636 Enabled if @option{-funit-at-a-time} is enabled.
4638 @item -fearly-inlining
4639 @opindex fearly-inlining
4640 Inline functions marked by @code{always_inline} and functions whose body seems
4641 smaller than the function call overhead early before doing
4642 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
4643 makes profiling significantly cheaper and usually inlining faster on programs
4644 having large chains of nested wrapper functions.
4648 @item -finline-limit=@var{n}
4649 @opindex finline-limit
4650 By default, GCC limits the size of functions that can be inlined. This flag
4651 allows the control of this limit for functions that are explicitly marked as
4652 inline (i.e., marked with the inline keyword or defined within the class
4653 definition in c++). @var{n} is the size of functions that can be inlined in
4654 number of pseudo instructions (not counting parameter handling). The default
4655 value of @var{n} is 600.
4656 Increasing this value can result in more inlined code at
4657 the cost of compilation time and memory consumption. Decreasing usually makes
4658 the compilation faster and less code will be inlined (which presumably
4659 means slower programs). This option is particularly useful for programs that
4660 use inlining heavily such as those based on recursive templates with C++.
4662 Inlining is actually controlled by a number of parameters, which may be
4663 specified individually by using @option{--param @var{name}=@var{value}}.
4664 The @option{-finline-limit=@var{n}} option sets some of these parameters
4668 @item max-inline-insns-single
4669 is set to @var{n}/2.
4670 @item max-inline-insns-auto
4671 is set to @var{n}/2.
4672 @item min-inline-insns
4673 is set to 130 or @var{n}/4, whichever is smaller.
4674 @item max-inline-insns-rtl
4678 See below for a documentation of the individual
4679 parameters controlling inlining.
4681 @emph{Note:} pseudo instruction represents, in this particular context, an
4682 abstract measurement of function's size. In no way does it represent a count
4683 of assembly instructions and as such its exact meaning might change from one
4684 release to an another.
4686 @item -fkeep-inline-functions
4687 @opindex fkeep-inline-functions
4688 In C, emit @code{static} functions that are declared @code{inline}
4689 into the object file, even if the function has been inlined into all
4690 of its callers. This switch does not affect functions using the
4691 @code{extern inline} extension in GNU C@. In C++, emit any and all
4692 inline functions into the object file.
4694 @item -fkeep-static-consts
4695 @opindex fkeep-static-consts
4696 Emit variables declared @code{static const} when optimization isn't turned
4697 on, even if the variables aren't referenced.
4699 GCC enables this option by default. If you want to force the compiler to
4700 check if the variable was referenced, regardless of whether or not
4701 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4703 @item -fmerge-constants
4704 Attempt to merge identical constants (string constants and floating point
4705 constants) across compilation units.
4707 This option is the default for optimized compilation if the assembler and
4708 linker support it. Use @option{-fno-merge-constants} to inhibit this
4711 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4713 @item -fmerge-all-constants
4714 Attempt to merge identical constants and identical variables.
4716 This option implies @option{-fmerge-constants}. In addition to
4717 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4718 arrays or initialized constant variables with integral or floating point
4719 types. Languages like C or C++ require each non-automatic variable to
4720 have distinct location, so using this option will result in non-conforming
4723 @item -fmodulo-sched
4724 @opindex fmodulo-sched
4725 Perform swing modulo scheduling immediately before the first scheduling
4726 pass. This pass looks at innermost loops and reorders their
4727 instructions by overlapping different iterations.
4729 @item -fno-branch-count-reg
4730 @opindex fno-branch-count-reg
4731 Do not use ``decrement and branch'' instructions on a count register,
4732 but instead generate a sequence of instructions that decrement a
4733 register, compare it against zero, then branch based upon the result.
4734 This option is only meaningful on architectures that support such
4735 instructions, which include x86, PowerPC, IA-64 and S/390.
4737 The default is @option{-fbranch-count-reg}.
4739 @item -fno-function-cse
4740 @opindex fno-function-cse
4741 Do not put function addresses in registers; make each instruction that
4742 calls a constant function contain the function's address explicitly.
4744 This option results in less efficient code, but some strange hacks
4745 that alter the assembler output may be confused by the optimizations
4746 performed when this option is not used.
4748 The default is @option{-ffunction-cse}
4750 @item -fno-zero-initialized-in-bss
4751 @opindex fno-zero-initialized-in-bss
4752 If the target supports a BSS section, GCC by default puts variables that
4753 are initialized to zero into BSS@. This can save space in the resulting
4756 This option turns off this behavior because some programs explicitly
4757 rely on variables going to the data section. E.g., so that the
4758 resulting executable can find the beginning of that section and/or make
4759 assumptions based on that.
4761 The default is @option{-fzero-initialized-in-bss}.
4763 @item -fbounds-check
4764 @opindex fbounds-check
4765 For front-ends that support it, generate additional code to check that
4766 indices used to access arrays are within the declared range. This is
4767 currently only supported by the Java and Fortran front-ends, where
4768 this option defaults to true and false respectively.
4770 @item -fmudflap -fmudflapth -fmudflapir
4774 @cindex bounds checking
4776 For front-ends that support it (C and C++), instrument all risky
4777 pointer/array dereferencing operations, some standard library
4778 string/heap functions, and some other associated constructs with
4779 range/validity tests. Modules so instrumented should be immune to
4780 buffer overflows, invalid heap use, and some other classes of C/C++
4781 programming errors. The instrumentation relies on a separate runtime
4782 library (@file{libmudflap}), which will be linked into a program if
4783 @option{-fmudflap} is given at link time. Run-time behavior of the
4784 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4785 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4788 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4789 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4790 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4791 instrumentation should ignore pointer reads. This produces less
4792 instrumentation (and therefore faster execution) and still provides
4793 some protection against outright memory corrupting writes, but allows
4794 erroneously read data to propagate within a program.
4796 @item -fthread-jumps
4797 @opindex fthread-jumps
4798 Perform optimizations where we check to see if a jump branches to a
4799 location where another comparison subsumed by the first is found. If
4800 so, the first branch is redirected to either the destination of the
4801 second branch or a point immediately following it, depending on whether
4802 the condition is known to be true or false.
4804 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4806 @item -fcse-follow-jumps
4807 @opindex fcse-follow-jumps
4808 In common subexpression elimination, scan through jump instructions
4809 when the target of the jump is not reached by any other path. For
4810 example, when CSE encounters an @code{if} statement with an
4811 @code{else} clause, CSE will follow the jump when the condition
4814 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4816 @item -fcse-skip-blocks
4817 @opindex fcse-skip-blocks
4818 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4819 follow jumps which conditionally skip over blocks. When CSE
4820 encounters a simple @code{if} statement with no else clause,
4821 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4822 body of the @code{if}.
4824 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4826 @item -frerun-cse-after-loop
4827 @opindex frerun-cse-after-loop
4828 Re-run common subexpression elimination after loop optimizations has been
4831 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4835 Perform a global common subexpression elimination pass.
4836 This pass also performs global constant and copy propagation.
4838 @emph{Note:} When compiling a program using computed gotos, a GCC
4839 extension, you may get better runtime performance if you disable
4840 the global common subexpression elimination pass by adding
4841 @option{-fno-gcse} to the command line.
4843 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4847 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4848 attempt to move loads which are only killed by stores into themselves. This
4849 allows a loop containing a load/store sequence to be changed to a load outside
4850 the loop, and a copy/store within the loop.
4852 Enabled by default when gcse is enabled.
4856 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4857 global common subexpression elimination. This pass will attempt to move
4858 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4859 loops containing a load/store sequence can be changed to a load before
4860 the loop and a store after the loop.
4862 Not enabled at any optimization level.
4866 When @option{-fgcse-las} is enabled, the global common subexpression
4867 elimination pass eliminates redundant loads that come after stores to the
4868 same memory location (both partial and full redundancies).
4870 Not enabled at any optimization level.
4872 @item -fgcse-after-reload
4873 @opindex fgcse-after-reload
4874 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4875 pass is performed after reload. The purpose of this pass is to cleanup
4878 @item -funsafe-loop-optimizations
4879 @opindex funsafe-loop-optimizations
4880 If given, the loop optimizer will assume that loop indices do not
4881 overflow, and that the loops with nontrivial exit condition are not
4882 infinite. This enables a wider range of loop optimizations even if
4883 the loop optimizer itself cannot prove that these assumptions are valid.
4884 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
4885 if it finds this kind of loop.
4887 @item -fcrossjumping
4888 @opindex crossjumping
4889 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4890 resulting code may or may not perform better than without cross-jumping.
4892 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4894 @item -fif-conversion
4895 @opindex if-conversion
4896 Attempt to transform conditional jumps into branch-less equivalents. This
4897 include use of conditional moves, min, max, set flags and abs instructions, and
4898 some tricks doable by standard arithmetics. The use of conditional execution
4899 on chips where it is available is controlled by @code{if-conversion2}.
4901 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4903 @item -fif-conversion2
4904 @opindex if-conversion2
4905 Use conditional execution (where available) to transform conditional jumps into
4906 branch-less equivalents.
4908 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4910 @item -fdelete-null-pointer-checks
4911 @opindex fdelete-null-pointer-checks
4912 Use global dataflow analysis to identify and eliminate useless checks
4913 for null pointers. The compiler assumes that dereferencing a null
4914 pointer would have halted the program. If a pointer is checked after
4915 it has already been dereferenced, it cannot be null.
4917 In some environments, this assumption is not true, and programs can
4918 safely dereference null pointers. Use
4919 @option{-fno-delete-null-pointer-checks} to disable this optimization
4920 for programs which depend on that behavior.
4922 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4924 @item -fexpensive-optimizations
4925 @opindex fexpensive-optimizations
4926 Perform a number of minor optimizations that are relatively expensive.
4928 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4930 @item -foptimize-register-move
4932 @opindex foptimize-register-move
4934 Attempt to reassign register numbers in move instructions and as
4935 operands of other simple instructions in order to maximize the amount of
4936 register tying. This is especially helpful on machines with two-operand
4939 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4942 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4944 @item -fdelayed-branch
4945 @opindex fdelayed-branch
4946 If supported for the target machine, attempt to reorder instructions
4947 to exploit instruction slots available after delayed branch
4950 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4952 @item -fschedule-insns
4953 @opindex fschedule-insns
4954 If supported for the target machine, attempt to reorder instructions to
4955 eliminate execution stalls due to required data being unavailable. This
4956 helps machines that have slow floating point or memory load instructions
4957 by allowing other instructions to be issued until the result of the load
4958 or floating point instruction is required.
4960 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4962 @item -fschedule-insns2
4963 @opindex fschedule-insns2
4964 Similar to @option{-fschedule-insns}, but requests an additional pass of
4965 instruction scheduling after register allocation has been done. This is
4966 especially useful on machines with a relatively small number of
4967 registers and where memory load instructions take more than one cycle.
4969 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4971 @item -fno-sched-interblock
4972 @opindex fno-sched-interblock
4973 Don't schedule instructions across basic blocks. This is normally
4974 enabled by default when scheduling before register allocation, i.e.@:
4975 with @option{-fschedule-insns} or at @option{-O2} or higher.
4977 @item -fno-sched-spec
4978 @opindex fno-sched-spec
4979 Don't allow speculative motion of non-load instructions. This is normally
4980 enabled by default when scheduling before register allocation, i.e.@:
4981 with @option{-fschedule-insns} or at @option{-O2} or higher.
4983 @item -fsched-spec-load
4984 @opindex fsched-spec-load
4985 Allow speculative motion of some load instructions. This only makes
4986 sense when scheduling before register allocation, i.e.@: with
4987 @option{-fschedule-insns} or at @option{-O2} or higher.
4989 @item -fsched-spec-load-dangerous
4990 @opindex fsched-spec-load-dangerous
4991 Allow speculative motion of more load instructions. This only makes
4992 sense when scheduling before register allocation, i.e.@: with
4993 @option{-fschedule-insns} or at @option{-O2} or higher.
4995 @item -fsched-stalled-insns=@var{n}
4996 @opindex fsched-stalled-insns
4997 Define how many insns (if any) can be moved prematurely from the queue
4998 of stalled insns into the ready list, during the second scheduling pass.
5000 @item -fsched-stalled-insns-dep=@var{n}
5001 @opindex fsched-stalled-insns-dep
5002 Define how many insn groups (cycles) will be examined for a dependency
5003 on a stalled insn that is candidate for premature removal from the queue
5004 of stalled insns. Has an effect only during the second scheduling pass,
5005 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
5007 @item -fsched2-use-superblocks
5008 @opindex fsched2-use-superblocks
5009 When scheduling after register allocation, do use superblock scheduling
5010 algorithm. Superblock scheduling allows motion across basic block boundaries
5011 resulting on faster schedules. This option is experimental, as not all machine
5012 descriptions used by GCC model the CPU closely enough to avoid unreliable
5013 results from the algorithm.
5015 This only makes sense when scheduling after register allocation, i.e.@: with
5016 @option{-fschedule-insns2} or at @option{-O2} or higher.
5018 @item -fsched2-use-traces
5019 @opindex fsched2-use-traces
5020 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
5021 allocation and additionally perform code duplication in order to increase the
5022 size of superblocks using tracer pass. See @option{-ftracer} for details on
5025 This mode should produce faster but significantly longer programs. Also
5026 without @option{-fbranch-probabilities} the traces constructed may not
5027 match the reality and hurt the performance. This only makes
5028 sense when scheduling after register allocation, i.e.@: with
5029 @option{-fschedule-insns2} or at @option{-O2} or higher.
5033 Eliminates redundant extension instructions and move the non redundant
5034 ones to optimal placement using LCM.
5036 @item -freschedule-modulo-scheduled-loops
5037 @opindex fscheduling-in-modulo-scheduled-loops
5038 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
5039 we may want to prevent the later scheduling passes from changing its schedule, we use this
5040 option to control that.
5042 @item -fcaller-saves
5043 @opindex fcaller-saves
5044 Enable values to be allocated in registers that will be clobbered by
5045 function calls, by emitting extra instructions to save and restore the
5046 registers around such calls. Such allocation is done only when it
5047 seems to result in better code than would otherwise be produced.
5049 This option is always enabled by default on certain machines, usually
5050 those which have no call-preserved registers to use instead.
5052 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5055 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
5056 enabled by default at @option{-O2} and @option{-O3}.
5059 Perform Full Redundancy Elimination (FRE) on trees. The difference
5060 between FRE and PRE is that FRE only considers expressions
5061 that are computed on all paths leading to the redundant computation.
5062 This analysis faster than PRE, though it exposes fewer redundancies.
5063 This flag is enabled by default at @option{-O} and higher.
5065 @item -ftree-copy-prop
5066 Perform copy propagation on trees. This pass eliminates unnecessary
5067 copy operations. This flag is enabled by default at @option{-O} and
5070 @item -ftree-store-copy-prop
5071 Perform copy propagation of memory loads and stores. This pass
5072 eliminates unnecessary copy operations in memory references
5073 (structures, global variables, arrays, etc). This flag is enabled by
5074 default at @option{-O2} and higher.
5077 Perform structural alias analysis on trees. This flag
5078 is enabled by default at @option{-O} and higher.
5081 Perform interprocedural pointer analysis.
5084 Perform forward store motion on trees. This flag is
5085 enabled by default at @option{-O} and higher.
5088 Perform sparse conditional constant propagation (CCP) on trees. This
5089 pass only operates on local scalar variables and is enabled by default
5090 at @option{-O} and higher.
5092 @item -ftree-store-ccp
5093 Perform sparse conditional constant propagation (CCP) on trees. This
5094 pass operates on both local scalar variables and memory stores and
5095 loads (global variables, structures, arrays, etc). This flag is
5096 enabled by default at @option{-O2} and higher.
5099 Perform dead code elimination (DCE) on trees. This flag is enabled by
5100 default at @option{-O} and higher.
5102 @item -ftree-dominator-opts
5103 Perform a variety of simple scalar cleanups (constant/copy
5104 propagation, redundancy elimination, range propagation and expression
5105 simplification) based on a dominator tree traversal. This also
5106 performs jump threading (to reduce jumps to jumps). This flag is
5107 enabled by default at @option{-O} and higher.
5110 Perform loop header copying on trees. This is beneficial since it increases
5111 effectiveness of code motion optimizations. It also saves one jump. This flag
5112 is enabled by default at @option{-O} and higher. It is not enabled
5113 for @option{-Os}, since it usually increases code size.
5115 @item -ftree-loop-optimize
5116 Perform loop optimizations on trees. This flag is enabled by default
5117 at @option{-O} and higher.
5119 @item -ftree-loop-linear
5120 Perform linear loop transformations on tree. This flag can improve cache
5121 performance and allow further loop optimizations to take place.
5123 @item -ftree-loop-im
5124 Perform loop invariant motion on trees. This pass moves only invariants that
5125 would be hard to handle at RTL level (function calls, operations that expand to
5126 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
5127 operands of conditions that are invariant out of the loop, so that we can use
5128 just trivial invariantness analysis in loop unswitching. The pass also includes
5131 @item -ftree-loop-ivcanon
5132 Create a canonical counter for number of iterations in the loop for that
5133 determining number of iterations requires complicated analysis. Later
5134 optimizations then may determine the number easily. Useful especially
5135 in connection with unrolling.
5138 Perform induction variable optimizations (strength reduction, induction
5139 variable merging and induction variable elimination) on trees.
5142 Perform scalar replacement of aggregates. This pass replaces structure
5143 references with scalars to prevent committing structures to memory too
5144 early. This flag is enabled by default at @option{-O} and higher.
5146 @item -ftree-copyrename
5147 Perform copy renaming on trees. This pass attempts to rename compiler
5148 temporaries to other variables at copy locations, usually resulting in
5149 variable names which more closely resemble the original variables. This flag
5150 is enabled by default at @option{-O} and higher.
5153 Perform temporary expression replacement during the SSA->normal phase. Single
5154 use/single def temporaries are replaced at their use location with their
5155 defining expression. This results in non-GIMPLE code, but gives the expanders
5156 much more complex trees to work on resulting in better RTL generation. This is
5157 enabled by default at @option{-O} and higher.
5160 Perform live range splitting during the SSA->normal phase. Distinct live
5161 ranges of a variable are split into unique variables, allowing for better
5162 optimization later. This is enabled by default at @option{-O} and higher.
5164 @item -ftree-vectorize
5165 Perform loop vectorization on trees.
5167 @item -ftree-vect-loop-version
5168 @opindex ftree-vect-loop-version
5169 Perform loop versioning when doing loop vectorization on trees. When a loop
5170 appears to be vectorizable except that data alignment or data dependence cannot
5171 be determined at compile time then vectorized and non-vectorized versions of
5172 the loop are generated along with runtime checks for alignment or dependence
5173 to control which version is executed. This option is enabled by default
5174 except at level @option{-Os} where it is disabled.
5177 Perform Value Range Propagation on trees. This is similar to the
5178 constant propagation pass, but instead of values, ranges of values are
5179 propagated. This allows the optimizers to remove unnecessary range
5180 checks like array bound checks and null pointer checks. This is
5181 enabled by default at @option{-O2} and higher. Null pointer check
5182 elimination is only done if @option{-fdelete-null-pointer-checks} is
5187 Perform tail duplication to enlarge superblock size. This transformation
5188 simplifies the control flow of the function allowing other optimizations to do
5191 @item -funroll-loops
5192 @opindex funroll-loops
5193 Unroll loops whose number of iterations can be determined at compile
5194 time or upon entry to the loop. @option{-funroll-loops} implies
5195 @option{-frerun-cse-after-loop}. This option makes code larger,
5196 and may or may not make it run faster.
5198 @item -funroll-all-loops
5199 @opindex funroll-all-loops
5200 Unroll all loops, even if their number of iterations is uncertain when
5201 the loop is entered. This usually makes programs run more slowly.
5202 @option{-funroll-all-loops} implies the same options as
5203 @option{-funroll-loops},
5205 @item -fsplit-ivs-in-unroller
5206 @opindex -fsplit-ivs-in-unroller
5207 Enables expressing of values of induction variables in later iterations
5208 of the unrolled loop using the value in the first iteration. This breaks
5209 long dependency chains, thus improving efficiency of the scheduling passes.
5211 Combination of @option{-fweb} and CSE is often sufficient to obtain the
5212 same effect. However in cases the loop body is more complicated than
5213 a single basic block, this is not reliable. It also does not work at all
5214 on some of the architectures due to restrictions in the CSE pass.
5216 This optimization is enabled by default.
5218 @item -fvariable-expansion-in-unroller
5219 @opindex -fvariable-expansion-in-unroller
5220 With this option, the compiler will create multiple copies of some
5221 local variables when unrolling a loop which can result in superior code.
5223 @item -fprefetch-loop-arrays
5224 @opindex fprefetch-loop-arrays
5225 If supported by the target machine, generate instructions to prefetch
5226 memory to improve the performance of loops that access large arrays.
5228 This option may generate better or worse code; results are highly
5229 dependent on the structure of loops within the source code.
5231 Disabled at level @option{-Os}.
5234 @itemx -fno-peephole2
5235 @opindex fno-peephole
5236 @opindex fno-peephole2
5237 Disable any machine-specific peephole optimizations. The difference
5238 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
5239 are implemented in the compiler; some targets use one, some use the
5240 other, a few use both.
5242 @option{-fpeephole} is enabled by default.
5243 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5245 @item -fno-guess-branch-probability
5246 @opindex fno-guess-branch-probability
5247 Do not guess branch probabilities using heuristics.
5249 GCC will use heuristics to guess branch probabilities if they are
5250 not provided by profiling feedback (@option{-fprofile-arcs}). These
5251 heuristics are based on the control flow graph. If some branch probabilities
5252 are specified by @samp{__builtin_expect}, then the heuristics will be
5253 used to guess branch probabilities for the rest of the control flow graph,
5254 taking the @samp{__builtin_expect} info into account. The interactions
5255 between the heuristics and @samp{__builtin_expect} can be complex, and in
5256 some cases, it may be useful to disable the heuristics so that the effects
5257 of @samp{__builtin_expect} are easier to understand.
5259 The default is @option{-fguess-branch-probability} at levels
5260 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5262 @item -freorder-blocks
5263 @opindex freorder-blocks
5264 Reorder basic blocks in the compiled function in order to reduce number of
5265 taken branches and improve code locality.
5267 Enabled at levels @option{-O2}, @option{-O3}.
5269 @item -freorder-blocks-and-partition
5270 @opindex freorder-blocks-and-partition
5271 In addition to reordering basic blocks in the compiled function, in order
5272 to reduce number of taken branches, partitions hot and cold basic blocks
5273 into separate sections of the assembly and .o files, to improve
5274 paging and cache locality performance.
5276 This optimization is automatically turned off in the presence of
5277 exception handling, for linkonce sections, for functions with a user-defined
5278 section attribute and on any architecture that does not support named
5281 @item -freorder-functions
5282 @opindex freorder-functions
5283 Reorder functions in the object file in order to
5284 improve code locality. This is implemented by using special
5285 subsections @code{.text.hot} for most frequently executed functions and
5286 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
5287 the linker so object file format must support named sections and linker must
5288 place them in a reasonable way.
5290 Also profile feedback must be available in to make this option effective. See
5291 @option{-fprofile-arcs} for details.
5293 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5295 @item -fstrict-aliasing
5296 @opindex fstrict-aliasing
5297 Allows the compiler to assume the strictest aliasing rules applicable to
5298 the language being compiled. For C (and C++), this activates
5299 optimizations based on the type of expressions. In particular, an
5300 object of one type is assumed never to reside at the same address as an
5301 object of a different type, unless the types are almost the same. For
5302 example, an @code{unsigned int} can alias an @code{int}, but not a
5303 @code{void*} or a @code{double}. A character type may alias any other
5306 Pay special attention to code like this:
5319 The practice of reading from a different union member than the one most
5320 recently written to (called ``type-punning'') is common. Even with
5321 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
5322 is accessed through the union type. So, the code above will work as
5323 expected. However, this code might not:
5334 Every language that wishes to perform language-specific alias analysis
5335 should define a function that computes, given an @code{tree}
5336 node, an alias set for the node. Nodes in different alias sets are not
5337 allowed to alias. For an example, see the C front-end function
5338 @code{c_get_alias_set}.
5340 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5342 @item -falign-functions
5343 @itemx -falign-functions=@var{n}
5344 @opindex falign-functions
5345 Align the start of functions to the next power-of-two greater than
5346 @var{n}, skipping up to @var{n} bytes. For instance,
5347 @option{-falign-functions=32} aligns functions to the next 32-byte
5348 boundary, but @option{-falign-functions=24} would align to the next
5349 32-byte boundary only if this can be done by skipping 23 bytes or less.
5351 @option{-fno-align-functions} and @option{-falign-functions=1} are
5352 equivalent and mean that functions will not be aligned.
5354 Some assemblers only support this flag when @var{n} is a power of two;
5355 in that case, it is rounded up.
5357 If @var{n} is not specified or is zero, use a machine-dependent default.
5359 Enabled at levels @option{-O2}, @option{-O3}.
5361 @item -falign-labels
5362 @itemx -falign-labels=@var{n}
5363 @opindex falign-labels
5364 Align all branch targets to a power-of-two boundary, skipping up to
5365 @var{n} bytes like @option{-falign-functions}. This option can easily
5366 make code slower, because it must insert dummy operations for when the
5367 branch target is reached in the usual flow of the code.
5369 @option{-fno-align-labels} and @option{-falign-labels=1} are
5370 equivalent and mean that labels will not be aligned.
5372 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
5373 are greater than this value, then their values are used instead.
5375 If @var{n} is not specified or is zero, use a machine-dependent default
5376 which is very likely to be @samp{1}, meaning no alignment.
5378 Enabled at levels @option{-O2}, @option{-O3}.
5381 @itemx -falign-loops=@var{n}
5382 @opindex falign-loops
5383 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
5384 like @option{-falign-functions}. The hope is that the loop will be
5385 executed many times, which will make up for any execution of the dummy
5388 @option{-fno-align-loops} and @option{-falign-loops=1} are
5389 equivalent and mean that loops will not be aligned.
5391 If @var{n} is not specified or is zero, use a machine-dependent default.
5393 Enabled at levels @option{-O2}, @option{-O3}.
5396 @itemx -falign-jumps=@var{n}
5397 @opindex falign-jumps
5398 Align branch targets to a power-of-two boundary, for branch targets
5399 where the targets can only be reached by jumping, skipping up to @var{n}
5400 bytes like @option{-falign-functions}. In this case, no dummy operations
5403 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
5404 equivalent and mean that loops will not be aligned.
5406 If @var{n} is not specified or is zero, use a machine-dependent default.
5408 Enabled at levels @option{-O2}, @option{-O3}.
5410 @item -funit-at-a-time
5411 @opindex funit-at-a-time
5412 Parse the whole compilation unit before starting to produce code.
5413 This allows some extra optimizations to take place but consumes
5414 more memory (in general). There are some compatibility issues
5415 with @emph{unit-at-a-time} mode:
5418 enabling @emph{unit-at-a-time} mode may change the order
5419 in which functions, variables, and top-level @code{asm} statements
5420 are emitted, and will likely break code relying on some particular
5421 ordering. The majority of such top-level @code{asm} statements,
5422 though, can be replaced by @code{section} attributes. The
5423 @option{fno-toplevel-reorder} option may be used to keep the ordering
5424 used in the input file, at the cost of some optimizations.
5427 @emph{unit-at-a-time} mode removes unreferenced static variables
5428 and functions. This may result in undefined references
5429 when an @code{asm} statement refers directly to variables or functions
5430 that are otherwise unused. In that case either the variable/function
5431 shall be listed as an operand of the @code{asm} statement operand or,
5432 in the case of top-level @code{asm} statements the attribute @code{used}
5433 shall be used on the declaration.
5436 Static functions now can use non-standard passing conventions that
5437 may break @code{asm} statements calling functions directly. Again,
5438 attribute @code{used} will prevent this behavior.
5441 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5442 but this scheme may not be supported by future releases of GCC@.
5444 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5446 @item -fno-toplevel-reorder
5447 Do not reorder top-level functions, variables, and @code{asm}
5448 statements. Output them in the same order that they appear in the
5449 input file. When this option is used, unreferenced static variables
5450 will not be removed. This option is intended to support existing code
5451 which relies on a particular ordering. For new code, it is better to
5456 Constructs webs as commonly used for register allocation purposes and assign
5457 each web individual pseudo register. This allows the register allocation pass
5458 to operate on pseudos directly, but also strengthens several other optimization
5459 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5460 however, make debugging impossible, since variables will no longer stay in a
5463 Enabled by default with @option{-funroll-loops}.
5465 @item -fwhole-program
5466 @opindex fwhole-program
5467 Assume that the current compilation unit represents whole program being
5468 compiled. All public functions and variables with the exception of @code{main}
5469 and those merged by attribute @code{externally_visible} become static functions
5470 and in a affect gets more aggressively optimized by interprocedural optimizers.
5471 While this option is equivalent to proper use of @code{static} keyword for
5472 programs consisting of single file, in combination with option
5473 @option{--combine} this flag can be used to compile most of smaller scale C
5474 programs since the functions and variables become local for the whole combined
5475 compilation unit, not for the single source file itself.
5478 @item -fno-cprop-registers
5479 @opindex fno-cprop-registers
5480 After register allocation and post-register allocation instruction splitting,
5481 we perform a copy-propagation pass to try to reduce scheduling dependencies
5482 and occasionally eliminate the copy.
5484 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5486 @item -fprofile-generate
5487 @opindex fprofile-generate
5489 Enable options usually used for instrumenting application to produce
5490 profile useful for later recompilation with profile feedback based
5491 optimization. You must use @option{-fprofile-generate} both when
5492 compiling and when linking your program.
5494 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5497 @opindex fprofile-use
5498 Enable profile feedback directed optimizations, and optimizations
5499 generally profitable only with profile feedback available.
5501 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
5502 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
5506 The following options control compiler behavior regarding floating
5507 point arithmetic. These options trade off between speed and
5508 correctness. All must be specifically enabled.
5512 @opindex ffloat-store
5513 Do not store floating point variables in registers, and inhibit other
5514 options that might change whether a floating point value is taken from a
5517 @cindex floating point precision
5518 This option prevents undesirable excess precision on machines such as
5519 the 68000 where the floating registers (of the 68881) keep more
5520 precision than a @code{double} is supposed to have. Similarly for the
5521 x86 architecture. For most programs, the excess precision does only
5522 good, but a few programs rely on the precise definition of IEEE floating
5523 point. Use @option{-ffloat-store} for such programs, after modifying
5524 them to store all pertinent intermediate computations into variables.
5528 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5529 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5530 @option{-fno-rounding-math}, @option{-fno-signaling-nans}
5531 and @option{fcx-limited-range}.
5533 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5535 This option should never be turned on by any @option{-O} option since
5536 it can result in incorrect output for programs which depend on
5537 an exact implementation of IEEE or ISO rules/specifications for
5540 @item -fno-math-errno
5541 @opindex fno-math-errno
5542 Do not set ERRNO after calling math functions that are executed
5543 with a single instruction, e.g., sqrt. A program that relies on
5544 IEEE exceptions for math error handling may want to use this flag
5545 for speed while maintaining IEEE arithmetic compatibility.
5547 This option should never be turned on by any @option{-O} option since
5548 it can result in incorrect output for programs which depend on
5549 an exact implementation of IEEE or ISO rules/specifications for
5552 The default is @option{-fmath-errno}.
5554 On Darwin systems, the math library never sets @code{errno}. There is therefore
5555 no reason for the compiler to consider the possibility that it might,
5556 and @option{-fno-math-errno} is the default.
5558 @item -funsafe-math-optimizations
5559 @opindex funsafe-math-optimizations
5560 Allow optimizations for floating-point arithmetic that (a) assume
5561 that arguments and results are valid and (b) may violate IEEE or
5562 ANSI standards. When used at link-time, it may include libraries
5563 or startup files that change the default FPU control word or other
5564 similar optimizations.
5566 This option should never be turned on by any @option{-O} option since
5567 it can result in incorrect output for programs which depend on
5568 an exact implementation of IEEE or ISO rules/specifications for
5571 The default is @option{-fno-unsafe-math-optimizations}.
5573 @item -ffinite-math-only
5574 @opindex ffinite-math-only
5575 Allow optimizations for floating-point arithmetic that assume
5576 that arguments and results are not NaNs or +-Infs.
5578 This option should never be turned on by any @option{-O} option since
5579 it can result in incorrect output for programs which depend on
5580 an exact implementation of IEEE or ISO rules/specifications.
5582 The default is @option{-fno-finite-math-only}.
5584 @item -fno-trapping-math
5585 @opindex fno-trapping-math
5586 Compile code assuming that floating-point operations cannot generate
5587 user-visible traps. These traps include division by zero, overflow,
5588 underflow, inexact result and invalid operation. This option implies
5589 @option{-fno-signaling-nans}. Setting this option may allow faster
5590 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5592 This option should never be turned on by any @option{-O} option since
5593 it can result in incorrect output for programs which depend on
5594 an exact implementation of IEEE or ISO rules/specifications for
5597 The default is @option{-ftrapping-math}.
5599 @item -frounding-math
5600 @opindex frounding-math
5601 Disable transformations and optimizations that assume default floating
5602 point rounding behavior. This is round-to-zero for all floating point
5603 to integer conversions, and round-to-nearest for all other arithmetic
5604 truncations. This option should be specified for programs that change
5605 the FP rounding mode dynamically, or that may be executed with a
5606 non-default rounding mode. This option disables constant folding of
5607 floating point expressions at compile-time (which may be affected by
5608 rounding mode) and arithmetic transformations that are unsafe in the
5609 presence of sign-dependent rounding modes.
5611 The default is @option{-fno-rounding-math}.
5613 This option is experimental and does not currently guarantee to
5614 disable all GCC optimizations that are affected by rounding mode.
5615 Future versions of GCC may provide finer control of this setting
5616 using C99's @code{FENV_ACCESS} pragma. This command line option
5617 will be used to specify the default state for @code{FENV_ACCESS}.
5619 @item -frtl-abstract-sequences
5620 @opindex frtl-abstract-sequences
5621 It is a size optimization method. This option is to find identical
5622 sequences of code, which can be turned into pseudo-procedures and
5623 then replace all occurrences with calls to the newly created
5624 subroutine. It is kind of an opposite of @option{-finline-functions}.
5625 This optimization runs at RTL level.
5627 @item -fsignaling-nans
5628 @opindex fsignaling-nans
5629 Compile code assuming that IEEE signaling NaNs may generate user-visible
5630 traps during floating-point operations. Setting this option disables
5631 optimizations that may change the number of exceptions visible with
5632 signaling NaNs. This option implies @option{-ftrapping-math}.
5634 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5637 The default is @option{-fno-signaling-nans}.
5639 This option is experimental and does not currently guarantee to
5640 disable all GCC optimizations that affect signaling NaN behavior.
5642 @item -fsingle-precision-constant
5643 @opindex fsingle-precision-constant
5644 Treat floating point constant as single precision constant instead of
5645 implicitly converting it to double precision constant.
5647 @item -fcx-limited-range
5648 @itemx -fno-cx-limited-range
5649 @opindex fcx-limited-range
5650 @opindex fno-cx-limited-range
5651 When enabled, this option states that a range reduction step is not
5652 needed when performing complex division. The default is
5653 @option{-fno-cx-limited-range}, but is enabled by @option{-ffast-math}.
5655 This option controls the default setting of the ISO C99
5656 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
5661 The following options control optimizations that may improve
5662 performance, but are not enabled by any @option{-O} options. This
5663 section includes experimental options that may produce broken code.
5666 @item -fbranch-probabilities
5667 @opindex fbranch-probabilities
5668 After running a program compiled with @option{-fprofile-arcs}
5669 (@pxref{Debugging Options,, Options for Debugging Your Program or
5670 @command{gcc}}), you can compile it a second time using
5671 @option{-fbranch-probabilities}, to improve optimizations based on
5672 the number of times each branch was taken. When the program
5673 compiled with @option{-fprofile-arcs} exits it saves arc execution
5674 counts to a file called @file{@var{sourcename}.gcda} for each source
5675 file The information in this data file is very dependent on the
5676 structure of the generated code, so you must use the same source code
5677 and the same optimization options for both compilations.
5679 With @option{-fbranch-probabilities}, GCC puts a
5680 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5681 These can be used to improve optimization. Currently, they are only
5682 used in one place: in @file{reorg.c}, instead of guessing which path a
5683 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5684 exactly determine which path is taken more often.
5686 @item -fprofile-values
5687 @opindex fprofile-values
5688 If combined with @option{-fprofile-arcs}, it adds code so that some
5689 data about values of expressions in the program is gathered.
5691 With @option{-fbranch-probabilities}, it reads back the data gathered
5692 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5693 notes to instructions for their later usage in optimizations.
5695 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5699 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5700 a code to gather information about values of expressions.
5702 With @option{-fbranch-probabilities}, it reads back the data gathered
5703 and actually performs the optimizations based on them.
5704 Currently the optimizations include specialization of division operation
5705 using the knowledge about the value of the denominator.
5707 @item -frename-registers
5708 @opindex frename-registers
5709 Attempt to avoid false dependencies in scheduled code by making use
5710 of registers left over after register allocation. This optimization
5711 will most benefit processors with lots of registers. Depending on the
5712 debug information format adopted by the target, however, it can
5713 make debugging impossible, since variables will no longer stay in
5714 a ``home register''.
5716 Enabled by default with @option{-funroll-loops}.
5720 Perform tail duplication to enlarge superblock size. This transformation
5721 simplifies the control flow of the function allowing other optimizations to do
5724 Enabled with @option{-fprofile-use}.
5726 @item -funroll-loops
5727 @opindex funroll-loops
5728 Unroll loops whose number of iterations can be determined at compile time or
5729 upon entry to the loop. @option{-funroll-loops} implies
5730 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
5731 It also turns on complete loop peeling (i.e.@: complete removal of loops with
5732 small constant number of iterations). This option makes code larger, and may
5733 or may not make it run faster.
5735 Enabled with @option{-fprofile-use}.
5737 @item -funroll-all-loops
5738 @opindex funroll-all-loops
5739 Unroll all loops, even if their number of iterations is uncertain when
5740 the loop is entered. This usually makes programs run more slowly.
5741 @option{-funroll-all-loops} implies the same options as
5742 @option{-funroll-loops}.
5745 @opindex fpeel-loops
5746 Peels the loops for that there is enough information that they do not
5747 roll much (from profile feedback). It also turns on complete loop peeling
5748 (i.e.@: complete removal of loops with small constant number of iterations).
5750 Enabled with @option{-fprofile-use}.
5752 @item -fmove-loop-invariants
5753 @opindex fmove-loop-invariants
5754 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
5755 at level @option{-O1}
5757 @item -funswitch-loops
5758 @opindex funswitch-loops
5759 Move branches with loop invariant conditions out of the loop, with duplicates
5760 of the loop on both branches (modified according to result of the condition).
5762 @item -ffunction-sections
5763 @itemx -fdata-sections
5764 @opindex ffunction-sections
5765 @opindex fdata-sections
5766 Place each function or data item into its own section in the output
5767 file if the target supports arbitrary sections. The name of the
5768 function or the name of the data item determines the section's name
5771 Use these options on systems where the linker can perform optimizations
5772 to improve locality of reference in the instruction space. Most systems
5773 using the ELF object format and SPARC processors running Solaris 2 have
5774 linkers with such optimizations. AIX may have these optimizations in
5777 Only use these options when there are significant benefits from doing
5778 so. When you specify these options, the assembler and linker will
5779 create larger object and executable files and will also be slower.
5780 You will not be able to use @code{gprof} on all systems if you
5781 specify this option and you may have problems with debugging if
5782 you specify both this option and @option{-g}.
5784 @item -fbranch-target-load-optimize
5785 @opindex fbranch-target-load-optimize
5786 Perform branch target register load optimization before prologue / epilogue
5788 The use of target registers can typically be exposed only during reload,
5789 thus hoisting loads out of loops and doing inter-block scheduling needs
5790 a separate optimization pass.
5792 @item -fbranch-target-load-optimize2
5793 @opindex fbranch-target-load-optimize2
5794 Perform branch target register load optimization after prologue / epilogue
5797 @item -fbtr-bb-exclusive
5798 @opindex fbtr-bb-exclusive
5799 When performing branch target register load optimization, don't reuse
5800 branch target registers in within any basic block.
5802 @item -fstack-protector
5803 Emit extra code to check for buffer overflows, such as stack smashing
5804 attacks. This is done by adding a guard variable to functions with
5805 vulnerable objects. This includes functions that call alloca, and
5806 functions with buffers larger than 8 bytes. The guards are initialized
5807 when a function is entered and then checked when the function exits.
5808 If a guard check fails, an error message is printed and the program exits.
5810 @item -fstack-protector-all
5811 Like @option{-fstack-protector} except that all functions are protected.
5813 @item -fsection-anchors
5814 @opindex fsection-anchors
5815 Try to reduce the number of symbolic address calculations by using
5816 shared ``anchor'' symbols to address nearby objects. This transformation
5817 can help to reduce the number of GOT entries and GOT accesses on some
5820 For example, the implementation of the following function @code{foo}:
5824 int foo (void) @{ return a + b + c; @}
5827 would usually calculate the addresses of all three variables, but if you
5828 compile it with @option{-fsection-anchors}, it will access the variables
5829 from a common anchor point instead. The effect is similar to the
5830 following pseudocode (which isn't valid C):
5835 register int *xr = &x;
5836 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5840 Not all targets support this option.
5842 @item --param @var{name}=@var{value}
5844 In some places, GCC uses various constants to control the amount of
5845 optimization that is done. For example, GCC will not inline functions
5846 that contain more that a certain number of instructions. You can
5847 control some of these constants on the command-line using the
5848 @option{--param} option.
5850 The names of specific parameters, and the meaning of the values, are
5851 tied to the internals of the compiler, and are subject to change
5852 without notice in future releases.
5854 In each case, the @var{value} is an integer. The allowable choices for
5855 @var{name} are given in the following table:
5858 @item salias-max-implicit-fields
5859 The maximum number of fields in a variable without direct
5860 structure accesses for which structure aliasing will consider trying
5861 to track each field. The default is 5
5863 @item salias-max-array-elements
5864 The maximum number of elements an array can have and its elements
5865 still be tracked individually by structure aliasing. The default is 4
5867 @item sra-max-structure-size
5868 The maximum structure size, in bytes, at which the scalar replacement
5869 of aggregates (SRA) optimization will perform block copies. The
5870 default value, 0, implies that GCC will select the most appropriate
5873 @item sra-field-structure-ratio
5874 The threshold ratio (as a percentage) between instantiated fields and
5875 the complete structure size. We say that if the ratio of the number
5876 of bytes in instantiated fields to the number of bytes in the complete
5877 structure exceeds this parameter, then block copies are not used. The
5880 @item max-crossjump-edges
5881 The maximum number of incoming edges to consider for crossjumping.
5882 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5883 the number of edges incoming to each block. Increasing values mean
5884 more aggressive optimization, making the compile time increase with
5885 probably small improvement in executable size.
5887 @item min-crossjump-insns
5888 The minimum number of instructions which must be matched at the end
5889 of two blocks before crossjumping will be performed on them. This
5890 value is ignored in the case where all instructions in the block being
5891 crossjumped from are matched. The default value is 5.
5893 @item max-grow-copy-bb-insns
5894 The maximum code size expansion factor when copying basic blocks
5895 instead of jumping. The expansion is relative to a jump instruction.
5896 The default value is 8.
5898 @item max-goto-duplication-insns
5899 The maximum number of instructions to duplicate to a block that jumps
5900 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5901 passes, GCC factors computed gotos early in the compilation process,
5902 and unfactors them as late as possible. Only computed jumps at the
5903 end of a basic blocks with no more than max-goto-duplication-insns are
5904 unfactored. The default value is 8.
5906 @item max-delay-slot-insn-search
5907 The maximum number of instructions to consider when looking for an
5908 instruction to fill a delay slot. If more than this arbitrary number of
5909 instructions is searched, the time savings from filling the delay slot
5910 will be minimal so stop searching. Increasing values mean more
5911 aggressive optimization, making the compile time increase with probably
5912 small improvement in executable run time.
5914 @item max-delay-slot-live-search
5915 When trying to fill delay slots, the maximum number of instructions to
5916 consider when searching for a block with valid live register
5917 information. Increasing this arbitrarily chosen value means more
5918 aggressive optimization, increasing the compile time. This parameter
5919 should be removed when the delay slot code is rewritten to maintain the
5922 @item max-gcse-memory
5923 The approximate maximum amount of memory that will be allocated in
5924 order to perform the global common subexpression elimination
5925 optimization. If more memory than specified is required, the
5926 optimization will not be done.
5928 @item max-gcse-passes
5929 The maximum number of passes of GCSE to run. The default is 1.
5931 @item max-pending-list-length
5932 The maximum number of pending dependencies scheduling will allow
5933 before flushing the current state and starting over. Large functions
5934 with few branches or calls can create excessively large lists which
5935 needlessly consume memory and resources.
5937 @item max-inline-insns-single
5938 Several parameters control the tree inliner used in gcc.
5939 This number sets the maximum number of instructions (counted in GCC's
5940 internal representation) in a single function that the tree inliner
5941 will consider for inlining. This only affects functions declared
5942 inline and methods implemented in a class declaration (C++).
5943 The default value is 450.
5945 @item max-inline-insns-auto
5946 When you use @option{-finline-functions} (included in @option{-O3}),
5947 a lot of functions that would otherwise not be considered for inlining
5948 by the compiler will be investigated. To those functions, a different
5949 (more restrictive) limit compared to functions declared inline can
5951 The default value is 90.
5953 @item large-function-insns
5954 The limit specifying really large functions. For functions larger than this
5955 limit after inlining inlining is constrained by
5956 @option{--param large-function-growth}. This parameter is useful primarily
5957 to avoid extreme compilation time caused by non-linear algorithms used by the
5959 This parameter is ignored when @option{-funit-at-a-time} is not used.
5960 The default value is 2700.
5962 @item large-function-growth
5963 Specifies maximal growth of large function caused by inlining in percents.
5964 This parameter is ignored when @option{-funit-at-a-time} is not used.
5965 The default value is 100 which limits large function growth to 2.0 times
5968 @item large-unit-insns
5969 The limit specifying large translation unit. Growth caused by inlining of
5970 units larger than this limit is limited by @option{--param inline-unit-growth}.
5971 For small units this might be too tight (consider unit consisting of function A
5972 that is inline and B that just calls A three time. If B is small relative to
5973 A, the growth of unit is 300\% and yet such inlining is very sane. For very
5974 large units consisting of small inlininable functions however the overall unit
5975 growth limit is needed to avoid exponential explosion of code size. Thus for
5976 smaller units, the size is increased to @option{--param large-unit-insns}
5977 before applying @option{--param inline-unit-growth}. The default is 10000
5979 @item inline-unit-growth
5980 Specifies maximal overall growth of the compilation unit caused by inlining.
5981 This parameter is ignored when @option{-funit-at-a-time} is not used.
5982 The default value is 50 which limits unit growth to 1.5 times the original
5985 @item max-inline-insns-recursive
5986 @itemx max-inline-insns-recursive-auto
5987 Specifies maximum number of instructions out-of-line copy of self recursive inline
5988 function can grow into by performing recursive inlining.
5990 For functions declared inline @option{--param max-inline-insns-recursive} is
5991 taken into account. For function not declared inline, recursive inlining
5992 happens only when @option{-finline-functions} (included in @option{-O3}) is
5993 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5994 default value is 450.
5996 @item max-inline-recursive-depth
5997 @itemx max-inline-recursive-depth-auto
5998 Specifies maximum recursion depth used by the recursive inlining.
6000 For functions declared inline @option{--param max-inline-recursive-depth} is
6001 taken into account. For function not declared inline, recursive inlining
6002 happens only when @option{-finline-functions} (included in @option{-O3}) is
6003 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
6004 default value is 450.
6006 @item min-inline-recursive-probability
6007 Recursive inlining is profitable only for function having deep recursion
6008 in average and can hurt for function having little recursion depth by
6009 increasing the prologue size or complexity of function body to other
6012 When profile feedback is available (see @option{-fprofile-generate}) the actual
6013 recursion depth can be guessed from probability that function will recurse via
6014 given call expression. This parameter limits inlining only to call expression
6015 whose probability exceeds given threshold (in percents). The default value is
6018 @item inline-call-cost
6019 Specify cost of call instruction relative to simple arithmetics operations
6020 (having cost of 1). Increasing this cost disqualifies inlining of non-leaf
6021 functions and at the same time increases size of leaf function that is believed to
6022 reduce function size by being inlined. In effect it increases amount of
6023 inlining for code having large abstraction penalty (many functions that just
6024 pass the arguments to other functions) and decrease inlining for code with low
6025 abstraction penalty. The default value is 16.
6027 @item max-unrolled-insns
6028 The maximum number of instructions that a loop should have if that loop
6029 is unrolled, and if the loop is unrolled, it determines how many times
6030 the loop code is unrolled.
6032 @item max-average-unrolled-insns
6033 The maximum number of instructions biased by probabilities of their execution
6034 that a loop should have if that loop is unrolled, and if the loop is unrolled,
6035 it determines how many times the loop code is unrolled.
6037 @item max-unroll-times
6038 The maximum number of unrollings of a single loop.
6040 @item max-peeled-insns
6041 The maximum number of instructions that a loop should have if that loop
6042 is peeled, and if the loop is peeled, it determines how many times
6043 the loop code is peeled.
6045 @item max-peel-times
6046 The maximum number of peelings of a single loop.
6048 @item max-completely-peeled-insns
6049 The maximum number of insns of a completely peeled loop.
6051 @item max-completely-peel-times
6052 The maximum number of iterations of a loop to be suitable for complete peeling.
6054 @item max-unswitch-insns
6055 The maximum number of insns of an unswitched loop.
6057 @item max-unswitch-level
6058 The maximum number of branches unswitched in a single loop.
6061 The minimum cost of an expensive expression in the loop invariant motion.
6063 @item iv-consider-all-candidates-bound
6064 Bound on number of candidates for induction variables below that
6065 all candidates are considered for each use in induction variable
6066 optimizations. Only the most relevant candidates are considered
6067 if there are more candidates, to avoid quadratic time complexity.
6069 @item iv-max-considered-uses
6070 The induction variable optimizations give up on loops that contain more
6071 induction variable uses.
6073 @item iv-always-prune-cand-set-bound
6074 If number of candidates in the set is smaller than this value,
6075 we always try to remove unnecessary ivs from the set during its
6076 optimization when a new iv is added to the set.
6078 @item scev-max-expr-size
6079 Bound on size of expressions used in the scalar evolutions analyzer.
6080 Large expressions slow the analyzer.
6082 @item vect-max-version-checks
6083 The maximum number of runtime checks that can be performed when doing
6084 loop versioning in the vectorizer. See option ftree-vect-loop-version
6085 for more information.
6087 @item max-iterations-to-track
6089 The maximum number of iterations of a loop the brute force algorithm
6090 for analysis of # of iterations of the loop tries to evaluate.
6092 @item hot-bb-count-fraction
6093 Select fraction of the maximal count of repetitions of basic block in program
6094 given basic block needs to have to be considered hot.
6096 @item hot-bb-frequency-fraction
6097 Select fraction of the maximal frequency of executions of basic block in
6098 function given basic block needs to have to be considered hot
6100 @item max-predicted-iterations
6101 The maximum number of loop iterations we predict statically. This is useful
6102 in cases where function contain single loop with known bound and other loop
6103 with unknown. We predict the known number of iterations correctly, while
6104 the unknown number of iterations average to roughly 10. This means that the
6105 loop without bounds would appear artificially cold relative to the other one.
6107 @item tracer-dynamic-coverage
6108 @itemx tracer-dynamic-coverage-feedback
6110 This value is used to limit superblock formation once the given percentage of
6111 executed instructions is covered. This limits unnecessary code size
6114 The @option{tracer-dynamic-coverage-feedback} is used only when profile
6115 feedback is available. The real profiles (as opposed to statically estimated
6116 ones) are much less balanced allowing the threshold to be larger value.
6118 @item tracer-max-code-growth
6119 Stop tail duplication once code growth has reached given percentage. This is
6120 rather hokey argument, as most of the duplicates will be eliminated later in
6121 cross jumping, so it may be set to much higher values than is the desired code
6124 @item tracer-min-branch-ratio
6126 Stop reverse growth when the reverse probability of best edge is less than this
6127 threshold (in percent).
6129 @item tracer-min-branch-ratio
6130 @itemx tracer-min-branch-ratio-feedback
6132 Stop forward growth if the best edge do have probability lower than this
6135 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
6136 compilation for profile feedback and one for compilation without. The value
6137 for compilation with profile feedback needs to be more conservative (higher) in
6138 order to make tracer effective.
6140 @item max-cse-path-length
6142 Maximum number of basic blocks on path that cse considers. The default is 10.
6145 The maximum instructions CSE process before flushing. The default is 1000.
6147 @item global-var-threshold
6149 Counts the number of function calls (@var{n}) and the number of
6150 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
6151 single artificial variable will be created to represent all the
6152 call-clobbered variables at function call sites. This artificial
6153 variable will then be made to alias every call-clobbered variable.
6154 (done as @code{int * size_t} on the host machine; beware overflow).
6156 @item max-aliased-vops
6158 Maximum number of virtual operands allowed to represent aliases
6159 before triggering the alias grouping heuristic. Alias grouping
6160 reduces compile times and memory consumption needed for aliasing at
6161 the expense of precision loss in alias information.
6163 @item ggc-min-expand
6165 GCC uses a garbage collector to manage its own memory allocation. This
6166 parameter specifies the minimum percentage by which the garbage
6167 collector's heap should be allowed to expand between collections.
6168 Tuning this may improve compilation speed; it has no effect on code
6171 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
6172 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
6173 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
6174 GCC is not able to calculate RAM on a particular platform, the lower
6175 bound of 30% is used. Setting this parameter and
6176 @option{ggc-min-heapsize} to zero causes a full collection to occur at
6177 every opportunity. This is extremely slow, but can be useful for
6180 @item ggc-min-heapsize
6182 Minimum size of the garbage collector's heap before it begins bothering
6183 to collect garbage. The first collection occurs after the heap expands
6184 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
6185 tuning this may improve compilation speed, and has no effect on code
6188 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
6189 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
6190 with a lower bound of 4096 (four megabytes) and an upper bound of
6191 131072 (128 megabytes). If GCC is not able to calculate RAM on a
6192 particular platform, the lower bound is used. Setting this parameter
6193 very large effectively disables garbage collection. Setting this
6194 parameter and @option{ggc-min-expand} to zero causes a full collection
6195 to occur at every opportunity.
6197 @item max-reload-search-insns
6198 The maximum number of instruction reload should look backward for equivalent
6199 register. Increasing values mean more aggressive optimization, making the
6200 compile time increase with probably slightly better performance. The default
6203 @item max-cselib-memory-locations
6204 The maximum number of memory locations cselib should take into account.
6205 Increasing values mean more aggressive optimization, making the compile time
6206 increase with probably slightly better performance. The default value is 500.
6208 @item max-flow-memory-locations
6209 Similar as @option{max-cselib-memory-locations} but for dataflow liveness.
6210 The default value is 100.
6212 @item reorder-blocks-duplicate
6213 @itemx reorder-blocks-duplicate-feedback
6215 Used by basic block reordering pass to decide whether to use unconditional
6216 branch or duplicate the code on its destination. Code is duplicated when its
6217 estimated size is smaller than this value multiplied by the estimated size of
6218 unconditional jump in the hot spots of the program.
6220 The @option{reorder-block-duplicate-feedback} is used only when profile
6221 feedback is available and may be set to higher values than
6222 @option{reorder-block-duplicate} since information about the hot spots is more
6225 @item max-sched-ready-insns
6226 The maximum number of instructions ready to be issued the scheduler should
6227 consider at any given time during the first scheduling pass. Increasing
6228 values mean more thorough searches, making the compilation time increase
6229 with probably little benefit. The default value is 100.
6231 @item max-sched-region-blocks
6232 The maximum number of blocks in a region to be considered for
6233 interblock scheduling. The default value is 10.
6235 @item max-sched-region-insns
6236 The maximum number of insns in a region to be considered for
6237 interblock scheduling. The default value is 100.
6240 The minimum probability (in percents) of reaching a source block
6241 for interblock speculative scheduling. The default value is 40.
6243 @item max-sched-extend-regions-iters
6244 The maximum number of iterations through CFG to extend regions.
6245 0 - disable region extension,
6246 N - do at most N iterations.
6247 The default value is 0.
6249 @item max-sched-insn-conflict-delay
6250 The maximum conflict delay for an insn to be considered for speculative motion.
6251 The default value is 3.
6253 @item sched-spec-prob-cutoff
6254 The minimal probability of speculation success (in percents), so that
6255 speculative insn will be scheduled.
6256 The default value is 40.
6258 @item max-last-value-rtl
6260 The maximum size measured as number of RTLs that can be recorded in an expression
6261 in combiner for a pseudo register as last known value of that register. The default
6264 @item integer-share-limit
6265 Small integer constants can use a shared data structure, reducing the
6266 compiler's memory usage and increasing its speed. This sets the maximum
6267 value of a shared integer constant's. The default value is 256.
6269 @item min-virtual-mappings
6270 Specifies the minimum number of virtual mappings in the incremental
6271 SSA updater that should be registered to trigger the virtual mappings
6272 heuristic defined by virtual-mappings-ratio. The default value is
6275 @item virtual-mappings-ratio
6276 If the number of virtual mappings is virtual-mappings-ratio bigger
6277 than the number of virtual symbols to be updated, then the incremental
6278 SSA updater switches to a full update for those symbols. The default
6281 @item ssp-buffer-size
6282 The minimum size of buffers (i.e. arrays) that will receive stack smashing
6283 protection when @option{-fstack-protection} is used.
6285 @item max-jump-thread-duplication-stmts
6286 Maximum number of statements allowed in a block that needs to be
6287 duplicated when threading jumps.
6289 @item max-fields-for-field-sensitive
6290 Maximum number of fields in a structure we will treat in
6291 a field sensitive manner during pointer analysis.
6296 @node Preprocessor Options
6297 @section Options Controlling the Preprocessor
6298 @cindex preprocessor options
6299 @cindex options, preprocessor
6301 These options control the C preprocessor, which is run on each C source
6302 file before actual compilation.
6304 If you use the @option{-E} option, nothing is done except preprocessing.
6305 Some of these options make sense only together with @option{-E} because
6306 they cause the preprocessor output to be unsuitable for actual
6311 You can use @option{-Wp,@var{option}} to bypass the compiler driver
6312 and pass @var{option} directly through to the preprocessor. If
6313 @var{option} contains commas, it is split into multiple options at the
6314 commas. However, many options are modified, translated or interpreted
6315 by the compiler driver before being passed to the preprocessor, and
6316 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
6317 interface is undocumented and subject to change, so whenever possible
6318 you should avoid using @option{-Wp} and let the driver handle the
6321 @item -Xpreprocessor @var{option}
6322 @opindex preprocessor
6323 Pass @var{option} as an option to the preprocessor. You can use this to
6324 supply system-specific preprocessor options which GCC does not know how to
6327 If you want to pass an option that takes an argument, you must use
6328 @option{-Xpreprocessor} twice, once for the option and once for the argument.
6331 @include cppopts.texi
6333 @node Assembler Options
6334 @section Passing Options to the Assembler
6336 @c prevent bad page break with this line
6337 You can pass options to the assembler.
6340 @item -Wa,@var{option}
6342 Pass @var{option} as an option to the assembler. If @var{option}
6343 contains commas, it is split into multiple options at the commas.
6345 @item -Xassembler @var{option}
6347 Pass @var{option} as an option to the assembler. You can use this to
6348 supply system-specific assembler options which GCC does not know how to
6351 If you want to pass an option that takes an argument, you must use
6352 @option{-Xassembler} twice, once for the option and once for the argument.
6357 @section Options for Linking
6358 @cindex link options
6359 @cindex options, linking
6361 These options come into play when the compiler links object files into
6362 an executable output file. They are meaningless if the compiler is
6363 not doing a link step.
6367 @item @var{object-file-name}
6368 A file name that does not end in a special recognized suffix is
6369 considered to name an object file or library. (Object files are
6370 distinguished from libraries by the linker according to the file
6371 contents.) If linking is done, these object files are used as input
6380 If any of these options is used, then the linker is not run, and
6381 object file names should not be used as arguments. @xref{Overall
6385 @item -l@var{library}
6386 @itemx -l @var{library}
6388 Search the library named @var{library} when linking. (The second
6389 alternative with the library as a separate argument is only for
6390 POSIX compliance and is not recommended.)
6392 It makes a difference where in the command you write this option; the
6393 linker searches and processes libraries and object files in the order they
6394 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
6395 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
6396 to functions in @samp{z}, those functions may not be loaded.
6398 The linker searches a standard list of directories for the library,
6399 which is actually a file named @file{lib@var{library}.a}. The linker
6400 then uses this file as if it had been specified precisely by name.
6402 The directories searched include several standard system directories
6403 plus any that you specify with @option{-L}.
6405 Normally the files found this way are library files---archive files
6406 whose members are object files. The linker handles an archive file by
6407 scanning through it for members which define symbols that have so far
6408 been referenced but not defined. But if the file that is found is an
6409 ordinary object file, it is linked in the usual fashion. The only
6410 difference between using an @option{-l} option and specifying a file name
6411 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
6412 and searches several directories.
6416 You need this special case of the @option{-l} option in order to
6417 link an Objective-C or Objective-C++ program.
6420 @opindex nostartfiles
6421 Do not use the standard system startup files when linking.
6422 The standard system libraries are used normally, unless @option{-nostdlib}
6423 or @option{-nodefaultlibs} is used.
6425 @item -nodefaultlibs
6426 @opindex nodefaultlibs
6427 Do not use the standard system libraries when linking.
6428 Only the libraries you specify will be passed to the linker.
6429 The standard startup files are used normally, unless @option{-nostartfiles}
6430 is used. The compiler may generate calls to @code{memcmp},
6431 @code{memset}, @code{memcpy} and @code{memmove}.
6432 These entries are usually resolved by entries in
6433 libc. These entry points should be supplied through some other
6434 mechanism when this option is specified.
6438 Do not use the standard system startup files or libraries when linking.
6439 No startup files and only the libraries you specify will be passed to
6440 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
6441 @code{memcpy} and @code{memmove}.
6442 These entries are usually resolved by entries in
6443 libc. These entry points should be supplied through some other
6444 mechanism when this option is specified.
6446 @cindex @option{-lgcc}, use with @option{-nostdlib}
6447 @cindex @option{-nostdlib} and unresolved references
6448 @cindex unresolved references and @option{-nostdlib}
6449 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
6450 @cindex @option{-nodefaultlibs} and unresolved references
6451 @cindex unresolved references and @option{-nodefaultlibs}
6452 One of the standard libraries bypassed by @option{-nostdlib} and
6453 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
6454 that GCC uses to overcome shortcomings of particular machines, or special
6455 needs for some languages.
6456 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
6457 Collection (GCC) Internals},
6458 for more discussion of @file{libgcc.a}.)
6459 In most cases, you need @file{libgcc.a} even when you want to avoid
6460 other standard libraries. In other words, when you specify @option{-nostdlib}
6461 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
6462 This ensures that you have no unresolved references to internal GCC
6463 library subroutines. (For example, @samp{__main}, used to ensure C++
6464 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
6465 GNU Compiler Collection (GCC) Internals}.)
6469 Produce a position independent executable on targets which support it.
6470 For predictable results, you must also specify the same set of options
6471 that were used to generate code (@option{-fpie}, @option{-fPIE},
6472 or model suboptions) when you specify this option.
6476 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
6477 that support it. This instructs the linker to add all symbols, not
6478 only used ones, to the dynamic symbol table. This option is needed
6479 for some uses of @code{dlopen} or to allow obtaining backtraces
6480 from within a program.
6484 Remove all symbol table and relocation information from the executable.
6488 On systems that support dynamic linking, this prevents linking with the shared
6489 libraries. On other systems, this option has no effect.
6493 Produce a shared object which can then be linked with other objects to
6494 form an executable. Not all systems support this option. For predictable
6495 results, you must also specify the same set of options that were used to
6496 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
6497 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
6498 needs to build supplementary stub code for constructors to work. On
6499 multi-libbed systems, @samp{gcc -shared} must select the correct support
6500 libraries to link against. Failing to supply the correct flags may lead
6501 to subtle defects. Supplying them in cases where they are not necessary
6504 @item -shared-libgcc
6505 @itemx -static-libgcc
6506 @opindex shared-libgcc
6507 @opindex static-libgcc
6508 On systems that provide @file{libgcc} as a shared library, these options
6509 force the use of either the shared or static version respectively.
6510 If no shared version of @file{libgcc} was built when the compiler was
6511 configured, these options have no effect.
6513 There are several situations in which an application should use the
6514 shared @file{libgcc} instead of the static version. The most common
6515 of these is when the application wishes to throw and catch exceptions
6516 across different shared libraries. In that case, each of the libraries
6517 as well as the application itself should use the shared @file{libgcc}.
6519 Therefore, the G++ and GCJ drivers automatically add
6520 @option{-shared-libgcc} whenever you build a shared library or a main
6521 executable, because C++ and Java programs typically use exceptions, so
6522 this is the right thing to do.
6524 If, instead, you use the GCC driver to create shared libraries, you may
6525 find that they will not always be linked with the shared @file{libgcc}.
6526 If GCC finds, at its configuration time, that you have a non-GNU linker
6527 or a GNU linker that does not support option @option{--eh-frame-hdr},
6528 it will link the shared version of @file{libgcc} into shared libraries
6529 by default. Otherwise, it will take advantage of the linker and optimize
6530 away the linking with the shared version of @file{libgcc}, linking with
6531 the static version of libgcc by default. This allows exceptions to
6532 propagate through such shared libraries, without incurring relocation
6533 costs at library load time.
6535 However, if a library or main executable is supposed to throw or catch
6536 exceptions, you must link it using the G++ or GCJ driver, as appropriate
6537 for the languages used in the program, or using the option
6538 @option{-shared-libgcc}, such that it is linked with the shared
6543 Bind references to global symbols when building a shared object. Warn
6544 about any unresolved references (unless overridden by the link editor
6545 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
6548 @item -Xlinker @var{option}
6550 Pass @var{option} as an option to the linker. You can use this to
6551 supply system-specific linker options which GCC does not know how to
6554 If you want to pass an option that takes an argument, you must use
6555 @option{-Xlinker} twice, once for the option and once for the argument.
6556 For example, to pass @option{-assert definitions}, you must write
6557 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
6558 @option{-Xlinker "-assert definitions"}, because this passes the entire
6559 string as a single argument, which is not what the linker expects.
6561 @item -Wl,@var{option}
6563 Pass @var{option} as an option to the linker. If @var{option} contains
6564 commas, it is split into multiple options at the commas.
6566 @item -u @var{symbol}
6568 Pretend the symbol @var{symbol} is undefined, to force linking of
6569 library modules to define it. You can use @option{-u} multiple times with
6570 different symbols to force loading of additional library modules.
6573 @node Directory Options
6574 @section Options for Directory Search
6575 @cindex directory options
6576 @cindex options, directory search
6579 These options specify directories to search for header files, for
6580 libraries and for parts of the compiler:
6585 Add the directory @var{dir} to the head of the list of directories to be
6586 searched for header files. This can be used to override a system header
6587 file, substituting your own version, since these directories are
6588 searched before the system header file directories. However, you should
6589 not use this option to add directories that contain vendor-supplied
6590 system header files (use @option{-isystem} for that). If you use more than
6591 one @option{-I} option, the directories are scanned in left-to-right
6592 order; the standard system directories come after.
6594 If a standard system include directory, or a directory specified with
6595 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
6596 option will be ignored. The directory will still be searched but as a
6597 system directory at its normal position in the system include chain.
6598 This is to ensure that GCC's procedure to fix buggy system headers and
6599 the ordering for the include_next directive are not inadvertently changed.
6600 If you really need to change the search order for system directories,
6601 use the @option{-nostdinc} and/or @option{-isystem} options.
6603 @item -iquote@var{dir}
6605 Add the directory @var{dir} to the head of the list of directories to
6606 be searched for header files only for the case of @samp{#include
6607 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
6608 otherwise just like @option{-I}.
6612 Add directory @var{dir} to the list of directories to be searched
6615 @item -B@var{prefix}
6617 This option specifies where to find the executables, libraries,
6618 include files, and data files of the compiler itself.
6620 The compiler driver program runs one or more of the subprograms
6621 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6622 @var{prefix} as a prefix for each program it tries to run, both with and
6623 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6625 For each subprogram to be run, the compiler driver first tries the
6626 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6627 was not specified, the driver tries two standard prefixes, which are
6628 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6629 those results in a file name that is found, the unmodified program
6630 name is searched for using the directories specified in your
6631 @env{PATH} environment variable.
6633 The compiler will check to see if the path provided by the @option{-B}
6634 refers to a directory, and if necessary it will add a directory
6635 separator character at the end of the path.
6637 @option{-B} prefixes that effectively specify directory names also apply
6638 to libraries in the linker, because the compiler translates these
6639 options into @option{-L} options for the linker. They also apply to
6640 includes files in the preprocessor, because the compiler translates these
6641 options into @option{-isystem} options for the preprocessor. In this case,
6642 the compiler appends @samp{include} to the prefix.
6644 The run-time support file @file{libgcc.a} can also be searched for using
6645 the @option{-B} prefix, if needed. If it is not found there, the two
6646 standard prefixes above are tried, and that is all. The file is left
6647 out of the link if it is not found by those means.
6649 Another way to specify a prefix much like the @option{-B} prefix is to use
6650 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6653 As a special kludge, if the path provided by @option{-B} is
6654 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6655 9, then it will be replaced by @file{[dir/]include}. This is to help
6656 with boot-strapping the compiler.
6658 @item -specs=@var{file}
6660 Process @var{file} after the compiler reads in the standard @file{specs}
6661 file, in order to override the defaults that the @file{gcc} driver
6662 program uses when determining what switches to pass to @file{cc1},
6663 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6664 @option{-specs=@var{file}} can be specified on the command line, and they
6665 are processed in order, from left to right.
6667 @item --sysroot=@var{dir}
6669 Use @var{dir} as the logical root directory for headers and libraries.
6670 For example, if the compiler would normally search for headers in
6671 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
6672 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
6674 If you use both this option and the @option{-isysroot} option, then
6675 the @option{--sysroot} option will apply to libraries, but the
6676 @option{-isysroot} option will apply to header files.
6678 The GNU linker (beginning with version 2.16) has the necessary support
6679 for this option. If your linker does not support this option, the
6680 header file aspect of @option{--sysroot} will still work, but the
6681 library aspect will not.
6685 This option has been deprecated. Please use @option{-iquote} instead for
6686 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6687 Any directories you specify with @option{-I} options before the @option{-I-}
6688 option are searched only for the case of @samp{#include "@var{file}"};
6689 they are not searched for @samp{#include <@var{file}>}.
6691 If additional directories are specified with @option{-I} options after
6692 the @option{-I-}, these directories are searched for all @samp{#include}
6693 directives. (Ordinarily @emph{all} @option{-I} directories are used
6696 In addition, the @option{-I-} option inhibits the use of the current
6697 directory (where the current input file came from) as the first search
6698 directory for @samp{#include "@var{file}"}. There is no way to
6699 override this effect of @option{-I-}. With @option{-I.} you can specify
6700 searching the directory which was current when the compiler was
6701 invoked. That is not exactly the same as what the preprocessor does
6702 by default, but it is often satisfactory.
6704 @option{-I-} does not inhibit the use of the standard system directories
6705 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6712 @section Specifying subprocesses and the switches to pass to them
6715 @command{gcc} is a driver program. It performs its job by invoking a
6716 sequence of other programs to do the work of compiling, assembling and
6717 linking. GCC interprets its command-line parameters and uses these to
6718 deduce which programs it should invoke, and which command-line options
6719 it ought to place on their command lines. This behavior is controlled
6720 by @dfn{spec strings}. In most cases there is one spec string for each
6721 program that GCC can invoke, but a few programs have multiple spec
6722 strings to control their behavior. The spec strings built into GCC can
6723 be overridden by using the @option{-specs=} command-line switch to specify
6726 @dfn{Spec files} are plaintext files that are used to construct spec
6727 strings. They consist of a sequence of directives separated by blank
6728 lines. The type of directive is determined by the first non-whitespace
6729 character on the line and it can be one of the following:
6732 @item %@var{command}
6733 Issues a @var{command} to the spec file processor. The commands that can
6737 @item %include <@var{file}>
6739 Search for @var{file} and insert its text at the current point in the
6742 @item %include_noerr <@var{file}>
6743 @cindex %include_noerr
6744 Just like @samp{%include}, but do not generate an error message if the include
6745 file cannot be found.
6747 @item %rename @var{old_name} @var{new_name}
6749 Rename the spec string @var{old_name} to @var{new_name}.
6753 @item *[@var{spec_name}]:
6754 This tells the compiler to create, override or delete the named spec
6755 string. All lines after this directive up to the next directive or
6756 blank line are considered to be the text for the spec string. If this
6757 results in an empty string then the spec will be deleted. (Or, if the
6758 spec did not exist, then nothing will happened.) Otherwise, if the spec
6759 does not currently exist a new spec will be created. If the spec does
6760 exist then its contents will be overridden by the text of this
6761 directive, unless the first character of that text is the @samp{+}
6762 character, in which case the text will be appended to the spec.
6764 @item [@var{suffix}]:
6765 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6766 and up to the next directive or blank line are considered to make up the
6767 spec string for the indicated suffix. When the compiler encounters an
6768 input file with the named suffix, it will processes the spec string in
6769 order to work out how to compile that file. For example:
6776 This says that any input file whose name ends in @samp{.ZZ} should be
6777 passed to the program @samp{z-compile}, which should be invoked with the
6778 command-line switch @option{-input} and with the result of performing the
6779 @samp{%i} substitution. (See below.)
6781 As an alternative to providing a spec string, the text that follows a
6782 suffix directive can be one of the following:
6785 @item @@@var{language}
6786 This says that the suffix is an alias for a known @var{language}. This is
6787 similar to using the @option{-x} command-line switch to GCC to specify a
6788 language explicitly. For example:
6795 Says that .ZZ files are, in fact, C++ source files.
6798 This causes an error messages saying:
6801 @var{name} compiler not installed on this system.
6805 GCC already has an extensive list of suffixes built into it.
6806 This directive will add an entry to the end of the list of suffixes, but
6807 since the list is searched from the end backwards, it is effectively
6808 possible to override earlier entries using this technique.
6812 GCC has the following spec strings built into it. Spec files can
6813 override these strings or create their own. Note that individual
6814 targets can also add their own spec strings to this list.
6817 asm Options to pass to the assembler
6818 asm_final Options to pass to the assembler post-processor
6819 cpp Options to pass to the C preprocessor
6820 cc1 Options to pass to the C compiler
6821 cc1plus Options to pass to the C++ compiler
6822 endfile Object files to include at the end of the link
6823 link Options to pass to the linker
6824 lib Libraries to include on the command line to the linker
6825 libgcc Decides which GCC support library to pass to the linker
6826 linker Sets the name of the linker
6827 predefines Defines to be passed to the C preprocessor
6828 signed_char Defines to pass to CPP to say whether @code{char} is signed
6830 startfile Object files to include at the start of the link
6833 Here is a small example of a spec file:
6839 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6842 This example renames the spec called @samp{lib} to @samp{old_lib} and
6843 then overrides the previous definition of @samp{lib} with a new one.
6844 The new definition adds in some extra command-line options before
6845 including the text of the old definition.
6847 @dfn{Spec strings} are a list of command-line options to be passed to their
6848 corresponding program. In addition, the spec strings can contain
6849 @samp{%}-prefixed sequences to substitute variable text or to
6850 conditionally insert text into the command line. Using these constructs
6851 it is possible to generate quite complex command lines.
6853 Here is a table of all defined @samp{%}-sequences for spec
6854 strings. Note that spaces are not generated automatically around the
6855 results of expanding these sequences. Therefore you can concatenate them
6856 together or combine them with constant text in a single argument.
6860 Substitute one @samp{%} into the program name or argument.
6863 Substitute the name of the input file being processed.
6866 Substitute the basename of the input file being processed.
6867 This is the substring up to (and not including) the last period
6868 and not including the directory.
6871 This is the same as @samp{%b}, but include the file suffix (text after
6875 Marks the argument containing or following the @samp{%d} as a
6876 temporary file name, so that that file will be deleted if GCC exits
6877 successfully. Unlike @samp{%g}, this contributes no text to the
6880 @item %g@var{suffix}
6881 Substitute a file name that has suffix @var{suffix} and is chosen
6882 once per compilation, and mark the argument in the same way as
6883 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6884 name is now chosen in a way that is hard to predict even when previously
6885 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6886 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6887 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6888 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6889 was simply substituted with a file name chosen once per compilation,
6890 without regard to any appended suffix (which was therefore treated
6891 just like ordinary text), making such attacks more likely to succeed.
6893 @item %u@var{suffix}
6894 Like @samp{%g}, but generates a new temporary file name even if
6895 @samp{%u@var{suffix}} was already seen.
6897 @item %U@var{suffix}
6898 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6899 new one if there is no such last file name. In the absence of any
6900 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6901 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6902 would involve the generation of two distinct file names, one
6903 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6904 simply substituted with a file name chosen for the previous @samp{%u},
6905 without regard to any appended suffix.
6907 @item %j@var{suffix}
6908 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6909 writable, and if save-temps is off; otherwise, substitute the name
6910 of a temporary file, just like @samp{%u}. This temporary file is not
6911 meant for communication between processes, but rather as a junk
6914 @item %|@var{suffix}
6915 @itemx %m@var{suffix}
6916 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6917 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6918 all. These are the two most common ways to instruct a program that it
6919 should read from standard input or write to standard output. If you
6920 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6921 construct: see for example @file{f/lang-specs.h}.
6923 @item %.@var{SUFFIX}
6924 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6925 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6926 terminated by the next space or %.
6929 Marks the argument containing or following the @samp{%w} as the
6930 designated output file of this compilation. This puts the argument
6931 into the sequence of arguments that @samp{%o} will substitute later.
6934 Substitutes the names of all the output files, with spaces
6935 automatically placed around them. You should write spaces
6936 around the @samp{%o} as well or the results are undefined.
6937 @samp{%o} is for use in the specs for running the linker.
6938 Input files whose names have no recognized suffix are not compiled
6939 at all, but they are included among the output files, so they will
6943 Substitutes the suffix for object files. Note that this is
6944 handled specially when it immediately follows @samp{%g, %u, or %U},
6945 because of the need for those to form complete file names. The
6946 handling is such that @samp{%O} is treated exactly as if it had already
6947 been substituted, except that @samp{%g, %u, and %U} do not currently
6948 support additional @var{suffix} characters following @samp{%O} as they would
6949 following, for example, @samp{.o}.
6952 Substitutes the standard macro predefinitions for the
6953 current target machine. Use this when running @code{cpp}.
6956 Like @samp{%p}, but puts @samp{__} before and after the name of each
6957 predefined macro, except for macros that start with @samp{__} or with
6958 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6962 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6963 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
6964 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6965 and @option{-imultilib} as necessary.
6968 Current argument is the name of a library or startup file of some sort.
6969 Search for that file in a standard list of directories and substitute
6970 the full name found.
6973 Print @var{str} as an error message. @var{str} is terminated by a newline.
6974 Use this when inconsistent options are detected.
6977 Substitute the contents of spec string @var{name} at this point.
6980 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6982 @item %x@{@var{option}@}
6983 Accumulate an option for @samp{%X}.
6986 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6990 Output the accumulated assembler options specified by @option{-Wa}.
6993 Output the accumulated preprocessor options specified by @option{-Wp}.
6996 Process the @code{asm} spec. This is used to compute the
6997 switches to be passed to the assembler.
7000 Process the @code{asm_final} spec. This is a spec string for
7001 passing switches to an assembler post-processor, if such a program is
7005 Process the @code{link} spec. This is the spec for computing the
7006 command line passed to the linker. Typically it will make use of the
7007 @samp{%L %G %S %D and %E} sequences.
7010 Dump out a @option{-L} option for each directory that GCC believes might
7011 contain startup files. If the target supports multilibs then the
7012 current multilib directory will be prepended to each of these paths.
7015 Process the @code{lib} spec. This is a spec string for deciding which
7016 libraries should be included on the command line to the linker.
7019 Process the @code{libgcc} spec. This is a spec string for deciding
7020 which GCC support library should be included on the command line to the linker.
7023 Process the @code{startfile} spec. This is a spec for deciding which
7024 object files should be the first ones passed to the linker. Typically
7025 this might be a file named @file{crt0.o}.
7028 Process the @code{endfile} spec. This is a spec string that specifies
7029 the last object files that will be passed to the linker.
7032 Process the @code{cpp} spec. This is used to construct the arguments
7033 to be passed to the C preprocessor.
7036 Process the @code{cc1} spec. This is used to construct the options to be
7037 passed to the actual C compiler (@samp{cc1}).
7040 Process the @code{cc1plus} spec. This is used to construct the options to be
7041 passed to the actual C++ compiler (@samp{cc1plus}).
7044 Substitute the variable part of a matched option. See below.
7045 Note that each comma in the substituted string is replaced by
7049 Remove all occurrences of @code{-S} from the command line. Note---this
7050 command is position dependent. @samp{%} commands in the spec string
7051 before this one will see @code{-S}, @samp{%} commands in the spec string
7052 after this one will not.
7054 @item %:@var{function}(@var{args})
7055 Call the named function @var{function}, passing it @var{args}.
7056 @var{args} is first processed as a nested spec string, then split
7057 into an argument vector in the usual fashion. The function returns
7058 a string which is processed as if it had appeared literally as part
7059 of the current spec.
7061 The following built-in spec functions are provided:
7064 @item @code{if-exists}
7065 The @code{if-exists} spec function takes one argument, an absolute
7066 pathname to a file. If the file exists, @code{if-exists} returns the
7067 pathname. Here is a small example of its usage:
7071 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
7074 @item @code{if-exists-else}
7075 The @code{if-exists-else} spec function is similar to the @code{if-exists}
7076 spec function, except that it takes two arguments. The first argument is
7077 an absolute pathname to a file. If the file exists, @code{if-exists-else}
7078 returns the pathname. If it does not exist, it returns the second argument.
7079 This way, @code{if-exists-else} can be used to select one file or another,
7080 based on the existence of the first. Here is a small example of its usage:
7084 crt0%O%s %:if-exists(crti%O%s) \
7085 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
7088 @item @code{replace-outfile}
7089 The @code{replace-outfile} spec function takes two arguments. It looks for the
7090 first argument in the outfiles array and replaces it with the second argument. Here
7091 is a small example of its usage:
7094 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
7100 Substitutes the @code{-S} switch, if that switch was given to GCC@.
7101 If that switch was not specified, this substitutes nothing. Note that
7102 the leading dash is omitted when specifying this option, and it is
7103 automatically inserted if the substitution is performed. Thus the spec
7104 string @samp{%@{foo@}} would match the command-line option @option{-foo}
7105 and would output the command line option @option{-foo}.
7107 @item %W@{@code{S}@}
7108 Like %@{@code{S}@} but mark last argument supplied within as a file to be
7111 @item %@{@code{S}*@}
7112 Substitutes all the switches specified to GCC whose names start
7113 with @code{-S}, but which also take an argument. This is used for
7114 switches like @option{-o}, @option{-D}, @option{-I}, etc.
7115 GCC considers @option{-o foo} as being
7116 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
7117 text, including the space. Thus two arguments would be generated.
7119 @item %@{@code{S}*&@code{T}*@}
7120 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
7121 (the order of @code{S} and @code{T} in the spec is not significant).
7122 There can be any number of ampersand-separated variables; for each the
7123 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
7125 @item %@{@code{S}:@code{X}@}
7126 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
7128 @item %@{!@code{S}:@code{X}@}
7129 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
7131 @item %@{@code{S}*:@code{X}@}
7132 Substitutes @code{X} if one or more switches whose names start with
7133 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
7134 once, no matter how many such switches appeared. However, if @code{%*}
7135 appears somewhere in @code{X}, then @code{X} will be substituted once
7136 for each matching switch, with the @code{%*} replaced by the part of
7137 that switch that matched the @code{*}.
7139 @item %@{.@code{S}:@code{X}@}
7140 Substitutes @code{X}, if processing a file with suffix @code{S}.
7142 @item %@{!.@code{S}:@code{X}@}
7143 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
7145 @item %@{@code{S}|@code{P}:@code{X}@}
7146 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
7147 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
7148 although they have a stronger binding than the @samp{|}. If @code{%*}
7149 appears in @code{X}, all of the alternatives must be starred, and only
7150 the first matching alternative is substituted.
7152 For example, a spec string like this:
7155 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
7158 will output the following command-line options from the following input
7159 command-line options:
7164 -d fred.c -foo -baz -boggle
7165 -d jim.d -bar -baz -boggle
7168 @item %@{S:X; T:Y; :D@}
7170 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
7171 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
7172 be as many clauses as you need. This may be combined with @code{.},
7173 @code{!}, @code{|}, and @code{*} as needed.
7178 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
7179 construct may contain other nested @samp{%} constructs or spaces, or
7180 even newlines. They are processed as usual, as described above.
7181 Trailing white space in @code{X} is ignored. White space may also
7182 appear anywhere on the left side of the colon in these constructs,
7183 except between @code{.} or @code{*} and the corresponding word.
7185 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
7186 handled specifically in these constructs. If another value of
7187 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
7188 @option{-W} switch is found later in the command line, the earlier
7189 switch value is ignored, except with @{@code{S}*@} where @code{S} is
7190 just one letter, which passes all matching options.
7192 The character @samp{|} at the beginning of the predicate text is used to
7193 indicate that a command should be piped to the following command, but
7194 only if @option{-pipe} is specified.
7196 It is built into GCC which switches take arguments and which do not.
7197 (You might think it would be useful to generalize this to allow each
7198 compiler's spec to say which switches take arguments. But this cannot
7199 be done in a consistent fashion. GCC cannot even decide which input
7200 files have been specified without knowing which switches take arguments,
7201 and it must know which input files to compile in order to tell which
7204 GCC also knows implicitly that arguments starting in @option{-l} are to be
7205 treated as compiler output files, and passed to the linker in their
7206 proper position among the other output files.
7208 @c man begin OPTIONS
7210 @node Target Options
7211 @section Specifying Target Machine and Compiler Version
7212 @cindex target options
7213 @cindex cross compiling
7214 @cindex specifying machine version
7215 @cindex specifying compiler version and target machine
7216 @cindex compiler version, specifying
7217 @cindex target machine, specifying
7219 The usual way to run GCC is to run the executable called @file{gcc}, or
7220 @file{<machine>-gcc} when cross-compiling, or
7221 @file{<machine>-gcc-<version>} to run a version other than the one that
7222 was installed last. Sometimes this is inconvenient, so GCC provides
7223 options that will switch to another cross-compiler or version.
7226 @item -b @var{machine}
7228 The argument @var{machine} specifies the target machine for compilation.
7230 The value to use for @var{machine} is the same as was specified as the
7231 machine type when configuring GCC as a cross-compiler. For
7232 example, if a cross-compiler was configured with @samp{configure
7233 arm-elf}, meaning to compile for an arm processor with elf binaries,
7234 then you would specify @option{-b arm-elf} to run that cross compiler.
7235 Because there are other options beginning with @option{-b}, the
7236 configuration must contain a hyphen.
7238 @item -V @var{version}
7240 The argument @var{version} specifies which version of GCC to run.
7241 This is useful when multiple versions are installed. For example,
7242 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
7245 The @option{-V} and @option{-b} options work by running the
7246 @file{<machine>-gcc-<version>} executable, so there's no real reason to
7247 use them if you can just run that directly.
7249 @node Submodel Options
7250 @section Hardware Models and Configurations
7251 @cindex submodel options
7252 @cindex specifying hardware config
7253 @cindex hardware models and configurations, specifying
7254 @cindex machine dependent options
7256 Earlier we discussed the standard option @option{-b} which chooses among
7257 different installed compilers for completely different target
7258 machines, such as VAX vs.@: 68000 vs.@: 80386.
7260 In addition, each of these target machine types can have its own
7261 special options, starting with @samp{-m}, to choose among various
7262 hardware models or configurations---for example, 68010 vs 68020,
7263 floating coprocessor or none. A single installed version of the
7264 compiler can compile for any model or configuration, according to the
7267 Some configurations of the compiler also support additional special
7268 options, usually for compatibility with other compilers on the same
7271 @c This list is ordered alphanumerically by subsection name.
7272 @c It should be the same order and spelling as these options are listed
7273 @c in Machine Dependent Options
7279 * Blackfin Options::
7283 * DEC Alpha Options::
7284 * DEC Alpha/VMS Options::
7286 * GNU/Linux Options::
7289 * i386 and x86-64 Options::
7302 * RS/6000 and PowerPC Options::
7303 * S/390 and zSeries Options::
7306 * System V Options::
7307 * TMS320C3x/C4x Options::
7311 * Xstormy16 Options::
7317 @subsection ARC Options
7320 These options are defined for ARC implementations:
7325 Compile code for little endian mode. This is the default.
7329 Compile code for big endian mode.
7332 @opindex mmangle-cpu
7333 Prepend the name of the cpu to all public symbol names.
7334 In multiple-processor systems, there are many ARC variants with different
7335 instruction and register set characteristics. This flag prevents code
7336 compiled for one cpu to be linked with code compiled for another.
7337 No facility exists for handling variants that are ``almost identical''.
7338 This is an all or nothing option.
7340 @item -mcpu=@var{cpu}
7342 Compile code for ARC variant @var{cpu}.
7343 Which variants are supported depend on the configuration.
7344 All variants support @option{-mcpu=base}, this is the default.
7346 @item -mtext=@var{text-section}
7347 @itemx -mdata=@var{data-section}
7348 @itemx -mrodata=@var{readonly-data-section}
7352 Put functions, data, and readonly data in @var{text-section},
7353 @var{data-section}, and @var{readonly-data-section} respectively
7354 by default. This can be overridden with the @code{section} attribute.
7355 @xref{Variable Attributes}.
7360 @subsection ARM Options
7363 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
7367 @item -mabi=@var{name}
7369 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
7370 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
7373 @opindex mapcs-frame
7374 Generate a stack frame that is compliant with the ARM Procedure Call
7375 Standard for all functions, even if this is not strictly necessary for
7376 correct execution of the code. Specifying @option{-fomit-frame-pointer}
7377 with this option will cause the stack frames not to be generated for
7378 leaf functions. The default is @option{-mno-apcs-frame}.
7382 This is a synonym for @option{-mapcs-frame}.
7385 @c not currently implemented
7386 @item -mapcs-stack-check
7387 @opindex mapcs-stack-check
7388 Generate code to check the amount of stack space available upon entry to
7389 every function (that actually uses some stack space). If there is
7390 insufficient space available then either the function
7391 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
7392 called, depending upon the amount of stack space required. The run time
7393 system is required to provide these functions. The default is
7394 @option{-mno-apcs-stack-check}, since this produces smaller code.
7396 @c not currently implemented
7398 @opindex mapcs-float
7399 Pass floating point arguments using the float point registers. This is
7400 one of the variants of the APCS@. This option is recommended if the
7401 target hardware has a floating point unit or if a lot of floating point
7402 arithmetic is going to be performed by the code. The default is
7403 @option{-mno-apcs-float}, since integer only code is slightly increased in
7404 size if @option{-mapcs-float} is used.
7406 @c not currently implemented
7407 @item -mapcs-reentrant
7408 @opindex mapcs-reentrant
7409 Generate reentrant, position independent code. The default is
7410 @option{-mno-apcs-reentrant}.
7413 @item -mthumb-interwork
7414 @opindex mthumb-interwork
7415 Generate code which supports calling between the ARM and Thumb
7416 instruction sets. Without this option the two instruction sets cannot
7417 be reliably used inside one program. The default is
7418 @option{-mno-thumb-interwork}, since slightly larger code is generated
7419 when @option{-mthumb-interwork} is specified.
7421 @item -mno-sched-prolog
7422 @opindex mno-sched-prolog
7423 Prevent the reordering of instructions in the function prolog, or the
7424 merging of those instruction with the instructions in the function's
7425 body. This means that all functions will start with a recognizable set
7426 of instructions (or in fact one of a choice from a small set of
7427 different function prologues), and this information can be used to
7428 locate the start if functions inside an executable piece of code. The
7429 default is @option{-msched-prolog}.
7432 @opindex mhard-float
7433 Generate output containing floating point instructions. This is the
7437 @opindex msoft-float
7438 Generate output containing library calls for floating point.
7439 @strong{Warning:} the requisite libraries are not available for all ARM
7440 targets. Normally the facilities of the machine's usual C compiler are
7441 used, but this cannot be done directly in cross-compilation. You must make
7442 your own arrangements to provide suitable library functions for
7445 @option{-msoft-float} changes the calling convention in the output file;
7446 therefore, it is only useful if you compile @emph{all} of a program with
7447 this option. In particular, you need to compile @file{libgcc.a}, the
7448 library that comes with GCC, with @option{-msoft-float} in order for
7451 @item -mfloat-abi=@var{name}
7453 Specifies which ABI to use for floating point values. Permissible values
7454 are: @samp{soft}, @samp{softfp} and @samp{hard}.
7456 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
7457 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
7458 of floating point instructions, but still uses the soft-float calling
7461 @item -mlittle-endian
7462 @opindex mlittle-endian
7463 Generate code for a processor running in little-endian mode. This is
7464 the default for all standard configurations.
7467 @opindex mbig-endian
7468 Generate code for a processor running in big-endian mode; the default is
7469 to compile code for a little-endian processor.
7471 @item -mwords-little-endian
7472 @opindex mwords-little-endian
7473 This option only applies when generating code for big-endian processors.
7474 Generate code for a little-endian word order but a big-endian byte
7475 order. That is, a byte order of the form @samp{32107654}. Note: this
7476 option should only be used if you require compatibility with code for
7477 big-endian ARM processors generated by versions of the compiler prior to
7480 @item -mcpu=@var{name}
7482 This specifies the name of the target ARM processor. GCC uses this name
7483 to determine what kind of instructions it can emit when generating
7484 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
7485 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
7486 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
7487 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
7488 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
7489 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
7490 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
7491 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
7492 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
7493 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
7494 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
7495 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
7496 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
7497 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
7500 @itemx -mtune=@var{name}
7502 This option is very similar to the @option{-mcpu=} option, except that
7503 instead of specifying the actual target processor type, and hence
7504 restricting which instructions can be used, it specifies that GCC should
7505 tune the performance of the code as if the target were of the type
7506 specified in this option, but still choosing the instructions that it
7507 will generate based on the cpu specified by a @option{-mcpu=} option.
7508 For some ARM implementations better performance can be obtained by using
7511 @item -march=@var{name}
7513 This specifies the name of the target ARM architecture. GCC uses this
7514 name to determine what kind of instructions it can emit when generating
7515 assembly code. This option can be used in conjunction with or instead
7516 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
7517 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
7518 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
7519 @samp{iwmmxt}, @samp{ep9312}.
7521 @item -mfpu=@var{name}
7522 @itemx -mfpe=@var{number}
7523 @itemx -mfp=@var{number}
7527 This specifies what floating point hardware (or hardware emulation) is
7528 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
7529 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
7530 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
7531 with older versions of GCC@.
7533 If @option{-msoft-float} is specified this specifies the format of
7534 floating point values.
7536 @item -mstructure-size-boundary=@var{n}
7537 @opindex mstructure-size-boundary
7538 The size of all structures and unions will be rounded up to a multiple
7539 of the number of bits set by this option. Permissible values are 8, 32
7540 and 64. The default value varies for different toolchains. For the COFF
7541 targeted toolchain the default value is 8. A value of 64 is only allowed
7542 if the underlying ABI supports it.
7544 Specifying the larger number can produce faster, more efficient code, but
7545 can also increase the size of the program. Different values are potentially
7546 incompatible. Code compiled with one value cannot necessarily expect to
7547 work with code or libraries compiled with another value, if they exchange
7548 information using structures or unions.
7550 @item -mabort-on-noreturn
7551 @opindex mabort-on-noreturn
7552 Generate a call to the function @code{abort} at the end of a
7553 @code{noreturn} function. It will be executed if the function tries to
7557 @itemx -mno-long-calls
7558 @opindex mlong-calls
7559 @opindex mno-long-calls
7560 Tells the compiler to perform function calls by first loading the
7561 address of the function into a register and then performing a subroutine
7562 call on this register. This switch is needed if the target function
7563 will lie outside of the 64 megabyte addressing range of the offset based
7564 version of subroutine call instruction.
7566 Even if this switch is enabled, not all function calls will be turned
7567 into long calls. The heuristic is that static functions, functions
7568 which have the @samp{short-call} attribute, functions that are inside
7569 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7570 definitions have already been compiled within the current compilation
7571 unit, will not be turned into long calls. The exception to this rule is
7572 that weak function definitions, functions with the @samp{long-call}
7573 attribute or the @samp{section} attribute, and functions that are within
7574 the scope of a @samp{#pragma long_calls} directive, will always be
7575 turned into long calls.
7577 This feature is not enabled by default. Specifying
7578 @option{-mno-long-calls} will restore the default behavior, as will
7579 placing the function calls within the scope of a @samp{#pragma
7580 long_calls_off} directive. Note these switches have no effect on how
7581 the compiler generates code to handle function calls via function
7584 @item -mnop-fun-dllimport
7585 @opindex mnop-fun-dllimport
7586 Disable support for the @code{dllimport} attribute.
7588 @item -msingle-pic-base
7589 @opindex msingle-pic-base
7590 Treat the register used for PIC addressing as read-only, rather than
7591 loading it in the prologue for each function. The run-time system is
7592 responsible for initializing this register with an appropriate value
7593 before execution begins.
7595 @item -mpic-register=@var{reg}
7596 @opindex mpic-register
7597 Specify the register to be used for PIC addressing. The default is R10
7598 unless stack-checking is enabled, when R9 is used.
7600 @item -mcirrus-fix-invalid-insns
7601 @opindex mcirrus-fix-invalid-insns
7602 @opindex mno-cirrus-fix-invalid-insns
7603 Insert NOPs into the instruction stream to in order to work around
7604 problems with invalid Maverick instruction combinations. This option
7605 is only valid if the @option{-mcpu=ep9312} option has been used to
7606 enable generation of instructions for the Cirrus Maverick floating
7607 point co-processor. This option is not enabled by default, since the
7608 problem is only present in older Maverick implementations. The default
7609 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7612 @item -mpoke-function-name
7613 @opindex mpoke-function-name
7614 Write the name of each function into the text section, directly
7615 preceding the function prologue. The generated code is similar to this:
7619 .ascii "arm_poke_function_name", 0
7622 .word 0xff000000 + (t1 - t0)
7623 arm_poke_function_name
7625 stmfd sp!, @{fp, ip, lr, pc@}
7629 When performing a stack backtrace, code can inspect the value of
7630 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7631 location @code{pc - 12} and the top 8 bits are set, then we know that
7632 there is a function name embedded immediately preceding this location
7633 and has length @code{((pc[-3]) & 0xff000000)}.
7637 Generate code for the 16-bit Thumb instruction set. The default is to
7638 use the 32-bit ARM instruction set.
7641 @opindex mtpcs-frame
7642 Generate a stack frame that is compliant with the Thumb Procedure Call
7643 Standard for all non-leaf functions. (A leaf function is one that does
7644 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7646 @item -mtpcs-leaf-frame
7647 @opindex mtpcs-leaf-frame
7648 Generate a stack frame that is compliant with the Thumb Procedure Call
7649 Standard for all leaf functions. (A leaf function is one that does
7650 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7652 @item -mcallee-super-interworking
7653 @opindex mcallee-super-interworking
7654 Gives all externally visible functions in the file being compiled an ARM
7655 instruction set header which switches to Thumb mode before executing the
7656 rest of the function. This allows these functions to be called from
7657 non-interworking code.
7659 @item -mcaller-super-interworking
7660 @opindex mcaller-super-interworking
7661 Allows calls via function pointers (including virtual functions) to
7662 execute correctly regardless of whether the target code has been
7663 compiled for interworking or not. There is a small overhead in the cost
7664 of executing a function pointer if this option is enabled.
7666 @item -mtp=@var{name}
7668 Specify the access model for the thread local storage pointer. The valid
7669 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
7670 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
7671 (supported in the arm6k architecture), and @option{auto}, which uses the
7672 best available method for the selected processor. The default setting is
7678 @subsection AVR Options
7681 These options are defined for AVR implementations:
7684 @item -mmcu=@var{mcu}
7686 Specify ATMEL AVR instruction set or MCU type.
7688 Instruction set avr1 is for the minimal AVR core, not supported by the C
7689 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7690 attiny11, attiny12, attiny15, attiny28).
7692 Instruction set avr2 (default) is for the classic AVR core with up to
7693 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7694 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7695 at90c8534, at90s8535).
7697 Instruction set avr3 is for the classic AVR core with up to 128K program
7698 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7700 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7701 memory space (MCU types: atmega8, atmega83, atmega85).
7703 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7704 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7705 atmega64, atmega128, at43usb355, at94k).
7709 Output instruction sizes to the asm file.
7711 @item -minit-stack=@var{N}
7712 @opindex minit-stack
7713 Specify the initial stack address, which may be a symbol or numeric value,
7714 @samp{__stack} is the default.
7716 @item -mno-interrupts
7717 @opindex mno-interrupts
7718 Generated code is not compatible with hardware interrupts.
7719 Code size will be smaller.
7721 @item -mcall-prologues
7722 @opindex mcall-prologues
7723 Functions prologues/epilogues expanded as call to appropriate
7724 subroutines. Code size will be smaller.
7726 @item -mno-tablejump
7727 @opindex mno-tablejump
7728 Do not generate tablejump insns which sometimes increase code size.
7731 @opindex mtiny-stack
7732 Change only the low 8 bits of the stack pointer.
7736 Assume int to be 8 bit integer. This affects the sizes of all types: A
7737 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7738 and long long will be 4 bytes. Please note that this option does not
7739 comply to the C standards, but it will provide you with smaller code
7743 @node Blackfin Options
7744 @subsection Blackfin Options
7745 @cindex Blackfin Options
7748 @item -momit-leaf-frame-pointer
7749 @opindex momit-leaf-frame-pointer
7750 Don't keep the frame pointer in a register for leaf functions. This
7751 avoids the instructions to save, set up and restore frame pointers and
7752 makes an extra register available in leaf functions. The option
7753 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7754 which might make debugging harder.
7756 @item -mspecld-anomaly
7757 @opindex mspecld-anomaly
7758 When enabled, the compiler will ensure that the generated code does not
7759 contain speculative loads after jump instructions. This option is enabled
7762 @item -mno-specld-anomaly
7763 @opindex mno-specld-anomaly
7764 Don't generate extra code to prevent speculative loads from occurring.
7766 @item -mcsync-anomaly
7767 @opindex mcsync-anomaly
7768 When enabled, the compiler will ensure that the generated code does not
7769 contain CSYNC or SSYNC instructions too soon after conditional branches.
7770 This option is enabled by default.
7772 @item -mno-csync-anomaly
7773 @opindex mno-csync-anomaly
7774 Don't generate extra code to prevent CSYNC or SSYNC instructions from
7775 occurring too soon after a conditional branch.
7779 When enabled, the compiler is free to take advantage of the knowledge that
7780 the entire program fits into the low 64k of memory.
7783 @opindex mno-low-64k
7784 Assume that the program is arbitrarily large. This is the default.
7786 @item -mid-shared-library
7787 @opindex mid-shared-library
7788 Generate code that supports shared libraries via the library ID method.
7789 This allows for execute in place and shared libraries in an environment
7790 without virtual memory management. This option implies @option{-fPIC}.
7792 @item -mno-id-shared-library
7793 @opindex mno-id-shared-library
7794 Generate code that doesn't assume ID based shared libraries are being used.
7795 This is the default.
7797 @item -mshared-library-id=n
7798 @opindex mshared-library-id
7799 Specified the identification number of the ID based shared library being
7800 compiled. Specifying a value of 0 will generate more compact code, specifying
7801 other values will force the allocation of that number to the current
7802 library but is no more space or time efficient than omitting this option.
7805 @itemx -mno-long-calls
7806 @opindex mlong-calls
7807 @opindex mno-long-calls
7808 Tells the compiler to perform function calls by first loading the
7809 address of the function into a register and then performing a subroutine
7810 call on this register. This switch is needed if the target function
7811 will lie outside of the 24 bit addressing range of the offset based
7812 version of subroutine call instruction.
7814 This feature is not enabled by default. Specifying
7815 @option{-mno-long-calls} will restore the default behavior. Note these
7816 switches have no effect on how the compiler generates code to handle
7817 function calls via function pointers.
7821 @subsection CRIS Options
7822 @cindex CRIS Options
7824 These options are defined specifically for the CRIS ports.
7827 @item -march=@var{architecture-type}
7828 @itemx -mcpu=@var{architecture-type}
7831 Generate code for the specified architecture. The choices for
7832 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7833 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7834 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7837 @item -mtune=@var{architecture-type}
7839 Tune to @var{architecture-type} everything applicable about the generated
7840 code, except for the ABI and the set of available instructions. The
7841 choices for @var{architecture-type} are the same as for
7842 @option{-march=@var{architecture-type}}.
7844 @item -mmax-stack-frame=@var{n}
7845 @opindex mmax-stack-frame
7846 Warn when the stack frame of a function exceeds @var{n} bytes.
7848 @item -melinux-stacksize=@var{n}
7849 @opindex melinux-stacksize
7850 Only available with the @samp{cris-axis-aout} target. Arranges for
7851 indications in the program to the kernel loader that the stack of the
7852 program should be set to @var{n} bytes.
7858 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7859 @option{-march=v3} and @option{-march=v8} respectively.
7861 @item -mmul-bug-workaround
7862 @itemx -mno-mul-bug-workaround
7863 @opindex mmul-bug-workaround
7864 @opindex mno-mul-bug-workaround
7865 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7866 models where it applies. This option is active by default.
7870 Enable CRIS-specific verbose debug-related information in the assembly
7871 code. This option also has the effect to turn off the @samp{#NO_APP}
7872 formatted-code indicator to the assembler at the beginning of the
7877 Do not use condition-code results from previous instruction; always emit
7878 compare and test instructions before use of condition codes.
7880 @item -mno-side-effects
7881 @opindex mno-side-effects
7882 Do not emit instructions with side-effects in addressing modes other than
7886 @itemx -mno-stack-align
7888 @itemx -mno-data-align
7889 @itemx -mconst-align
7890 @itemx -mno-const-align
7891 @opindex mstack-align
7892 @opindex mno-stack-align
7893 @opindex mdata-align
7894 @opindex mno-data-align
7895 @opindex mconst-align
7896 @opindex mno-const-align
7897 These options (no-options) arranges (eliminate arrangements) for the
7898 stack-frame, individual data and constants to be aligned for the maximum
7899 single data access size for the chosen CPU model. The default is to
7900 arrange for 32-bit alignment. ABI details such as structure layout are
7901 not affected by these options.
7909 Similar to the stack- data- and const-align options above, these options
7910 arrange for stack-frame, writable data and constants to all be 32-bit,
7911 16-bit or 8-bit aligned. The default is 32-bit alignment.
7913 @item -mno-prologue-epilogue
7914 @itemx -mprologue-epilogue
7915 @opindex mno-prologue-epilogue
7916 @opindex mprologue-epilogue
7917 With @option{-mno-prologue-epilogue}, the normal function prologue and
7918 epilogue that sets up the stack-frame are omitted and no return
7919 instructions or return sequences are generated in the code. Use this
7920 option only together with visual inspection of the compiled code: no
7921 warnings or errors are generated when call-saved registers must be saved,
7922 or storage for local variable needs to be allocated.
7928 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7929 instruction sequences that load addresses for functions from the PLT part
7930 of the GOT rather than (traditional on other architectures) calls to the
7931 PLT@. The default is @option{-mgotplt}.
7935 Legacy no-op option only recognized with the cris-axis-aout target.
7939 Legacy no-op option only recognized with the cris-axis-elf and
7940 cris-axis-linux-gnu targets.
7944 Only recognized with the cris-axis-aout target, where it selects a
7945 GNU/linux-like multilib, include files and instruction set for
7950 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7954 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7955 to link with input-output functions from a simulator library. Code,
7956 initialized data and zero-initialized data are allocated consecutively.
7960 Like @option{-sim}, but pass linker options to locate initialized data at
7961 0x40000000 and zero-initialized data at 0x80000000.
7965 @subsection CRX Options
7968 These options are defined specifically for the CRX ports.
7974 Enable the use of multiply-accumulate instructions. Disabled by default.
7978 Push instructions will be used to pass outgoing arguments when functions
7979 are called. Enabled by default.
7982 @node Darwin Options
7983 @subsection Darwin Options
7984 @cindex Darwin options
7986 These options are defined for all architectures running the Darwin operating
7989 FSF GCC on Darwin does not create ``fat'' object files; it will create
7990 an object file for the single architecture that it was built to
7991 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7992 @option{-arch} options are used; it does so by running the compiler or
7993 linker multiple times and joining the results together with
7996 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7997 @samp{i686}) is determined by the flags that specify the ISA
7998 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7999 @option{-force_cpusubtype_ALL} option can be used to override this.
8001 The Darwin tools vary in their behavior when presented with an ISA
8002 mismatch. The assembler, @file{as}, will only permit instructions to
8003 be used that are valid for the subtype of the file it is generating,
8004 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
8005 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
8006 and print an error if asked to create a shared library with a less
8007 restrictive subtype than its input files (for instance, trying to put
8008 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
8009 for executables, @file{ld}, will quietly give the executable the most
8010 restrictive subtype of any of its input files.
8015 Add the framework directory @var{dir} to the head of the list of
8016 directories to be searched for header files. These directories are
8017 interleaved with those specified by @option{-I} options and are
8018 scanned in a left-to-right order.
8020 A framework directory is a directory with frameworks in it. A
8021 framework is a directory with a @samp{"Headers"} and/or
8022 @samp{"PrivateHeaders"} directory contained directly in it that ends
8023 in @samp{".framework"}. The name of a framework is the name of this
8024 directory excluding the @samp{".framework"}. Headers associated with
8025 the framework are found in one of those two directories, with
8026 @samp{"Headers"} being searched first. A subframework is a framework
8027 directory that is in a framework's @samp{"Frameworks"} directory.
8028 Includes of subframework headers can only appear in a header of a
8029 framework that contains the subframework, or in a sibling subframework
8030 header. Two subframeworks are siblings if they occur in the same
8031 framework. A subframework should not have the same name as a
8032 framework, a warning will be issued if this is violated. Currently a
8033 subframework cannot have subframeworks, in the future, the mechanism
8034 may be extended to support this. The standard frameworks can be found
8035 in @samp{"/System/Library/Frameworks"} and
8036 @samp{"/Library/Frameworks"}. An example include looks like
8037 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
8038 the name of the framework and header.h is found in the
8039 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
8043 Emit debugging information for symbols that are used. For STABS
8044 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
8045 This is by default ON@.
8049 Emit debugging information for all symbols and types.
8051 @item -mmacosx-version-min=@var{version}
8052 The earliest version of MacOS X that this executable will run on
8053 is @var{version}. Typical values of @var{version} include @code{10.1},
8054 @code{10.2}, and @code{10.3.9}.
8056 The default for this option is to make choices that seem to be most
8059 @item -mone-byte-bool
8060 @opindex -mone-byte-bool
8061 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
8062 By default @samp{sizeof(bool)} is @samp{4} when compiling for
8063 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
8064 option has no effect on x86.
8066 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
8067 to generate code that is not binary compatible with code generated
8068 without that switch. Using this switch may require recompiling all
8069 other modules in a program, including system libraries. Use this
8070 switch to conform to a non-default data model.
8072 @item -mfix-and-continue
8073 @itemx -ffix-and-continue
8074 @itemx -findirect-data
8075 @opindex mfix-and-continue
8076 @opindex ffix-and-continue
8077 @opindex findirect-data
8078 Generate code suitable for fast turn around development. Needed to
8079 enable gdb to dynamically load @code{.o} files into already running
8080 programs. @option{-findirect-data} and @option{-ffix-and-continue}
8081 are provided for backwards compatibility.
8085 Loads all members of static archive libraries.
8086 See man ld(1) for more information.
8088 @item -arch_errors_fatal
8089 @opindex arch_errors_fatal
8090 Cause the errors having to do with files that have the wrong architecture
8094 @opindex bind_at_load
8095 Causes the output file to be marked such that the dynamic linker will
8096 bind all undefined references when the file is loaded or launched.
8100 Produce a Mach-o bundle format file.
8101 See man ld(1) for more information.
8103 @item -bundle_loader @var{executable}
8104 @opindex bundle_loader
8105 This option specifies the @var{executable} that will be loading the build
8106 output file being linked. See man ld(1) for more information.
8109 @opindex -dynamiclib
8110 When passed this option, GCC will produce a dynamic library instead of
8111 an executable when linking, using the Darwin @file{libtool} command.
8113 @item -force_cpusubtype_ALL
8114 @opindex -force_cpusubtype_ALL
8115 This causes GCC's output file to have the @var{ALL} subtype, instead of
8116 one controlled by the @option{-mcpu} or @option{-march} option.
8118 @item -allowable_client @var{client_name}
8120 @itemx -compatibility_version
8121 @itemx -current_version
8123 @itemx -dependency-file
8125 @itemx -dylinker_install_name
8127 @itemx -exported_symbols_list
8129 @itemx -flat_namespace
8130 @itemx -force_flat_namespace
8131 @itemx -headerpad_max_install_names
8134 @itemx -install_name
8135 @itemx -keep_private_externs
8136 @itemx -multi_module
8137 @itemx -multiply_defined
8138 @itemx -multiply_defined_unused
8140 @itemx -no_dead_strip_inits_and_terms
8141 @itemx -nofixprebinding
8144 @itemx -noseglinkedit
8145 @itemx -pagezero_size
8147 @itemx -prebind_all_twolevel_modules
8148 @itemx -private_bundle
8149 @itemx -read_only_relocs
8151 @itemx -sectobjectsymbols
8155 @itemx -sectobjectsymbols
8158 @itemx -segs_read_only_addr
8159 @itemx -segs_read_write_addr
8160 @itemx -seg_addr_table
8161 @itemx -seg_addr_table_filename
8164 @itemx -segs_read_only_addr
8165 @itemx -segs_read_write_addr
8166 @itemx -single_module
8169 @itemx -sub_umbrella
8170 @itemx -twolevel_namespace
8173 @itemx -unexported_symbols_list
8174 @itemx -weak_reference_mismatches
8177 @opindex allowable_client
8178 @opindex client_name
8179 @opindex compatibility_version
8180 @opindex current_version
8182 @opindex dependency-file
8184 @opindex dylinker_install_name
8186 @opindex exported_symbols_list
8188 @opindex flat_namespace
8189 @opindex force_flat_namespace
8190 @opindex headerpad_max_install_names
8193 @opindex install_name
8194 @opindex keep_private_externs
8195 @opindex multi_module
8196 @opindex multiply_defined
8197 @opindex multiply_defined_unused
8199 @opindex no_dead_strip_inits_and_terms
8200 @opindex nofixprebinding
8201 @opindex nomultidefs
8203 @opindex noseglinkedit
8204 @opindex pagezero_size
8206 @opindex prebind_all_twolevel_modules
8207 @opindex private_bundle
8208 @opindex read_only_relocs
8210 @opindex sectobjectsymbols
8214 @opindex sectobjectsymbols
8217 @opindex segs_read_only_addr
8218 @opindex segs_read_write_addr
8219 @opindex seg_addr_table
8220 @opindex seg_addr_table_filename
8221 @opindex seglinkedit
8223 @opindex segs_read_only_addr
8224 @opindex segs_read_write_addr
8225 @opindex single_module
8227 @opindex sub_library
8228 @opindex sub_umbrella
8229 @opindex twolevel_namespace
8232 @opindex unexported_symbols_list
8233 @opindex weak_reference_mismatches
8234 @opindex whatsloaded
8236 These options are passed to the Darwin linker. The Darwin linker man page
8237 describes them in detail.
8240 @node DEC Alpha Options
8241 @subsection DEC Alpha Options
8243 These @samp{-m} options are defined for the DEC Alpha implementations:
8246 @item -mno-soft-float
8248 @opindex mno-soft-float
8249 @opindex msoft-float
8250 Use (do not use) the hardware floating-point instructions for
8251 floating-point operations. When @option{-msoft-float} is specified,
8252 functions in @file{libgcc.a} will be used to perform floating-point
8253 operations. Unless they are replaced by routines that emulate the
8254 floating-point operations, or compiled in such a way as to call such
8255 emulations routines, these routines will issue floating-point
8256 operations. If you are compiling for an Alpha without floating-point
8257 operations, you must ensure that the library is built so as not to call
8260 Note that Alpha implementations without floating-point operations are
8261 required to have floating-point registers.
8266 @opindex mno-fp-regs
8267 Generate code that uses (does not use) the floating-point register set.
8268 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8269 register set is not used, floating point operands are passed in integer
8270 registers as if they were integers and floating-point results are passed
8271 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8272 so any function with a floating-point argument or return value called by code
8273 compiled with @option{-mno-fp-regs} must also be compiled with that
8276 A typical use of this option is building a kernel that does not use,
8277 and hence need not save and restore, any floating-point registers.
8281 The Alpha architecture implements floating-point hardware optimized for
8282 maximum performance. It is mostly compliant with the IEEE floating
8283 point standard. However, for full compliance, software assistance is
8284 required. This option generates code fully IEEE compliant code
8285 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8286 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8287 defined during compilation. The resulting code is less efficient but is
8288 able to correctly support denormalized numbers and exceptional IEEE
8289 values such as not-a-number and plus/minus infinity. Other Alpha
8290 compilers call this option @option{-ieee_with_no_inexact}.
8292 @item -mieee-with-inexact
8293 @opindex mieee-with-inexact
8294 This is like @option{-mieee} except the generated code also maintains
8295 the IEEE @var{inexact-flag}. Turning on this option causes the
8296 generated code to implement fully-compliant IEEE math. In addition to
8297 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8298 macro. On some Alpha implementations the resulting code may execute
8299 significantly slower than the code generated by default. Since there is
8300 very little code that depends on the @var{inexact-flag}, you should
8301 normally not specify this option. Other Alpha compilers call this
8302 option @option{-ieee_with_inexact}.
8304 @item -mfp-trap-mode=@var{trap-mode}
8305 @opindex mfp-trap-mode
8306 This option controls what floating-point related traps are enabled.
8307 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8308 The trap mode can be set to one of four values:
8312 This is the default (normal) setting. The only traps that are enabled
8313 are the ones that cannot be disabled in software (e.g., division by zero
8317 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8321 Like @samp{su}, but the instructions are marked to be safe for software
8322 completion (see Alpha architecture manual for details).
8325 Like @samp{su}, but inexact traps are enabled as well.
8328 @item -mfp-rounding-mode=@var{rounding-mode}
8329 @opindex mfp-rounding-mode
8330 Selects the IEEE rounding mode. Other Alpha compilers call this option
8331 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8336 Normal IEEE rounding mode. Floating point numbers are rounded towards
8337 the nearest machine number or towards the even machine number in case
8341 Round towards minus infinity.
8344 Chopped rounding mode. Floating point numbers are rounded towards zero.
8347 Dynamic rounding mode. A field in the floating point control register
8348 (@var{fpcr}, see Alpha architecture reference manual) controls the
8349 rounding mode in effect. The C library initializes this register for
8350 rounding towards plus infinity. Thus, unless your program modifies the
8351 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8354 @item -mtrap-precision=@var{trap-precision}
8355 @opindex mtrap-precision
8356 In the Alpha architecture, floating point traps are imprecise. This
8357 means without software assistance it is impossible to recover from a
8358 floating trap and program execution normally needs to be terminated.
8359 GCC can generate code that can assist operating system trap handlers
8360 in determining the exact location that caused a floating point trap.
8361 Depending on the requirements of an application, different levels of
8362 precisions can be selected:
8366 Program precision. This option is the default and means a trap handler
8367 can only identify which program caused a floating point exception.
8370 Function precision. The trap handler can determine the function that
8371 caused a floating point exception.
8374 Instruction precision. The trap handler can determine the exact
8375 instruction that caused a floating point exception.
8378 Other Alpha compilers provide the equivalent options called
8379 @option{-scope_safe} and @option{-resumption_safe}.
8381 @item -mieee-conformant
8382 @opindex mieee-conformant
8383 This option marks the generated code as IEEE conformant. You must not
8384 use this option unless you also specify @option{-mtrap-precision=i} and either
8385 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8386 is to emit the line @samp{.eflag 48} in the function prologue of the
8387 generated assembly file. Under DEC Unix, this has the effect that
8388 IEEE-conformant math library routines will be linked in.
8390 @item -mbuild-constants
8391 @opindex mbuild-constants
8392 Normally GCC examines a 32- or 64-bit integer constant to
8393 see if it can construct it from smaller constants in two or three
8394 instructions. If it cannot, it will output the constant as a literal and
8395 generate code to load it from the data segment at runtime.
8397 Use this option to require GCC to construct @emph{all} integer constants
8398 using code, even if it takes more instructions (the maximum is six).
8400 You would typically use this option to build a shared library dynamic
8401 loader. Itself a shared library, it must relocate itself in memory
8402 before it can find the variables and constants in its own data segment.
8408 Select whether to generate code to be assembled by the vendor-supplied
8409 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8427 Indicate whether GCC should generate code to use the optional BWX,
8428 CIX, FIX and MAX instruction sets. The default is to use the instruction
8429 sets supported by the CPU type specified via @option{-mcpu=} option or that
8430 of the CPU on which GCC was built if none was specified.
8435 @opindex mfloat-ieee
8436 Generate code that uses (does not use) VAX F and G floating point
8437 arithmetic instead of IEEE single and double precision.
8439 @item -mexplicit-relocs
8440 @itemx -mno-explicit-relocs
8441 @opindex mexplicit-relocs
8442 @opindex mno-explicit-relocs
8443 Older Alpha assemblers provided no way to generate symbol relocations
8444 except via assembler macros. Use of these macros does not allow
8445 optimal instruction scheduling. GNU binutils as of version 2.12
8446 supports a new syntax that allows the compiler to explicitly mark
8447 which relocations should apply to which instructions. This option
8448 is mostly useful for debugging, as GCC detects the capabilities of
8449 the assembler when it is built and sets the default accordingly.
8453 @opindex msmall-data
8454 @opindex mlarge-data
8455 When @option{-mexplicit-relocs} is in effect, static data is
8456 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
8457 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
8458 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
8459 16-bit relocations off of the @code{$gp} register. This limits the
8460 size of the small data area to 64KB, but allows the variables to be
8461 directly accessed via a single instruction.
8463 The default is @option{-mlarge-data}. With this option the data area
8464 is limited to just below 2GB@. Programs that require more than 2GB of
8465 data must use @code{malloc} or @code{mmap} to allocate the data in the
8466 heap instead of in the program's data segment.
8468 When generating code for shared libraries, @option{-fpic} implies
8469 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
8473 @opindex msmall-text
8474 @opindex mlarge-text
8475 When @option{-msmall-text} is used, the compiler assumes that the
8476 code of the entire program (or shared library) fits in 4MB, and is
8477 thus reachable with a branch instruction. When @option{-msmall-data}
8478 is used, the compiler can assume that all local symbols share the
8479 same @code{$gp} value, and thus reduce the number of instructions
8480 required for a function call from 4 to 1.
8482 The default is @option{-mlarge-text}.
8484 @item -mcpu=@var{cpu_type}
8486 Set the instruction set and instruction scheduling parameters for
8487 machine type @var{cpu_type}. You can specify either the @samp{EV}
8488 style name or the corresponding chip number. GCC supports scheduling
8489 parameters for the EV4, EV5 and EV6 family of processors and will
8490 choose the default values for the instruction set from the processor
8491 you specify. If you do not specify a processor type, GCC will default
8492 to the processor on which the compiler was built.
8494 Supported values for @var{cpu_type} are
8500 Schedules as an EV4 and has no instruction set extensions.
8504 Schedules as an EV5 and has no instruction set extensions.
8508 Schedules as an EV5 and supports the BWX extension.
8513 Schedules as an EV5 and supports the BWX and MAX extensions.
8517 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
8521 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
8524 @item -mtune=@var{cpu_type}
8526 Set only the instruction scheduling parameters for machine type
8527 @var{cpu_type}. The instruction set is not changed.
8529 @item -mmemory-latency=@var{time}
8530 @opindex mmemory-latency
8531 Sets the latency the scheduler should assume for typical memory
8532 references as seen by the application. This number is highly
8533 dependent on the memory access patterns used by the application
8534 and the size of the external cache on the machine.
8536 Valid options for @var{time} are
8540 A decimal number representing clock cycles.
8546 The compiler contains estimates of the number of clock cycles for
8547 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8548 (also called Dcache, Scache, and Bcache), as well as to main memory.
8549 Note that L3 is only valid for EV5.
8554 @node DEC Alpha/VMS Options
8555 @subsection DEC Alpha/VMS Options
8557 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8560 @item -mvms-return-codes
8561 @opindex mvms-return-codes
8562 Return VMS condition codes from main. The default is to return POSIX
8563 style condition (e.g.@ error) codes.
8567 @subsection FRV Options
8574 Only use the first 32 general purpose registers.
8579 Use all 64 general purpose registers.
8584 Use only the first 32 floating point registers.
8589 Use all 64 floating point registers
8592 @opindex mhard-float
8594 Use hardware instructions for floating point operations.
8597 @opindex msoft-float
8599 Use library routines for floating point operations.
8604 Dynamically allocate condition code registers.
8609 Do not try to dynamically allocate condition code registers, only
8610 use @code{icc0} and @code{fcc0}.
8615 Change ABI to use double word insns.
8620 Do not use double word instructions.
8625 Use floating point double instructions.
8630 Do not use floating point double instructions.
8635 Use media instructions.
8640 Do not use media instructions.
8645 Use multiply and add/subtract instructions.
8650 Do not use multiply and add/subtract instructions.
8655 Select the FDPIC ABI, that uses function descriptors to represent
8656 pointers to functions. Without any PIC/PIE-related options, it
8657 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
8658 assumes GOT entries and small data are within a 12-bit range from the
8659 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
8660 are computed with 32 bits.
8663 @opindex minline-plt
8665 Enable inlining of PLT entries in function calls to functions that are
8666 not known to bind locally. It has no effect without @option{-mfdpic}.
8667 It's enabled by default if optimizing for speed and compiling for
8668 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
8669 optimization option such as @option{-O3} or above is present in the
8675 Assume a large TLS segment when generating thread-local code.
8680 Do not assume a large TLS segment when generating thread-local code.
8685 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
8686 that is known to be in read-only sections. It's enabled by default,
8687 except for @option{-fpic} or @option{-fpie}: even though it may help
8688 make the global offset table smaller, it trades 1 instruction for 4.
8689 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
8690 one of which may be shared by multiple symbols, and it avoids the need
8691 for a GOT entry for the referenced symbol, so it's more likely to be a
8692 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
8694 @item -multilib-library-pic
8695 @opindex multilib-library-pic
8697 Link with the (library, not FD) pic libraries. It's implied by
8698 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
8699 @option{-fpic} without @option{-mfdpic}. You should never have to use
8705 Follow the EABI requirement of always creating a frame pointer whenever
8706 a stack frame is allocated. This option is enabled by default and can
8707 be disabled with @option{-mno-linked-fp}.
8710 @opindex mlong-calls
8712 Use indirect addressing to call functions outside the current
8713 compilation unit. This allows the functions to be placed anywhere
8714 within the 32-bit address space.
8716 @item -malign-labels
8717 @opindex malign-labels
8719 Try to align labels to an 8-byte boundary by inserting nops into the
8720 previous packet. This option only has an effect when VLIW packing
8721 is enabled. It doesn't create new packets; it merely adds nops to
8725 @opindex mlibrary-pic
8727 Generate position-independent EABI code.
8732 Use only the first four media accumulator registers.
8737 Use all eight media accumulator registers.
8742 Pack VLIW instructions.
8747 Do not pack VLIW instructions.
8752 Do not mark ABI switches in e_flags.
8757 Enable the use of conditional-move instructions (default).
8759 This switch is mainly for debugging the compiler and will likely be removed
8760 in a future version.
8762 @item -mno-cond-move
8763 @opindex mno-cond-move
8765 Disable the use of conditional-move instructions.
8767 This switch is mainly for debugging the compiler and will likely be removed
8768 in a future version.
8773 Enable the use of conditional set instructions (default).
8775 This switch is mainly for debugging the compiler and will likely be removed
8776 in a future version.
8781 Disable the use of conditional set instructions.
8783 This switch is mainly for debugging the compiler and will likely be removed
8784 in a future version.
8789 Enable the use of conditional execution (default).
8791 This switch is mainly for debugging the compiler and will likely be removed
8792 in a future version.
8794 @item -mno-cond-exec
8795 @opindex mno-cond-exec
8797 Disable the use of conditional execution.
8799 This switch is mainly for debugging the compiler and will likely be removed
8800 in a future version.
8803 @opindex mvliw-branch
8805 Run a pass to pack branches into VLIW instructions (default).
8807 This switch is mainly for debugging the compiler and will likely be removed
8808 in a future version.
8810 @item -mno-vliw-branch
8811 @opindex mno-vliw-branch
8813 Do not run a pass to pack branches into VLIW instructions.
8815 This switch is mainly for debugging the compiler and will likely be removed
8816 in a future version.
8818 @item -mmulti-cond-exec
8819 @opindex mmulti-cond-exec
8821 Enable optimization of @code{&&} and @code{||} in conditional execution
8824 This switch is mainly for debugging the compiler and will likely be removed
8825 in a future version.
8827 @item -mno-multi-cond-exec
8828 @opindex mno-multi-cond-exec
8830 Disable optimization of @code{&&} and @code{||} in conditional execution.
8832 This switch is mainly for debugging the compiler and will likely be removed
8833 in a future version.
8835 @item -mnested-cond-exec
8836 @opindex mnested-cond-exec
8838 Enable nested conditional execution optimizations (default).
8840 This switch is mainly for debugging the compiler and will likely be removed
8841 in a future version.
8843 @item -mno-nested-cond-exec
8844 @opindex mno-nested-cond-exec
8846 Disable nested conditional execution optimizations.
8848 This switch is mainly for debugging the compiler and will likely be removed
8849 in a future version.
8851 @item -moptimize-membar
8852 @opindex moptimize-membar
8854 This switch removes redundant @code{membar} instructions from the
8855 compiler generated code. It is enabled by default.
8857 @item -mno-optimize-membar
8858 @opindex mno-optimize-membar
8860 This switch disables the automatic removal of redundant @code{membar}
8861 instructions from the generated code.
8863 @item -mtomcat-stats
8864 @opindex mtomcat-stats
8866 Cause gas to print out tomcat statistics.
8868 @item -mcpu=@var{cpu}
8871 Select the processor type for which to generate code. Possible values are
8872 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8873 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8877 @node GNU/Linux Options
8878 @subsection GNU/Linux Options
8880 These @samp{-m} options are defined for GNU/Linux targets:
8885 Use the GNU C library instead of uClibc. This is the default except
8886 on @samp{*-*-linux-*uclibc*} targets.
8890 Use uClibc instead of the GNU C library. This is the default on
8891 @samp{*-*-linux-*uclibc*} targets.
8894 @node H8/300 Options
8895 @subsection H8/300 Options
8897 These @samp{-m} options are defined for the H8/300 implementations:
8902 Shorten some address references at link time, when possible; uses the
8903 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8904 ld, Using ld}, for a fuller description.
8908 Generate code for the H8/300H@.
8912 Generate code for the H8S@.
8916 Generate code for the H8S and H8/300H in the normal mode. This switch
8917 must be used either with @option{-mh} or @option{-ms}.
8921 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8925 Make @code{int} data 32 bits by default.
8929 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8930 The default for the H8/300H and H8S is to align longs and floats on 4
8932 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8933 This option has no effect on the H8/300.
8937 @subsection HPPA Options
8938 @cindex HPPA Options
8940 These @samp{-m} options are defined for the HPPA family of computers:
8943 @item -march=@var{architecture-type}
8945 Generate code for the specified architecture. The choices for
8946 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8947 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8948 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8949 architecture option for your machine. Code compiled for lower numbered
8950 architectures will run on higher numbered architectures, but not the
8954 @itemx -mpa-risc-1-1
8955 @itemx -mpa-risc-2-0
8956 @opindex mpa-risc-1-0
8957 @opindex mpa-risc-1-1
8958 @opindex mpa-risc-2-0
8959 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8962 @opindex mbig-switch
8963 Generate code suitable for big switch tables. Use this option only if
8964 the assembler/linker complain about out of range branches within a switch
8967 @item -mjump-in-delay
8968 @opindex mjump-in-delay
8969 Fill delay slots of function calls with unconditional jump instructions
8970 by modifying the return pointer for the function call to be the target
8971 of the conditional jump.
8973 @item -mdisable-fpregs
8974 @opindex mdisable-fpregs
8975 Prevent floating point registers from being used in any manner. This is
8976 necessary for compiling kernels which perform lazy context switching of
8977 floating point registers. If you use this option and attempt to perform
8978 floating point operations, the compiler will abort.
8980 @item -mdisable-indexing
8981 @opindex mdisable-indexing
8982 Prevent the compiler from using indexing address modes. This avoids some
8983 rather obscure problems when compiling MIG generated code under MACH@.
8985 @item -mno-space-regs
8986 @opindex mno-space-regs
8987 Generate code that assumes the target has no space registers. This allows
8988 GCC to generate faster indirect calls and use unscaled index address modes.
8990 Such code is suitable for level 0 PA systems and kernels.
8992 @item -mfast-indirect-calls
8993 @opindex mfast-indirect-calls
8994 Generate code that assumes calls never cross space boundaries. This
8995 allows GCC to emit code which performs faster indirect calls.
8997 This option will not work in the presence of shared libraries or nested
9000 @item -mfixed-range=@var{register-range}
9001 @opindex mfixed-range
9002 Generate code treating the given register range as fixed registers.
9003 A fixed register is one that the register allocator can not use. This is
9004 useful when compiling kernel code. A register range is specified as
9005 two registers separated by a dash. Multiple register ranges can be
9006 specified separated by a comma.
9008 @item -mlong-load-store
9009 @opindex mlong-load-store
9010 Generate 3-instruction load and store sequences as sometimes required by
9011 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9014 @item -mportable-runtime
9015 @opindex mportable-runtime
9016 Use the portable calling conventions proposed by HP for ELF systems.
9020 Enable the use of assembler directives only GAS understands.
9022 @item -mschedule=@var{cpu-type}
9024 Schedule code according to the constraints for the machine type
9025 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9026 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9027 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9028 proper scheduling option for your machine. The default scheduling is
9032 @opindex mlinker-opt
9033 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9034 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9035 linkers in which they give bogus error messages when linking some programs.
9038 @opindex msoft-float
9039 Generate output containing library calls for floating point.
9040 @strong{Warning:} the requisite libraries are not available for all HPPA
9041 targets. Normally the facilities of the machine's usual C compiler are
9042 used, but this cannot be done directly in cross-compilation. You must make
9043 your own arrangements to provide suitable library functions for
9044 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9045 does provide software floating point support.
9047 @option{-msoft-float} changes the calling convention in the output file;
9048 therefore, it is only useful if you compile @emph{all} of a program with
9049 this option. In particular, you need to compile @file{libgcc.a}, the
9050 library that comes with GCC, with @option{-msoft-float} in order for
9055 Generate the predefine, @code{_SIO}, for server IO@. The default is
9056 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9057 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
9058 options are available under HP-UX and HI-UX@.
9062 Use GNU ld specific options. This passes @option{-shared} to ld when
9063 building a shared library. It is the default when GCC is configured,
9064 explicitly or implicitly, with the GNU linker. This option does not
9065 have any affect on which ld is called, it only changes what parameters
9066 are passed to that ld. The ld that is called is determined by the
9067 @option{--with-ld} configure option, GCC's program search path, and
9068 finally by the user's @env{PATH}. The linker used by GCC can be printed
9069 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
9070 on the 64 bit HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9074 Use HP ld specific options. This passes @option{-b} to ld when building
9075 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9076 links. It is the default when GCC is configured, explicitly or
9077 implicitly, with the HP linker. This option does not have any affect on
9078 which ld is called, it only changes what parameters are passed to that
9079 ld. The ld that is called is determined by the @option{--with-ld}
9080 configure option, GCC's program search path, and finally by the user's
9081 @env{PATH}. The linker used by GCC can be printed using @samp{which
9082 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
9083 HP-UX GCC, i.e. configured with @samp{hppa*64*-*-hpux*}.
9086 @opindex mno-long-calls
9087 Generate code that uses long call sequences. This ensures that a call
9088 is always able to reach linker generated stubs. The default is to generate
9089 long calls only when the distance from the call site to the beginning
9090 of the function or translation unit, as the case may be, exceeds a
9091 predefined limit set by the branch type being used. The limits for
9092 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9093 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9096 Distances are measured from the beginning of functions when using the
9097 @option{-ffunction-sections} option, or when using the @option{-mgas}
9098 and @option{-mno-portable-runtime} options together under HP-UX with
9101 It is normally not desirable to use this option as it will degrade
9102 performance. However, it may be useful in large applications,
9103 particularly when partial linking is used to build the application.
9105 The types of long calls used depends on the capabilities of the
9106 assembler and linker, and the type of code being generated. The
9107 impact on systems that support long absolute calls, and long pic
9108 symbol-difference or pc-relative calls should be relatively small.
9109 However, an indirect call is used on 32-bit ELF systems in pic code
9110 and it is quite long.
9112 @item -munix=@var{unix-std}
9114 Generate compiler predefines and select a startfile for the specified
9115 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
9116 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
9117 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
9118 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
9119 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
9122 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
9123 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
9124 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
9125 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
9126 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
9127 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
9129 It is @emph{important} to note that this option changes the interfaces
9130 for various library routines. It also affects the operational behavior
9131 of the C library. Thus, @emph{extreme} care is needed in using this
9134 Library code that is intended to operate with more than one UNIX
9135 standard must test, set and restore the variable @var{__xpg4_extended_mask}
9136 as appropriate. Most GNU software doesn't provide this capability.
9140 Suppress the generation of link options to search libdld.sl when the
9141 @option{-static} option is specified on HP-UX 10 and later.
9145 The HP-UX implementation of setlocale in libc has a dependency on
9146 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9147 when the @option{-static} option is specified, special link options
9148 are needed to resolve this dependency.
9150 On HP-UX 10 and later, the GCC driver adds the necessary options to
9151 link with libdld.sl when the @option{-static} option is specified.
9152 This causes the resulting binary to be dynamic. On the 64-bit port,
9153 the linkers generate dynamic binaries by default in any case. The
9154 @option{-nolibdld} option can be used to prevent the GCC driver from
9155 adding these link options.
9159 Add support for multithreading with the @dfn{dce thread} library
9160 under HP-UX@. This option sets flags for both the preprocessor and
9164 @node i386 and x86-64 Options
9165 @subsection Intel 386 and AMD x86-64 Options
9166 @cindex i386 Options
9167 @cindex x86-64 Options
9168 @cindex Intel 386 Options
9169 @cindex AMD x86-64 Options
9171 These @samp{-m} options are defined for the i386 and x86-64 family of
9175 @item -mtune=@var{cpu-type}
9177 Tune to @var{cpu-type} everything applicable about the generated code, except
9178 for the ABI and the set of available instructions. The choices for
9182 Produce code optimized for the most common IA32/AMD64/EM64T processors.
9183 If you know the CPU on which your code will run, then you should use
9184 the corresponding @option{-mtune} option instead of
9185 @option{-mtune=generic}. But, if you do not know exactly what CPU users
9186 of your application will have, then you should use this option.
9188 As new processors are deployed in the marketplace, the behavior of this
9189 option will change. Therefore, if you upgrade to a newer version of
9190 GCC, the code generated option will change to reflect the processors
9191 that were most common when that version of GCC was released.
9193 There is no @option{-march=generic} option because @option{-march}
9194 indicates the instruction set the compiler can use, and there is no
9195 generic instruction set applicable to all processors. In contrast,
9196 @option{-mtune} indicates the processor (or, in this case, collection of
9197 processors) for which the code is optimized.
9199 This selects the CPU to tune for at compilation time by determining
9200 the processor type of the compiling machine. Using @option{-mtune=native}
9201 will produce code optimized for the local machine under the constraints
9202 of the selected instruction set. Using @option{-march=native} will
9203 enable all instruction subsets supported by the local machine (hence
9204 the result might not run on different machines).
9206 Original Intel's i386 CPU@.
9208 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
9210 Intel Pentium CPU with no MMX support.
9212 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
9214 Intel PentiumPro CPU@.
9216 Same as @code{generic}, but when used as @code{march} option, PentiumPro
9217 instruction set will be used, so the code will run on all i686 family chips.
9219 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
9220 @item pentium3, pentium3m
9221 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
9224 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
9225 support. Used by Centrino notebooks.
9226 @item pentium4, pentium4m
9227 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
9229 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
9232 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
9233 SSE2 and SSE3 instruction set support.
9235 AMD K6 CPU with MMX instruction set support.
9237 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
9238 @item athlon, athlon-tbird
9239 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
9241 @item athlon-4, athlon-xp, athlon-mp
9242 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
9243 instruction set support.
9244 @item k8, opteron, athlon64, athlon-fx
9245 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
9246 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
9248 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
9251 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
9252 instruction set support.
9254 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
9255 implemented for this chip.)
9257 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
9258 implemented for this chip.)
9261 While picking a specific @var{cpu-type} will schedule things appropriately
9262 for that particular chip, the compiler will not generate any code that
9263 does not run on the i386 without the @option{-march=@var{cpu-type}} option
9266 @item -march=@var{cpu-type}
9268 Generate instructions for the machine type @var{cpu-type}. The choices
9269 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
9270 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
9272 @item -mcpu=@var{cpu-type}
9274 A deprecated synonym for @option{-mtune}.
9283 @opindex mpentiumpro
9284 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
9285 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
9286 These synonyms are deprecated.
9288 @item -mfpmath=@var{unit}
9290 Generate floating point arithmetics for selected unit @var{unit}. The choices
9295 Use the standard 387 floating point coprocessor present majority of chips and
9296 emulated otherwise. Code compiled with this option will run almost everywhere.
9297 The temporary results are computed in 80bit precision instead of precision
9298 specified by the type resulting in slightly different results compared to most
9299 of other chips. See @option{-ffloat-store} for more detailed description.
9301 This is the default choice for i386 compiler.
9304 Use scalar floating point instructions present in the SSE instruction set.
9305 This instruction set is supported by Pentium3 and newer chips, in the AMD line
9306 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
9307 instruction set supports only single precision arithmetics, thus the double and
9308 extended precision arithmetics is still done using 387. Later version, present
9309 only in Pentium4 and the future AMD x86-64 chips supports double precision
9312 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
9313 or @option{-msse2} switches to enable SSE extensions and make this option
9314 effective. For the x86-64 compiler, these extensions are enabled by default.
9316 The resulting code should be considerably faster in the majority of cases and avoid
9317 the numerical instability problems of 387 code, but may break some existing
9318 code that expects temporaries to be 80bit.
9320 This is the default choice for the x86-64 compiler.
9323 Attempt to utilize both instruction sets at once. This effectively double the
9324 amount of available registers and on chips with separate execution units for
9325 387 and SSE the execution resources too. Use this option with care, as it is
9326 still experimental, because the GCC register allocator does not model separate
9327 functional units well resulting in instable performance.
9330 @item -masm=@var{dialect}
9331 @opindex masm=@var{dialect}
9332 Output asm instructions using selected @var{dialect}. Supported
9333 choices are @samp{intel} or @samp{att} (the default one). Darwin does
9334 not support @samp{intel}.
9339 @opindex mno-ieee-fp
9340 Control whether or not the compiler uses IEEE floating point
9341 comparisons. These handle correctly the case where the result of a
9342 comparison is unordered.
9345 @opindex msoft-float
9346 Generate output containing library calls for floating point.
9347 @strong{Warning:} the requisite libraries are not part of GCC@.
9348 Normally the facilities of the machine's usual C compiler are used, but
9349 this can't be done directly in cross-compilation. You must make your
9350 own arrangements to provide suitable library functions for
9353 On machines where a function returns floating point results in the 80387
9354 register stack, some floating point opcodes may be emitted even if
9355 @option{-msoft-float} is used.
9357 @item -mno-fp-ret-in-387
9358 @opindex mno-fp-ret-in-387
9359 Do not use the FPU registers for return values of functions.
9361 The usual calling convention has functions return values of types
9362 @code{float} and @code{double} in an FPU register, even if there
9363 is no FPU@. The idea is that the operating system should emulate
9366 The option @option{-mno-fp-ret-in-387} causes such values to be returned
9367 in ordinary CPU registers instead.
9369 @item -mno-fancy-math-387
9370 @opindex mno-fancy-math-387
9371 Some 387 emulators do not support the @code{sin}, @code{cos} and
9372 @code{sqrt} instructions for the 387. Specify this option to avoid
9373 generating those instructions. This option is the default on FreeBSD,
9374 OpenBSD and NetBSD@. This option is overridden when @option{-march}
9375 indicates that the target cpu will always have an FPU and so the
9376 instruction will not need emulation. As of revision 2.6.1, these
9377 instructions are not generated unless you also use the
9378 @option{-funsafe-math-optimizations} switch.
9380 @item -malign-double
9381 @itemx -mno-align-double
9382 @opindex malign-double
9383 @opindex mno-align-double
9384 Control whether GCC aligns @code{double}, @code{long double}, and
9385 @code{long long} variables on a two word boundary or a one word
9386 boundary. Aligning @code{double} variables on a two word boundary will
9387 produce code that runs somewhat faster on a @samp{Pentium} at the
9388 expense of more memory.
9390 On x86-64, @option{-malign-double} is enabled by default.
9392 @strong{Warning:} if you use the @option{-malign-double} switch,
9393 structures containing the above types will be aligned differently than
9394 the published application binary interface specifications for the 386
9395 and will not be binary compatible with structures in code compiled
9396 without that switch.
9398 @item -m96bit-long-double
9399 @itemx -m128bit-long-double
9400 @opindex m96bit-long-double
9401 @opindex m128bit-long-double
9402 These switches control the size of @code{long double} type. The i386
9403 application binary interface specifies the size to be 96 bits,
9404 so @option{-m96bit-long-double} is the default in 32 bit mode.
9406 Modern architectures (Pentium and newer) would prefer @code{long double}
9407 to be aligned to an 8 or 16 byte boundary. In arrays or structures
9408 conforming to the ABI, this would not be possible. So specifying a
9409 @option{-m128bit-long-double} will align @code{long double}
9410 to a 16 byte boundary by padding the @code{long double} with an additional
9413 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
9414 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
9416 Notice that neither of these options enable any extra precision over the x87
9417 standard of 80 bits for a @code{long double}.
9419 @strong{Warning:} if you override the default value for your target ABI, the
9420 structures and arrays containing @code{long double} variables will change
9421 their size as well as function calling convention for function taking
9422 @code{long double} will be modified. Hence they will not be binary
9423 compatible with arrays or structures in code compiled without that switch.
9425 @item -mmlarge-data-threshold=@var{number}
9426 @opindex mlarge-data-threshold=@var{number}
9427 When @option{-mcmodel=medium} is specified, the data greater than
9428 @var{threshold} are placed in large data section. This value must be the
9429 same across all object linked into the binary and defaults to 65535.
9432 @itemx -mno-svr3-shlib
9433 @opindex msvr3-shlib
9434 @opindex mno-svr3-shlib
9435 Control whether GCC places uninitialized local variables into the
9436 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
9437 into @code{bss}. These options are meaningful only on System V Release 3.
9441 Use a different function-calling convention, in which functions that
9442 take a fixed number of arguments return with the @code{ret} @var{num}
9443 instruction, which pops their arguments while returning. This saves one
9444 instruction in the caller since there is no need to pop the arguments
9447 You can specify that an individual function is called with this calling
9448 sequence with the function attribute @samp{stdcall}. You can also
9449 override the @option{-mrtd} option by using the function attribute
9450 @samp{cdecl}. @xref{Function Attributes}.
9452 @strong{Warning:} this calling convention is incompatible with the one
9453 normally used on Unix, so you cannot use it if you need to call
9454 libraries compiled with the Unix compiler.
9456 Also, you must provide function prototypes for all functions that
9457 take variable numbers of arguments (including @code{printf});
9458 otherwise incorrect code will be generated for calls to those
9461 In addition, seriously incorrect code will result if you call a
9462 function with too many arguments. (Normally, extra arguments are
9463 harmlessly ignored.)
9465 @item -mregparm=@var{num}
9467 Control how many registers are used to pass integer arguments. By
9468 default, no registers are used to pass arguments, and at most 3
9469 registers can be used. You can control this behavior for a specific
9470 function by using the function attribute @samp{regparm}.
9471 @xref{Function Attributes}.
9473 @strong{Warning:} if you use this switch, and
9474 @var{num} is nonzero, then you must build all modules with the same
9475 value, including any libraries. This includes the system libraries and
9479 @opindex msseregparm
9480 Use SSE register passing conventions for float and double arguments
9481 and return values. You can control this behavior for a specific
9482 function by using the function attribute @samp{sseregparm}.
9483 @xref{Function Attributes}.
9485 @strong{Warning:} if you use this switch then you must build all
9486 modules with the same value, including any libraries. This includes
9487 the system libraries and startup modules.
9489 @item -mstackrealign
9490 @opindex mstackrealign
9491 Realign the stack at entry. On the Intel x86, the
9492 @option{-mstackrealign} option will generate an alternate prologue and
9493 epilogue that realigns the runtime stack. This supports mixing legacy
9494 codes that keep a 4-byte aligned stack with modern codes that keep a
9495 16-byte stack for SSE compatibility. The alternate prologue and
9496 epilogue are slower and bigger than the regular ones, and the
9497 alternate prologue requires an extra scratch register; this lowers the
9498 number of registers available if used in conjunction with the
9499 @code{regparm} attribute. The @option{-mstackrealign} option is
9500 incompatible with the nested function prologue; this is considered a
9501 hard error. See also the attribute @code{force_align_arg_pointer},
9502 applicable to individual functions.
9504 @item -mpreferred-stack-boundary=@var{num}
9505 @opindex mpreferred-stack-boundary
9506 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
9507 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
9508 the default is 4 (16 bytes or 128 bits).
9510 On Pentium and PentiumPro, @code{double} and @code{long double} values
9511 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
9512 suffer significant run time performance penalties. On Pentium III, the
9513 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
9514 properly if it is not 16 byte aligned.
9516 To ensure proper alignment of this values on the stack, the stack boundary
9517 must be as aligned as that required by any value stored on the stack.
9518 Further, every function must be generated such that it keeps the stack
9519 aligned. Thus calling a function compiled with a higher preferred
9520 stack boundary from a function compiled with a lower preferred stack
9521 boundary will most likely misalign the stack. It is recommended that
9522 libraries that use callbacks always use the default setting.
9524 This extra alignment does consume extra stack space, and generally
9525 increases code size. Code that is sensitive to stack space usage, such
9526 as embedded systems and operating system kernels, may want to reduce the
9527 preferred alignment to @option{-mpreferred-stack-boundary=2}.
9545 These switches enable or disable the use of instructions in the MMX,
9546 SSE, SSE2 or 3DNow! extended instruction sets. These extensions are
9547 also available as built-in functions: see @ref{X86 Built-in Functions},
9548 for details of the functions enabled and disabled by these switches.
9550 To have SSE/SSE2 instructions generated automatically from floating-point
9551 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
9553 These options will enable GCC to use these extended instructions in
9554 generated code, even without @option{-mfpmath=sse}. Applications which
9555 perform runtime CPU detection must compile separate files for each
9556 supported architecture, using the appropriate flags. In particular,
9557 the file containing the CPU detection code should be compiled without
9561 @itemx -mno-push-args
9563 @opindex mno-push-args
9564 Use PUSH operations to store outgoing parameters. This method is shorter
9565 and usually equally fast as method using SUB/MOV operations and is enabled
9566 by default. In some cases disabling it may improve performance because of
9567 improved scheduling and reduced dependencies.
9569 @item -maccumulate-outgoing-args
9570 @opindex maccumulate-outgoing-args
9571 If enabled, the maximum amount of space required for outgoing arguments will be
9572 computed in the function prologue. This is faster on most modern CPUs
9573 because of reduced dependencies, improved scheduling and reduced stack usage
9574 when preferred stack boundary is not equal to 2. The drawback is a notable
9575 increase in code size. This switch implies @option{-mno-push-args}.
9579 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
9580 on thread-safe exception handling must compile and link all code with the
9581 @option{-mthreads} option. When compiling, @option{-mthreads} defines
9582 @option{-D_MT}; when linking, it links in a special thread helper library
9583 @option{-lmingwthrd} which cleans up per thread exception handling data.
9585 @item -mno-align-stringops
9586 @opindex mno-align-stringops
9587 Do not align destination of inlined string operations. This switch reduces
9588 code size and improves performance in case the destination is already aligned,
9589 but GCC doesn't know about it.
9591 @item -minline-all-stringops
9592 @opindex minline-all-stringops
9593 By default GCC inlines string operations only when destination is known to be
9594 aligned at least to 4 byte boundary. This enables more inlining, increase code
9595 size, but may improve performance of code that depends on fast memcpy, strlen
9596 and memset for short lengths.
9598 @item -momit-leaf-frame-pointer
9599 @opindex momit-leaf-frame-pointer
9600 Don't keep the frame pointer in a register for leaf functions. This
9601 avoids the instructions to save, set up and restore frame pointers and
9602 makes an extra register available in leaf functions. The option
9603 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9604 which might make debugging harder.
9606 @item -mtls-direct-seg-refs
9607 @itemx -mno-tls-direct-seg-refs
9608 @opindex mtls-direct-seg-refs
9609 Controls whether TLS variables may be accessed with offsets from the
9610 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
9611 or whether the thread base pointer must be added. Whether or not this
9612 is legal depends on the operating system, and whether it maps the
9613 segment to cover the entire TLS area.
9615 For systems that use GNU libc, the default is on.
9618 These @samp{-m} switches are supported in addition to the above
9619 on AMD x86-64 processors in 64-bit environments.
9626 Generate code for a 32-bit or 64-bit environment.
9627 The 32-bit environment sets int, long and pointer to 32 bits and
9628 generates code that runs on any i386 system.
9629 The 64-bit environment sets int to 32 bits and long and pointer
9630 to 64 bits and generates code for AMD's x86-64 architecture.
9633 @opindex no-red-zone
9634 Do not use a so called red zone for x86-64 code. The red zone is mandated
9635 by the x86-64 ABI, it is a 128-byte area beyond the location of the
9636 stack pointer that will not be modified by signal or interrupt handlers
9637 and therefore can be used for temporary data without adjusting the stack
9638 pointer. The flag @option{-mno-red-zone} disables this red zone.
9640 @item -mcmodel=small
9641 @opindex mcmodel=small
9642 Generate code for the small code model: the program and its symbols must
9643 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
9644 Programs can be statically or dynamically linked. This is the default
9647 @item -mcmodel=kernel
9648 @opindex mcmodel=kernel
9649 Generate code for the kernel code model. The kernel runs in the
9650 negative 2 GB of the address space.
9651 This model has to be used for Linux kernel code.
9653 @item -mcmodel=medium
9654 @opindex mcmodel=medium
9655 Generate code for the medium model: The program is linked in the lower 2
9656 GB of the address space but symbols can be located anywhere in the
9657 address space. Programs can be statically or dynamically linked, but
9658 building of shared libraries are not supported with the medium model.
9660 @item -mcmodel=large
9661 @opindex mcmodel=large
9662 Generate code for the large model: This model makes no assumptions
9663 about addresses and sizes of sections. Currently GCC does not implement
9668 @subsection IA-64 Options
9669 @cindex IA-64 Options
9671 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9675 @opindex mbig-endian
9676 Generate code for a big endian target. This is the default for HP-UX@.
9678 @item -mlittle-endian
9679 @opindex mlittle-endian
9680 Generate code for a little endian target. This is the default for AIX5
9687 Generate (or don't) code for the GNU assembler. This is the default.
9688 @c Also, this is the default if the configure option @option{--with-gnu-as}
9695 Generate (or don't) code for the GNU linker. This is the default.
9696 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9701 Generate code that does not use a global pointer register. The result
9702 is not position independent code, and violates the IA-64 ABI@.
9704 @item -mvolatile-asm-stop
9705 @itemx -mno-volatile-asm-stop
9706 @opindex mvolatile-asm-stop
9707 @opindex mno-volatile-asm-stop
9708 Generate (or don't) a stop bit immediately before and after volatile asm
9711 @item -mregister-names
9712 @itemx -mno-register-names
9713 @opindex mregister-names
9714 @opindex mno-register-names
9715 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9716 the stacked registers. This may make assembler output more readable.
9722 Disable (or enable) optimizations that use the small data section. This may
9723 be useful for working around optimizer bugs.
9726 @opindex mconstant-gp
9727 Generate code that uses a single constant global pointer value. This is
9728 useful when compiling kernel code.
9732 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9733 This is useful when compiling firmware code.
9735 @item -minline-float-divide-min-latency
9736 @opindex minline-float-divide-min-latency
9737 Generate code for inline divides of floating point values
9738 using the minimum latency algorithm.
9740 @item -minline-float-divide-max-throughput
9741 @opindex minline-float-divide-max-throughput
9742 Generate code for inline divides of floating point values
9743 using the maximum throughput algorithm.
9745 @item -minline-int-divide-min-latency
9746 @opindex minline-int-divide-min-latency
9747 Generate code for inline divides of integer values
9748 using the minimum latency algorithm.
9750 @item -minline-int-divide-max-throughput
9751 @opindex minline-int-divide-max-throughput
9752 Generate code for inline divides of integer values
9753 using the maximum throughput algorithm.
9755 @item -minline-sqrt-min-latency
9756 @opindex minline-sqrt-min-latency
9757 Generate code for inline square roots
9758 using the minimum latency algorithm.
9760 @item -minline-sqrt-max-throughput
9761 @opindex minline-sqrt-max-throughput
9762 Generate code for inline square roots
9763 using the maximum throughput algorithm.
9765 @item -mno-dwarf2-asm
9767 @opindex mno-dwarf2-asm
9768 @opindex mdwarf2-asm
9769 Don't (or do) generate assembler code for the DWARF2 line number debugging
9770 info. This may be useful when not using the GNU assembler.
9772 @item -mearly-stop-bits
9773 @itemx -mno-early-stop-bits
9774 @opindex mearly-stop-bits
9775 @opindex mno-early-stop-bits
9776 Allow stop bits to be placed earlier than immediately preceding the
9777 instruction that triggered the stop bit. This can improve instruction
9778 scheduling, but does not always do so.
9780 @item -mfixed-range=@var{register-range}
9781 @opindex mfixed-range
9782 Generate code treating the given register range as fixed registers.
9783 A fixed register is one that the register allocator can not use. This is
9784 useful when compiling kernel code. A register range is specified as
9785 two registers separated by a dash. Multiple register ranges can be
9786 specified separated by a comma.
9788 @item -mtls-size=@var{tls-size}
9790 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
9793 @item -mtune=@var{cpu-type}
9795 Tune the instruction scheduling for a particular CPU, Valid values are
9796 itanium, itanium1, merced, itanium2, and mckinley.
9802 Add support for multithreading using the POSIX threads library. This
9803 option sets flags for both the preprocessor and linker. It does
9804 not affect the thread safety of object code produced by the compiler or
9805 that of libraries supplied with it. These are HP-UX specific flags.
9811 Generate code for a 32-bit or 64-bit environment.
9812 The 32-bit environment sets int, long and pointer to 32 bits.
9813 The 64-bit environment sets int to 32 bits and long and pointer
9814 to 64 bits. These are HP-UX specific flags.
9816 @item -mno-sched-br-data-spec
9817 @itemx -msched-br-data-spec
9818 @opindex -mno-sched-br-data-spec
9819 @opindex -msched-br-data-spec
9820 (Dis/En)able data speculative scheduling before reload.
9821 This will result in generation of the ld.a instructions and
9822 the corresponding check instructions (ld.c / chk.a).
9823 The default is 'disable'.
9825 @item -msched-ar-data-spec
9826 @itemx -mno-sched-ar-data-spec
9827 @opindex -msched-ar-data-spec
9828 @opindex -mno-sched-ar-data-spec
9829 (En/Dis)able data speculative scheduling after reload.
9830 This will result in generation of the ld.a instructions and
9831 the corresponding check instructions (ld.c / chk.a).
9832 The default is 'enable'.
9834 @item -mno-sched-control-spec
9835 @itemx -msched-control-spec
9836 @opindex -mno-sched-control-spec
9837 @opindex -msched-control-spec
9838 (Dis/En)able control speculative scheduling. This feature is
9839 available only during region scheduling (i.e. before reload).
9840 This will result in generation of the ld.s instructions and
9841 the corresponding check instructions chk.s .
9842 The default is 'disable'.
9844 @item -msched-br-in-data-spec
9845 @itemx -mno-sched-br-in-data-spec
9846 @opindex -msched-br-in-data-spec
9847 @opindex -mno-sched-br-in-data-spec
9848 (En/Dis)able speculative scheduling of the instructions that
9849 are dependent on the data speculative loads before reload.
9850 This is effective only with @option{-msched-br-data-spec} enabled.
9851 The default is 'enable'.
9853 @item -msched-ar-in-data-spec
9854 @itemx -mno-sched-ar-in-data-spec
9855 @opindex -msched-ar-in-data-spec
9856 @opindex -mno-sched-ar-in-data-spec
9857 (En/Dis)able speculative scheduling of the instructions that
9858 are dependent on the data speculative loads after reload.
9859 This is effective only with @option{-msched-ar-data-spec} enabled.
9860 The default is 'enable'.
9862 @item -msched-in-control-spec
9863 @itemx -mno-sched-in-control-spec
9864 @opindex -msched-in-control-spec
9865 @opindex -mno-sched-in-control-spec
9866 (En/Dis)able speculative scheduling of the instructions that
9867 are dependent on the control speculative loads.
9868 This is effective only with @option{-msched-control-spec} enabled.
9869 The default is 'enable'.
9872 @itemx -mno-sched-ldc
9873 @opindex -msched-ldc
9874 @opindex -mno-sched-ldc
9875 (En/Dis)able use of simple data speculation checks ld.c .
9876 If disabled, only chk.a instructions will be emitted to check
9877 data speculative loads.
9878 The default is 'enable'.
9880 @item -mno-sched-control-ldc
9881 @itemx -msched-control-ldc
9882 @opindex -mno-sched-control-ldc
9883 @opindex -msched-control-ldc
9884 (Dis/En)able use of ld.c instructions to check control speculative loads.
9885 If enabled, in case of control speculative load with no speculatively
9886 scheduled dependent instructions this load will be emitted as ld.sa and
9887 ld.c will be used to check it.
9888 The default is 'disable'.
9890 @item -mno-sched-spec-verbose
9891 @itemx -msched-spec-verbose
9892 @opindex -mno-sched-spec-verbose
9893 @opindex -msched-spec-verbose
9894 (Dis/En)able printing of the information about speculative motions.
9896 @item -mno-sched-prefer-non-data-spec-insns
9897 @itemx -msched-prefer-non-data-spec-insns
9898 @opindex -mno-sched-prefer-non-data-spec-insns
9899 @opindex -msched-prefer-non-data-spec-insns
9900 If enabled, data speculative instructions will be chosen for schedule
9901 only if there are no other choices at the moment. This will make
9902 the use of the data speculation much more conservative.
9903 The default is 'disable'.
9905 @item -mno-sched-prefer-non-control-spec-insns
9906 @itemx -msched-prefer-non-control-spec-insns
9907 @opindex -mno-sched-prefer-non-control-spec-insns
9908 @opindex -msched-prefer-non-control-spec-insns
9909 If enabled, control speculative instructions will be chosen for schedule
9910 only if there are no other choices at the moment. This will make
9911 the use of the control speculation much more conservative.
9912 The default is 'disable'.
9914 @item -mno-sched-count-spec-in-critical-path
9915 @itemx -msched-count-spec-in-critical-path
9916 @opindex -mno-sched-count-spec-in-critical-path
9917 @opindex -msched-count-spec-in-critical-path
9918 If enabled, speculative dependencies will be considered during
9919 computation of the instructions priorities. This will make the use of the
9920 speculation a bit more conservative.
9921 The default is 'disable'.
9926 @subsection M32C Options
9927 @cindex M32C options
9930 @item -mcpu=@var{name}
9932 Select the CPU for which code is generated. @var{name} may be one of
9933 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
9934 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
9939 Specifies that the program will be run on the simulator. This causes
9940 an alternate runtime library to be linked in which supports, for
9941 example, file I/O. You must not use this option when generating
9942 programs that will run on real hardware; you must provide your own
9943 runtime library for whatever I/O functions are needed.
9945 @item -memregs=@var{number}
9947 Specifies the number of memory-based pseudo-registers GCC will use
9948 during code generation. These pseudo-registers will be used like real
9949 registers, so there is a tradeoff between GCC's ability to fit the
9950 code into available registers, and the performance penalty of using
9951 memory instead of registers. Note that all modules in a program must
9952 be compiled with the same value for this option. Because of that, you
9953 must not use this option with the default runtime libraries gcc
9958 @node M32R/D Options
9959 @subsection M32R/D Options
9960 @cindex M32R/D options
9962 These @option{-m} options are defined for Renesas M32R/D architectures:
9967 Generate code for the M32R/2@.
9971 Generate code for the M32R/X@.
9975 Generate code for the M32R@. This is the default.
9978 @opindex mmodel=small
9979 Assume all objects live in the lower 16MB of memory (so that their addresses
9980 can be loaded with the @code{ld24} instruction), and assume all subroutines
9981 are reachable with the @code{bl} instruction.
9982 This is the default.
9984 The addressability of a particular object can be set with the
9985 @code{model} attribute.
9987 @item -mmodel=medium
9988 @opindex mmodel=medium
9989 Assume objects may be anywhere in the 32-bit address space (the compiler
9990 will generate @code{seth/add3} instructions to load their addresses), and
9991 assume all subroutines are reachable with the @code{bl} instruction.
9994 @opindex mmodel=large
9995 Assume objects may be anywhere in the 32-bit address space (the compiler
9996 will generate @code{seth/add3} instructions to load their addresses), and
9997 assume subroutines may not be reachable with the @code{bl} instruction
9998 (the compiler will generate the much slower @code{seth/add3/jl}
9999 instruction sequence).
10002 @opindex msdata=none
10003 Disable use of the small data area. Variables will be put into
10004 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
10005 @code{section} attribute has been specified).
10006 This is the default.
10008 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
10009 Objects may be explicitly put in the small data area with the
10010 @code{section} attribute using one of these sections.
10012 @item -msdata=sdata
10013 @opindex msdata=sdata
10014 Put small global and static data in the small data area, but do not
10015 generate special code to reference them.
10018 @opindex msdata=use
10019 Put small global and static data in the small data area, and generate
10020 special instructions to reference them.
10024 @cindex smaller data references
10025 Put global and static objects less than or equal to @var{num} bytes
10026 into the small data or bss sections instead of the normal data or bss
10027 sections. The default value of @var{num} is 8.
10028 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
10029 for this option to have any effect.
10031 All modules should be compiled with the same @option{-G @var{num}} value.
10032 Compiling with different values of @var{num} may or may not work; if it
10033 doesn't the linker will give an error message---incorrect code will not be
10038 Makes the M32R specific code in the compiler display some statistics
10039 that might help in debugging programs.
10041 @item -malign-loops
10042 @opindex malign-loops
10043 Align all loops to a 32-byte boundary.
10045 @item -mno-align-loops
10046 @opindex mno-align-loops
10047 Do not enforce a 32-byte alignment for loops. This is the default.
10049 @item -missue-rate=@var{number}
10050 @opindex missue-rate=@var{number}
10051 Issue @var{number} instructions per cycle. @var{number} can only be 1
10054 @item -mbranch-cost=@var{number}
10055 @opindex mbranch-cost=@var{number}
10056 @var{number} can only be 1 or 2. If it is 1 then branches will be
10057 preferred over conditional code, if it is 2, then the opposite will
10060 @item -mflush-trap=@var{number}
10061 @opindex mflush-trap=@var{number}
10062 Specifies the trap number to use to flush the cache. The default is
10063 12. Valid numbers are between 0 and 15 inclusive.
10065 @item -mno-flush-trap
10066 @opindex mno-flush-trap
10067 Specifies that the cache cannot be flushed by using a trap.
10069 @item -mflush-func=@var{name}
10070 @opindex mflush-func=@var{name}
10071 Specifies the name of the operating system function to call to flush
10072 the cache. The default is @emph{_flush_cache}, but a function call
10073 will only be used if a trap is not available.
10075 @item -mno-flush-func
10076 @opindex mno-flush-func
10077 Indicates that there is no OS function for flushing the cache.
10081 @node M680x0 Options
10082 @subsection M680x0 Options
10083 @cindex M680x0 options
10085 These are the @samp{-m} options defined for the 68000 series. The default
10086 values for these options depends on which style of 68000 was selected when
10087 the compiler was configured; the defaults for the most common choices are
10095 Generate output for a 68000. This is the default
10096 when the compiler is configured for 68000-based systems.
10098 Use this option for microcontrollers with a 68000 or EC000 core,
10099 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
10105 Generate output for a 68020. This is the default
10106 when the compiler is configured for 68020-based systems.
10110 Generate output containing 68881 instructions for floating point.
10111 This is the default for most 68020 systems unless @option{--nfp} was
10112 specified when the compiler was configured.
10116 Generate output for a 68030. This is the default when the compiler is
10117 configured for 68030-based systems.
10121 Generate output for a 68040. This is the default when the compiler is
10122 configured for 68040-based systems.
10124 This option inhibits the use of 68881/68882 instructions that have to be
10125 emulated by software on the 68040. Use this option if your 68040 does not
10126 have code to emulate those instructions.
10130 Generate output for a 68060. This is the default when the compiler is
10131 configured for 68060-based systems.
10133 This option inhibits the use of 68020 and 68881/68882 instructions that
10134 have to be emulated by software on the 68060. Use this option if your 68060
10135 does not have code to emulate those instructions.
10139 Generate output for a CPU32. This is the default
10140 when the compiler is configured for CPU32-based systems.
10142 Use this option for microcontrollers with a
10143 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
10144 68336, 68340, 68341, 68349 and 68360.
10148 Generate output for a 520X ``coldfire'' family cpu. This is the default
10149 when the compiler is configured for 520X-based systems.
10151 Use this option for microcontroller with a 5200 core, including
10152 the MCF5202, MCF5203, MCF5204 and MCF5202.
10156 Generate output for a ColdFire V4e family cpu (e.g.@: 547x/548x).
10157 This includes use of hardware floating point instructions.
10161 Generate output for a 68040, without using any of the new instructions.
10162 This results in code which can run relatively efficiently on either a
10163 68020/68881 or a 68030 or a 68040. The generated code does use the
10164 68881 instructions that are emulated on the 68040.
10168 Generate output for a 68060, without using any of the new instructions.
10169 This results in code which can run relatively efficiently on either a
10170 68020/68881 or a 68030 or a 68040. The generated code does use the
10171 68881 instructions that are emulated on the 68060.
10174 @opindex msoft-float
10175 Generate output containing library calls for floating point.
10176 @strong{Warning:} the requisite libraries are not available for all m68k
10177 targets. Normally the facilities of the machine's usual C compiler are
10178 used, but this can't be done directly in cross-compilation. You must
10179 make your own arrangements to provide suitable library functions for
10180 cross-compilation. The embedded targets @samp{m68k-*-aout} and
10181 @samp{m68k-*-coff} do provide software floating point support.
10185 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10186 Additionally, parameters passed on the stack are also aligned to a
10187 16-bit boundary even on targets whose API mandates promotion to 32-bit.
10190 @opindex mnobitfield
10191 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
10192 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
10196 Do use the bit-field instructions. The @option{-m68020} option implies
10197 @option{-mbitfield}. This is the default if you use a configuration
10198 designed for a 68020.
10202 Use a different function-calling convention, in which functions
10203 that take a fixed number of arguments return with the @code{rtd}
10204 instruction, which pops their arguments while returning. This
10205 saves one instruction in the caller since there is no need to pop
10206 the arguments there.
10208 This calling convention is incompatible with the one normally
10209 used on Unix, so you cannot use it if you need to call libraries
10210 compiled with the Unix compiler.
10212 Also, you must provide function prototypes for all functions that
10213 take variable numbers of arguments (including @code{printf});
10214 otherwise incorrect code will be generated for calls to those
10217 In addition, seriously incorrect code will result if you call a
10218 function with too many arguments. (Normally, extra arguments are
10219 harmlessly ignored.)
10221 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
10222 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
10225 @itemx -mno-align-int
10226 @opindex malign-int
10227 @opindex mno-align-int
10228 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
10229 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
10230 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
10231 Aligning variables on 32-bit boundaries produces code that runs somewhat
10232 faster on processors with 32-bit busses at the expense of more memory.
10234 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
10235 align structures containing the above types differently than
10236 most published application binary interface specifications for the m68k.
10240 Use the pc-relative addressing mode of the 68000 directly, instead of
10241 using a global offset table. At present, this option implies @option{-fpic},
10242 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
10243 not presently supported with @option{-mpcrel}, though this could be supported for
10244 68020 and higher processors.
10246 @item -mno-strict-align
10247 @itemx -mstrict-align
10248 @opindex mno-strict-align
10249 @opindex mstrict-align
10250 Do not (do) assume that unaligned memory references will be handled by
10254 Generate code that allows the data segment to be located in a different
10255 area of memory from the text segment. This allows for execute in place in
10256 an environment without virtual memory management. This option implies
10259 @item -mno-sep-data
10260 Generate code that assumes that the data segment follows the text segment.
10261 This is the default.
10263 @item -mid-shared-library
10264 Generate code that supports shared libraries via the library ID method.
10265 This allows for execute in place and shared libraries in an environment
10266 without virtual memory management. This option implies @option{-fPIC}.
10268 @item -mno-id-shared-library
10269 Generate code that doesn't assume ID based shared libraries are being used.
10270 This is the default.
10272 @item -mshared-library-id=n
10273 Specified the identification number of the ID based shared library being
10274 compiled. Specifying a value of 0 will generate more compact code, specifying
10275 other values will force the allocation of that number to the current
10276 library but is no more space or time efficient than omitting this option.
10280 @node M68hc1x Options
10281 @subsection M68hc1x Options
10282 @cindex M68hc1x options
10284 These are the @samp{-m} options defined for the 68hc11 and 68hc12
10285 microcontrollers. The default values for these options depends on
10286 which style of microcontroller was selected when the compiler was configured;
10287 the defaults for the most common choices are given below.
10294 Generate output for a 68HC11. This is the default
10295 when the compiler is configured for 68HC11-based systems.
10301 Generate output for a 68HC12. This is the default
10302 when the compiler is configured for 68HC12-based systems.
10308 Generate output for a 68HCS12.
10310 @item -mauto-incdec
10311 @opindex mauto-incdec
10312 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
10319 Enable the use of 68HC12 min and max instructions.
10322 @itemx -mno-long-calls
10323 @opindex mlong-calls
10324 @opindex mno-long-calls
10325 Treat all calls as being far away (near). If calls are assumed to be
10326 far away, the compiler will use the @code{call} instruction to
10327 call a function and the @code{rtc} instruction for returning.
10331 Consider type @code{int} to be 16 bits wide, like @code{short int}.
10333 @item -msoft-reg-count=@var{count}
10334 @opindex msoft-reg-count
10335 Specify the number of pseudo-soft registers which are used for the
10336 code generation. The maximum number is 32. Using more pseudo-soft
10337 register may or may not result in better code depending on the program.
10338 The default is 4 for 68HC11 and 2 for 68HC12.
10342 @node MCore Options
10343 @subsection MCore Options
10344 @cindex MCore options
10346 These are the @samp{-m} options defined for the Motorola M*Core
10352 @itemx -mno-hardlit
10354 @opindex mno-hardlit
10355 Inline constants into the code stream if it can be done in two
10356 instructions or less.
10362 Use the divide instruction. (Enabled by default).
10364 @item -mrelax-immediate
10365 @itemx -mno-relax-immediate
10366 @opindex mrelax-immediate
10367 @opindex mno-relax-immediate
10368 Allow arbitrary sized immediates in bit operations.
10370 @item -mwide-bitfields
10371 @itemx -mno-wide-bitfields
10372 @opindex mwide-bitfields
10373 @opindex mno-wide-bitfields
10374 Always treat bit-fields as int-sized.
10376 @item -m4byte-functions
10377 @itemx -mno-4byte-functions
10378 @opindex m4byte-functions
10379 @opindex mno-4byte-functions
10380 Force all functions to be aligned to a four byte boundary.
10382 @item -mcallgraph-data
10383 @itemx -mno-callgraph-data
10384 @opindex mcallgraph-data
10385 @opindex mno-callgraph-data
10386 Emit callgraph information.
10389 @itemx -mno-slow-bytes
10390 @opindex mslow-bytes
10391 @opindex mno-slow-bytes
10392 Prefer word access when reading byte quantities.
10394 @item -mlittle-endian
10395 @itemx -mbig-endian
10396 @opindex mlittle-endian
10397 @opindex mbig-endian
10398 Generate code for a little endian target.
10404 Generate code for the 210 processor.
10408 @subsection MIPS Options
10409 @cindex MIPS options
10415 Generate big-endian code.
10419 Generate little-endian code. This is the default for @samp{mips*el-*-*}
10422 @item -march=@var{arch}
10424 Generate code that will run on @var{arch}, which can be the name of a
10425 generic MIPS ISA, or the name of a particular processor.
10427 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
10428 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
10429 The processor names are:
10430 @samp{4kc}, @samp{4km}, @samp{4kp},
10431 @samp{5kc}, @samp{5kf},
10433 @samp{24k}, @samp{24kc}, @samp{24kf}, @samp{24kx},
10436 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
10437 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
10438 @samp{rm7000}, @samp{rm9000},
10441 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
10442 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
10443 The special value @samp{from-abi} selects the
10444 most compatible architecture for the selected ABI (that is,
10445 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
10447 In processor names, a final @samp{000} can be abbreviated as @samp{k}
10448 (for example, @samp{-march=r2k}). Prefixes are optional, and
10449 @samp{vr} may be written @samp{r}.
10451 GCC defines two macros based on the value of this option. The first
10452 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
10453 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
10454 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
10455 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
10456 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
10458 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
10459 above. In other words, it will have the full prefix and will not
10460 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
10461 the macro names the resolved architecture (either @samp{"mips1"} or
10462 @samp{"mips3"}). It names the default architecture when no
10463 @option{-march} option is given.
10465 @item -mtune=@var{arch}
10467 Optimize for @var{arch}. Among other things, this option controls
10468 the way instructions are scheduled, and the perceived cost of arithmetic
10469 operations. The list of @var{arch} values is the same as for
10472 When this option is not used, GCC will optimize for the processor
10473 specified by @option{-march}. By using @option{-march} and
10474 @option{-mtune} together, it is possible to generate code that will
10475 run on a family of processors, but optimize the code for one
10476 particular member of that family.
10478 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
10479 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
10480 @samp{-march} ones described above.
10484 Equivalent to @samp{-march=mips1}.
10488 Equivalent to @samp{-march=mips2}.
10492 Equivalent to @samp{-march=mips3}.
10496 Equivalent to @samp{-march=mips4}.
10500 Equivalent to @samp{-march=mips32}.
10504 Equivalent to @samp{-march=mips32r2}.
10508 Equivalent to @samp{-march=mips64}.
10513 @opindex mno-mips16
10514 Generate (do not generate) MIPS16 code. If GCC is targetting a
10515 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
10527 Generate code for the given ABI@.
10529 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
10530 generates 64-bit code when you select a 64-bit architecture, but you
10531 can use @option{-mgp32} to get 32-bit code instead.
10533 For information about the O64 ABI, see
10534 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
10537 @itemx -mno-abicalls
10539 @opindex mno-abicalls
10540 Generate (do not generate) code that is suitable for SVR4-style
10541 dynamic objects. @option{-mabicalls} is the default for SVR4-based
10546 Generate (do not generate) code that is fully position-independent,
10547 and that can therefore be linked into shared libraries. This option
10548 only affects @option{-mabicalls}.
10550 All @option{-mabicalls} code has traditionally been position-independent,
10551 regardless of options like @option{-fPIC} and @option{-fpic}. However,
10552 as an extension, the GNU toolchain allows executables to use absolute
10553 accesses for locally-binding symbols. It can also use shorter GP
10554 initialization sequences and generate direct calls to locally-defined
10555 functions. This mode is selected by @option{-mno-shared}.
10557 @option{-mno-shared} depends on binutils 2.16 or higher and generates
10558 objects that can only be linked by the GNU linker. However, the option
10559 does not affect the ABI of the final executable; it only affects the ABI
10560 of relocatable objects. Using @option{-mno-shared} will generally make
10561 executables both smaller and quicker.
10563 @option{-mshared} is the default.
10569 Lift (do not lift) the usual restrictions on the size of the global
10572 GCC normally uses a single instruction to load values from the GOT@.
10573 While this is relatively efficient, it will only work if the GOT
10574 is smaller than about 64k. Anything larger will cause the linker
10575 to report an error such as:
10577 @cindex relocation truncated to fit (MIPS)
10579 relocation truncated to fit: R_MIPS_GOT16 foobar
10582 If this happens, you should recompile your code with @option{-mxgot}.
10583 It should then work with very large GOTs, although it will also be
10584 less efficient, since it will take three instructions to fetch the
10585 value of a global symbol.
10587 Note that some linkers can create multiple GOTs. If you have such a
10588 linker, you should only need to use @option{-mxgot} when a single object
10589 file accesses more than 64k's worth of GOT entries. Very few do.
10591 These options have no effect unless GCC is generating position
10596 Assume that general-purpose registers are 32 bits wide.
10600 Assume that general-purpose registers are 64 bits wide.
10604 Assume that floating-point registers are 32 bits wide.
10608 Assume that floating-point registers are 64 bits wide.
10611 @opindex mhard-float
10612 Use floating-point coprocessor instructions.
10615 @opindex msoft-float
10616 Do not use floating-point coprocessor instructions. Implement
10617 floating-point calculations using library calls instead.
10619 @item -msingle-float
10620 @opindex msingle-float
10621 Assume that the floating-point coprocessor only supports single-precision
10624 @itemx -mdouble-float
10625 @opindex mdouble-float
10626 Assume that the floating-point coprocessor supports double-precision
10627 operations. This is the default.
10633 Use (do not use) the MIPS DSP ASE. @xref{MIPS DSP Built-in Functions}.
10635 @itemx -mpaired-single
10636 @itemx -mno-paired-single
10637 @opindex mpaired-single
10638 @opindex mno-paired-single
10639 Use (do not use) paired-single floating-point instructions.
10640 @xref{MIPS Paired-Single Support}. This option can only be used
10641 when generating 64-bit code and requires hardware floating-point
10642 support to be enabled.
10647 @opindex mno-mips3d
10648 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
10649 The option @option{-mips3d} implies @option{-mpaired-single}.
10653 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
10654 an explanation of the default and the way that the pointer size is
10659 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
10661 The default size of @code{int}s, @code{long}s and pointers depends on
10662 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
10663 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
10664 32-bit @code{long}s. Pointers are the same size as @code{long}s,
10665 or the same size as integer registers, whichever is smaller.
10671 Assume (do not assume) that all symbols have 32-bit values, regardless
10672 of the selected ABI@. This option is useful in combination with
10673 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
10674 to generate shorter and faster references to symbolic addresses.
10678 @cindex smaller data references (MIPS)
10679 @cindex gp-relative references (MIPS)
10680 Put global and static items less than or equal to @var{num} bytes into
10681 the small data or bss section instead of the normal data or bss section.
10682 This allows the data to be accessed using a single instruction.
10684 All modules should be compiled with the same @option{-G @var{num}}
10687 @item -membedded-data
10688 @itemx -mno-embedded-data
10689 @opindex membedded-data
10690 @opindex mno-embedded-data
10691 Allocate variables to the read-only data section first if possible, then
10692 next in the small data section if possible, otherwise in data. This gives
10693 slightly slower code than the default, but reduces the amount of RAM required
10694 when executing, and thus may be preferred for some embedded systems.
10696 @item -muninit-const-in-rodata
10697 @itemx -mno-uninit-const-in-rodata
10698 @opindex muninit-const-in-rodata
10699 @opindex mno-uninit-const-in-rodata
10700 Put uninitialized @code{const} variables in the read-only data section.
10701 This option is only meaningful in conjunction with @option{-membedded-data}.
10703 @item -msplit-addresses
10704 @itemx -mno-split-addresses
10705 @opindex msplit-addresses
10706 @opindex mno-split-addresses
10707 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
10708 relocation operators. This option has been superseded by
10709 @option{-mexplicit-relocs} but is retained for backwards compatibility.
10711 @item -mexplicit-relocs
10712 @itemx -mno-explicit-relocs
10713 @opindex mexplicit-relocs
10714 @opindex mno-explicit-relocs
10715 Use (do not use) assembler relocation operators when dealing with symbolic
10716 addresses. The alternative, selected by @option{-mno-explicit-relocs},
10717 is to use assembler macros instead.
10719 @option{-mexplicit-relocs} is the default if GCC was configured
10720 to use an assembler that supports relocation operators.
10722 @item -mcheck-zero-division
10723 @itemx -mno-check-zero-division
10724 @opindex mcheck-zero-division
10725 @opindex mno-check-zero-division
10726 Trap (do not trap) on integer division by zero. The default is
10727 @option{-mcheck-zero-division}.
10729 @item -mdivide-traps
10730 @itemx -mdivide-breaks
10731 @opindex mdivide-traps
10732 @opindex mdivide-breaks
10733 MIPS systems check for division by zero by generating either a
10734 conditional trap or a break instruction. Using traps results in
10735 smaller code, but is only supported on MIPS II and later. Also, some
10736 versions of the Linux kernel have a bug that prevents trap from
10737 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
10738 allow conditional traps on architectures that support them and
10739 @option{-mdivide-breaks} to force the use of breaks.
10741 The default is usually @option{-mdivide-traps}, but this can be
10742 overridden at configure time using @option{--with-divide=breaks}.
10743 Divide-by-zero checks can be completely disabled using
10744 @option{-mno-check-zero-division}.
10749 @opindex mno-memcpy
10750 Force (do not force) the use of @code{memcpy()} for non-trivial block
10751 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
10752 most constant-sized copies.
10755 @itemx -mno-long-calls
10756 @opindex mlong-calls
10757 @opindex mno-long-calls
10758 Disable (do not disable) use of the @code{jal} instruction. Calling
10759 functions using @code{jal} is more efficient but requires the caller
10760 and callee to be in the same 256 megabyte segment.
10762 This option has no effect on abicalls code. The default is
10763 @option{-mno-long-calls}.
10769 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
10770 instructions, as provided by the R4650 ISA@.
10773 @itemx -mno-fused-madd
10774 @opindex mfused-madd
10775 @opindex mno-fused-madd
10776 Enable (disable) use of the floating point multiply-accumulate
10777 instructions, when they are available. The default is
10778 @option{-mfused-madd}.
10780 When multiply-accumulate instructions are used, the intermediate
10781 product is calculated to infinite precision and is not subject to
10782 the FCSR Flush to Zero bit. This may be undesirable in some
10787 Tell the MIPS assembler to not run its preprocessor over user
10788 assembler files (with a @samp{.s} suffix) when assembling them.
10791 @itemx -mno-fix-r4000
10792 @opindex mfix-r4000
10793 @opindex mno-fix-r4000
10794 Work around certain R4000 CPU errata:
10797 A double-word or a variable shift may give an incorrect result if executed
10798 immediately after starting an integer division.
10800 A double-word or a variable shift may give an incorrect result if executed
10801 while an integer multiplication is in progress.
10803 An integer division may give an incorrect result if started in a delay slot
10804 of a taken branch or a jump.
10808 @itemx -mno-fix-r4400
10809 @opindex mfix-r4400
10810 @opindex mno-fix-r4400
10811 Work around certain R4400 CPU errata:
10814 A double-word or a variable shift may give an incorrect result if executed
10815 immediately after starting an integer division.
10819 @itemx -mno-fix-vr4120
10820 @opindex mfix-vr4120
10821 Work around certain VR4120 errata:
10824 @code{dmultu} does not always produce the correct result.
10826 @code{div} and @code{ddiv} do not always produce the correct result if one
10827 of the operands is negative.
10829 The workarounds for the division errata rely on special functions in
10830 @file{libgcc.a}. At present, these functions are only provided by
10831 the @code{mips64vr*-elf} configurations.
10833 Other VR4120 errata require a nop to be inserted between certain pairs of
10834 instructions. These errata are handled by the assembler, not by GCC itself.
10837 @opindex mfix-vr4130
10838 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
10839 workarounds are implemented by the assembler rather than by GCC,
10840 although GCC will avoid using @code{mflo} and @code{mfhi} if the
10841 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
10842 instructions are available instead.
10845 @itemx -mno-fix-sb1
10847 Work around certain SB-1 CPU core errata.
10848 (This flag currently works around the SB-1 revision 2
10849 ``F1'' and ``F2'' floating point errata.)
10851 @item -mflush-func=@var{func}
10852 @itemx -mno-flush-func
10853 @opindex mflush-func
10854 Specifies the function to call to flush the I and D caches, or to not
10855 call any such function. If called, the function must take the same
10856 arguments as the common @code{_flush_func()}, that is, the address of the
10857 memory range for which the cache is being flushed, the size of the
10858 memory range, and the number 3 (to flush both caches). The default
10859 depends on the target GCC was configured for, but commonly is either
10860 @samp{_flush_func} or @samp{__cpu_flush}.
10862 @item -mbranch-likely
10863 @itemx -mno-branch-likely
10864 @opindex mbranch-likely
10865 @opindex mno-branch-likely
10866 Enable or disable use of Branch Likely instructions, regardless of the
10867 default for the selected architecture. By default, Branch Likely
10868 instructions may be generated if they are supported by the selected
10869 architecture. An exception is for the MIPS32 and MIPS64 architectures
10870 and processors which implement those architectures; for those, Branch
10871 Likely instructions will not be generated by default because the MIPS32
10872 and MIPS64 architectures specifically deprecate their use.
10874 @item -mfp-exceptions
10875 @itemx -mno-fp-exceptions
10876 @opindex mfp-exceptions
10877 Specifies whether FP exceptions are enabled. This affects how we schedule
10878 FP instructions for some processors. The default is that FP exceptions are
10881 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
10882 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
10885 @item -mvr4130-align
10886 @itemx -mno-vr4130-align
10887 @opindex mvr4130-align
10888 The VR4130 pipeline is two-way superscalar, but can only issue two
10889 instructions together if the first one is 8-byte aligned. When this
10890 option is enabled, GCC will align pairs of instructions that it
10891 thinks should execute in parallel.
10893 This option only has an effect when optimizing for the VR4130.
10894 It normally makes code faster, but at the expense of making it bigger.
10895 It is enabled by default at optimization level @option{-O3}.
10899 @subsection MMIX Options
10900 @cindex MMIX Options
10902 These options are defined for the MMIX:
10906 @itemx -mno-libfuncs
10908 @opindex mno-libfuncs
10909 Specify that intrinsic library functions are being compiled, passing all
10910 values in registers, no matter the size.
10913 @itemx -mno-epsilon
10915 @opindex mno-epsilon
10916 Generate floating-point comparison instructions that compare with respect
10917 to the @code{rE} epsilon register.
10919 @item -mabi=mmixware
10921 @opindex mabi-mmixware
10923 Generate code that passes function parameters and return values that (in
10924 the called function) are seen as registers @code{$0} and up, as opposed to
10925 the GNU ABI which uses global registers @code{$231} and up.
10927 @item -mzero-extend
10928 @itemx -mno-zero-extend
10929 @opindex mzero-extend
10930 @opindex mno-zero-extend
10931 When reading data from memory in sizes shorter than 64 bits, use (do not
10932 use) zero-extending load instructions by default, rather than
10933 sign-extending ones.
10936 @itemx -mno-knuthdiv
10938 @opindex mno-knuthdiv
10939 Make the result of a division yielding a remainder have the same sign as
10940 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10941 remainder follows the sign of the dividend. Both methods are
10942 arithmetically valid, the latter being almost exclusively used.
10944 @item -mtoplevel-symbols
10945 @itemx -mno-toplevel-symbols
10946 @opindex mtoplevel-symbols
10947 @opindex mno-toplevel-symbols
10948 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10949 code can be used with the @code{PREFIX} assembly directive.
10953 Generate an executable in the ELF format, rather than the default
10954 @samp{mmo} format used by the @command{mmix} simulator.
10956 @item -mbranch-predict
10957 @itemx -mno-branch-predict
10958 @opindex mbranch-predict
10959 @opindex mno-branch-predict
10960 Use (do not use) the probable-branch instructions, when static branch
10961 prediction indicates a probable branch.
10963 @item -mbase-addresses
10964 @itemx -mno-base-addresses
10965 @opindex mbase-addresses
10966 @opindex mno-base-addresses
10967 Generate (do not generate) code that uses @emph{base addresses}. Using a
10968 base address automatically generates a request (handled by the assembler
10969 and the linker) for a constant to be set up in a global register. The
10970 register is used for one or more base address requests within the range 0
10971 to 255 from the value held in the register. The generally leads to short
10972 and fast code, but the number of different data items that can be
10973 addressed is limited. This means that a program that uses lots of static
10974 data may require @option{-mno-base-addresses}.
10976 @item -msingle-exit
10977 @itemx -mno-single-exit
10978 @opindex msingle-exit
10979 @opindex mno-single-exit
10980 Force (do not force) generated code to have a single exit point in each
10984 @node MN10300 Options
10985 @subsection MN10300 Options
10986 @cindex MN10300 options
10988 These @option{-m} options are defined for Matsushita MN10300 architectures:
10993 Generate code to avoid bugs in the multiply instructions for the MN10300
10994 processors. This is the default.
10996 @item -mno-mult-bug
10997 @opindex mno-mult-bug
10998 Do not generate code to avoid bugs in the multiply instructions for the
10999 MN10300 processors.
11003 Generate code which uses features specific to the AM33 processor.
11007 Do not generate code which uses features specific to the AM33 processor. This
11010 @item -mreturn-pointer-on-d0
11011 @opindex mreturn-pointer-on-d0
11012 When generating a function which returns a pointer, return the pointer
11013 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
11014 only in a0, and attempts to call such functions without a prototype
11015 would result in errors. Note that this option is on by default; use
11016 @option{-mno-return-pointer-on-d0} to disable it.
11020 Do not link in the C run-time initialization object file.
11024 Indicate to the linker that it should perform a relaxation optimization pass
11025 to shorten branches, calls and absolute memory addresses. This option only
11026 has an effect when used on the command line for the final link step.
11028 This option makes symbolic debugging impossible.
11032 @subsection MT Options
11035 These @option{-m} options are defined for Morpho MT architectures:
11039 @item -march=@var{cpu-type}
11041 Generate code that will run on @var{cpu-type}, which is the name of a system
11042 representing a certain processor type. Possible values for
11043 @var{cpu-type} are @samp{ms1-64-001}, @samp{ms1-16-002},
11044 @samp{ms1-16-003} and @samp{ms2}.
11046 When this option is not used, the default is @option{-march=ms1-16-002}.
11050 Use byte loads and stores when generating code.
11054 Do not use byte loads and stores when generating code.
11058 Use simulator runtime
11062 Do not link in the C run-time initialization object file
11063 @file{crti.o}. Other run-time initialization and termination files
11064 such as @file{startup.o} and @file{exit.o} are still included on the
11065 linker command line.
11069 @node PDP-11 Options
11070 @subsection PDP-11 Options
11071 @cindex PDP-11 Options
11073 These options are defined for the PDP-11:
11078 Use hardware FPP floating point. This is the default. (FIS floating
11079 point on the PDP-11/40 is not supported.)
11082 @opindex msoft-float
11083 Do not use hardware floating point.
11087 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
11091 Return floating-point results in memory. This is the default.
11095 Generate code for a PDP-11/40.
11099 Generate code for a PDP-11/45. This is the default.
11103 Generate code for a PDP-11/10.
11105 @item -mbcopy-builtin
11106 @opindex bcopy-builtin
11107 Use inline @code{movmemhi} patterns for copying memory. This is the
11112 Do not use inline @code{movmemhi} patterns for copying memory.
11118 Use 16-bit @code{int}. This is the default.
11124 Use 32-bit @code{int}.
11127 @itemx -mno-float32
11129 @opindex mno-float32
11130 Use 64-bit @code{float}. This is the default.
11133 @itemx -mno-float64
11135 @opindex mno-float64
11136 Use 32-bit @code{float}.
11140 Use @code{abshi2} pattern. This is the default.
11144 Do not use @code{abshi2} pattern.
11146 @item -mbranch-expensive
11147 @opindex mbranch-expensive
11148 Pretend that branches are expensive. This is for experimenting with
11149 code generation only.
11151 @item -mbranch-cheap
11152 @opindex mbranch-cheap
11153 Do not pretend that branches are expensive. This is the default.
11157 Generate code for a system with split I&D@.
11161 Generate code for a system without split I&D@. This is the default.
11165 Use Unix assembler syntax. This is the default when configured for
11166 @samp{pdp11-*-bsd}.
11170 Use DEC assembler syntax. This is the default when configured for any
11171 PDP-11 target other than @samp{pdp11-*-bsd}.
11174 @node PowerPC Options
11175 @subsection PowerPC Options
11176 @cindex PowerPC options
11178 These are listed under @xref{RS/6000 and PowerPC Options}.
11180 @node RS/6000 and PowerPC Options
11181 @subsection IBM RS/6000 and PowerPC Options
11182 @cindex RS/6000 and PowerPC Options
11183 @cindex IBM RS/6000 and PowerPC Options
11185 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
11192 @itemx -mno-powerpc
11193 @itemx -mpowerpc-gpopt
11194 @itemx -mno-powerpc-gpopt
11195 @itemx -mpowerpc-gfxopt
11196 @itemx -mno-powerpc-gfxopt
11198 @itemx -mno-powerpc64
11202 @itemx -mno-popcntb
11208 @opindex mno-power2
11210 @opindex mno-powerpc
11211 @opindex mpowerpc-gpopt
11212 @opindex mno-powerpc-gpopt
11213 @opindex mpowerpc-gfxopt
11214 @opindex mno-powerpc-gfxopt
11215 @opindex mpowerpc64
11216 @opindex mno-powerpc64
11220 @opindex mno-popcntb
11223 GCC supports two related instruction set architectures for the
11224 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
11225 instructions supported by the @samp{rios} chip set used in the original
11226 RS/6000 systems and the @dfn{PowerPC} instruction set is the
11227 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
11228 the IBM 4xx, 6xx, and follow-on microprocessors.
11230 Neither architecture is a subset of the other. However there is a
11231 large common subset of instructions supported by both. An MQ
11232 register is included in processors supporting the POWER architecture.
11234 You use these options to specify which instructions are available on the
11235 processor you are using. The default value of these options is
11236 determined when configuring GCC@. Specifying the
11237 @option{-mcpu=@var{cpu_type}} overrides the specification of these
11238 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
11239 rather than the options listed above.
11241 The @option{-mpower} option allows GCC to generate instructions that
11242 are found only in the POWER architecture and to use the MQ register.
11243 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
11244 to generate instructions that are present in the POWER2 architecture but
11245 not the original POWER architecture.
11247 The @option{-mpowerpc} option allows GCC to generate instructions that
11248 are found only in the 32-bit subset of the PowerPC architecture.
11249 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
11250 GCC to use the optional PowerPC architecture instructions in the
11251 General Purpose group, including floating-point square root. Specifying
11252 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
11253 use the optional PowerPC architecture instructions in the Graphics
11254 group, including floating-point select.
11256 The @option{-mmfcrf} option allows GCC to generate the move from
11257 condition register field instruction implemented on the POWER4
11258 processor and other processors that support the PowerPC V2.01
11260 The @option{-mpopcntb} option allows GCC to generate the popcount and
11261 double precision FP reciprocal estimate instruction implemented on the
11262 POWER5 processor and other processors that support the PowerPC V2.02
11264 The @option{-mfprnd} option allows GCC to generate the FP round to
11265 integer instructions implemented on the POWER5+ processor and other
11266 processors that support the PowerPC V2.03 architecture.
11268 The @option{-mpowerpc64} option allows GCC to generate the additional
11269 64-bit instructions that are found in the full PowerPC64 architecture
11270 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
11271 @option{-mno-powerpc64}.
11273 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
11274 will use only the instructions in the common subset of both
11275 architectures plus some special AIX common-mode calls, and will not use
11276 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
11277 permits GCC to use any instruction from either architecture and to
11278 allow use of the MQ register; specify this for the Motorola MPC601.
11280 @item -mnew-mnemonics
11281 @itemx -mold-mnemonics
11282 @opindex mnew-mnemonics
11283 @opindex mold-mnemonics
11284 Select which mnemonics to use in the generated assembler code. With
11285 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
11286 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
11287 assembler mnemonics defined for the POWER architecture. Instructions
11288 defined in only one architecture have only one mnemonic; GCC uses that
11289 mnemonic irrespective of which of these options is specified.
11291 GCC defaults to the mnemonics appropriate for the architecture in
11292 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
11293 value of these option. Unless you are building a cross-compiler, you
11294 should normally not specify either @option{-mnew-mnemonics} or
11295 @option{-mold-mnemonics}, but should instead accept the default.
11297 @item -mcpu=@var{cpu_type}
11299 Set architecture type, register usage, choice of mnemonics, and
11300 instruction scheduling parameters for machine type @var{cpu_type}.
11301 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
11302 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
11303 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
11304 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
11305 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
11306 @samp{860}, @samp{970}, @samp{8540}, @samp{ec603e}, @samp{G3},
11307 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
11308 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6},
11309 @samp{common}, @samp{powerpc}, @samp{powerpc64},
11310 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
11312 @option{-mcpu=common} selects a completely generic processor. Code
11313 generated under this option will run on any POWER or PowerPC processor.
11314 GCC will use only the instructions in the common subset of both
11315 architectures, and will not use the MQ register. GCC assumes a generic
11316 processor model for scheduling purposes.
11318 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
11319 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
11320 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
11321 types, with an appropriate, generic processor model assumed for
11322 scheduling purposes.
11324 The other options specify a specific processor. Code generated under
11325 those options will run best on that processor, and may not run at all on
11328 The @option{-mcpu} options automatically enable or disable the
11329 following options: @option{-maltivec}, @option{-mfprnd},
11330 @option{-mhard-float}, @option{-mmfcrf}, @option{-mmultiple},
11331 @option{-mnew-mnemonics}, @option{-mpopcntb}, @option{-mpower},
11332 @option{-mpower2}, @option{-mpowerpc64}, @option{-mpowerpc-gpopt},
11333 @option{-mpowerpc-gfxopt}, @option{-mstring}, @option{-mmulhw}, @option{dlmzb}.
11334 The particular options
11335 set for any particular CPU will vary between compiler versions,
11336 depending on what setting seems to produce optimal code for that CPU;
11337 it doesn't necessarily reflect the actual hardware's capabilities. If
11338 you wish to set an individual option to a particular value, you may
11339 specify it after the @option{-mcpu} option, like @samp{-mcpu=970
11342 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
11343 not enabled or disabled by the @option{-mcpu} option at present because
11344 AIX does not have full support for these options. You may still
11345 enable or disable them individually if you're sure it'll work in your
11348 @item -mtune=@var{cpu_type}
11350 Set the instruction scheduling parameters for machine type
11351 @var{cpu_type}, but do not set the architecture type, register usage, or
11352 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
11353 values for @var{cpu_type} are used for @option{-mtune} as for
11354 @option{-mcpu}. If both are specified, the code generated will use the
11355 architecture, registers, and mnemonics set by @option{-mcpu}, but the
11356 scheduling parameters set by @option{-mtune}.
11362 Generate code to compute division as reciprocal estimate and iterative
11363 refinement, creating opportunities for increased throughput. This
11364 feature requires: optional PowerPC Graphics instruction set for single
11365 precision and FRE instruction for double precision, assuming divides
11366 cannot generate user-visible traps, and the domain values not include
11367 Infinities, denormals or zero denominator.
11370 @itemx -mno-altivec
11372 @opindex mno-altivec
11373 Generate code that uses (does not use) AltiVec instructions, and also
11374 enable the use of built-in functions that allow more direct access to
11375 the AltiVec instruction set. You may also need to set
11376 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
11382 @opindex mno-vrsave
11383 Generate VRSAVE instructions when generating AltiVec code.
11386 @opindex msecure-plt
11387 Generate code that allows ld and ld.so to build executables and shared
11388 libraries with non-exec .plt and .got sections. This is a PowerPC
11389 32-bit SYSV ABI option.
11393 Generate code that uses a BSS .plt section that ld.so fills in, and
11394 requires .plt and .got sections that are both writable and executable.
11395 This is a PowerPC 32-bit SYSV ABI option.
11401 This switch enables or disables the generation of ISEL instructions.
11403 @item -misel=@var{yes/no}
11404 This switch has been deprecated. Use @option{-misel} and
11405 @option{-mno-isel} instead.
11411 This switch enables or disables the generation of SPE simd
11414 @item -mspe=@var{yes/no}
11415 This option has been deprecated. Use @option{-mspe} and
11416 @option{-mno-spe} instead.
11418 @item -mfloat-gprs=@var{yes/single/double/no}
11419 @itemx -mfloat-gprs
11420 @opindex mfloat-gprs
11421 This switch enables or disables the generation of floating point
11422 operations on the general purpose registers for architectures that
11425 The argument @var{yes} or @var{single} enables the use of
11426 single-precision floating point operations.
11428 The argument @var{double} enables the use of single and
11429 double-precision floating point operations.
11431 The argument @var{no} disables floating point operations on the
11432 general purpose registers.
11434 This option is currently only available on the MPC854x.
11440 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
11441 targets (including GNU/Linux). The 32-bit environment sets int, long
11442 and pointer to 32 bits and generates code that runs on any PowerPC
11443 variant. The 64-bit environment sets int to 32 bits and long and
11444 pointer to 64 bits, and generates code for PowerPC64, as for
11445 @option{-mpowerpc64}.
11448 @itemx -mno-fp-in-toc
11449 @itemx -mno-sum-in-toc
11450 @itemx -mminimal-toc
11452 @opindex mno-fp-in-toc
11453 @opindex mno-sum-in-toc
11454 @opindex mminimal-toc
11455 Modify generation of the TOC (Table Of Contents), which is created for
11456 every executable file. The @option{-mfull-toc} option is selected by
11457 default. In that case, GCC will allocate at least one TOC entry for
11458 each unique non-automatic variable reference in your program. GCC
11459 will also place floating-point constants in the TOC@. However, only
11460 16,384 entries are available in the TOC@.
11462 If you receive a linker error message that saying you have overflowed
11463 the available TOC space, you can reduce the amount of TOC space used
11464 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
11465 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
11466 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
11467 generate code to calculate the sum of an address and a constant at
11468 run-time instead of putting that sum into the TOC@. You may specify one
11469 or both of these options. Each causes GCC to produce very slightly
11470 slower and larger code at the expense of conserving TOC space.
11472 If you still run out of space in the TOC even when you specify both of
11473 these options, specify @option{-mminimal-toc} instead. This option causes
11474 GCC to make only one TOC entry for every file. When you specify this
11475 option, GCC will produce code that is slower and larger but which
11476 uses extremely little TOC space. You may wish to use this option
11477 only on files that contain less frequently executed code.
11483 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
11484 @code{long} type, and the infrastructure needed to support them.
11485 Specifying @option{-maix64} implies @option{-mpowerpc64} and
11486 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
11487 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
11490 @itemx -mno-xl-compat
11491 @opindex mxl-compat
11492 @opindex mno-xl-compat
11493 Produce code that conforms more closely to IBM XL compiler semantics
11494 when using AIX-compatible ABI. Pass floating-point arguments to
11495 prototyped functions beyond the register save area (RSA) on the stack
11496 in addition to argument FPRs. Do not assume that most significant
11497 double in 128-bit long double value is properly rounded when comparing
11498 values and converting to double. Use XL symbol names for long double
11501 The AIX calling convention was extended but not initially documented to
11502 handle an obscure K&R C case of calling a function that takes the
11503 address of its arguments with fewer arguments than declared. IBM XL
11504 compilers access floating point arguments which do not fit in the
11505 RSA from the stack when a subroutine is compiled without
11506 optimization. Because always storing floating-point arguments on the
11507 stack is inefficient and rarely needed, this option is not enabled by
11508 default and only is necessary when calling subroutines compiled by IBM
11509 XL compilers without optimization.
11513 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
11514 application written to use message passing with special startup code to
11515 enable the application to run. The system must have PE installed in the
11516 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
11517 must be overridden with the @option{-specs=} option to specify the
11518 appropriate directory location. The Parallel Environment does not
11519 support threads, so the @option{-mpe} option and the @option{-pthread}
11520 option are incompatible.
11522 @item -malign-natural
11523 @itemx -malign-power
11524 @opindex malign-natural
11525 @opindex malign-power
11526 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
11527 @option{-malign-natural} overrides the ABI-defined alignment of larger
11528 types, such as floating-point doubles, on their natural size-based boundary.
11529 The option @option{-malign-power} instructs GCC to follow the ABI-specified
11530 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
11532 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
11536 @itemx -mhard-float
11537 @opindex msoft-float
11538 @opindex mhard-float
11539 Generate code that does not use (uses) the floating-point register set.
11540 Software floating point emulation is provided if you use the
11541 @option{-msoft-float} option, and pass the option to GCC when linking.
11544 @itemx -mno-multiple
11546 @opindex mno-multiple
11547 Generate code that uses (does not use) the load multiple word
11548 instructions and the store multiple word instructions. These
11549 instructions are generated by default on POWER systems, and not
11550 generated on PowerPC systems. Do not use @option{-mmultiple} on little
11551 endian PowerPC systems, since those instructions do not work when the
11552 processor is in little endian mode. The exceptions are PPC740 and
11553 PPC750 which permit the instructions usage in little endian mode.
11558 @opindex mno-string
11559 Generate code that uses (does not use) the load string instructions
11560 and the store string word instructions to save multiple registers and
11561 do small block moves. These instructions are generated by default on
11562 POWER systems, and not generated on PowerPC systems. Do not use
11563 @option{-mstring} on little endian PowerPC systems, since those
11564 instructions do not work when the processor is in little endian mode.
11565 The exceptions are PPC740 and PPC750 which permit the instructions
11566 usage in little endian mode.
11571 @opindex mno-update
11572 Generate code that uses (does not use) the load or store instructions
11573 that update the base register to the address of the calculated memory
11574 location. These instructions are generated by default. If you use
11575 @option{-mno-update}, there is a small window between the time that the
11576 stack pointer is updated and the address of the previous frame is
11577 stored, which means code that walks the stack frame across interrupts or
11578 signals may get corrupted data.
11581 @itemx -mno-fused-madd
11582 @opindex mfused-madd
11583 @opindex mno-fused-madd
11584 Generate code that uses (does not use) the floating point multiply and
11585 accumulate instructions. These instructions are generated by default if
11586 hardware floating is used.
11592 Generate code that uses (does not use) the half-word multiply and
11593 multiply-accumulate instructions on the IBM 405 and 440 processors.
11594 These instructions are generated by default when targetting those
11601 Generate code that uses (does not use) the string-search @samp{dlmzb}
11602 instruction on the IBM 405 and 440 processors. This instruction is
11603 generated by default when targetting those processors.
11605 @item -mno-bit-align
11607 @opindex mno-bit-align
11608 @opindex mbit-align
11609 On System V.4 and embedded PowerPC systems do not (do) force structures
11610 and unions that contain bit-fields to be aligned to the base type of the
11613 For example, by default a structure containing nothing but 8
11614 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
11615 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
11616 the structure would be aligned to a 1 byte boundary and be one byte in
11619 @item -mno-strict-align
11620 @itemx -mstrict-align
11621 @opindex mno-strict-align
11622 @opindex mstrict-align
11623 On System V.4 and embedded PowerPC systems do not (do) assume that
11624 unaligned memory references will be handled by the system.
11626 @item -mrelocatable
11627 @itemx -mno-relocatable
11628 @opindex mrelocatable
11629 @opindex mno-relocatable
11630 On embedded PowerPC systems generate code that allows (does not allow)
11631 the program to be relocated to a different address at runtime. If you
11632 use @option{-mrelocatable} on any module, all objects linked together must
11633 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
11635 @item -mrelocatable-lib
11636 @itemx -mno-relocatable-lib
11637 @opindex mrelocatable-lib
11638 @opindex mno-relocatable-lib
11639 On embedded PowerPC systems generate code that allows (does not allow)
11640 the program to be relocated to a different address at runtime. Modules
11641 compiled with @option{-mrelocatable-lib} can be linked with either modules
11642 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
11643 with modules compiled with the @option{-mrelocatable} options.
11649 On System V.4 and embedded PowerPC systems do not (do) assume that
11650 register 2 contains a pointer to a global area pointing to the addresses
11651 used in the program.
11654 @itemx -mlittle-endian
11656 @opindex mlittle-endian
11657 On System V.4 and embedded PowerPC systems compile code for the
11658 processor in little endian mode. The @option{-mlittle-endian} option is
11659 the same as @option{-mlittle}.
11662 @itemx -mbig-endian
11664 @opindex mbig-endian
11665 On System V.4 and embedded PowerPC systems compile code for the
11666 processor in big endian mode. The @option{-mbig-endian} option is
11667 the same as @option{-mbig}.
11669 @item -mdynamic-no-pic
11670 @opindex mdynamic-no-pic
11671 On Darwin and Mac OS X systems, compile code so that it is not
11672 relocatable, but that its external references are relocatable. The
11673 resulting code is suitable for applications, but not shared
11676 @item -mprioritize-restricted-insns=@var{priority}
11677 @opindex mprioritize-restricted-insns
11678 This option controls the priority that is assigned to
11679 dispatch-slot restricted instructions during the second scheduling
11680 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
11681 @var{no/highest/second-highest} priority to dispatch slot restricted
11684 @item -msched-costly-dep=@var{dependence_type}
11685 @opindex msched-costly-dep
11686 This option controls which dependences are considered costly
11687 by the target during instruction scheduling. The argument
11688 @var{dependence_type} takes one of the following values:
11689 @var{no}: no dependence is costly,
11690 @var{all}: all dependences are costly,
11691 @var{true_store_to_load}: a true dependence from store to load is costly,
11692 @var{store_to_load}: any dependence from store to load is costly,
11693 @var{number}: any dependence which latency >= @var{number} is costly.
11695 @item -minsert-sched-nops=@var{scheme}
11696 @opindex minsert-sched-nops
11697 This option controls which nop insertion scheme will be used during
11698 the second scheduling pass. The argument @var{scheme} takes one of the
11700 @var{no}: Don't insert nops.
11701 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
11702 according to the scheduler's grouping.
11703 @var{regroup_exact}: Insert nops to force costly dependent insns into
11704 separate groups. Insert exactly as many nops as needed to force an insn
11705 to a new group, according to the estimated processor grouping.
11706 @var{number}: Insert nops to force costly dependent insns into
11707 separate groups. Insert @var{number} nops to force an insn to a new group.
11710 @opindex mcall-sysv
11711 On System V.4 and embedded PowerPC systems compile code using calling
11712 conventions that adheres to the March 1995 draft of the System V
11713 Application Binary Interface, PowerPC processor supplement. This is the
11714 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
11716 @item -mcall-sysv-eabi
11717 @opindex mcall-sysv-eabi
11718 Specify both @option{-mcall-sysv} and @option{-meabi} options.
11720 @item -mcall-sysv-noeabi
11721 @opindex mcall-sysv-noeabi
11722 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
11724 @item -mcall-solaris
11725 @opindex mcall-solaris
11726 On System V.4 and embedded PowerPC systems compile code for the Solaris
11730 @opindex mcall-linux
11731 On System V.4 and embedded PowerPC systems compile code for the
11732 Linux-based GNU system.
11736 On System V.4 and embedded PowerPC systems compile code for the
11737 Hurd-based GNU system.
11739 @item -mcall-netbsd
11740 @opindex mcall-netbsd
11741 On System V.4 and embedded PowerPC systems compile code for the
11742 NetBSD operating system.
11744 @item -maix-struct-return
11745 @opindex maix-struct-return
11746 Return all structures in memory (as specified by the AIX ABI)@.
11748 @item -msvr4-struct-return
11749 @opindex msvr4-struct-return
11750 Return structures smaller than 8 bytes in registers (as specified by the
11753 @item -mabi=@var{abi-type}
11755 Extend the current ABI with a particular extension, or remove such extension.
11756 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
11757 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
11761 Extend the current ABI with SPE ABI extensions. This does not change
11762 the default ABI, instead it adds the SPE ABI extensions to the current
11766 @opindex mabi=no-spe
11767 Disable Booke SPE ABI extensions for the current ABI@.
11769 @item -mabi=ibmlongdouble
11770 @opindex mabi=ibmlongdouble
11771 Change the current ABI to use IBM extended precision long double.
11772 This is a PowerPC 32-bit SYSV ABI option.
11774 @item -mabi=ieeelongdouble
11775 @opindex mabi=ieeelongdouble
11776 Change the current ABI to use IEEE extended precision long double.
11777 This is a PowerPC 32-bit Linux ABI option.
11780 @itemx -mno-prototype
11781 @opindex mprototype
11782 @opindex mno-prototype
11783 On System V.4 and embedded PowerPC systems assume that all calls to
11784 variable argument functions are properly prototyped. Otherwise, the
11785 compiler must insert an instruction before every non prototyped call to
11786 set or clear bit 6 of the condition code register (@var{CR}) to
11787 indicate whether floating point values were passed in the floating point
11788 registers in case the function takes a variable arguments. With
11789 @option{-mprototype}, only calls to prototyped variable argument functions
11790 will set or clear the bit.
11794 On embedded PowerPC systems, assume that the startup module is called
11795 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
11796 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
11801 On embedded PowerPC systems, assume that the startup module is called
11802 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
11807 On embedded PowerPC systems, assume that the startup module is called
11808 @file{crt0.o} and the standard C libraries are @file{libads.a} and
11811 @item -myellowknife
11812 @opindex myellowknife
11813 On embedded PowerPC systems, assume that the startup module is called
11814 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
11819 On System V.4 and embedded PowerPC systems, specify that you are
11820 compiling for a VxWorks system.
11824 Specify that you are compiling for the WindISS simulation environment.
11828 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
11829 header to indicate that @samp{eabi} extended relocations are used.
11835 On System V.4 and embedded PowerPC systems do (do not) adhere to the
11836 Embedded Applications Binary Interface (eabi) which is a set of
11837 modifications to the System V.4 specifications. Selecting @option{-meabi}
11838 means that the stack is aligned to an 8 byte boundary, a function
11839 @code{__eabi} is called to from @code{main} to set up the eabi
11840 environment, and the @option{-msdata} option can use both @code{r2} and
11841 @code{r13} to point to two separate small data areas. Selecting
11842 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
11843 do not call an initialization function from @code{main}, and the
11844 @option{-msdata} option will only use @code{r13} to point to a single
11845 small data area. The @option{-meabi} option is on by default if you
11846 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
11849 @opindex msdata=eabi
11850 On System V.4 and embedded PowerPC systems, put small initialized
11851 @code{const} global and static data in the @samp{.sdata2} section, which
11852 is pointed to by register @code{r2}. Put small initialized
11853 non-@code{const} global and static data in the @samp{.sdata} section,
11854 which is pointed to by register @code{r13}. Put small uninitialized
11855 global and static data in the @samp{.sbss} section, which is adjacent to
11856 the @samp{.sdata} section. The @option{-msdata=eabi} option is
11857 incompatible with the @option{-mrelocatable} option. The
11858 @option{-msdata=eabi} option also sets the @option{-memb} option.
11861 @opindex msdata=sysv
11862 On System V.4 and embedded PowerPC systems, put small global and static
11863 data in the @samp{.sdata} section, which is pointed to by register
11864 @code{r13}. Put small uninitialized global and static data in the
11865 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
11866 The @option{-msdata=sysv} option is incompatible with the
11867 @option{-mrelocatable} option.
11869 @item -msdata=default
11871 @opindex msdata=default
11873 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
11874 compile code the same as @option{-msdata=eabi}, otherwise compile code the
11875 same as @option{-msdata=sysv}.
11878 @opindex msdata-data
11879 On System V.4 and embedded PowerPC systems, put small global
11880 data in the @samp{.sdata} section. Put small uninitialized global
11881 data in the @samp{.sbss} section. Do not use register @code{r13}
11882 to address small data however. This is the default behavior unless
11883 other @option{-msdata} options are used.
11887 @opindex msdata=none
11889 On embedded PowerPC systems, put all initialized global and static data
11890 in the @samp{.data} section, and all uninitialized data in the
11891 @samp{.bss} section.
11895 @cindex smaller data references (PowerPC)
11896 @cindex .sdata/.sdata2 references (PowerPC)
11897 On embedded PowerPC systems, put global and static items less than or
11898 equal to @var{num} bytes into the small data or bss sections instead of
11899 the normal data or bss section. By default, @var{num} is 8. The
11900 @option{-G @var{num}} switch is also passed to the linker.
11901 All modules should be compiled with the same @option{-G @var{num}} value.
11904 @itemx -mno-regnames
11906 @opindex mno-regnames
11907 On System V.4 and embedded PowerPC systems do (do not) emit register
11908 names in the assembly language output using symbolic forms.
11911 @itemx -mno-longcall
11913 @opindex mno-longcall
11914 By default assume that all calls are far away so that a longer more
11915 expensive calling sequence is required. This is required for calls
11916 further than 32 megabytes (33,554,432 bytes) from the current location.
11917 A short call will be generated if the compiler knows
11918 the call cannot be that far away. This setting can be overridden by
11919 the @code{shortcall} function attribute, or by @code{#pragma
11922 Some linkers are capable of detecting out-of-range calls and generating
11923 glue code on the fly. On these systems, long calls are unnecessary and
11924 generate slower code. As of this writing, the AIX linker can do this,
11925 as can the GNU linker for PowerPC/64. It is planned to add this feature
11926 to the GNU linker for 32-bit PowerPC systems as well.
11928 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
11929 callee, L42'', plus a ``branch island'' (glue code). The two target
11930 addresses represent the callee and the ``branch island''. The
11931 Darwin/PPC linker will prefer the first address and generate a ``bl
11932 callee'' if the PPC ``bl'' instruction will reach the callee directly;
11933 otherwise, the linker will generate ``bl L42'' to call the ``branch
11934 island''. The ``branch island'' is appended to the body of the
11935 calling function; it computes the full 32-bit address of the callee
11938 On Mach-O (Darwin) systems, this option directs the compiler emit to
11939 the glue for every direct call, and the Darwin linker decides whether
11940 to use or discard it.
11942 In the future, we may cause GCC to ignore all longcall specifications
11943 when the linker is known to generate glue.
11947 Adds support for multithreading with the @dfn{pthreads} library.
11948 This option sets flags for both the preprocessor and linker.
11952 @node S/390 and zSeries Options
11953 @subsection S/390 and zSeries Options
11954 @cindex S/390 and zSeries Options
11956 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
11960 @itemx -msoft-float
11961 @opindex mhard-float
11962 @opindex msoft-float
11963 Use (do not use) the hardware floating-point instructions and registers
11964 for floating-point operations. When @option{-msoft-float} is specified,
11965 functions in @file{libgcc.a} will be used to perform floating-point
11966 operations. When @option{-mhard-float} is specified, the compiler
11967 generates IEEE floating-point instructions. This is the default.
11969 @item -mlong-double-64
11970 @itemx -mlong-double-128
11971 @opindex mlong-double-64
11972 @opindex mlong-double-128
11973 These switches control the size of @code{long double} type. A size
11974 of 64bit makes the @code{long double} type equivalent to the @code{double}
11975 type. This is the default.
11978 @itemx -mno-backchain
11979 @opindex mbackchain
11980 @opindex mno-backchain
11981 Store (do not store) the address of the caller's frame as backchain pointer
11982 into the callee's stack frame.
11983 A backchain may be needed to allow debugging using tools that do not understand
11984 DWARF-2 call frame information.
11985 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
11986 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
11987 the backchain is placed into the topmost word of the 96/160 byte register
11990 In general, code compiled with @option{-mbackchain} is call-compatible with
11991 code compiled with @option{-mmo-backchain}; however, use of the backchain
11992 for debugging purposes usually requires that the whole binary is built with
11993 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
11994 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
11995 to build a linux kernel use @option{-msoft-float}.
11997 The default is to not maintain the backchain.
11999 @item -mpacked-stack
12000 @item -mno-packed-stack
12001 @opindex mpacked-stack
12002 @opindex mno-packed-stack
12003 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
12004 specified, the compiler uses the all fields of the 96/160 byte register save
12005 area only for their default purpose; unused fields still take up stack space.
12006 When @option{-mpacked-stack} is specified, register save slots are densely
12007 packed at the top of the register save area; unused space is reused for other
12008 purposes, allowing for more efficient use of the available stack space.
12009 However, when @option{-mbackchain} is also in effect, the topmost word of
12010 the save area is always used to store the backchain, and the return address
12011 register is always saved two words below the backchain.
12013 As long as the stack frame backchain is not used, code generated with
12014 @option{-mpacked-stack} is call-compatible with code generated with
12015 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
12016 S/390 or zSeries generated code that uses the stack frame backchain at run
12017 time, not just for debugging purposes. Such code is not call-compatible
12018 with code compiled with @option{-mpacked-stack}. Also, note that the
12019 combination of @option{-mbackchain},
12020 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
12021 to build a linux kernel use @option{-msoft-float}.
12023 The default is to not use the packed stack layout.
12026 @itemx -mno-small-exec
12027 @opindex msmall-exec
12028 @opindex mno-small-exec
12029 Generate (or do not generate) code using the @code{bras} instruction
12030 to do subroutine calls.
12031 This only works reliably if the total executable size does not
12032 exceed 64k. The default is to use the @code{basr} instruction instead,
12033 which does not have this limitation.
12039 When @option{-m31} is specified, generate code compliant to the
12040 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
12041 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
12042 particular to generate 64-bit instructions. For the @samp{s390}
12043 targets, the default is @option{-m31}, while the @samp{s390x}
12044 targets default to @option{-m64}.
12050 When @option{-mzarch} is specified, generate code using the
12051 instructions available on z/Architecture.
12052 When @option{-mesa} is specified, generate code using the
12053 instructions available on ESA/390. Note that @option{-mesa} is
12054 not possible with @option{-m64}.
12055 When generating code compliant to the GNU/Linux for S/390 ABI,
12056 the default is @option{-mesa}. When generating code compliant
12057 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
12063 Generate (or do not generate) code using the @code{mvcle} instruction
12064 to perform block moves. When @option{-mno-mvcle} is specified,
12065 use a @code{mvc} loop instead. This is the default unless optimizing for
12072 Print (or do not print) additional debug information when compiling.
12073 The default is to not print debug information.
12075 @item -march=@var{cpu-type}
12077 Generate code that will run on @var{cpu-type}, which is the name of a system
12078 representing a certain processor type. Possible values for
12079 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
12080 When generating code using the instructions available on z/Architecture,
12081 the default is @option{-march=z900}. Otherwise, the default is
12082 @option{-march=g5}.
12084 @item -mtune=@var{cpu-type}
12086 Tune to @var{cpu-type} everything applicable about the generated code,
12087 except for the ABI and the set of available instructions.
12088 The list of @var{cpu-type} values is the same as for @option{-march}.
12089 The default is the value used for @option{-march}.
12092 @itemx -mno-tpf-trace
12093 @opindex mtpf-trace
12094 @opindex mno-tpf-trace
12095 Generate code that adds (does not add) in TPF OS specific branches to trace
12096 routines in the operating system. This option is off by default, even
12097 when compiling for the TPF OS@.
12100 @itemx -mno-fused-madd
12101 @opindex mfused-madd
12102 @opindex mno-fused-madd
12103 Generate code that uses (does not use) the floating point multiply and
12104 accumulate instructions. These instructions are generated by default if
12105 hardware floating point is used.
12107 @item -mwarn-framesize=@var{framesize}
12108 @opindex mwarn-framesize
12109 Emit a warning if the current function exceeds the given frame size. Because
12110 this is a compile time check it doesn't need to be a real problem when the program
12111 runs. It is intended to identify functions which most probably cause
12112 a stack overflow. It is useful to be used in an environment with limited stack
12113 size e.g.@: the linux kernel.
12115 @item -mwarn-dynamicstack
12116 @opindex mwarn-dynamicstack
12117 Emit a warning if the function calls alloca or uses dynamically
12118 sized arrays. This is generally a bad idea with a limited stack size.
12120 @item -mstack-guard=@var{stack-guard}
12121 @item -mstack-size=@var{stack-size}
12122 @opindex mstack-guard
12123 @opindex mstack-size
12124 These arguments always have to be used in conjunction. If they are present the s390
12125 back end emits additional instructions in the function prologue which trigger a trap
12126 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
12127 (remember that the stack on s390 grows downward). These options are intended to
12128 be used to help debugging stack overflow problems. The additionally emitted code
12129 causes only little overhead and hence can also be used in production like systems
12130 without greater performance degradation. The given values have to be exact
12131 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard} without
12133 In order to be efficient the extra code makes the assumption that the stack starts
12134 at an address aligned to the value given by @var{stack-size}.
12138 @subsection SH Options
12140 These @samp{-m} options are defined for the SH implementations:
12145 Generate code for the SH1.
12149 Generate code for the SH2.
12152 Generate code for the SH2e.
12156 Generate code for the SH3.
12160 Generate code for the SH3e.
12164 Generate code for the SH4 without a floating-point unit.
12166 @item -m4-single-only
12167 @opindex m4-single-only
12168 Generate code for the SH4 with a floating-point unit that only
12169 supports single-precision arithmetic.
12173 Generate code for the SH4 assuming the floating-point unit is in
12174 single-precision mode by default.
12178 Generate code for the SH4.
12182 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
12183 floating-point unit is not used.
12185 @item -m4a-single-only
12186 @opindex m4a-single-only
12187 Generate code for the SH4a, in such a way that no double-precision
12188 floating point operations are used.
12191 @opindex m4a-single
12192 Generate code for the SH4a assuming the floating-point unit is in
12193 single-precision mode by default.
12197 Generate code for the SH4a.
12201 Same as @option{-m4a-nofpu}, except that it implicitly passes
12202 @option{-dsp} to the assembler. GCC doesn't generate any DSP
12203 instructions at the moment.
12207 Compile code for the processor in big endian mode.
12211 Compile code for the processor in little endian mode.
12215 Align doubles at 64-bit boundaries. Note that this changes the calling
12216 conventions, and thus some functions from the standard C library will
12217 not work unless you recompile it first with @option{-mdalign}.
12221 Shorten some address references at link time, when possible; uses the
12222 linker option @option{-relax}.
12226 Use 32-bit offsets in @code{switch} tables. The default is to use
12231 Enable the use of the instruction @code{fmovd}.
12235 Comply with the calling conventions defined by Renesas.
12239 Comply with the calling conventions defined by Renesas.
12243 Comply with the calling conventions defined for GCC before the Renesas
12244 conventions were available. This option is the default for all
12245 targets of the SH toolchain except for @samp{sh-symbianelf}.
12248 @opindex mnomacsave
12249 Mark the @code{MAC} register as call-clobbered, even if
12250 @option{-mhitachi} is given.
12254 Increase IEEE-compliance of floating-point code.
12255 At the moment, this is equivalent to @option{-fno-finite-math-only}.
12256 When generating 16 bit SH opcodes, getting IEEE-conforming results for
12257 comparisons of NANs / infinities incurs extra overhead in every
12258 floating point comparison, therefore the default is set to
12259 @option{-ffinite-math-only}.
12263 Dump instruction size and location in the assembly code.
12266 @opindex mpadstruct
12267 This option is deprecated. It pads structures to multiple of 4 bytes,
12268 which is incompatible with the SH ABI@.
12272 Optimize for space instead of speed. Implied by @option{-Os}.
12275 @opindex mprefergot
12276 When generating position-independent code, emit function calls using
12277 the Global Offset Table instead of the Procedure Linkage Table.
12281 Generate a library function call to invalidate instruction cache
12282 entries, after fixing up a trampoline. This library function call
12283 doesn't assume it can write to the whole memory address space. This
12284 is the default when the target is @code{sh-*-linux*}.
12286 @item -multcost=@var{number}
12287 @opindex multcost=@var{number}
12288 Set the cost to assume for a multiply insn.
12290 @item -mdiv=@var{strategy}
12291 @opindex mdiv=@var{strategy}
12292 Set the division strategy to use for SHmedia code. @var{strategy} must be
12293 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
12294 inv:call2, inv:fp .
12295 "fp" performs the operation in floating point. This has a very high latency,
12296 but needs only a few instructions, so it might be a good choice if
12297 your code has enough easily exploitable ILP to allow the compiler to
12298 schedule the floating point instructions together with other instructions.
12299 Division by zero causes a floating point exception.
12300 "inv" uses integer operations to calculate the inverse of the divisor,
12301 and then multiplies the dividend with the inverse. This strategy allows
12302 cse and hoisting of the inverse calculation. Division by zero calculates
12303 an unspecified result, but does not trap.
12304 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
12305 have been found, or if the entire operation has been hoisted to the same
12306 place, the last stages of the inverse calculation are intertwined with the
12307 final multiply to reduce the overall latency, at the expense of using a few
12308 more instructions, and thus offering fewer scheduling opportunities with
12310 "call" calls a library function that usually implements the inv:minlat
12312 This gives high code density for m5-*media-nofpu compilations.
12313 "call2" uses a different entry point of the same library function, where it
12314 assumes that a pointer to a lookup table has already been set up, which
12315 exposes the pointer load to cse / code hoisting optimizations.
12316 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
12317 code generation, but if the code stays unoptimized, revert to the "call",
12318 "call2", or "fp" strategies, respectively. Note that the
12319 potentially-trapping side effect of division by zero is carried by a
12320 separate instruction, so it is possible that all the integer instructions
12321 are hoisted out, but the marker for the side effect stays where it is.
12322 A recombination to fp operations or a call is not possible in that case.
12323 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
12324 that the inverse calculation was nor separated from the multiply, they speed
12325 up division where the dividend fits into 20 bits (plus sign where applicable),
12326 by inserting a test to skip a number of operations in this case; this test
12327 slows down the case of larger dividends. inv20u assumes the case of a such
12328 a small dividend to be unlikely, and inv20l assumes it to be likely.
12330 @item -mdivsi3_libfunc=@var{name}
12331 @opindex mdivsi3_libfunc=@var{name}
12332 Set the name of the library function used for 32 bit signed division to
12333 @var{name}. This only affect the name used in the call and inv:call
12334 division strategies, and the compiler will still expect the same
12335 sets of input/output/clobbered registers as if this option was not present.
12337 @item -madjust-unroll
12338 @opindex madjust-unroll
12339 Throttle unrolling to avoid thrashing target registers.
12340 This option only has an effect if the gcc code base supports the
12341 TARGET_ADJUST_UNROLL_MAX target hook.
12343 @item -mindexed-addressing
12344 @opindex mindexed-addressing
12345 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
12346 This is only safe if the hardware and/or OS implement 32 bit wrap-around
12347 semantics for the indexed addressing mode. The architecture allows the
12348 implementation of processors with 64 bit MMU, which the OS could use to
12349 get 32 bit addressing, but since no current hardware implementation supports
12350 this or any other way to make the indexed addressing mode safe to use in
12351 the 32 bit ABI, the default is -mno-indexed-addressing.
12353 @item -mgettrcost=@var{number}
12354 @opindex mgettrcost=@var{number}
12355 Set the cost assumed for the gettr instruction to @var{number}.
12356 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
12360 Assume pt* instructions won't trap. This will generally generate better
12361 scheduled code, but is unsafe on current hardware. The current architecture
12362 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
12363 This has the unintentional effect of making it unsafe to schedule ptabs /
12364 ptrel before a branch, or hoist it out of a loop. For example,
12365 __do_global_ctors, a part of libgcc that runs constructors at program
12366 startup, calls functions in a list which is delimited by -1. With the
12367 -mpt-fixed option, the ptabs will be done before testing against -1.
12368 That means that all the constructors will be run a bit quicker, but when
12369 the loop comes to the end of the list, the program crashes because ptabs
12370 loads -1 into a target register. Since this option is unsafe for any
12371 hardware implementing the current architecture specification, the default
12372 is -mno-pt-fixed. Unless the user specifies a specific cost with
12373 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
12374 this deters register allocation using target registers for storing
12377 @item -minvalid-symbols
12378 @opindex minvalid-symbols
12379 Assume symbols might be invalid. Ordinary function symbols generated by
12380 the compiler will always be valid to load with movi/shori/ptabs or
12381 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
12382 to generate symbols that will cause ptabs / ptrel to trap.
12383 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
12384 It will then prevent cross-basic-block cse, hoisting and most scheduling
12385 of symbol loads. The default is @option{-mno-invalid-symbols}.
12388 @node SPARC Options
12389 @subsection SPARC Options
12390 @cindex SPARC options
12392 These @samp{-m} options are supported on the SPARC:
12395 @item -mno-app-regs
12397 @opindex mno-app-regs
12399 Specify @option{-mapp-regs} to generate output using the global registers
12400 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
12403 To be fully SVR4 ABI compliant at the cost of some performance loss,
12404 specify @option{-mno-app-regs}. You should compile libraries and system
12405 software with this option.
12408 @itemx -mhard-float
12410 @opindex mhard-float
12411 Generate output containing floating point instructions. This is the
12415 @itemx -msoft-float
12417 @opindex msoft-float
12418 Generate output containing library calls for floating point.
12419 @strong{Warning:} the requisite libraries are not available for all SPARC
12420 targets. Normally the facilities of the machine's usual C compiler are
12421 used, but this cannot be done directly in cross-compilation. You must make
12422 your own arrangements to provide suitable library functions for
12423 cross-compilation. The embedded targets @samp{sparc-*-aout} and
12424 @samp{sparclite-*-*} do provide software floating point support.
12426 @option{-msoft-float} changes the calling convention in the output file;
12427 therefore, it is only useful if you compile @emph{all} of a program with
12428 this option. In particular, you need to compile @file{libgcc.a}, the
12429 library that comes with GCC, with @option{-msoft-float} in order for
12432 @item -mhard-quad-float
12433 @opindex mhard-quad-float
12434 Generate output containing quad-word (long double) floating point
12437 @item -msoft-quad-float
12438 @opindex msoft-quad-float
12439 Generate output containing library calls for quad-word (long double)
12440 floating point instructions. The functions called are those specified
12441 in the SPARC ABI@. This is the default.
12443 As of this writing, there are no SPARC implementations that have hardware
12444 support for the quad-word floating point instructions. They all invoke
12445 a trap handler for one of these instructions, and then the trap handler
12446 emulates the effect of the instruction. Because of the trap handler overhead,
12447 this is much slower than calling the ABI library routines. Thus the
12448 @option{-msoft-quad-float} option is the default.
12450 @item -mno-unaligned-doubles
12451 @itemx -munaligned-doubles
12452 @opindex mno-unaligned-doubles
12453 @opindex munaligned-doubles
12454 Assume that doubles have 8 byte alignment. This is the default.
12456 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
12457 alignment only if they are contained in another type, or if they have an
12458 absolute address. Otherwise, it assumes they have 4 byte alignment.
12459 Specifying this option avoids some rare compatibility problems with code
12460 generated by other compilers. It is not the default because it results
12461 in a performance loss, especially for floating point code.
12463 @item -mno-faster-structs
12464 @itemx -mfaster-structs
12465 @opindex mno-faster-structs
12466 @opindex mfaster-structs
12467 With @option{-mfaster-structs}, the compiler assumes that structures
12468 should have 8 byte alignment. This enables the use of pairs of
12469 @code{ldd} and @code{std} instructions for copies in structure
12470 assignment, in place of twice as many @code{ld} and @code{st} pairs.
12471 However, the use of this changed alignment directly violates the SPARC
12472 ABI@. Thus, it's intended only for use on targets where the developer
12473 acknowledges that their resulting code will not be directly in line with
12474 the rules of the ABI@.
12476 @item -mimpure-text
12477 @opindex mimpure-text
12478 @option{-mimpure-text}, used in addition to @option{-shared}, tells
12479 the compiler to not pass @option{-z text} to the linker when linking a
12480 shared object. Using this option, you can link position-dependent
12481 code into a shared object.
12483 @option{-mimpure-text} suppresses the ``relocations remain against
12484 allocatable but non-writable sections'' linker error message.
12485 However, the necessary relocations will trigger copy-on-write, and the
12486 shared object is not actually shared across processes. Instead of
12487 using @option{-mimpure-text}, you should compile all source code with
12488 @option{-fpic} or @option{-fPIC}.
12490 This option is only available on SunOS and Solaris.
12492 @item -mcpu=@var{cpu_type}
12494 Set the instruction set, register set, and instruction scheduling parameters
12495 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
12496 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
12497 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
12498 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
12499 @samp{ultrasparc3}, and @samp{niagara}.
12501 Default instruction scheduling parameters are used for values that select
12502 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
12503 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
12505 Here is a list of each supported architecture and their supported
12510 v8: supersparc, hypersparc
12511 sparclite: f930, f934, sparclite86x
12513 v9: ultrasparc, ultrasparc3, niagara
12516 By default (unless configured otherwise), GCC generates code for the V7
12517 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
12518 additionally optimizes it for the Cypress CY7C602 chip, as used in the
12519 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
12520 SPARCStation 1, 2, IPX etc.
12522 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
12523 architecture. The only difference from V7 code is that the compiler emits
12524 the integer multiply and integer divide instructions which exist in SPARC-V8
12525 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
12526 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
12529 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
12530 the SPARC architecture. This adds the integer multiply, integer divide step
12531 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
12532 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
12533 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
12534 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
12535 MB86934 chip, which is the more recent SPARClite with FPU@.
12537 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
12538 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
12539 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
12540 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
12541 optimizes it for the TEMIC SPARClet chip.
12543 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
12544 architecture. This adds 64-bit integer and floating-point move instructions,
12545 3 additional floating-point condition code registers and conditional move
12546 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
12547 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
12548 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
12549 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
12550 @option{-mcpu=niagara}, the compiler additionally optimizes it for
12551 Sun UltraSPARC T1 chips.
12553 @item -mtune=@var{cpu_type}
12555 Set the instruction scheduling parameters for machine type
12556 @var{cpu_type}, but do not set the instruction set or register set that the
12557 option @option{-mcpu=@var{cpu_type}} would.
12559 The same values for @option{-mcpu=@var{cpu_type}} can be used for
12560 @option{-mtune=@var{cpu_type}}, but the only useful values are those
12561 that select a particular cpu implementation. Those are @samp{cypress},
12562 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
12563 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
12564 @samp{ultrasparc3}, and @samp{niagara}.
12569 @opindex mno-v8plus
12570 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
12571 difference from the V8 ABI is that the global and out registers are
12572 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
12573 mode for all SPARC-V9 processors.
12579 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
12580 Visual Instruction Set extensions. The default is @option{-mno-vis}.
12583 These @samp{-m} options are supported in addition to the above
12584 on SPARC-V9 processors in 64-bit environments:
12587 @item -mlittle-endian
12588 @opindex mlittle-endian
12589 Generate code for a processor running in little-endian mode. It is only
12590 available for a few configurations and most notably not on Solaris and Linux.
12596 Generate code for a 32-bit or 64-bit environment.
12597 The 32-bit environment sets int, long and pointer to 32 bits.
12598 The 64-bit environment sets int to 32 bits and long and pointer
12601 @item -mcmodel=medlow
12602 @opindex mcmodel=medlow
12603 Generate code for the Medium/Low code model: 64-bit addresses, programs
12604 must be linked in the low 32 bits of memory. Programs can be statically
12605 or dynamically linked.
12607 @item -mcmodel=medmid
12608 @opindex mcmodel=medmid
12609 Generate code for the Medium/Middle code model: 64-bit addresses, programs
12610 must be linked in the low 44 bits of memory, the text and data segments must
12611 be less than 2GB in size and the data segment must be located within 2GB of
12614 @item -mcmodel=medany
12615 @opindex mcmodel=medany
12616 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
12617 may be linked anywhere in memory, the text and data segments must be less
12618 than 2GB in size and the data segment must be located within 2GB of the
12621 @item -mcmodel=embmedany
12622 @opindex mcmodel=embmedany
12623 Generate code for the Medium/Anywhere code model for embedded systems:
12624 64-bit addresses, the text and data segments must be less than 2GB in
12625 size, both starting anywhere in memory (determined at link time). The
12626 global register %g4 points to the base of the data segment. Programs
12627 are statically linked and PIC is not supported.
12630 @itemx -mno-stack-bias
12631 @opindex mstack-bias
12632 @opindex mno-stack-bias
12633 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
12634 frame pointer if present, are offset by @minus{}2047 which must be added back
12635 when making stack frame references. This is the default in 64-bit mode.
12636 Otherwise, assume no such offset is present.
12639 These switches are supported in addition to the above on Solaris:
12644 Add support for multithreading using the Solaris threads library. This
12645 option sets flags for both the preprocessor and linker. This option does
12646 not affect the thread safety of object code produced by the compiler or
12647 that of libraries supplied with it.
12651 Add support for multithreading using the POSIX threads library. This
12652 option sets flags for both the preprocessor and linker. This option does
12653 not affect the thread safety of object code produced by the compiler or
12654 that of libraries supplied with it.
12658 This is a synonym for @option{-pthreads}.
12661 @node System V Options
12662 @subsection Options for System V
12664 These additional options are available on System V Release 4 for
12665 compatibility with other compilers on those systems:
12670 Create a shared object.
12671 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
12675 Identify the versions of each tool used by the compiler, in a
12676 @code{.ident} assembler directive in the output.
12680 Refrain from adding @code{.ident} directives to the output file (this is
12683 @item -YP,@var{dirs}
12685 Search the directories @var{dirs}, and no others, for libraries
12686 specified with @option{-l}.
12688 @item -Ym,@var{dir}
12690 Look in the directory @var{dir} to find the M4 preprocessor.
12691 The assembler uses this option.
12692 @c This is supposed to go with a -Yd for predefined M4 macro files, but
12693 @c the generic assembler that comes with Solaris takes just -Ym.
12696 @node TMS320C3x/C4x Options
12697 @subsection TMS320C3x/C4x Options
12698 @cindex TMS320C3x/C4x Options
12700 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
12704 @item -mcpu=@var{cpu_type}
12706 Set the instruction set, register set, and instruction scheduling
12707 parameters for machine type @var{cpu_type}. Supported values for
12708 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
12709 @samp{c44}. The default is @samp{c40} to generate code for the
12714 @itemx -msmall-memory
12716 @opindex mbig-memory
12718 @opindex msmall-memory
12720 Generates code for the big or small memory model. The small memory
12721 model assumed that all data fits into one 64K word page. At run-time
12722 the data page (DP) register must be set to point to the 64K page
12723 containing the .bss and .data program sections. The big memory model is
12724 the default and requires reloading of the DP register for every direct
12731 Allow (disallow) allocation of general integer operands into the block
12732 count register BK@.
12738 Enable (disable) generation of code using decrement and branch,
12739 DBcond(D), instructions. This is enabled by default for the C4x. To be
12740 on the safe side, this is disabled for the C3x, since the maximum
12741 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
12742 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
12743 that it can utilize the decrement and branch instruction, but will give
12744 up if there is more than one memory reference in the loop. Thus a loop
12745 where the loop counter is decremented can generate slightly more
12746 efficient code, in cases where the RPTB instruction cannot be utilized.
12748 @item -mdp-isr-reload
12750 @opindex mdp-isr-reload
12752 Force the DP register to be saved on entry to an interrupt service
12753 routine (ISR), reloaded to point to the data section, and restored on
12754 exit from the ISR@. This should not be required unless someone has
12755 violated the small memory model by modifying the DP register, say within
12762 For the C3x use the 24-bit MPYI instruction for integer multiplies
12763 instead of a library call to guarantee 32-bit results. Note that if one
12764 of the operands is a constant, then the multiplication will be performed
12765 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
12766 then squaring operations are performed inline instead of a library call.
12769 @itemx -mno-fast-fix
12771 @opindex mno-fast-fix
12772 The C3x/C4x FIX instruction to convert a floating point value to an
12773 integer value chooses the nearest integer less than or equal to the
12774 floating point value rather than to the nearest integer. Thus if the
12775 floating point number is negative, the result will be incorrectly
12776 truncated an additional code is necessary to detect and correct this
12777 case. This option can be used to disable generation of the additional
12778 code required to correct the result.
12784 Enable (disable) generation of repeat block sequences using the RPTB
12785 instruction for zero overhead looping. The RPTB construct is only used
12786 for innermost loops that do not call functions or jump across the loop
12787 boundaries. There is no advantage having nested RPTB loops due to the
12788 overhead required to save and restore the RC, RS, and RE registers.
12789 This is enabled by default with @option{-O2}.
12791 @item -mrpts=@var{count}
12795 Enable (disable) the use of the single instruction repeat instruction
12796 RPTS@. If a repeat block contains a single instruction, and the loop
12797 count can be guaranteed to be less than the value @var{count}, GCC will
12798 emit a RPTS instruction instead of a RPTB@. If no value is specified,
12799 then a RPTS will be emitted even if the loop count cannot be determined
12800 at compile time. Note that the repeated instruction following RPTS does
12801 not have to be reloaded from memory each iteration, thus freeing up the
12802 CPU buses for operands. However, since interrupts are blocked by this
12803 instruction, it is disabled by default.
12805 @item -mloop-unsigned
12806 @itemx -mno-loop-unsigned
12807 @opindex mloop-unsigned
12808 @opindex mno-loop-unsigned
12809 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
12810 is @math{2^{31} + 1} since these instructions test if the iteration count is
12811 negative to terminate the loop. If the iteration count is unsigned
12812 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
12813 exceeded. This switch allows an unsigned iteration count.
12817 Try to emit an assembler syntax that the TI assembler (asm30) is happy
12818 with. This also enforces compatibility with the API employed by the TI
12819 C3x C compiler. For example, long doubles are passed as structures
12820 rather than in floating point registers.
12826 Generate code that uses registers (stack) for passing arguments to functions.
12827 By default, arguments are passed in registers where possible rather
12828 than by pushing arguments on to the stack.
12830 @item -mparallel-insns
12831 @itemx -mno-parallel-insns
12832 @opindex mparallel-insns
12833 @opindex mno-parallel-insns
12834 Allow the generation of parallel instructions. This is enabled by
12835 default with @option{-O2}.
12837 @item -mparallel-mpy
12838 @itemx -mno-parallel-mpy
12839 @opindex mparallel-mpy
12840 @opindex mno-parallel-mpy
12841 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
12842 provided @option{-mparallel-insns} is also specified. These instructions have
12843 tight register constraints which can pessimize the code generation
12844 of large functions.
12849 @subsection V850 Options
12850 @cindex V850 Options
12852 These @samp{-m} options are defined for V850 implementations:
12856 @itemx -mno-long-calls
12857 @opindex mlong-calls
12858 @opindex mno-long-calls
12859 Treat all calls as being far away (near). If calls are assumed to be
12860 far away, the compiler will always load the functions address up into a
12861 register, and call indirect through the pointer.
12867 Do not optimize (do optimize) basic blocks that use the same index
12868 pointer 4 or more times to copy pointer into the @code{ep} register, and
12869 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
12870 option is on by default if you optimize.
12872 @item -mno-prolog-function
12873 @itemx -mprolog-function
12874 @opindex mno-prolog-function
12875 @opindex mprolog-function
12876 Do not use (do use) external functions to save and restore registers
12877 at the prologue and epilogue of a function. The external functions
12878 are slower, but use less code space if more than one function saves
12879 the same number of registers. The @option{-mprolog-function} option
12880 is on by default if you optimize.
12884 Try to make the code as small as possible. At present, this just turns
12885 on the @option{-mep} and @option{-mprolog-function} options.
12887 @item -mtda=@var{n}
12889 Put static or global variables whose size is @var{n} bytes or less into
12890 the tiny data area that register @code{ep} points to. The tiny data
12891 area can hold up to 256 bytes in total (128 bytes for byte references).
12893 @item -msda=@var{n}
12895 Put static or global variables whose size is @var{n} bytes or less into
12896 the small data area that register @code{gp} points to. The small data
12897 area can hold up to 64 kilobytes.
12899 @item -mzda=@var{n}
12901 Put static or global variables whose size is @var{n} bytes or less into
12902 the first 32 kilobytes of memory.
12906 Specify that the target processor is the V850.
12909 @opindex mbig-switch
12910 Generate code suitable for big switch tables. Use this option only if
12911 the assembler/linker complain about out of range branches within a switch
12916 This option will cause r2 and r5 to be used in the code generated by
12917 the compiler. This setting is the default.
12919 @item -mno-app-regs
12920 @opindex mno-app-regs
12921 This option will cause r2 and r5 to be treated as fixed registers.
12925 Specify that the target processor is the V850E1. The preprocessor
12926 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
12927 this option is used.
12931 Specify that the target processor is the V850E@. The preprocessor
12932 constant @samp{__v850e__} will be defined if this option is used.
12934 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
12935 are defined then a default target processor will be chosen and the
12936 relevant @samp{__v850*__} preprocessor constant will be defined.
12938 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
12939 defined, regardless of which processor variant is the target.
12941 @item -mdisable-callt
12942 @opindex mdisable-callt
12943 This option will suppress generation of the CALLT instruction for the
12944 v850e and v850e1 flavors of the v850 architecture. The default is
12945 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
12950 @subsection VAX Options
12951 @cindex VAX options
12953 These @samp{-m} options are defined for the VAX:
12958 Do not output certain jump instructions (@code{aobleq} and so on)
12959 that the Unix assembler for the VAX cannot handle across long
12964 Do output those jump instructions, on the assumption that you
12965 will assemble with the GNU assembler.
12969 Output code for g-format floating point numbers instead of d-format.
12972 @node x86-64 Options
12973 @subsection x86-64 Options
12974 @cindex x86-64 options
12976 These are listed under @xref{i386 and x86-64 Options}.
12978 @node Xstormy16 Options
12979 @subsection Xstormy16 Options
12980 @cindex Xstormy16 Options
12982 These options are defined for Xstormy16:
12987 Choose startup files and linker script suitable for the simulator.
12990 @node Xtensa Options
12991 @subsection Xtensa Options
12992 @cindex Xtensa Options
12994 These options are supported for Xtensa targets:
12998 @itemx -mno-const16
13000 @opindex mno-const16
13001 Enable or disable use of @code{CONST16} instructions for loading
13002 constant values. The @code{CONST16} instruction is currently not a
13003 standard option from Tensilica. When enabled, @code{CONST16}
13004 instructions are always used in place of the standard @code{L32R}
13005 instructions. The use of @code{CONST16} is enabled by default only if
13006 the @code{L32R} instruction is not available.
13009 @itemx -mno-fused-madd
13010 @opindex mfused-madd
13011 @opindex mno-fused-madd
13012 Enable or disable use of fused multiply/add and multiply/subtract
13013 instructions in the floating-point option. This has no effect if the
13014 floating-point option is not also enabled. Disabling fused multiply/add
13015 and multiply/subtract instructions forces the compiler to use separate
13016 instructions for the multiply and add/subtract operations. This may be
13017 desirable in some cases where strict IEEE 754-compliant results are
13018 required: the fused multiply add/subtract instructions do not round the
13019 intermediate result, thereby producing results with @emph{more} bits of
13020 precision than specified by the IEEE standard. Disabling fused multiply
13021 add/subtract instructions also ensures that the program output is not
13022 sensitive to the compiler's ability to combine multiply and add/subtract
13025 @item -mtext-section-literals
13026 @itemx -mno-text-section-literals
13027 @opindex mtext-section-literals
13028 @opindex mno-text-section-literals
13029 Control the treatment of literal pools. The default is
13030 @option{-mno-text-section-literals}, which places literals in a separate
13031 section in the output file. This allows the literal pool to be placed
13032 in a data RAM/ROM, and it also allows the linker to combine literal
13033 pools from separate object files to remove redundant literals and
13034 improve code size. With @option{-mtext-section-literals}, the literals
13035 are interspersed in the text section in order to keep them as close as
13036 possible to their references. This may be necessary for large assembly
13039 @item -mtarget-align
13040 @itemx -mno-target-align
13041 @opindex mtarget-align
13042 @opindex mno-target-align
13043 When this option is enabled, GCC instructs the assembler to
13044 automatically align instructions to reduce branch penalties at the
13045 expense of some code density. The assembler attempts to widen density
13046 instructions to align branch targets and the instructions following call
13047 instructions. If there are not enough preceding safe density
13048 instructions to align a target, no widening will be performed. The
13049 default is @option{-mtarget-align}. These options do not affect the
13050 treatment of auto-aligned instructions like @code{LOOP}, which the
13051 assembler will always align, either by widening density instructions or
13052 by inserting no-op instructions.
13055 @itemx -mno-longcalls
13056 @opindex mlongcalls
13057 @opindex mno-longcalls
13058 When this option is enabled, GCC instructs the assembler to translate
13059 direct calls to indirect calls unless it can determine that the target
13060 of a direct call is in the range allowed by the call instruction. This
13061 translation typically occurs for calls to functions in other source
13062 files. Specifically, the assembler translates a direct @code{CALL}
13063 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
13064 The default is @option{-mno-longcalls}. This option should be used in
13065 programs where the call target can potentially be out of range. This
13066 option is implemented in the assembler, not the compiler, so the
13067 assembly code generated by GCC will still show direct call
13068 instructions---look at the disassembled object code to see the actual
13069 instructions. Note that the assembler will use an indirect call for
13070 every cross-file call, not just those that really will be out of range.
13073 @node zSeries Options
13074 @subsection zSeries Options
13075 @cindex zSeries options
13077 These are listed under @xref{S/390 and zSeries Options}.
13079 @node Code Gen Options
13080 @section Options for Code Generation Conventions
13081 @cindex code generation conventions
13082 @cindex options, code generation
13083 @cindex run-time options
13085 These machine-independent options control the interface conventions
13086 used in code generation.
13088 Most of them have both positive and negative forms; the negative form
13089 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
13090 one of the forms is listed---the one which is not the default. You
13091 can figure out the other form by either removing @samp{no-} or adding
13095 @item -fbounds-check
13096 @opindex fbounds-check
13097 For front-ends that support it, generate additional code to check that
13098 indices used to access arrays are within the declared range. This is
13099 currently only supported by the Java and Fortran front-ends, where
13100 this option defaults to true and false respectively.
13104 This option generates traps for signed overflow on addition, subtraction,
13105 multiplication operations.
13109 This option instructs the compiler to assume that signed arithmetic
13110 overflow of addition, subtraction and multiplication wraps around
13111 using twos-complement representation. This flag enables some optimizations
13112 and disables others. This option is enabled by default for the Java
13113 front-end, as required by the Java language specification.
13116 @opindex fexceptions
13117 Enable exception handling. Generates extra code needed to propagate
13118 exceptions. For some targets, this implies GCC will generate frame
13119 unwind information for all functions, which can produce significant data
13120 size overhead, although it does not affect execution. If you do not
13121 specify this option, GCC will enable it by default for languages like
13122 C++ which normally require exception handling, and disable it for
13123 languages like C that do not normally require it. However, you may need
13124 to enable this option when compiling C code that needs to interoperate
13125 properly with exception handlers written in C++. You may also wish to
13126 disable this option if you are compiling older C++ programs that don't
13127 use exception handling.
13129 @item -fnon-call-exceptions
13130 @opindex fnon-call-exceptions
13131 Generate code that allows trapping instructions to throw exceptions.
13132 Note that this requires platform-specific runtime support that does
13133 not exist everywhere. Moreover, it only allows @emph{trapping}
13134 instructions to throw exceptions, i.e.@: memory references or floating
13135 point instructions. It does not allow exceptions to be thrown from
13136 arbitrary signal handlers such as @code{SIGALRM}.
13138 @item -funwind-tables
13139 @opindex funwind-tables
13140 Similar to @option{-fexceptions}, except that it will just generate any needed
13141 static data, but will not affect the generated code in any other way.
13142 You will normally not enable this option; instead, a language processor
13143 that needs this handling would enable it on your behalf.
13145 @item -fasynchronous-unwind-tables
13146 @opindex fasynchronous-unwind-tables
13147 Generate unwind table in dwarf2 format, if supported by target machine. The
13148 table is exact at each instruction boundary, so it can be used for stack
13149 unwinding from asynchronous events (such as debugger or garbage collector).
13151 @item -fpcc-struct-return
13152 @opindex fpcc-struct-return
13153 Return ``short'' @code{struct} and @code{union} values in memory like
13154 longer ones, rather than in registers. This convention is less
13155 efficient, but it has the advantage of allowing intercallability between
13156 GCC-compiled files and files compiled with other compilers, particularly
13157 the Portable C Compiler (pcc).
13159 The precise convention for returning structures in memory depends
13160 on the target configuration macros.
13162 Short structures and unions are those whose size and alignment match
13163 that of some integer type.
13165 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
13166 switch is not binary compatible with code compiled with the
13167 @option{-freg-struct-return} switch.
13168 Use it to conform to a non-default application binary interface.
13170 @item -freg-struct-return
13171 @opindex freg-struct-return
13172 Return @code{struct} and @code{union} values in registers when possible.
13173 This is more efficient for small structures than
13174 @option{-fpcc-struct-return}.
13176 If you specify neither @option{-fpcc-struct-return} nor
13177 @option{-freg-struct-return}, GCC defaults to whichever convention is
13178 standard for the target. If there is no standard convention, GCC
13179 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
13180 the principal compiler. In those cases, we can choose the standard, and
13181 we chose the more efficient register return alternative.
13183 @strong{Warning:} code compiled with the @option{-freg-struct-return}
13184 switch is not binary compatible with code compiled with the
13185 @option{-fpcc-struct-return} switch.
13186 Use it to conform to a non-default application binary interface.
13188 @item -fshort-enums
13189 @opindex fshort-enums
13190 Allocate to an @code{enum} type only as many bytes as it needs for the
13191 declared range of possible values. Specifically, the @code{enum} type
13192 will be equivalent to the smallest integer type which has enough room.
13194 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
13195 code that is not binary compatible with code generated without that switch.
13196 Use it to conform to a non-default application binary interface.
13198 @item -fshort-double
13199 @opindex fshort-double
13200 Use the same size for @code{double} as for @code{float}.
13202 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
13203 code that is not binary compatible with code generated without that switch.
13204 Use it to conform to a non-default application binary interface.
13206 @item -fshort-wchar
13207 @opindex fshort-wchar
13208 Override the underlying type for @samp{wchar_t} to be @samp{short
13209 unsigned int} instead of the default for the target. This option is
13210 useful for building programs to run under WINE@.
13212 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
13213 code that is not binary compatible with code generated without that switch.
13214 Use it to conform to a non-default application binary interface.
13217 @opindex fno-common
13218 In C, allocate even uninitialized global variables in the data section of the
13219 object file, rather than generating them as common blocks. This has the
13220 effect that if the same variable is declared (without @code{extern}) in
13221 two different compilations, you will get an error when you link them.
13222 The only reason this might be useful is if you wish to verify that the
13223 program will work on other systems which always work this way.
13227 Ignore the @samp{#ident} directive.
13229 @item -finhibit-size-directive
13230 @opindex finhibit-size-directive
13231 Don't output a @code{.size} assembler directive, or anything else that
13232 would cause trouble if the function is split in the middle, and the
13233 two halves are placed at locations far apart in memory. This option is
13234 used when compiling @file{crtstuff.c}; you should not need to use it
13237 @item -fverbose-asm
13238 @opindex fverbose-asm
13239 Put extra commentary information in the generated assembly code to
13240 make it more readable. This option is generally only of use to those
13241 who actually need to read the generated assembly code (perhaps while
13242 debugging the compiler itself).
13244 @option{-fno-verbose-asm}, the default, causes the
13245 extra information to be omitted and is useful when comparing two assembler
13250 @cindex global offset table
13252 Generate position-independent code (PIC) suitable for use in a shared
13253 library, if supported for the target machine. Such code accesses all
13254 constant addresses through a global offset table (GOT)@. The dynamic
13255 loader resolves the GOT entries when the program starts (the dynamic
13256 loader is not part of GCC; it is part of the operating system). If
13257 the GOT size for the linked executable exceeds a machine-specific
13258 maximum size, you get an error message from the linker indicating that
13259 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
13260 instead. (These maximums are 8k on the SPARC and 32k
13261 on the m68k and RS/6000. The 386 has no such limit.)
13263 Position-independent code requires special support, and therefore works
13264 only on certain machines. For the 386, GCC supports PIC for System V
13265 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
13266 position-independent.
13268 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13273 If supported for the target machine, emit position-independent code,
13274 suitable for dynamic linking and avoiding any limit on the size of the
13275 global offset table. This option makes a difference on the m68k,
13276 PowerPC and SPARC@.
13278 Position-independent code requires special support, and therefore works
13279 only on certain machines.
13281 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
13288 These options are similar to @option{-fpic} and @option{-fPIC}, but
13289 generated position independent code can be only linked into executables.
13290 Usually these options are used when @option{-pie} GCC option will be
13291 used during linking.
13293 @item -fno-jump-tables
13294 @opindex fno-jump-tables
13295 Do not use jump tables for switch statements even where it would be
13296 more efficient than other code generation strategies. This option is
13297 of use in conjunction with @option{-fpic} or @option{-fPIC} for
13298 building code which forms part of a dynamic linker and cannot
13299 reference the address of a jump table. On some targets, jump tables
13300 do not require a GOT and this option is not needed.
13302 @item -ffixed-@var{reg}
13304 Treat the register named @var{reg} as a fixed register; generated code
13305 should never refer to it (except perhaps as a stack pointer, frame
13306 pointer or in some other fixed role).
13308 @var{reg} must be the name of a register. The register names accepted
13309 are machine-specific and are defined in the @code{REGISTER_NAMES}
13310 macro in the machine description macro file.
13312 This flag does not have a negative form, because it specifies a
13315 @item -fcall-used-@var{reg}
13316 @opindex fcall-used
13317 Treat the register named @var{reg} as an allocable register that is
13318 clobbered by function calls. It may be allocated for temporaries or
13319 variables that do not live across a call. Functions compiled this way
13320 will not save and restore the register @var{reg}.
13322 It is an error to used this flag with the frame pointer or stack pointer.
13323 Use of this flag for other registers that have fixed pervasive roles in
13324 the machine's execution model will produce disastrous results.
13326 This flag does not have a negative form, because it specifies a
13329 @item -fcall-saved-@var{reg}
13330 @opindex fcall-saved
13331 Treat the register named @var{reg} as an allocable register saved by
13332 functions. It may be allocated even for temporaries or variables that
13333 live across a call. Functions compiled this way will save and restore
13334 the register @var{reg} if they use it.
13336 It is an error to used this flag with the frame pointer or stack pointer.
13337 Use of this flag for other registers that have fixed pervasive roles in
13338 the machine's execution model will produce disastrous results.
13340 A different sort of disaster will result from the use of this flag for
13341 a register in which function values may be returned.
13343 This flag does not have a negative form, because it specifies a
13346 @item -fpack-struct[=@var{n}]
13347 @opindex fpack-struct
13348 Without a value specified, pack all structure members together without
13349 holes. When a value is specified (which must be a small power of two), pack
13350 structure members according to this value, representing the maximum
13351 alignment (that is, objects with default alignment requirements larger than
13352 this will be output potentially unaligned at the next fitting location.
13354 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
13355 code that is not binary compatible with code generated without that switch.
13356 Additionally, it makes the code suboptimal.
13357 Use it to conform to a non-default application binary interface.
13359 @item -finstrument-functions
13360 @opindex finstrument-functions
13361 Generate instrumentation calls for entry and exit to functions. Just
13362 after function entry and just before function exit, the following
13363 profiling functions will be called with the address of the current
13364 function and its call site. (On some platforms,
13365 @code{__builtin_return_address} does not work beyond the current
13366 function, so the call site information may not be available to the
13367 profiling functions otherwise.)
13370 void __cyg_profile_func_enter (void *this_fn,
13372 void __cyg_profile_func_exit (void *this_fn,
13376 The first argument is the address of the start of the current function,
13377 which may be looked up exactly in the symbol table.
13379 This instrumentation is also done for functions expanded inline in other
13380 functions. The profiling calls will indicate where, conceptually, the
13381 inline function is entered and exited. This means that addressable
13382 versions of such functions must be available. If all your uses of a
13383 function are expanded inline, this may mean an additional expansion of
13384 code size. If you use @samp{extern inline} in your C code, an
13385 addressable version of such functions must be provided. (This is
13386 normally the case anyways, but if you get lucky and the optimizer always
13387 expands the functions inline, you might have gotten away without
13388 providing static copies.)
13390 A function may be given the attribute @code{no_instrument_function}, in
13391 which case this instrumentation will not be done. This can be used, for
13392 example, for the profiling functions listed above, high-priority
13393 interrupt routines, and any functions from which the profiling functions
13394 cannot safely be called (perhaps signal handlers, if the profiling
13395 routines generate output or allocate memory).
13397 @item -fstack-check
13398 @opindex fstack-check
13399 Generate code to verify that you do not go beyond the boundary of the
13400 stack. You should specify this flag if you are running in an
13401 environment with multiple threads, but only rarely need to specify it in
13402 a single-threaded environment since stack overflow is automatically
13403 detected on nearly all systems if there is only one stack.
13405 Note that this switch does not actually cause checking to be done; the
13406 operating system must do that. The switch causes generation of code
13407 to ensure that the operating system sees the stack being extended.
13409 @item -fstack-limit-register=@var{reg}
13410 @itemx -fstack-limit-symbol=@var{sym}
13411 @itemx -fno-stack-limit
13412 @opindex fstack-limit-register
13413 @opindex fstack-limit-symbol
13414 @opindex fno-stack-limit
13415 Generate code to ensure that the stack does not grow beyond a certain value,
13416 either the value of a register or the address of a symbol. If the stack
13417 would grow beyond the value, a signal is raised. For most targets,
13418 the signal is raised before the stack overruns the boundary, so
13419 it is possible to catch the signal without taking special precautions.
13421 For instance, if the stack starts at absolute address @samp{0x80000000}
13422 and grows downwards, you can use the flags
13423 @option{-fstack-limit-symbol=__stack_limit} and
13424 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
13425 of 128KB@. Note that this may only work with the GNU linker.
13427 @cindex aliasing of parameters
13428 @cindex parameters, aliased
13429 @item -fargument-alias
13430 @itemx -fargument-noalias
13431 @itemx -fargument-noalias-global
13432 @itemx -fargument-noalias-anything
13433 @opindex fargument-alias
13434 @opindex fargument-noalias
13435 @opindex fargument-noalias-global
13436 @opindex fargument-noalias-anything
13437 Specify the possible relationships among parameters and between
13438 parameters and global data.
13440 @option{-fargument-alias} specifies that arguments (parameters) may
13441 alias each other and may alias global storage.@*
13442 @option{-fargument-noalias} specifies that arguments do not alias
13443 each other, but may alias global storage.@*
13444 @option{-fargument-noalias-global} specifies that arguments do not
13445 alias each other and do not alias global storage.
13446 @option{-fargument-noalias-anything} specifies that arguments do not
13447 alias any other storage.
13449 Each language will automatically use whatever option is required by
13450 the language standard. You should not need to use these options yourself.
13452 @item -fleading-underscore
13453 @opindex fleading-underscore
13454 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
13455 change the way C symbols are represented in the object file. One use
13456 is to help link with legacy assembly code.
13458 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
13459 generate code that is not binary compatible with code generated without that
13460 switch. Use it to conform to a non-default application binary interface.
13461 Not all targets provide complete support for this switch.
13463 @item -ftls-model=@var{model}
13464 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
13465 The @var{model} argument should be one of @code{global-dynamic},
13466 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
13468 The default without @option{-fpic} is @code{initial-exec}; with
13469 @option{-fpic} the default is @code{global-dynamic}.
13471 @item -fvisibility=@var{default|internal|hidden|protected}
13472 @opindex fvisibility
13473 Set the default ELF image symbol visibility to the specified option---all
13474 symbols will be marked with this unless overridden within the code.
13475 Using this feature can very substantially improve linking and
13476 load times of shared object libraries, produce more optimized
13477 code, provide near-perfect API export and prevent symbol clashes.
13478 It is @strong{strongly} recommended that you use this in any shared objects
13481 Despite the nomenclature, @code{default} always means public ie;
13482 available to be linked against from outside the shared object.
13483 @code{protected} and @code{internal} are pretty useless in real-world
13484 usage so the only other commonly used option will be @code{hidden}.
13485 The default if @option{-fvisibility} isn't specified is
13486 @code{default}, i.e., make every
13487 symbol public---this causes the same behavior as previous versions of
13490 A good explanation of the benefits offered by ensuring ELF
13491 symbols have the correct visibility is given by ``How To Write
13492 Shared Libraries'' by Ulrich Drepper (which can be found at
13493 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
13494 solution made possible by this option to marking things hidden when
13495 the default is public is to make the default hidden and mark things
13496 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
13497 and @code{__attribute__ ((visibility("default")))} instead of
13498 @code{__declspec(dllexport)} you get almost identical semantics with
13499 identical syntax. This is a great boon to those working with
13500 cross-platform projects.
13502 For those adding visibility support to existing code, you may find
13503 @samp{#pragma GCC visibility} of use. This works by you enclosing
13504 the declarations you wish to set visibility for with (for example)
13505 @samp{#pragma GCC visibility push(hidden)} and
13506 @samp{#pragma GCC visibility pop}.
13507 Bear in mind that symbol visibility should be viewed @strong{as
13508 part of the API interface contract} and thus all new code should
13509 always specify visibility when it is not the default ie; declarations
13510 only for use within the local DSO should @strong{always} be marked explicitly
13511 as hidden as so to avoid PLT indirection overheads---making this
13512 abundantly clear also aids readability and self-documentation of the code.
13513 Note that due to ISO C++ specification requirements, operator new and
13514 operator delete must always be of default visibility.
13516 Be aware that headers from outside your project, in particular system
13517 headers and headers from any other library you use, may not be
13518 expecting to be compiled with visibility other than the default. You
13519 may need to explicitly say @samp{#pragma GCC visibility push(default)}
13520 before including any such headers.
13522 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
13523 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
13524 no modifications. However, this means that calls to @samp{extern}
13525 functions with no explicit visibility will use the PLT, so it is more
13526 effective to use @samp{__attribute ((visibility))} and/or
13527 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
13528 declarations should be treated as hidden.
13530 Note that @samp{-fvisibility} does affect C++ vague linkage
13531 entities. This means that, for instance, an exception class that will
13532 be thrown between DSOs must be explicitly marked with default
13533 visibility so that the @samp{type_info} nodes will be unified between
13536 An overview of these techniques, their benefits and how to use them
13537 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
13543 @node Environment Variables
13544 @section Environment Variables Affecting GCC
13545 @cindex environment variables
13547 @c man begin ENVIRONMENT
13548 This section describes several environment variables that affect how GCC
13549 operates. Some of them work by specifying directories or prefixes to use
13550 when searching for various kinds of files. Some are used to specify other
13551 aspects of the compilation environment.
13553 Note that you can also specify places to search using options such as
13554 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
13555 take precedence over places specified using environment variables, which
13556 in turn take precedence over those specified by the configuration of GCC@.
13557 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
13558 GNU Compiler Collection (GCC) Internals}.
13563 @c @itemx LC_COLLATE
13565 @c @itemx LC_MONETARY
13566 @c @itemx LC_NUMERIC
13571 @c @findex LC_COLLATE
13572 @findex LC_MESSAGES
13573 @c @findex LC_MONETARY
13574 @c @findex LC_NUMERIC
13578 These environment variables control the way that GCC uses
13579 localization information that allow GCC to work with different
13580 national conventions. GCC inspects the locale categories
13581 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
13582 so. These locale categories can be set to any value supported by your
13583 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
13584 Kingdom encoded in UTF-8.
13586 The @env{LC_CTYPE} environment variable specifies character
13587 classification. GCC uses it to determine the character boundaries in
13588 a string; this is needed for some multibyte encodings that contain quote
13589 and escape characters that would otherwise be interpreted as a string
13592 The @env{LC_MESSAGES} environment variable specifies the language to
13593 use in diagnostic messages.
13595 If the @env{LC_ALL} environment variable is set, it overrides the value
13596 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
13597 and @env{LC_MESSAGES} default to the value of the @env{LANG}
13598 environment variable. If none of these variables are set, GCC
13599 defaults to traditional C English behavior.
13603 If @env{TMPDIR} is set, it specifies the directory to use for temporary
13604 files. GCC uses temporary files to hold the output of one stage of
13605 compilation which is to be used as input to the next stage: for example,
13606 the output of the preprocessor, which is the input to the compiler
13609 @item GCC_EXEC_PREFIX
13610 @findex GCC_EXEC_PREFIX
13611 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
13612 names of the subprograms executed by the compiler. No slash is added
13613 when this prefix is combined with the name of a subprogram, but you can
13614 specify a prefix that ends with a slash if you wish.
13616 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
13617 an appropriate prefix to use based on the pathname it was invoked with.
13619 If GCC cannot find the subprogram using the specified prefix, it
13620 tries looking in the usual places for the subprogram.
13622 The default value of @env{GCC_EXEC_PREFIX} is
13623 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
13624 of @code{prefix} when you ran the @file{configure} script.
13626 Other prefixes specified with @option{-B} take precedence over this prefix.
13628 This prefix is also used for finding files such as @file{crt0.o} that are
13631 In addition, the prefix is used in an unusual way in finding the
13632 directories to search for header files. For each of the standard
13633 directories whose name normally begins with @samp{/usr/local/lib/gcc}
13634 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
13635 replacing that beginning with the specified prefix to produce an
13636 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
13637 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
13638 These alternate directories are searched first; the standard directories
13641 @item COMPILER_PATH
13642 @findex COMPILER_PATH
13643 The value of @env{COMPILER_PATH} is a colon-separated list of
13644 directories, much like @env{PATH}. GCC tries the directories thus
13645 specified when searching for subprograms, if it can't find the
13646 subprograms using @env{GCC_EXEC_PREFIX}.
13649 @findex LIBRARY_PATH
13650 The value of @env{LIBRARY_PATH} is a colon-separated list of
13651 directories, much like @env{PATH}. When configured as a native compiler,
13652 GCC tries the directories thus specified when searching for special
13653 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
13654 using GCC also uses these directories when searching for ordinary
13655 libraries for the @option{-l} option (but directories specified with
13656 @option{-L} come first).
13660 @cindex locale definition
13661 This variable is used to pass locale information to the compiler. One way in
13662 which this information is used is to determine the character set to be used
13663 when character literals, string literals and comments are parsed in C and C++.
13664 When the compiler is configured to allow multibyte characters,
13665 the following values for @env{LANG} are recognized:
13669 Recognize JIS characters.
13671 Recognize SJIS characters.
13673 Recognize EUCJP characters.
13676 If @env{LANG} is not defined, or if it has some other value, then the
13677 compiler will use mblen and mbtowc as defined by the default locale to
13678 recognize and translate multibyte characters.
13682 Some additional environments variables affect the behavior of the
13685 @include cppenv.texi
13689 @node Precompiled Headers
13690 @section Using Precompiled Headers
13691 @cindex precompiled headers
13692 @cindex speed of compilation
13694 Often large projects have many header files that are included in every
13695 source file. The time the compiler takes to process these header files
13696 over and over again can account for nearly all of the time required to
13697 build the project. To make builds faster, GCC allows users to
13698 `precompile' a header file; then, if builds can use the precompiled
13699 header file they will be much faster.
13701 To create a precompiled header file, simply compile it as you would any
13702 other file, if necessary using the @option{-x} option to make the driver
13703 treat it as a C or C++ header file. You will probably want to use a
13704 tool like @command{make} to keep the precompiled header up-to-date when
13705 the headers it contains change.
13707 A precompiled header file will be searched for when @code{#include} is
13708 seen in the compilation. As it searches for the included file
13709 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
13710 compiler looks for a precompiled header in each directory just before it
13711 looks for the include file in that directory. The name searched for is
13712 the name specified in the @code{#include} with @samp{.gch} appended. If
13713 the precompiled header file can't be used, it is ignored.
13715 For instance, if you have @code{#include "all.h"}, and you have
13716 @file{all.h.gch} in the same directory as @file{all.h}, then the
13717 precompiled header file will be used if possible, and the original
13718 header will be used otherwise.
13720 Alternatively, you might decide to put the precompiled header file in a
13721 directory and use @option{-I} to ensure that directory is searched
13722 before (or instead of) the directory containing the original header.
13723 Then, if you want to check that the precompiled header file is always
13724 used, you can put a file of the same name as the original header in this
13725 directory containing an @code{#error} command.
13727 This also works with @option{-include}. So yet another way to use
13728 precompiled headers, good for projects not designed with precompiled
13729 header files in mind, is to simply take most of the header files used by
13730 a project, include them from another header file, precompile that header
13731 file, and @option{-include} the precompiled header. If the header files
13732 have guards against multiple inclusion, they will be skipped because
13733 they've already been included (in the precompiled header).
13735 If you need to precompile the same header file for different
13736 languages, targets, or compiler options, you can instead make a
13737 @emph{directory} named like @file{all.h.gch}, and put each precompiled
13738 header in the directory, perhaps using @option{-o}. It doesn't matter
13739 what you call the files in the directory, every precompiled header in
13740 the directory will be considered. The first precompiled header
13741 encountered in the directory that is valid for this compilation will
13742 be used; they're searched in no particular order.
13744 There are many other possibilities, limited only by your imagination,
13745 good sense, and the constraints of your build system.
13747 A precompiled header file can be used only when these conditions apply:
13751 Only one precompiled header can be used in a particular compilation.
13754 A precompiled header can't be used once the first C token is seen. You
13755 can have preprocessor directives before a precompiled header; you can
13756 even include a precompiled header from inside another header, so long as
13757 there are no C tokens before the @code{#include}.
13760 The precompiled header file must be produced for the same language as
13761 the current compilation. You can't use a C precompiled header for a C++
13765 The precompiled header file must have been produced by the same compiler
13766 binary as the current compilation is using.
13769 Any macros defined before the precompiled header is included must
13770 either be defined in the same way as when the precompiled header was
13771 generated, or must not affect the precompiled header, which usually
13772 means that they don't appear in the precompiled header at all.
13774 The @option{-D} option is one way to define a macro before a
13775 precompiled header is included; using a @code{#define} can also do it.
13776 There are also some options that define macros implicitly, like
13777 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
13780 @item If debugging information is output when using the precompiled
13781 header, using @option{-g} or similar, the same kind of debugging information
13782 must have been output when building the precompiled header. However,
13783 a precompiled header built using @option{-g} can be used in a compilation
13784 when no debugging information is being output.
13786 @item The same @option{-m} options must generally be used when building
13787 and using the precompiled header. @xref{Submodel Options},
13788 for any cases where this rule is relaxed.
13790 @item Each of the following options must be the same when building and using
13791 the precompiled header:
13793 @gccoptlist{-fexceptions -funit-at-a-time}
13796 Some other command-line options starting with @option{-f},
13797 @option{-p}, or @option{-O} must be defined in the same way as when
13798 the precompiled header was generated. At present, it's not clear
13799 which options are safe to change and which are not; the safest choice
13800 is to use exactly the same options when generating and using the
13801 precompiled header. The following are known to be safe:
13803 @gccoptlist{-fmessage-length= -fpreprocessed
13804 -fsched-interblock -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous
13805 -fsched-verbose=<number> -fschedule-insns -fvisibility=
13810 For all of these except the last, the compiler will automatically
13811 ignore the precompiled header if the conditions aren't met. If you
13812 find an option combination that doesn't work and doesn't cause the
13813 precompiled header to be ignored, please consider filing a bug report,
13816 If you do use differing options when generating and using the
13817 precompiled header, the actual behavior will be a mixture of the
13818 behavior for the options. For instance, if you use @option{-g} to
13819 generate the precompiled header but not when using it, you may or may
13820 not get debugging information for routines in the precompiled header.
13822 @node Running Protoize
13823 @section Running Protoize
13825 The program @code{protoize} is an optional part of GCC@. You can use
13826 it to add prototypes to a program, thus converting the program to ISO
13827 C in one respect. The companion program @code{unprotoize} does the
13828 reverse: it removes argument types from any prototypes that are found.
13830 When you run these programs, you must specify a set of source files as
13831 command line arguments. The conversion programs start out by compiling
13832 these files to see what functions they define. The information gathered
13833 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
13835 After scanning comes actual conversion. The specified files are all
13836 eligible to be converted; any files they include (whether sources or
13837 just headers) are eligible as well.
13839 But not all the eligible files are converted. By default,
13840 @code{protoize} and @code{unprotoize} convert only source and header
13841 files in the current directory. You can specify additional directories
13842 whose files should be converted with the @option{-d @var{directory}}
13843 option. You can also specify particular files to exclude with the
13844 @option{-x @var{file}} option. A file is converted if it is eligible, its
13845 directory name matches one of the specified directory names, and its
13846 name within the directory has not been excluded.
13848 Basic conversion with @code{protoize} consists of rewriting most
13849 function definitions and function declarations to specify the types of
13850 the arguments. The only ones not rewritten are those for varargs
13853 @code{protoize} optionally inserts prototype declarations at the
13854 beginning of the source file, to make them available for any calls that
13855 precede the function's definition. Or it can insert prototype
13856 declarations with block scope in the blocks where undeclared functions
13859 Basic conversion with @code{unprotoize} consists of rewriting most
13860 function declarations to remove any argument types, and rewriting
13861 function definitions to the old-style pre-ISO form.
13863 Both conversion programs print a warning for any function declaration or
13864 definition that they can't convert. You can suppress these warnings
13867 The output from @code{protoize} or @code{unprotoize} replaces the
13868 original source file. The original file is renamed to a name ending
13869 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
13870 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
13871 for DOS) file already exists, then the source file is simply discarded.
13873 @code{protoize} and @code{unprotoize} both depend on GCC itself to
13874 scan the program and collect information about the functions it uses.
13875 So neither of these programs will work until GCC is installed.
13877 Here is a table of the options you can use with @code{protoize} and
13878 @code{unprotoize}. Each option works with both programs unless
13882 @item -B @var{directory}
13883 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
13884 usual directory (normally @file{/usr/local/lib}). This file contains
13885 prototype information about standard system functions. This option
13886 applies only to @code{protoize}.
13888 @item -c @var{compilation-options}
13889 Use @var{compilation-options} as the options when running @command{gcc} to
13890 produce the @samp{.X} files. The special option @option{-aux-info} is
13891 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
13893 Note that the compilation options must be given as a single argument to
13894 @code{protoize} or @code{unprotoize}. If you want to specify several
13895 @command{gcc} options, you must quote the entire set of compilation options
13896 to make them a single word in the shell.
13898 There are certain @command{gcc} arguments that you cannot use, because they
13899 would produce the wrong kind of output. These include @option{-g},
13900 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
13901 the @var{compilation-options}, they are ignored.
13904 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
13905 systems) instead of @samp{.c}. This is convenient if you are converting
13906 a C program to C++. This option applies only to @code{protoize}.
13909 Add explicit global declarations. This means inserting explicit
13910 declarations at the beginning of each source file for each function
13911 that is called in the file and was not declared. These declarations
13912 precede the first function definition that contains a call to an
13913 undeclared function. This option applies only to @code{protoize}.
13915 @item -i @var{string}
13916 Indent old-style parameter declarations with the string @var{string}.
13917 This option applies only to @code{protoize}.
13919 @code{unprotoize} converts prototyped function definitions to old-style
13920 function definitions, where the arguments are declared between the
13921 argument list and the initial @samp{@{}. By default, @code{unprotoize}
13922 uses five spaces as the indentation. If you want to indent with just
13923 one space instead, use @option{-i " "}.
13926 Keep the @samp{.X} files. Normally, they are deleted after conversion
13930 Add explicit local declarations. @code{protoize} with @option{-l} inserts
13931 a prototype declaration for each function in each block which calls the
13932 function without any declaration. This option applies only to
13936 Make no real changes. This mode just prints information about the conversions
13937 that would have been done without @option{-n}.
13940 Make no @samp{.save} files. The original files are simply deleted.
13941 Use this option with caution.
13943 @item -p @var{program}
13944 Use the program @var{program} as the compiler. Normally, the name
13945 @file{gcc} is used.
13948 Work quietly. Most warnings are suppressed.
13951 Print the version number, just like @option{-v} for @command{gcc}.
13954 If you need special compiler options to compile one of your program's
13955 source files, then you should generate that file's @samp{.X} file
13956 specially, by running @command{gcc} on that source file with the
13957 appropriate options and the option @option{-aux-info}. Then run
13958 @code{protoize} on the entire set of files. @code{protoize} will use
13959 the existing @samp{.X} file because it is newer than the source file.
13963 gcc -Dfoo=bar file1.c -aux-info file1.X
13968 You need to include the special files along with the rest in the
13969 @code{protoize} command, even though their @samp{.X} files already
13970 exist, because otherwise they won't get converted.
13972 @xref{Protoize Caveats}, for more information on how to use
13973 @code{protoize} successfully.